The objective of this work was to evaluate, through principal component analysis, the reduction in the dimensionality of soil chemical and physical attributes in order to understand the spatial and temporal variability of grain crop yield. The experimental area of 54 ha has been managed under precision agriculture for eight years. Based on six grain yield maps (soybean - 2000/2001 harvest season; maize - 2001/2002; soybean - 2002/2003; wheat - 2003; soybean - 2003/2004; and maize - 2004/2005), the area was split in three grain yield zones (high, medium, and low). Fifteen representative geo‑referentiated points were defined in order to determine soil chemical and physical attributes, totaling 63 analyzed variables. Among chemical attributes, the high K content in soil is the variable that best explains the spatial variability of grain crop yield, probably due to the imbalance of the Ca:K and Mg:K relations. The low‑yield zone had low soil physical quality. In this case, soil water infiltration, solely, is the variable that best explains the performance of the grain crops. Principal components analysis of soil chemical and physical attributes is an efficient strategy to explain the spatial and temporal variability of grain crop yield.
precision agriculture; multivariate analysis; spatial variability; temporal variability; management zones
