Land‐Use Intensity Reversed the Role of Cropland in Ecological Restoration Over the World's Most Severe Soil Erosion Region

Long‐term extending cultivation activities resulted in the world's worst soil erosion on the Chinese Loess Plateau (LP). By converting cropland into vegetated land, the Grain for Green Project (GfGP)—the world's largest investment revegetation project—effectively alleviates the soil erosion on the LP. However, during the GfGP implementation, the positive effect of cropland to the revegetation and soil erosion control has been underestimated to date, hindering a comprehensive evaluation to the effect of cropland on ecological restoration. Here, we evaluated the effect of the GfGP on soil erosion control across the LP, analyzed the dominant driver of the LP vegetation greening, and further identified the contributions of croplands to this world's largest revegetation project. We found that the vegetation of the LP was significantly improved and its leaf area increased by 1.23 × 105 km2 after the implementation of the GfGP, which contributed 42% to the decrease of the LP soil loss. Among them, our results show that cropland contributed 39.3% to the increased leaf areas of the LP, higher than grassland (36.3%) and forestland (14.3%). With the reduction of agricultural area, the contribution of cropland to the increased leaf areas in the LP was still the largest, which was mainly due to the increase in cropland utilization intensity. This study highlights the significance of the GfGP in soil erosion control and revises our understanding of the role of cropland in ecological restoration and society development.

soil erosion will also result in considerable harm to environment and society, such as desertification, landslides, decreased land productivity, water pollution, food security, poverty, etc (Poesen, 2018;Vanacker et al., 2003). Considering the negative influences caused by the soil erosion of the LP, the Chinese government has been taking active actions to effectively control its serious erosion. Thus, diverse approaches and policies have been promoted by the policy-makers to control soil erosion of the LP . Currently, the erosion levels of the LP have been restored to the state of the agriculture period in 740 AD, and the annual sediment transport of the YR has decreased from 1.5 billion tons in 1950 to 200 million tons in 2013 (Chen et al., 2015).
The Grain for Green Project (GfGP) is the largest revegetation program ever implemented in the world, and the most impressive implementation in controlling soil erosion and reducing sediment on the LP (Hua et al., 2016;Ren et al., 2015). Essentially, it is an experiment in planting vegetation on a large scale to stabilize soil and reduce soil erosion (Feng et al., 2016), at a cost of $8.7 billion from 1999 to 2015 financed by the Chinese government. To date, vegetation greening in varying degrees has been witnessed across the LP after the GfGP implementation (Chen et al., 2015;Sun et al., 2013;Zhang et al., 2021), and numerous studies have indisputably attested that soil erosion and sediment discharge over the LP have been significantly reduced (Karnieli et al., 2014;Li et al., 2016;Spracklen et al., 2012). However, the alleviated soil erosion is the result of the multifactor intervention, including revegetation, terrace farming, irrigation system, dams, etc . Therefore, in order to assess the effectiveness of the GfGP in controlling soil erosion and to determine its contribution to environmental well-being, attention needs to be taken to ascertain the revegetation-resulted alleviation in soil erosion.
Intensive reclamation activities for feeding a growing population have led to the destruction of surface vegetation (Chazdon, 2008;Lamb et al., 2005;Zhou et al., 2023) and further to the LP's worst soil erosion . Accordingly, for controlling soil erosion, the GfGP takes considerable efforts to improve vegetation coverage by converting cropland to grassland and forest (Liu et al., 2008). However, there is disagreement as to the negative effects of cropland on surface vegetation destruction. Satellite data show that from 2000 to 2017, cropland accounted for 32% of the net increase of leaf areas in China (Chen et al., 2019), second only to forestland. Yet, with respect to the revegetation of the LP, the focus of prior research is dedicated to the grassland and forest land (Li et al., 2016;Liu et al., 2021;Sun et al., 2014). The contribution of cropland to the revegetation process may have been largely downplayed. In particular, with the government's plans to expand the GfGP initiative (Hua et al., 2016), there is a great need to identify which land use is the main force behind the LP vegetation greening, which can make sure a successful adjustment of the GfGP priorities and methods.
In this study, we aim to evaluate the effectiveness of revegetation promoted by the GfGP in controlling soil erosion, and identify the dominant land use of the LP vegetation greening, while further understanding the importance of cropland in the revegetation process. Based on the Revised Universal Soil Loss Equation (RUSLE), soil loss on the LP during 1995-2015 was evaluated. On this basis, we then accessed the contribution of revegetation to the soil erosion control of the LP. To identify the dominant land use of the LP vegetation greening, the leaf area index (LAI) was utilized to assess the contribution of forestland, grassland, and cropland in the revegetation process after the GfGP. The recent leading role of cropland in vegetation greening on the LP after the GfGP, could clarify the potential of cropland in future revegetation management.

