Modelling the spatial variability of soil micronutrients for site specific nutrient management in a semi-arid tropical environment

Abstract

Deficiency of micronutrients in the soils of semi-arid tropical (SAT) environments is a major limiting factor for crop production. This study aims to model the spatial variability of soil micronutrients in croplands of Thimmajipet block, Mahabubnagar, Telangana, India. We collected 1508 georeferenced surface (0–15 cm) soil samples at a grid interval of 325 × 325 m and analysed them for DTPA extractable fractions of iron (Fe), manganese (Mn), copper (Cu), zinc (Zn) and boron (B). Landform and topographic variables were derived from the digital elevation model of the study area. Results showed that all the micronutrients varied highly (CV > 35%) in the order of Zn > Fe > Cu > B > Mn. The Q–Q plots indicated non-normal frequency distribution for all the micronutrients, and log transformation improved them. We compared spherical, circular, exponential, and Gaussian models of ordinary kriging interpolation to define the spatial variability structure. The spherical model with the lowest RMSE and better r2 values was selected to map the spatial distribution of the micronutrients. The nugget to sill ratio (N:S > 0.75) indicated that the micronutrients were weakly spatial dependent. Multiple regression analysis (p < 0.05) indicated that the micronutrient concentration and spatial distribution were controlled by (1) parent material; (2) CaCO3; (3) OC; and (4) antagonistic nutrient interaction. Further, critical deficiency zones of Zn, Cu, B and Fe were delineated by integrating cadastral map on the spatial distribution maps. Our study demonstrates that the large scale spatial variability mapping of soil micronutrients is a prerequisite for implementing site-specific nutrient management in the SAT regions.