Abstract
Inappropriate land management and fertilizer application may lead to nutrient deficiency soil degradation in the long run. The soil degradation in terms of changes in soil physical and chemical properties is the main impairment of crop productivity. Therefore, management practices based on soil spatial variability are now inevitable for enhancing agricultural productivity, food safety, and environmental modelling. The present study conducted to characterize the field-scale spatial variability of soil physical (sand, silt, clay, water content at field capacity (FC), and permanent wilting point (PWP), mean weight diameter (MWD)) and soil chemical properties (pH, EC, soil organic carbon (SOC), available phosphorous (Av-P), and available potassium (Av-K)) in soybean-wheat belts in Vertisols of central India. These belts are intensively cultivated and followed the uniform management practices without considering soil spatial variability. A total of 260 geocoded soil surface (0–20 cm) samples were randomly collected from the study area. The values of soil pH, EC, SOC, Av-P, Av-K, sand, silt, clay, FC water content, PWP water content and MWD varied from 6.09 to 8.56, 0.04 to 0.43 dS m− 1, 0.15 to 1.26%, 1.87 to 60.84 kg ha− 1, 62.16 to 669.76 kg ha− 1, 33.21 to 55.80%, 13.28 to 30.28% 22.00 to 46.72%, 21.06 to 40.95% 10.57 to 26.10% and 0.77 to 1.34 mm, respectively. The statistical analysis showed high spatial variability across the study area for soil EC, Av-P, and Av-K, as indicated by its coefficient of variations value of 47.09%, 59.31%, and 37.27%, respectively. The lowest variability was observed for the soil pH (CV = 6.35%). However, SOC (CV = 28.62%) and MWD (CV = 30.10%) fall under the moderate category of variability. Correlation analysis showed that SOC was significantly correlated with Av-P (r = 0.25*), Av-K (r = 0.25**), MWD (r = 0.46**), sand (r= -0.32*), silt (r = 0.32*), clay (r = 0.45*), and FC (r = 0.25*). The surface map of soil physio-chemical properties was generated through ordinary kriging techniques. Based on the lowest values of root mean square error (RMSE), the exponential model was found to be the best fit for pH, EC, SOC, Av-P, sand, and MWD, while the Gaussian model was found to be the best fit for Av-k and FC. The silt and clay distribution were well explained by the spherical model; PWP followed the circular model. The SOC, Av-P, and MWD showed strong spatial dependency (nugget/sill > 0.25). The sand and clay content showed weak spatial dependency. The remaining properties exhibited moderate spatial dependency. Further, the positive value of goodness of prediction ‘G’ indicated that developed semivariogram parameters could be used for prediction of soil value at unsampled locations. The present study exhibit that the geostatistical models are useful in addressing the soil spatial variability and will help farmers and decision-makers for improving land management practices.