Study Region
The LP (100°54′-114°33′E and 33°43′-41°16′N) is located in the middle reaches of the YR basin, and covers an area of 620,000 km 2 (Figure 1a). Its annual average temperature is 3.7-14.0°C. The annual average precipitation shows a growing trend from northwest to southeast with a value of 200-750 mm (Zhang et al., 2018). The terrain varies from 75 to 5,149 m (Figure 1), and the surface is covered by extremely erodible loess with an average thickness of 100 m. Moreover, the LP surface is one of the world's most eroded areas (Figure 1b) because of high-intensity rainstorms in summer, erodible loess soil, and steep topography (Zhang & Liu, 2005). The main type of land use of this Plateau is grassland, followed by cropland (Figure 1c).

Data
The annual normalized differential vegetation index (NDVI) data at 1 km spatial resolution is used to calculate the C factor and derived from SPOT-VGT PROBA-V 1 KM PRODUCTS. The land use/land cover data is provided by Landsat Data Series (TM, ETM) and used to evaluate the P factor. The DEM data at 30 m spatial resolution is used to assess the LS factor. The R factor is estimated by annual precipitation data at 1 km spatial resolution. The R factor and K factor are calculated using the monthly precipitation data and soil attribute data at 1 km spatial resolution. Sediment load data from the gauging stations of Tangnaihai (TNH), Shizuishan (SZS), Longmen (LM), Tongguan (TG) are used to validate the soil erosion simulation results based on the RUSLE. The LAI products with 0.05° spatial resolution are used to quantify the LP vegetation greening and browning from 1995 to 2015. Cropping system data, grain production area, and yield data from 1995 to 2010 are used to analyze the intensity of cropland utilization.

Evaluation of Soil Erosion on the Loess Plateau
In this study, the RUSLE model was adopted to evaluate the soil erosion on the LP from 1995 to 2015. This model is a mathematical soil erosion model which is based on the quantitative relationship between soil erosion and its main influencing factors. It is an empirical soil erosion prediction equation which is established based on experimental observation data combined with statistical analysis and generalization of soil erosion influencing factors to quantitatively predict annual average soil loss. The calculation principle is to multiply the mathematical expressions of four factors, in order to quantify and estimate the annual average soil erosion caused by rainfall and surface flow. Above four factors are the rainfall erosivity (R factor), soil erodibility (K factor), topography (LS factor) and land use (C factor and P factor), respectively. The RUSLE model (Renard et al., 1991) is commonly utilized to evaluate the soil erosion rate at long time series under different management practices (Li et al., 2022;Rajbanshi & Bhattacharya, 2020;Thomas et al., 2018). Here, the RUSLE model was expressed as follows: where A (t hm −1 yr −1 ) stands for the potential soil erosion rate; R (MJ mm hm −2 h −1 yr −1 ) represents the precipitation erosivity factor; K (t hm 2 h MJ −1 mm −1 hm −2 ) is the soil erodibility factor; LS (dimensionless), the slope length and steepness factor; C (dimensionless) expresses the land cover and management factor; P (dimensionless) is the support practice factor. For details about how to calculate the above factors are presented in the next (from Sections 2.3.1-2.3.5).

R Factor
The R factor stands for the rainfall erosion intensity, which is the main source of soil erosion dynamics. The R factor in this study was determined by a precipitation erosivity model according to monthly precipitation (Wischmeier & Smith, 1978): where P i (mm) stands for the monthly precipitation, and P (mm) stands for the annual precipitation. The R A (100 f t in a −1 h −1 yr −1 ) is the annual precipitation erosivity in United Stated units. In detail, f is the feet, t stands for the short tons, i represents the inches and a is the acres. Therefore, R A (100 f t in a −1 h −1 yr −1 ) needs to be converted into R (MJ mm hm −2 h −1 yr −1 ) by multiplying by 17.

K Factor
The K factor reflects the difference in soil erosion caused by different soil textures. It actually is an important representative of soil resistance in RUSLE model, represents the intrinsic properties of soil sensitive to soil erosion. Different soil properties and types show significant differences in their sensitivity to water flow velocity. The larger the K value is, the more easily soil erosion occurs. In this study, it was assessed based on the EPIC model (Williams, 1990 where S a (%), S i (%), C l (%), and OM (%) stand for the content of sand, silt, clay, and organic matter; SN = 1 − Sa/100. The K (t hm 2 h MJ −1 mm −1 hm −2 ) is the soil erodibility.

