Changes in Runoff and Sediment Load and Potential Causes in the Malian River Basin on the Loess Plateau

The loessial tableland is a unique landform type on the Loess Plateau in China. Long-term soil erosion has led to the retreat of gullies and the rapid reduction of fertile arable land, which has further decreased agricultural production. In this study, we chose the Malian River basin to analyze the temporal and spatial variation of its runoff and sediment load, as well as the potential causes. The annual runoff and sediment load at six hydrological stations in the study area were collected for the period between 1960 and 2016. The Mann−Kendall and Pettitt tests were respectively applied to detect temporal variations and abrupt changes in the runoff and sediment loads. The results showed that an abrupt change in the runoff and sediment loads occurred in 2003. The average annual runoff in the Malian River was 4.42× 108 m3 yr−1 from 1960 to 2002, and decreased to 3.32× 108 m3 yr−1 in 2003–2016. The average annual sediment load was 1.27 × 108 t yr−1 in 1960–2002, and decreased to 0.65 × 108 t yr−1 in 2003–2016. The spatial patterns in the sediment load suggested that the Hongde sub-basin contributed a higher sediment count to the Malian River, which may require additional attention for soil and water conservation in the future. Anthropogenic activities significantly affected runoff and sediment load reduction according to the double-mass curve method, accounting for 90.7% and 78.7%, respectively, whereas rainfall changes were 9.3% and 21.3%, respectively. As such, the present study analyzed the loessial tableland runoff and sediment load characteristics of the Malian River basin for soil and water erosion management.


Introduction
The loessial tableland is an important agricultural production region on the Loess Plateau in China. It has a flat terrain and is surrounded by valleys due to a high deposition of loess soil [1][2][3]. Thousands of years of human habitation and agricultural activities have increased soil erosion in the tableland, which not only affects the management and development of the Yellow River, but also fragments the Loess Plateau [4][5][6]. Significant sediment discharge into the Yellow River has been observed due to extreme soil erosion, resulting in high-risk sedimentation and flooding in the lower reaches of the Yellow River [7,8]. As a result, many studies have investigated the loessial tableland runoff and sediment load as a means to further understand soil erosion changes across the Loess Plateau [9].
River runoff and sediment load connect marine and terrestrial ecosystems, and have important effects on surface material transport, geomorphology, and ocean circulation [10][11][12]. In recent decades, numerous large rivers have exhibited runoff and sediment loads, including the Ebor River, Mississippi River, Colorado River, and Columbia River [13][14][15][16]. Therefore, river protection and the promotion of sustainable basin development has been of great significance in understanding the variation of sediment load and runoff, as well as their probable sources. Lowered runoff and sediment load have been significantly affected by climate change and human activities such as the implementation of reservoirs, check dams, soil and water conservation, and afforestation. Climate change stimulates the water cycle and changes the distribution and quantity of runoff, especially rainfall. Labat et al. reported increasing runoff in 221 rivers across the world following an increase in rainfall [17]. However, other research suggests that anthropogenic activities predominantly affected runoff and sediment loads. Walling observed significant decreased annual runoff and sediment loads in 145 international rivers due to severe anthropogenic activities [18]. Wang et al. indicated that sediment load decreased with an increasing percentage of vegetation cover in the Yellow River [19]. In addition, the construction of terraces and check dams also led to runoff and sediment load reduction in related areas [20][21][22].
The Yellow River, the second-largest river in China, is among the highest sedimentladen rivers [23]. Driven by a series of soil and water conservation measures, the sharp decrease in runoff and sediment load in the Yellow River and its tributaries has attracted wide attention. Wei et al. investigated the runoff changes of eight hydrological stations across the Yellow River mainstream, to which the annual average streamflow exhibited an M-type spatial pattern and a parabolic annual average suspended sediment discharge curve. A distinct decreasing streamflow and suspended sediment discharge was observed between 1950 and 2013 [24]. Mu et al. conducted a systematic sediment change analysis of the Yellow River from 1919 to 2008 by using the anomaly accumulation method and double-mass curve, and anthropogenic activities were deemed the primary cause of the significant sediment decrease in the Yellow River [25]. Yao et al. observed a decreasing annual average runoff and sediment load trend in the Yellow River across 50 years of recent data, primarily due to anthropogenic activities [26]. Similarly, Gao et al. analyzed the linkage between runoff and sediment load in 14 Loess Plateau sub-basins; the results indicated that check dams and reservoirs significantly affected sediment load reduction, while vegetation restoration was the main factor after 2000 [27].
The Malian River, a third Yellow River tributary, is located on the loessial tableland. It contributed an annual average of 1.34 × 10 8 t sediment to the Yellow River, accounting for 8% of the annual total in the whole basin [28,29]. Existing literature has reported significant reductions in runoff and sediment load in a majority of rivers except those on the loessial tableland. A number of studies have investigated the temporal and spatial variations in runoff and sediment loads on the entire Loess Plateau; however, these studies did not deeply investigate the entirety of the loessial tableland, such as the Malian River. Most of the existing research has focused on the erosion model simulations and isotope labeling from soil erosion studies, but no analysis of the long-term series data has been conducted using the runoff and sediment loads. Furthermore, few studies have analyzed the potential sources of these regional runoff and sediment load changes. Therefore, our study sought to investigate the spatial-temporal variations in annual runoff and sediment load in the Malian River basin and characterize the underling factors that affect these runoff and sediment load variations. We checked and revised the data homogenization process and described the characteristics of runoff and sediment loads in the Malian River basin in detail. This provides a theoretical basis for soil and water conservation in the gully region of the Loess Plateau.

