Spatial predictions of the permanent wilting point in arid and semi-arid regions of Northeast China

The permanent wilting point (PWP) is the minimal point of soil moisture a plant requires not to wilt. If moisture decreases to this, or any lower point, a plant wilts, and can no longer recover its turgidity when placed in a saturated atmosphere for 12 h. The PWP is an essential physical property that has a powerful influence on other soil properties and is important for agricultural production, including irrigation water use efficiency and crop yield. However, there are few published datasets on the PWP of cultivated land in the arid and semi-arid regions of Northeast China. Direct measurements of PWP are time-consuming and expensive; therefore, the aim of this study was to evaluate how environmental variables could be used in spatial predictions of PWP in an arid and semi-arid region based on boosted regression tree (BRT) and multiple linear stepwise regression (MLSR) analyses. Seventy soil samples and nine covariates (including topography and vegetation variables) were collected and analyzed. Cross-validation procedure was used to evaluate model performance and uncertainty. Accuracy evaluation results showed that the BRT model had better predictive performance than the MLSR. PWP content decreased from the southwest to the northeast of the study area, and average values of both models were 26%. BRT and MLSR models should be compared and calibrated to obtain the best prediction effect of PWP spatial distribution in similar areas. Elevation, slope gradient, profile curvature, topographic wetness index, and land-use variables are the main environmental indicators that should be included when generating PWP maps of regions with a dry, continental climate. Such information could aid local land managers and government agencies with evaluating soil quality and water sequestration potential. In conclusion, this study could be used as a reference for predicting the hydrological parameters of the topsoil in other similar ecological environments.