Spatially-defined soil and plant properties contribute to better planning of experiments and commercial fields. This research had as objective the study of spatial variability of some physical and chemical properties of the soil and their relationship with soybean productivity. The experiment was conducted in the agricultural year of 2000, at the Federal University of Santa Maria, RS, Brazil, on land-leveled paddy soil. An area of 160 x 88m was sampled at an interval of 8 x 8m, resulting in 240 sampling points. The plant properties studied were soybean yield and plant height. Several chemical and physical properties of the topsoil (0 to 0.15m) and subsoil (0.15 to 0.30m) were determined. Geostatistical analysis was done only for the soil layer that presented correlation with the plant properties (0-0.15m), and the space dependence of the attributes was evaluated by scaled semivariograms. Except for soybean production, particle density and flocculation degree, all soil properties presented moderate (0.64 to 0.75) to strong (>0.75) space dependence. The production of soybean was influenced by space variability of the analyzed soil properties. The spatial variability of the soil physical and chemical properties in land-leveled paddy soil affected the production of soybean. The properties were classified in two different categories, based on semivariogram model and range: (i) exponential model and range smaller than 40m and (ii) gaussian model and range smaller than 67m. A gaussian model with range smaller than 45m described the spatial variation of soybean yeild.
Planosols; lowlands; geoestatistics; soil leveling