LS Factor
The LS factor stands for the impact of topographic relief on soil erosion, which restricts the redistribution of surface matter and energy, and determines the movement state and direction of surface runoff. The L factor refers to the ratio of sediment yield of a slope with a specific slope length to that of a standard plot with a slope length under the same other conditions. The S factor refers to the ratio of slope sediment yield of a specific slope to that of a standard plot under the same other conditions. In watershed or regional scale soil erosion assessment, the LS factor can be directly calculated by DEM. Considering that the LP has more steep slopes, the L and S factors are obtained by Equations 4 and 5, respectively (Liu et al., 1994;Mccool et al., 1987;Wischmeier & Smith, 1978): where L (dimensionless), the slope length factor; S (dimensionless), the steepness factor; λ (m), the slope length; θ (°), the slope gradient; m (dimensionless), the slope length.

C Factor
Vegetation cover and land cover protect soil primarily by trapping rainfall, increasing infiltration, and slowing runoff rates, thereby reducing soil erosion. Vegetation cover, vegetation type and management mode are different and have different effects on soil erosion. In the RUSLE model, the C factor is used to reflect the influence of vegetation cover and management measures on soil erosion. It is defined as the ratio of soil loss of specific vegetation cover and field management land to soil loss of control land under the same conditions under timely ploughing and continuous fallow. In this study, it was evaluated by Equations 6 and 7, which was established based on the NDVI data and fractional vegetation coverage (Cai et al., 2000): where FVC (%) stands for the fractional vegetation coverage, NDVI min is the NDVI minimum in this area, and NDVI max is the NDVI maximum in this area.

P Factor
The P factor refers to the effects of soil and water conservation measures on reducing the amount of soil erosion. It is defined as the ratio of the amount of erosion on the land with a specific measure to that on the slope. In large-scale watershed or regional soil erosion studies, P values are usually determined by assigning different land use types. Here, according to the available land use/land cover data and the assigned values in previous studies (Table 1), ArcGIS software was used to obtain the support practice factor values.

Evaluation of the Role of Vegetation Dynamics in Soil Erosion Control
The GfGP was initiated in 2000 across the LP. To quantify the influence of vegetation dynamics on soil erosion rate change in the LP, three scenarios (P1, P2 and P3) were set up based on the year 2000 as the dividing point in this study, as follows: P3 = 2000−2015 × 1995−1999 × 1995−1999 Land use/land cover P value Reference   2000-2015, 1995−1999 and 1995−1999 are the annual average C factor and annual average P factor respectively during 1995-1999 (both of C and P indicate the influences of vegetation dynamics). Therefore, P3 stands for the annual average soil erosion without considering vegetation dynamics during 2000-2015. The total variation of soil erosion between P1 and P2 is represented by ΔA, the variation in soil erosion due to vegetation dynamics is represented by ΔA veg , and the contribution of vegetation dynamics to soil erosion control on the LP can be obtained by ΔA veg /ΔA.

The Performance Evaluation of the RUSLE Model
Based on the Pearson correlation coefficient, the observed annual sediment load data from Tangnaihan, Toudaoguai, Longmen and Tonguan gauging stations were taken to evaluate the performance of RUSLE simulated results in our study, as follows: where n is the observation times of sediment load from gauging stations; S j and O j are the simulated and the observed sediment load, respectively; and are the mean value of simulated and observed sediment load, respectively; ρ M,O ∈[−1,1]. The more ρ M,O closes to 1, the better the simulation performance of the model.

The Net Changes in Leaf Areas
The estimation of net changes in leaf area from 1995 to 2015 was based on the trends of annual LAI. In our study, the trends of annual LAI were suggested to be linear (Chen et al., 2019). The net changes of leaf area at grid scale were divided into vegetation greening and vegetation browning. The area of each pattern could be obtained by counting the area of the grid, as shown in follows: where k stands for a grid from the data of the trends of annual LAI, t represents the grids number of the LP, T k is the trend for the grid of k, G k is the area for the grid of k, and N yr is the study period duration (21 was set in this study).

