Plant traits in influencing soil moisture in semiarid grasslands of the Loess Plateau, China.

Large-scale vegetation restoration projects pose threats to water resource security in water-limited regions. Thus, the quantification of how vegetation cover affects soil moisture is of key importance to support effective restoration schemes in drylands. However, the current understanding of such effects remains poor. For this study, an in-situ vegetation-removal experiment was conducted at 36 herbaceous grassland sites having different community compositions and topographical conditions in two adjacent loess watersheds of the Loess Plateau, China. The effects of vegetation cover (vegetation effects) on soil moisture were analyzed across soil profiles (0-180 cm) and two growing seasons. Overall, 13 plant traits and 7 topographic and soil properties were employed to evaluate how community compositions modulated vegetation effects on soil moisture. The results showed that vegetation cover increased soil moisture in the surface layer (0-20 cm) by 6.81% during wet periods (semi-monthly rainfall >30 mm) relative to an in-situ unvegetated control, but primarily induced a decline of soil moisture in the deep soil layer (20-180 cm) by 19.44% across two growing seasons. Redundancy analysis (RDA) and structural equation modeling (SEM) suggested that these vegetation effects on soil moisture were significantly correlated with vegetative height, leaf area, shallow root allocation, and slope gradient. Our study revealed that tall, small-leaved, and shallow-rooted plants on flat topographies were beneficial to soil water retention and replenishment. This implied that current restoration strategies may be significantly improved through the development of optimal communities and diverse terracing measures. Our findings are anticipated to provide effective guidance for soil water conservation, as well as ecosystem rehabilitation in dry and degraded regions.