ABSTRACT
Demarcating soil management zones can be useful, for instance, delimiting homogeneous areas and selecting attributes that are generally correlated with plant productivity, but doing so involves several different steps. The objective of this study was to identify the chemical and physical attributes of soil and soybean plants that explain crop productivity, in addition to suggesting and testing a methodological procedure for defining soil management zones. The procedure consisted of six steps: sample collection, data filtering, variable selection, interpolation, grouping, and evaluation of management zones. The samples were collected in an experimental area of 12.5 ha cultivated with soybean during the 2013/14 crop in Dystrophic Red Latosol, in Mato Grosso, Brazil. A total of 117 pairs of plant and soil samples were collected. Student’s t-test was used (α = 0.02) to verify that the number of samples was adequate for correlation analysis. Results showed that only the P and Mn content in the grains explained (based on R2 values) the variation in soybean grain productivity the area. Based on the interpolation of these contents by ordinary kriging, the fuzzy C-means algorithm was used to separate them into groups by similarity. Division into two groups was the best option, which could be differentiated by Mann-Whitney test (P < 0.05), resulting in a map with 10 management zones.
Keywords:
Glycine max L.; Direct seeding; Precision agriculture.
Thumbnail
Thumbnail