Study Area
The Malian River is situated in the central reaches of the Yellow River (between 35 • 14 -37 • 23 N and 106 • 40 -108 • 35 E). The total length of the main stream is 374.8 km, with a basin area of 19,086 km 2 . The basin has a temperate continental climate that is hot and rainy in the summer and cold and dry in the winter. The average annual rainfall is 531.12 mm, and the average annual air temperature is 9.14 • C. The northern part of the basin is mainly low mountains, and the middle and downstream of the basin are flat loessial tableland. This basin has the typical characteristics of the Loess Plateau, with basic geomorphic units such as loessial tableland, loessial ridge, and loessial hillock, which is representative of the gully region on the Loess Plateau [30]. The average specific sediment yield of the basin is 5862.94 t km −2 yr −1 .

Data Sources
The annual runoff and sediment load from 1960 to 2016 were collected from six hydrologic stations: the Hongde, Qingyang, and Yuluoping stations on the main stream, and the Yuele, Banqiao, and Jiaqiao stations on the tributaries of the river (Table 1 and Figure 1). The Yuluoping hydrologic station is at the outlet of the basin, which has a drainage area of 19,019 km 2 . The presented hydrological data were provided by the Yellow River Water Conservancy Commission (YRCC) of the Ministry of Water Resources, China. The continuous annual rainfall and air temperatures were collected from eight meteorological stations between 1960 and 2016 (Chinese Climate Center; http://data.cma. cn/site/index.html). Most importantly, the consistency, homogeneity, and quality of the data were checked prior to their release. 531.12 mm, and the average annual air temperature is 9.14 °C. The northern part of the basin is mainly low mountains, and the middle and downstream of the basin are flat loessial tableland. This basin has the typical characteristics of the Loess Plateau, with basic geomorphic units such as loessial tableland, loessial ridge, and loessial hillock, which is representative of the gully region on the Loess Plateau [30]. The average specific sediment yield of the basin is 5862.94 t km −2 yr −1 .

Data Sources
The annual runoff and sediment load from 1960 to 2016 were collected from six hydrologic stations: the Hongde, Qingyang, and Yuluoping stations on the main stream, and the Yuele, Banqiao, and Jiaqiao stations on the tributaries of the river (Table 1 and Figure  1). The Yuluoping hydrologic station is at the outlet of the basin, which has a drainage area of 19,019 km 2 . The presented hydrological data were provided by the Yellow River Water Conservancy Commission (YRCC) of the Ministry of Water Resources, China. The continuous annual rainfall and air temperatures were collected from eight meteorological stations between 1960 and 2016 (Chinese Climate Center; http://data.cma.cn/site/index.html). Most importantly, the consistency, homogeneity, and quality of the data were checked prior to their release.

