Comparison of three methods for delineating management zones for site-specific crop management

The goal of this study was to evaluate different strategies for defining homogeneous zones, and to verify their usefulness by comparing soil moisture contents and crop yields. Data were obtained from a field experiment near General Pico, La Pampa, Argentina, with a sampling grid of 598 points. Soil texture, organic matter, available phosphorus, pH, and electrical conductivity were determined. Altimetry was obtained by a GPS (Garmin 19X). Corn yields and available soil moisture contents were determined in a 184 point grid. Treatments were early and late planting date, and control and 100 kg N ha−1. MULTISPATI-PCA was used to carry out for multivariate spatial analyses of soil attributes. The synthetic variables (sPC) were analyzed, mapped and used for determination of homogeneous zones by three strategies: S1-natural rupture method on sPC1, S2-Fuzzy k-means cluster analysis on soil variables and S3-Fuzzy k-means cluster analysis on all sPC. Zones were compared for means of soil variables, available soil water and maize yields. The three tested statistical strategies divided the field in a very similar manner when the minimum number of zones was considered, with coincidence in the classification of 98% of the sampled grid points. The different strategies resulted in different number of zones with different characteristics. The choice of the method depends on the objectives for determination of management zones. In our case, for fertilization management zonification might prioritize the differentiation of OM and available P contents and use S3, while if water was the main limiting factor, the management zones would be two according to S1 or S2, responding to textural and altimetry differences.