Comparison of Superiority of Indigenous Lentil Rhizobia under Drought Stress

Conventional soil survey methods for soils within the watersheds in Iran require a significant budget with many soil surveyors and much time. Additionally, no accurate and reliable information exists on the spatial variability of superface soil parameters in order to predict the soil loss by different models (RUSLE, PISAC, EUPOSEM, MORGAN). Also information on planning and management activities is lacking. These limitations call for methods of estimating soil properties using minimum sampling derived from important terrain parameters. This study was performed to develop soil-landscape models in three geological units (E2Sc, Ku, Plc), in a part of Mehr- watershed, Sabzevar. Six soil variables selected for this study were topsoil clay, gravel, sand, organic matter content, field capacity and bulk density measured at 316 sites on a regular 100m grid. Topographic attributes were calculated by a digital elevation model with 100m spacing. Finally, multiple linear regression analyses relating soil to topographic attributes were performed and then models were validated by additional sample points (78 of 316). The developed regression models showed significant relationships between surface soil properties and topographic attributes such as elevation, slope, aspect, wetness index, stream power index and sediment transport index. The mean errors and root mean square errors in the validation of the models were low and acceptable. The regression equations could explain only 26 to 72 % of the variability measured in the soil attributes in the watershed scale with 100m spacing.