Soil sampling is an important part of soil physical and chemical characterization underlying management and fertilization practices. For this purpose, composite sampling is largely used in order to represent the area by the homogenization of single samples. Consequently, the subdivision of the area in homogeneous plots plays an important role. The aim of this study was to evaluate the variability of soil chemical (pH; Ca2+; Mg2+; Al3+; K; P; H; organic carbon) and physical properties (coarse sand, fine sand, silt, and clay) in a sample set at different subdivision levels and to verify the reliability of the composite sample of the area. The study area is part of the Experimental Farm Canguiri of the Federal University of Paraná, Curitiba (PR). Soil sampling was performed in an area of 12.88 ha at points 30 m away from each other, resulting in 135 simple samples. Data were analyzed by dispersion and normality parameters, using the following subdivision levels: (a) total area (n=135); (b) geomorphological compartments: lower (n=38), medium (n=60), upper (n=37); (c) compartment division in mapping units: lower - 3 units, medium - 4 units, upper - 3 units. Al3+, P and K+ were the properties with highest variability. With increasing area subdivision, the statistical parameters showed less data variation. For Al3+, for example, at the highest level of subdivision, the amplitude values (percentage difference between highest and lowest values) decreased from 3.895 to 294 %, indicating the importance of the criteria underlying plot division. In areas with higher slope, even the composite sampling based on geomorphological compartments or mapping units did not ensure the sampling quality. For most properties in the total area and geomorphological compartments, the data distribution was not normal, limiting the use of composite sampling under these conditions, although the opposite was observed when the area was subdivided into mapping units.
composite sampling; sample variability; coefficient of variation; normal distribution
