Influence of Soil Moisture in Semi-Fixed Sand Dunes of the Tengger Desert, China, Based on PLS-SEM and SHAP Models

Abstract

Drought stress significantly limits the function and stability of desert ecosystems. This research examines the distribution characteristics of soil moisture across different microtopographic types in the semi-fixed dunes located at the southeastern edge of the Tengger Desert. We constructed a path model to examine the direct and indirect impacts of topography, shrub vegetation, and herbaceous vegetation. The data encompassed soil moisture, topography, and vegetation variables, which were collected from field experiments to ensure their accuracy and relevance. Furthermore, SHAP models based on machine learning algorithms were utilized to elucidate the specific mechanisms through which key factors influence soil moisture. The results of the descriptive statistics indicate the highest surface soil moisture content, recorded at 1.21%, was observed at the bottom of the dunes, while the leeward slopes demonstrated elevated moisture levels in the middle and deep soil layers, with measurements of 2.25% and 2.43%, respectively. Soil moisture at different depths initially decreases and then increases with greater herbaceous cover and slope direction, while surface soil moisture follows a similar trend in terms of height difference, with 3 m serving as the boundary for trend changes. Middle and deep soil moistures initially increase and then decrease with greater biomass and shrub coverage, with 30 g and 40% serving as the boundary for trend changes respectively. This study elucidates the spatial distribution patterns and influencing factors of soil moisture in semi-fixed dunes, offering valuable references for the establishment of sand-stabilizing vegetation in desert regions.