Delineation of Soil Management Zones: Comparison of Two Proximal Soil Sensor Systems.

Subdividing agricultural fields into soil management zones (MZ) with relatively homogeneous soil properties and yield potential can be used as the basis of site-specific nutrient management, and offers an effective alternative to uniform and variable rate applications. This study evaluated the efficiency of two proximal soil sensors to delineate MZs in a 21-ha commercial potato (Solanum tuberosum L.) field located in St. André (NB). The Dualem 21-S (D21S) was utilized for the measurement of the soil electrical conductivity in fall of 2015. Ground penetrating radar (GPR; GSSI SIR-3000 model; 400 MHz antenna; 30 sec^-1 scan) was utilized for the measurement of the soil dielectric constant in February 2016. Data from the GPR and D21S were used to stratify the fields into MZs using an unsupervised fuzzy k-means clustering algorithm. Soil samples (0-15 cm) collected from 154 georeferenced sampling points were analyzed for physico-chemical properties (SOM, pH, Mehlich-3 extractable elements). The soil particle size analysis was completed only on 41 soil samples. Potato tuber yields were measured using a yield monitor in 2013, 2014, and 2016. The D21S was found to be effective in delineating MZ whereas the GPR was not effective. The two MZs delineated with GPR did not show significant differences in potato yields. Based on the D21S soil EC dataset, two MZs were found to be optimal to subdivide the potato field. The MZs showed significant differences in soil water regime (soil moisture: 23.4% vs 28.5%) and in some soil physico-chemical properties (clay: 141 g kg^-1 vs 189 g kg^-1). Significant differences in potato yields (4.8, 7.6 and 10.1 Mg ha^-1 for 2013, 2014 and 2016, respectively) between the two MZs were attributed to differing water supply associated with differing soil texture and soil drainage characteristics.