Revegetation Strikingly Alleviates Soil Erosion on the LP
Vegetation greening and browning are defined as the increase and the decrease of annual LAI trends at grid scale from 1995 to 2015, respectively. According to the LAI data, 78% of the LP shows greening and 9.8% is browning (Figure 2a and Figure S1 in Supporting Information S1). Vegetation greening of the LP is equivalent to an increase of 1.2 × 10 5 km 2 of new leaf area over the 21-year (Table S2 in Supporting Information S2). The greening is found mainly in the area along the river as well as cropland, grassland, and forest land. The annual average LAI over the LP increased from 0.511 m 2 m −2 in 1995 to 0.686 m 2 m −2 in 2015, with an increased rate of 0.008 m 2 m −2 yr −1 . Vegetation coverage of the LP was greatly improved, and greening was dominant on the LP from 1995 to 2015.
Soil erosion on the LP was obviously alleviated at different degrees during 1995-2015. The soil erosion estimated results (validated against the observed sediment load data, Figures S4a and S4b in Supporting Information S1) show that the trends of soil erosion rates on the LP during 1995-2015 ( Figure 3a) range from −63.61 t km −2 yr −2 to 33.93 t km −2 yr −2 . Generally, soil erosion mitigation (the decreased trend in soil erosion rate) accounted for 78.14% of this area (Figure 3b), which mainly focused on forest land and grassland along rivers in the east and southwest of the LP. Soil erosion intensification occurred in 14.17% of the study area, mainly in hilly and mountain areas located in the east of the LP. Likewise, 7.69% of the study area showed no obvious change. Moreover, the trends of soil erosion rate between 0 t km −2 yr −2 and 10 t km −2 yr −2 (including 10 t km −2 yr −2 ) covered 14.09% area of the LP, while its values between −10 t km −2 yr −2 and 0 t km −2 yr −2 (including −10 t km −2 yr −2 ) occupied almost 77.25% area of the LP.
A marked alleviation had been achieved in the soil erosion of the LP from 1995 to 2015. We use P1 (soil erosion rates: based on 1995-1999) and P2 (soil erosion rates: based on 2000-2015) to stand for two soil erosion scenarios: before and after vegetation restoration, respectively. Spatially, the soil erosion rate in P1 (452.23 t km −2 yr −1 on average) was obviously higher than that in P2 (297.23 t km −2 yr −1 on average). Compared to P1, there were  significant declines in soil erosion in P2, mainly located in the hilly and gully areas and loess tableland areas of the eastern and southeastern LP (Figure 4a and 4b). To further reveal the contribution of revegetation to the soil erosion variation of the LP, we use P3 (Figure 4c) to stand for the soil erosion scenario lacking large-scale revegetation, by evaluating the average annual soil erosion rates during 2000-2015 with the absence of vegetation dynamics. We find that after the large-scale revegetation, the soil loss decreases by 32.38% (−1.24 × 10 8 t, P2−P1) compared with P1 (3.83 × 10 8 t). Meanwhile, revegetation contributed 42% (−0.52 × 10 8 t, (P2−P3)/ (P2−P1)) to soil erosion amount changes (Figure 4d and 4e, Table S1 in Supporting Information S2).

Cropland Dominates the Greening of the LP
According to the statistics of net changes in leaf areas, it is interesting to find that the cropland was the main driving factor of the vegetation greening on the LP ( Figure 5 and Table S2 in Supporting Information S2). From 1995 to 2015, the new leaf areas of the LP increased by 1.23 × 10 5 km 2 over the 21-year period, which is equivalent to more than 19% of the LP. All that matters is that cropland contributed the most (0.48 × 10 5 km 2 ) to the increased new leaf areas, followed by grassland (0.45 × 10 5 km 2 ). In other words, 75.6% of the increased new leaf areas (the increased 1.23 × 10 5 km 2 of leaf areas mainly came from the 78% area of the LP) in the LP were from cropland and grassland in about equal measure (39.3% and 36.3%, respectively) ( Figure S5 in Supporting Information S1), but cropland was prominently dominant. Forest land and other vegetated land contributed about 0.18 × 10 5 km 2 and 0.12 × 10 5 km 2 of new leaf areas respectively, equivalent to 14.3% and 10.1% of the increased new leaf areas in the LP. To sum up, vegetation greening over the LP mainly came from the cropland and grassland. Compared with vegetation greening area, vegetation browning was just 0.03 × 10 5 km 2 and equivalent to 0.44% of the LP.
In fact, in 1995-2000, 2000-2005, 2005-2010, and 2010-2015 (Figure 5b), forest land showed a net increase (increased by 4.12 × 10 3 km 2 ), while cropland area appeared to have a net decrease (decreased by 7.13 × 10 3 km 2 ). Meanwhile, although the reduction took place in grassland during 2000-2005 and 2010-2015, its area increased by 3.05 × 10 3 km 2 from 1995 to 2015. Figure 5d shows the trends of LAI for cropland, forest land, and grassland in the LP during 1995-2015. However, an annual average trend of LAI was highest in cropland (growth rate: 0.0118 m 2 m −2 yr −1 ) and lowest in grassland (growth rate: 0.0082 m 2 m −2 yr −1 ). The annual average trend of LAI in forest land is only slightly higher than that in grassland, with a growth rate of 0.0088 m 2 m −2 yr −1 . Overall, cropland contributed the most to vegetation greening based on its decreased land-use area.

