Towards site specific management zones delineation in rotational cropping system: Application of multivariate spatial clustering model based on soil properties

Abstract

Site|specific management for irrigation purposes remains a great challenge especially when dealing with mixed cropping systems due to spatial variability of soil properties. This variability generates crop yield heterogeneity within agricultural fields.In this study, rice-corn, cotton rotational cropping system was cultivated in Tolima region-Colombia. A total of 72 geo-referenced representative surface soil samples (0–20 cm depth) were collected covering a total area of 5 ha of Inceptisols. The aim was to delineate field management zones (MZs) using three methods M1, M2 and M3, and evaluate their usefulness for a further site-specific management. Georeferenced soil samples were analyzed for soil texture, bulk density (BD), field capacity (FC), soil available water (AW), mesopores percentage (Mes), soil penetration resistance (PR), organic matter (OM), available phosphorus (P), and exchangeable bases content (Ca + Mg/K). Spatial variability of soil properties was analyzed with geostatistics approach. Further, spatial principal component (sPC), were performed to delineate the MZs using: M1- Fisher-Jenks algorithm with the first sPC; M2- Fuzzy k-means cluster analysis with two sPC; and M3- Fuzzy k-means with soil variables. MZs delineation results were validated by comparing the zones to soil properties and yield data coming from the rice-corn, cotton system. Two MZs were recommended using M2 due to its lower fragmentation compared to other methodologies. This delineation could be suggested for fertilization management since Clay + Silt, OM, P, and Ca + Mg/K content differentiate when delineating one and two MZs. M2 provided differences for AW, Mes, and PR when considering three MZs which can be used as a basis for irrigation site-specific management in precision agriculture.