Journal of the Indian Society of Soil Science

Field studies on the influence of landscape and soil characteristics under two management levds (Low and Optimum) on wheat were conducted on 26 varying soil units for three years each (from 1979 to 1995) in different watersheds and/or villages around Nagpur during winter seasons (November-March). The objectives of the study were to establish the land use requirements (suitability criteria) and to correlate properties of swell-shrink soils with potential wheat yields, Per cent slope among site parameters and soil depth, per cent silt, plant available waterholding capacity (PAWC) and length of growing period (LGP) among the soil parameters were observed to be the main yield determinants. For achieving maximum yield, the optimum values computed from the quadratic curves were soil on <2% slope, 120–150 cm soil depth, 58 to 68% clay, 228 mm PAWC, 230 days LGP. As per the nature of relationship of soil-site parameters with grain yield and based on the optimum and/or critical values, the limits of soilsite requirements have been defined. At optimum crop management level, among the 10 multiple linear regression (MLR) models developed the one, namely, Grain yield in q ha⁻¹ = -119.63 + 0.162 depth in cm -0.544 coarse fragments in % volume -0,256 si1t% -21.509 organic carbon % -0.101 CEC in emol (p⁺) kg⁻¹ + 0,289 CaCO₃% + 216.601 EC_e in dS m⁻¹ + 14.535 pH + 4.669 ESP -0.060 PAWC in nun was most reliable, as it explained 98% (R² = 0.980) variability in yield. At low crop management level, another model, namely, Grain yield in q ha_-1 = -87.50 + 0.318 slope % -2.304 drainage ratings -0.538 coarse fragments in % volume -0.237 sand% 0,804 silt % -17.119 organic carbon % -0.206 CEC in cmol (p⁺) kg⁻¹ + 0.736 CaCO₃% + 71.629 EC_e + 13.444 pH + 5.070 ESP -0.024 PAWC in mm + 0.142 LGP in days appeared to be more appropriate (R² = 0.989) out of the five models established. However, the user can adopt any of the models developed under each management level, bearing the risk of the corresponding standard errors and the index of multicollinearity.