The Causes of Cropland Greening
The intensity of cropland utilization in the LP obviously improved from the1990s to 2015. The cropping rotation data, standing for the cropping frequency of cropland, were used to quantify the intensity of cropland utilization. Spatially, the area with increased cropping frequency accounted for 30% of the LP, while the area with decreased cropping frequency just accounted for 14% of the LP (Figure 6a). In addition, the cropping frequency variation of the LP in the 1990s, 2000s, and 2015 indicates that the high-intensity (double cropping per year) and med-intensity  (triple cropping for 2 years) area increased by 43% and 37%, respectively ( Figure 6b). The improved utilization intensity explains why cropland contributed the most to vegetation greening with decreased land-use area.
To figure out the effect of the improved intensity of cropland utilization, we selected the typical agricultural provinces (Shanxi, Shaanxi, and Gansu) in the LP to analyze their crop production area, and grain yield (Figure 7). From 1995 to 2015, the crop production area of Shanxi and Gansu decreased by 18.42% and 2.73% respectively, while the grain yield of Shanxi, and Gansu increased by 33.33% and 85.71% respectively. Thus, the two provinces show an obvious upward trend in yield per 1 km 2 at a growth rate of 718 t yr −1 and 793 t yr −1 . In addition, Shannxi still achieved an incremental trend in yield per 1 km 2 with a growth rate of 542 t yr −1 , even though its crop production area increased by 4.13%, due to its growing grain yield (increased by 37.36%). In general, the improvement of cropland utilization not only contributes greatly to vegetation greening but also significantly promotes grain yield.

The Drivers of Soil Erosion Control on the LP
The LP is vulnerable to soil erosion due to bare surface, steep topography, and concentrated rainfalls. It is the most sediment-rich region in the world, making the YR the largest sediment carrier in the world (Zhang & Liu, 2005). However, since 1995, the annual sediment transport of LP has shown an obvious downward trend. In addition, it was found that sediment transport of the LP decreased, especially after GfGP (Sun et al., 2014;Wang et al., 2015).
Vegetation cover improvement and terrace farming are the major anthropogenic factors to the LP soil erosion control. Owing to the implementation of a series of vegetation rehabilitation projects, especially the GfGP since 1999, vegetation on the LP improved obviously (Chen et al., 2015;Fu et al., 2011). The improved vegetation covers lowered loess erodibility on the LP by intercepting raindrops, reducing the velocity of raindrops, slowing overland water flow, and decreasing water discharge (Mohammad & Adam, 2010;Wang et al., 2015;Wei et al., 2010). Moreover, terrace farming has the dual functions of agricultural production and ecological restoration, and can reduce the influence of slope and valley runoff and sediment yield, so it has become an important means to reduce the sediment flux of rivers in low altitude areas on the LP Wang et al., 2015). Erosive precipitation is a significant dynamic factor of precipitation-runoff erosivity, and soil erosion is closely related to it (Wischmeier & Smith, 1978). Several studies have indicated that the erosive precipitation and erosive power of precipitation on the LP have decreased in recent 50 years (Liu et al., 2018;Wang et al., 2015;Xin et al., 2010). The decline in erosive precipitation covers most areas of the LP, with a consequent reduction in sediment production.

