1. INTRODUCTION
Soil erosion is wide-ranging worldwide and has irreversible effects on all-natural and artificial ecosystems (Fu et al., 2011). It is not only the primary cause of soil deterioration (Marques et al., 2008), land productivity decline (Lantican et al., 2003), and degradation of rivers, lakes and estuaries but also often carries sediments and pollutants. Soil erosion has accelerated approximately 85% of global land degradation and resulted in a 17% reduction in food production (Tang et al., 2015), which has become one of the most serious societal and environmental problems in the world.
The middle and upper reaches of the Yellow River basin are located on the Loess Plateau of China. Owing to its unique soil characteristics, climate conditions and extremely fragile ecosystem (Wang et al., 2017; Xin et al., 2012), the Loess Plateau is a global hot spots of soil erosion. Approximately 40% of the whole Loess Plateau suffers from extremely high erosion, and the annual soil erosion modulus is more than 5000 t/km2 (Fu et al., 2011), which increases the risk of soil degradation and restricts the sustainable development of the ecosystem (Zhao et al., 2016). The high-risk areas are mainly concentrated in hilly and gully loess areas (Chen et al., 2007), which are the main source of Yellow River sediment. Since the 1950s, to control severe soil erosion, the government has issued many policies and implemented a series of soil conservation measures for the Loess Plateau. The control of soil and water loss on the Loess Plateau can be divided into two parts, namely, slope vegetation restoration (i.e., reforestation) and gully control (i.e., check dams).
As a main measure for reducing erosion, vegetation restoration has been widely used on the Loess Plateau, especially since the implementation of the ‘Grain for Green’ project. The increase in vegetation coverage can effectively control soil erosion through root consolidation of soil, increased infiltration, and reduced surface runoff and water flow velocity (Bruijnzeel, 2004), improving the resistance of the soil erosion and reducing water erosion capacity and sediment transport capacity (Rey et al., 2005; Vanacker et al., 2007). At the same time, the variation in vegetation cover will directly lead to changes in land use. In fact, land use changes will be affected by human activities, such as the implementation of policies and urban development and its impact on soil erosion is bidirectional. Changes in land use may not only lead to sediment reduction (Choukri et al., 2020) but also increase soil erosion and further lead to land degradation (Aneseyee et al., 2020). For example, Quiñonero-Rubio et al. (2016) used the WaTEM/SEDEM model to simulate the Upper Taibilla catchment of Spain and indicated that afforestation reduced sediment yield by 13.9% (1956-2000). Aneseyee et al. (2020) demonstrated that land use changes increased soil loss by 26.25% and sediment export by 3.45% (1988-2018), mainly due to the expansion of urbanization and reclamation.
The construction of check dams has become an effective soil and water conservation measure for sediment control. A large number of check dams have been constructed in the gully channel to intercept the sediment, block floods, and reduce downstream scouring (Ran et al., 2008). When the check dams are fully filled, the main component of the land behind check dams is surface soil that is rich in nutrients, and its fertility is much higher than that of sloped farmland. According to reports, approximately 3200 km2of dam croplands were formed in 2002 (Jin et al., 2012). Moreover, check dams effectively prevent sediment from entering the Yellow River, thus reducing the sediment concentration of the Yellow River (Ran et al., 2008). Although not widely used worldwide, check dams have been applied to control soil erosion and have been reported in France, the United States, Spain, China and elsewhere (Abedini et al., 2012; Bellin et al., 2011; Borja et al., 2018; Castillo et al., 2007; Fang, 2017). Polyakov et al. (2014) carried out field sampling in the Santa Rita Experimental Range in the United States and found that the check dams retained 50% of sediment yield. By measuring the sediment deposition within the gully channels of the Loreto catchment in the Andean Mountains, Borja et al. (2018) indicated that check dams reduced the sediment exported by more than 70%. Boix-Fayos et al. (2008) applied the WATEM-SEDEM model to the Rogativa catchment of Spain and showed that check dams reduced the sediment load by approximately 77% without land use changes. Xu et al. (2013) explored the interception benefit of check dams in the Yanhe River watershed and revealed that the proportional reduction in sediment reached from 34.6-48.0% in the rainy season (1984-1987) and increased to between 79.4 and 85.5% from 2006-2008.
In the past few decades, the Loess Plateau has also undergone varying degrees of climate change, in which precipitation variation plays a leading role in soil erosion. Sun et al. (2015) analyzed the precipitation changes of the Loess Plateau from 1961 to 2011, and found that the amount of precipitation in most areas showed a downward trend. In addition, over the past six decades, the sediment concentration of the Yellow River has decreased substantially (Wu et al., 2020), and the main tributaries in the middle reaches of the Yellow River have reported that sediment load has decreased rapidly (Gao et al., 2017; Rustomji et al., 2008; Yue et al., 2014). Hence, it is particularly important to understand the driving factors and mechanisms behind the changes in soil erosion and sediment load, which is also a prerequisite for sustainable watershed management (Montgomery, 2007).
To quantitatively identify the impact of the three factors (precipitation variation, land use changes and check dams), the Kuye River watershed, with an area of 8651 km2 on the Loess Plateau, was selected as the study area by combining field investigations with model simulation. The main objectives of this study were to (1) explore the characteristics of soil erosion and sediment yield and to (2) quantify the contribution of precipitation variation, check dam construction and man-made land use changes to sediment reduction in the Kuye River watershed, which is a typical watershed in the coarse sandy hilly catchment region of the Yellow River basin.