Methodologies
A simple linear regression test and a Mann-Kendall test were applied to characterize any variations in the annual runoff and sediment load trends at the different stations. The abrupt changes of runoff and sediment load were detected using the Pettitt test method [31][32][33]. Additionally, a double-mass curve and a sediment identity factor method were implemented to evaluate the potential runoff and the effects of the sediment load changes.

Mann-Kendall Trend Test
The Mann-Kendall non-parametric statistical test (M-K test) can effectively detect variations in hydrological data over time, and has been widely employed in the fields of hydrology and climate [34][35][36]. The method assumes a stable time series, and a random independence of the time series is required. This method reflects a changing trend by constructing statistics. A positive standardized statistic value indicates an increasing sequence trend; otherwise, a decreasing statistical trend is observed. At a significance level of 0.05, the critical value of the statistical parameters was ±1.96. At a significance level of 0.01, the critical value of the statistical parameters was ±2.58.

Change Point Analysis
The Pettitt test determines the mutation point of a sequence based on the change of trend of a long time series [37]. For a time series X with a length of T, the change point is t. Then, the hypothesis sequence is divided into two segments. The two samples are (x 1 , x 2 , . . . , x t ) and (x t+1 , x t+2 , . . . , x T ). The formula for |U t | is written as follows: where sgn() is determined as follows: The possible abrupt change points of a time series are determined by the maximum value of |U t |.

Runoff Depth and Specific Sediment Yield
The runoff and sediment load spatial patterns in the Malian River basin were characterized based on the runoff depth and specific sediment yield. The runoff depth (R) was calculated as: The area between two hydrologic stations was calculated as follows: where V represents the annual runoff (m 3 ) and A represents the control area between the two hydrological stations (km 2 ). The specific sediment yield (t km −2 yr −1 ) exhibited similarities to the runoff depth, and was thus calculated based on the observed sediment load-to-control area ratio.

Double-Mass Curve Method
In general, the double-mass curve method is applied to investigate consistencies and/or variations in two parameters [25,38]. A relationship line is drawn in a rectangular coordinate system between the individual continuous cumulative values of two separate variables over the same period. The slope is generally analyzed after a double-mass curve of runoff and sediment load and rainfall is drawn. If the slope of the straight line does not deviate significantly, it means that human activities have no significant effect on runoff and sediment load; otherwise, the results indicate the significance of anthropogenic activities [39].

Sediment Identity Factor Analysis
The sediment identity factor method was first applied in economics. Wang et al. applied this method to characterize various conditions and their potential contributions in reducing the sediment load in the Yellow River basin [40]. According to their results, the basin sediment load variations were significantly dependent upon the rainfall, runoff coefficient, and sediment concentration. As such, the river sediment load was defined as: where S is the annual sediment load of the basin, R is the annual runoff, P is the annual rainfall, Cr is the ratio of annual runoff to rainfall (runoff coefficient), and SSC is the suspended sediment concentration.
When defining a function as r(X) = X −1 dX/dt with time, the counterpart for sediment load decrease rate was: Therefore, the variation of sediment load in the basin was regarded as the attribution from rainfall, the runoff coefficient, and the sediment concentration. The reduction rate of the sediment load was calculated as the sum of the reduction rates of these three factors, and then their relative contributions to reducing sediment in the basin were estimated [41].

