干旱区生态环境知识资源中心

Soil erosion accelerates the degradation of semiarid ecosystems. How to explain the erosion processes is increasingly attracting scientists’ and practitioners’ attention. In this study, we hypothesized the intensity and uncertainty of runoff and sediment events as the ecological function of soil erosion by using structural equation modeling (SEM), and integrated observed influencing factors to represent the ecological structure of soil erosion. We further assessed the interactions of ecological structures and functions in four typical patch covers containing bare patch (B type), Andropogon patch (G type), Artemisia coparia patch (H type) and Spiraea pubescens patch (J type) in the Loess Plateau over three rainy seasons. We found that (1) the highest and lowest probabilities of runoff (approaching to 0.37 and 0.22) and sediment (approaching to 0.27 and 0.03) events were in B and J types respectively. B type produced the highest runoff coefficients (0.004) and yielded 2-3 times more sediment than other three vegetation patch covers. (2) Antecedent soil moisture and crown width were two relative more crucial indicators to simultaneously affect the uncertainty and intensity of soil erosion in different patch cover patterns. (3) Soil saturated hydraulic conductivity connected the soil and plant latent variables to indicate the complexity of ecological structure. (4) The measurement error of sediment yield became the main source of system error of the soil erosion assessment by SEM. These results suggest that the uncertainty of runoff and sediment event was an important descriptor to improve the interpretation of ecological function of soil erosion. The logistic regression and SEM form a new analysis framework to explore the interaction between ecological structure and function of soil erosion. This framework could enhance the understanding of erosion processes, and be beneficial to managing the restoration vegetation and utilizing the available water and soil resources in semiarid environment.