The Drivers of Vegetation Greening on the LP
Vegetation improvement has been regarded as an effective measure to minimize soil erosion . LAI data attest to an obvious increase in vegetation greening on the LP during 1995-2015 (increase rate: 0.008 m 2 m −2 yr −1 ), and the area of a new leaf was increased by 1.23 × 10 5 km 2 . Several previous studies confirm this vegetation greening trend Zhou et al., 2019). Overall, the above results can be attributed to large-scale vegetation restoration projects, intensified agriculture, and climate factor (Kou et al., 2021;Schwarzel et al., 2020).
The Chinese government strongly supports large-scale vegetation restoration projects represented by the GfGP to control soil erosion and to improve the ecological environment. Those projects promote afforestation, reforestation, and grass regeneration by large-scale land-use change. Therefore, during 1995-2015, the area of forest land and grassland in the LP increased by 4.12 × 10 3 km 2 and 3.05 × 10 3 km 2 , respectively. However, we noted that since 1995, cropland has made the largest contribution to the increase of new leaf. Apparently, cropland proves a greater role in vegetation greening than previously thought (Chen et al., 2019). It is hard to ignore the dominant driver cropland plays in vegetation greening on the LP. In addition, the key climatic factors that determine vegetation growth, such as temperature, precipitation and CO 2 concentration, all promoted vegetation growth to varying degrees during the same period (Lan et al., 2021;Li et al., 2016). Meanwhile, climate warming as well as CO 2 fertilization effects can lengthen the growing season and increase the maximum rate of productivity, particularly in the water-limited areas (Donohue et al., 2013;McMurtrie et al., 2008;Shen et al., 2015). Therefore, the critical factors to achieve large-scale greening on the LP can be attributed to human-induced land use and climate change.

Cropland Brings Well-Being to Society and the Environment
Improving vegetation coverage can effectively control soil erosion on the LP, which is seen as the fundamental purpose of the GfGP. This project successfully returned erosion levels to historic values and made sediment discharge from the LP into the YR decline to AD 740 (Chen et al., 2015). Evidently, the role that forestland and grassland played in the GfGP has been focused on and recognized by more studies. But it should be noticed that compared to forestland and grassland, cropland made the greatest contribution to the vegetation greening on the LP during 1995-2015. With its limited area, cropland performs a prominent role in improving vegetation coverage, which cannot be neglected in the objective evaluation of ecological effects in the GfGP.
In addition, economic development and ecological conservation are both important goals that the Chinese government is committed to advancing. Therefore, as the Chinese government pushes for sustainable and stable urbanization, the reform centered on improving the environment and promoting development is increasing. The GfGP is a typical example. However, it has been noticed that the accelerated expansion of cities, towns, and villages to various degrees, has come at the cost of cropland (Kou et al., 2021;Li et al., 2022). In particular, after the implementation of GfGP, forests and grasslands have recovered somewhat, but at the expense of arable land. As a result, the area of cropland has been greatly reduced, and even the local food supply is insufficient and even extends to the surrounding areas. However, for China with its population ranked at the top in the world, sufficient cropland has always been the most effective guarantee for food self-sufficiency and security. Therefore, the concerns about the availability of cropland and food supply are standing out (Chen et al., 2015). Fortunately, this food supply problem due to decreased cropland can be alleviated by multiple cropping, fertilizer use, farm mechanization, improved irrigation system, quick-growing hybrid cultivars, farm intensification, crop insurance programmes, and some climate factors (Chen et al., 2019;Keenan et al., 2016;Kou et al., 2021). In other words, food production deficits as a result of cropland redaction can be compensated by the enhancement of yield per unit area. The above results emphasize that cropland is beneficial in both improving environmental performance and society's sustainable development.

Conclusions
The annual sediment load of the Yellow River (YR) reached its historical peak, due to extensive vegetation destruction caused by intensive cropland reclamation activities on the Chinese Loess Plateau (LP). In order to restore vegetation and further control soil erosion, the largest ecological restoration project (the GfGP) was promoted across the LP since 1999. This project mainly took the measure of returning cropland to forestland and grassland, for the sake of improving vegetation and further controlling soil erosion on the LP.
Satellite information shows that, vegetation greening occurred in 78% of the LP area during 1995-2015, and the leaf area increased by 1.23 × 10 5 km 2 over the 21 years. In addition, 78.14% of the LP area showed a decreasing trend in soil erosion rates. The above results demonstrate that the GfGP has been successful in improving vegetation coverage and controlling soil erosion.
It is worth noting that from 1995 to 2015, although the area of cropland decreased gradually, the contribution of cultivated land to vegetation greening was the largest. This was mainly due to the improvement of cropland utilization intensity, which was promoted by multiple cropping, fertilizer use, irrigation system improvement and so on. Nevertheless, the effect of cropland on ecological restoration has been downplayed. Moreover, cropland also shoulders the responsibilities of local food supply and food security. Based on its significant impacts on environment protection and society development, the positive effects of cropland should be comprehensively considered in the process of promoting the GfGP.