Temporal Variation of Runoff And Sediment Load
The annual runoff and sediment load varied distinctly within different decades in the Malian River basin. The average annual runoff in the Malian River basin was 4.13 × 10 8 yr −1 , ranging from 2.15 × 10 8 m 3 (1972) to 9.62 × 10 8 m 3 (1964). The average annual sediment load was 1.12 × 10 8 t yr −1 , ranging from 0. 16   The M-K test trend analysis was applied to investigate the runoff and sediment loads at six Malian River basin hydrological stations from 1960 to 2016 ( Table 2). The annual runoff at the Yuluoping and Banqiao Stations showed significant decreasing trends (p < 0.05).  The M-K test trend analysis was applied to investigate the runoff and sediment loads at six Malian River basin hydrological stations from 1960 to 2016 ( Table 2). The annual runoff at the Yuluoping and Banqiao Stations showed significant decreasing trends (p < 0.05). The M-K test trend analysis was applied to investigate the runoff and sediment loads at six Malian River basin hydrological stations from 1960 to 2016 ( Table 2). The annual runoff at the Yuluoping and Banqiao Stations showed significant decreasing trends (p < 0.05).      Figure 5 shows the spatial distribution of runoff depth in the Malian River basin during different decades. Overall, the runoff depth showed a decreasing trend from south to north during different decades in the basin. The annual average runoff depth indicated a decreasing trend from 1960 to 2016 at a rate of 0.14 mm yr −1 . From the 1960s to 1980s, the runoff depth varied moderately. In the 1990s, the runoff depth in most sub-basins was less than 30 mm, except for that controlled by the Yuele Station. The runoff depth decreased significantly in the downstream of the Malian River basin from 1990-1999. After 2000, the runoff depth continued to decrease, and an extremely low runoff depth was detected in the area controlled by Hongde (10 mm yr −1 ) between 2010 and 2016.  Figure 5 shows the spatial distribution of runoff depth in the Malian River basin during different decades. Overall, the runoff depth showed a decreasing trend from south to north during different decades in the basin. The annual average runoff depth indicated a decreasing trend from 1960 to 2016 at a rate of 0.14 mm yr −1 . From the 1960s to 1980s, the runoff depth varied moderately. In the 1990s, the runoff depth in most sub-basins was less than 30 mm, except for that controlled by the Yuele Station. The runoff depth decreased significantly in the downstream of the Malian River basin from 1990-1999. After 2000, the runoff depth continued to decrease, and an extremely low runoff depth was detected in the area controlled by Hongde (10 mm yr −1 ) between 2010 and 2016. Sustainability 2021, 13, x FOR PEER REVIEW 8 of 16 The specific sediment yield reflected soil erosion rates in the sub-basin. An overall decreasing annual sediment yield was observed between 1960 and 2016 at a rate of 53.50 t km −2 yr −1 . Figure 6 shows a higher specific sediment yield in the upstream of the basin (controlled by Hongde station). In the 1960s, the specific sediment yield decreased gradually from upstream to downstream at an average of 6721.7 t km −2 yr −1 in the entire basin. From the 1970s to 1990s, however, the specific sediment yield varied in different regions, with some higher than 5000 t km −2 yr −1 . After 2000, no sub-basins exhibited a specific sediment yield larger than 5000 t km −2 yr −1 , and an extremely low specific sediment yield was observed downstream of the Malian River basin between 2010 and 2016. The specific sediment yield reflected soil erosion rates in the sub-basin. An overall decreasing annual sediment yield was observed between 1960 and 2016 at a rate of 53.50 t km −2 yr −1 . Figure 6 shows a higher specific sediment yield in the upstream of the basin (controlled by Hongde station). In the 1960s, the specific sediment yield decreased gradually from upstream to downstream at an average of 6721.7 t km −2 yr −1 in the entire basin. From the 1970s to 1990s, however, the specific sediment yield varied in different regions, with some higher than 5000 t km −2 yr −1 . After 2000, no sub-basins exhibited a specific sediment yield larger than 5000 t km −2 yr −1 , and an extremely low specific sediment yield was observed downstream of the Malian River basin between 2010 and 2016.

Contributions of Human Activities and Rainfall to Runoff and Sediment Load Reduction
Double-mass curves were drawn to further quantify changes in the runoff and sediment load before and after the abrupt change (Figure 7). The results suggested that rainfall contributed 9.3% to the decrease in runoff, and human activities contributed to 90.7% of the runoff. For the sediment load, the changes in the annual rainfall accounted for 21.3% of the decrease, and human activities contributed to 78.7%. The analysis indicates that human activities had significant impacts on the runoff and sediment-load reduction in the Malian River basin (Table 3).

Contributions of Human Activities and Rainfall to Runoff and Sediment Load Reduction
Double-mass curves were drawn to further quantify changes in the runoff and sediment load before and after the abrupt change (Figure 7). The results suggested that rainfall contributed 9.3% to the decrease in runoff, and human activities contributed to 90.7% of the runoff. For the sediment load, the changes in the annual rainfall accounted for 21.3% of the decrease, and human activities contributed to 78.7%. The analysis indicates that human activities had significant impacts on the runoff and sediment-load reduction in the Malian River basin (Table 3).    The sediment identity factor method indicated that the sediment concentration, runoff coefficient, and rainfall contributed an average of 24.5%, 57.8%, and 17.7% to the sediment load reduction, respectively (Figure 7). Therefore, the sediment load reduction was mainly achieved by reducing the sediment concentration and the runoff coefficient. When we compared the attributions from rainfall changes, both the double-mass curve and sediment identity factor method showed similar results.

Impacts of Climate Change on Runoff and Sediment Load
Previous studies have examined changes in the runoff and sediment loads of different rivers and the influencing factors on the Loess Plateau [42]. Climate changes affected the runoff and sediment load based on the rainfall and air temperatures [24,[42][43][44]. Figure  8 shows the changing trends of annual average air temperature and annual average rainfall in the Malian River basin. The annual rainfall decreased by 0.074 mm yr −1 , and the annual air temperature increased by 0.031 °C yr −1 . This suggested that the study area experienced a relatively drier and warmer period from 1960 to 2016, which was similar to the results of the previous studies of the Yellow River basin [45][46][47]. The sediment identity factor method indicated that the sediment concentration, runoff coefficient, and rainfall contributed an average of 24.5%, 57.8%, and 17.7% to the sediment load reduction, respectively (Figure 7). Therefore, the sediment load reduction was mainly achieved by reducing the sediment concentration and the runoff coefficient. When we compared the attributions from rainfall changes, both the double-mass curve and sediment identity factor method showed similar results.

Impacts of Climate Change on Runoff and Sediment Load
Previous studies have examined changes in the runoff and sediment loads of different rivers and the influencing factors on the Loess Plateau [42]. Climate changes affected the runoff and sediment load based on the rainfall and air temperatures [24,[42][43][44]. Figure 8 shows the changing trends of annual average air temperature and annual average rainfall in the Malian River basin. The annual rainfall decreased by 0.074 mm yr −1 , and the annual air temperature increased by 0.031 • C yr −1 . This suggested that the study area experienced a relatively drier and warmer period from 1960 to 2016, which was similar to the results of the previous studies of the Yellow River basin [45][46][47]. In our study, rainfall changes were responsible for 9.3% of the runoff and 21.3% of the sediment load reduction. A comparison with other tributaries on the Loess Plateau indicated that changes in the runoff and sediment loads in the Malian River basin were In our study, rainfall changes were responsible for 9.3% of the runoff and 21.3% of the sediment load reduction. A comparison with other tributaries on the Loess Plateau indicated that changes in the runoff and sediment loads in the Malian River basin were insignificant. However, the attributions were consistent with other rivers. As shown in Table 4, most of the studies suggested that anthropogenic activities significantly reduced the runoff and sediment load in most rivers surrounding the Yellow River basin [43,[48][49][50][51][52][53]. For example, Wang et al. reported that 30% of the sediment load reduction observed between 1969 and 2005 in the Yellow River basin area was produced by rainfall changes. Similar attributions were also documented in other tributaries of the Yellow River basin [51]. Note: * means the impact of soil and water conservation.

Impacts of Human Activities on Runoff and Sediment Load
Anthropogenic activities produced a significant runoff and sediment load decrease in the Yellow River basin, particularly in the Loess Plateau region. In particular, large-scale water and soil conservation measures have been implemented in the Yellow River basin region since the 1950s, including afforestation, fish-scale pits, terraces, and check dam and reservoir applications [54][55][56]. Numerous studies have reported the significance of soil and water conservation measures in drastically reducing runoff and sediment loads in the Yellow River [57][58][59][60] and many other rivers around the world. Meade and Moody showed sediment discharge changes in the Missouri-Mississippi River from 1987 to 2006, and their results indicated that check-dam retention characteristics in the 1950s significantly reduced the sediment in the muddy part of the river [14]. Sharda et al. analyzed the runoff and sediment load reduction benefits of terrace fields in India, and showed that the efficiency of water and sediment reduction could reach approximately 80% and 90%, respectively [61]. In addition, reservoir construction also significantly affected the runoff and sediment load reduction. Naik and Jay showed that the decrease in runoff and sediment load in the Columbia River over the past 150 years was due to reservoir regulation and irrigation water control [16].
Previous literature has indicated that the 80% runoff reduction and 70% sediment load decrease in the Yellow River basin were a result of anthropogenic activities. Check dams are an effective measure to trap large amounts of sediment [62,63]. During the past 70 years, approximately 56,000 check dams have been established in the Loess Plateau region, including 5500 large dams [64]. However, the number of check dams in the Malian River basin is limited. Statistics have indicated the presence of 211 check dams (storage capacity > 0.005 × 10 8 m 3 ) in the Malian River basin, with a total storage capacity of 1.75 × 10 8 m 3 . Reservoirs can reduce the flood peaks and increase flow during the dry period by impounding upstream incoming water [65,66]. Only two medium-sized reservoirs are present in the Malian River basin, thereby limiting the trapping effects of reservoirs to streamflow reduction in the basin.

Conclusions
The present study examined the spatial and temporal variations of runoff and sediment load at six hydrological stations in the Malian River basin between 1960 and 2016.The Malian River basin exhibited a decreasing annual runoff and sediment load trend of 0.027 × 10 8 m 3 yr −1 and 0.0102 × 10 8 t yr −1 , respectively. A higher runoff depth was observed downstream, while the sediment yield was higher upstream of the basin. Rainfall and anthropogenic activity contributed to runoff and sediment load reductions in the Malian River basin based on the double-mass curve. Specifically, anthropogenic activities accounted for 90.7% and 78.7% of the runoff and sediment load reduction, respectively. The sediment identity factor analysis indicated that the sediment concentration, runoff coefficient, and rainfall contributed an average of 24.5%, 57.8%, and 17.7% to the sediment load reduction, respectively. The sediment yield was still higher in the Malian River basin compared to other watersheds on the Loess Plateau; however, it exhibited an evident decreasing trend. The spatial distribution results exhibited a high sediment yield upstream of the study area, possibly due to significant implementations of soil and water conservation in this region.  After the Grain for Green policy was instituted, the region of forestland and grassland in the basin gradually increased, which is an important factor for the reduction of runoff and sediment loads. Due to the geographical location, water resources are restricted in the northern region, where the vegetation restoration is far less vigorous than in the southern forest region, and the sediment load is still the most serious across the basin. In addition, numerous gullies in the northern region hinder vegetation, and since the vegetation is mostly herbaceous, runoff and sediment load reductions were limited compared to the use of macrophanerophytes in the southern region. Future research should be conducted to examine soil and water conservation in the northern basin region, which is controlled by Hongde Station.

Conclusions
The present study examined the spatial and temporal variations of runoff and sediment load at six hydrological stations in the Malian River basin between 1960 and 2016.The Malian River basin exhibited a decreasing annual runoff and sediment load trend of 0.027 × 10 8 m 3 yr −1 and 0.0102 × 10 8 t yr −1 , respectively. A higher runoff depth was observed downstream, while the sediment yield was higher upstream of the basin. Rainfall and anthropogenic activity contributed to runoff and sediment load reductions in the Malian River basin based on the double-mass curve. Specifically, anthropogenic activities accounted for 90.7% and 78.7% of the runoff and sediment load reduction, respectively.
The sediment identity factor analysis indicated that the sediment concentration, runoff coefficient, and rainfall contributed an average of 24.5%, 57.8%, and 17.7% to the sediment load reduction, respectively. The sediment yield was still higher in the Malian River basin compared to other watersheds on the Loess Plateau; however, it exhibited an evident decreasing trend. The spatial distribution results exhibited a high sediment yield upstream of the study area, possibly due to significant implementations of soil and water conservation in this region.
Author Contributions: The research content was written by M.D. and X.M. The revision of the manuscript was completed by G.Z., P.G. and W.S. All authors have read and agreed to the published version of the manuscript.