Special Issue: Human-environment interactions and Ecosystems

Dynamics and interactions of water-related ecosystem services in the Yellow River Basin, China

  • WANG Peng , 1, 2, 3 ,
  • XU Mingxiang , 1, 2, 3, 4, *
  • 1. The Research Center of Soil and Water Conservation and Ecological Environment, CAS and Ministry of Education, Yangling 712100, Shaanxi, China
  • 2. Institute of Soil and Water Conservation, CAS and Ministry of Water Resources, Yangling 712100, Shaanxi, China
  • 3. University of Chinese Academy of Sciences, Beijing 100049, China
  • 4. Institute of Soil and Water Conservation, Northwest A&F University, Yangling 712100, Shaanxi, China
* Xu Mingxiang, Professor, specialized in ecosystem services. E-mail:

Wang Peng, PhD Candidate, specialized in ecosystem services. E-mail:

Received date: 2022-11-18

  Accepted date: 2023-04-11

  Online published: 2023-08-29

Supported by

The National Key Research and Development Program of China(2022YFF1300802)

National Natural Science Foundation of China(42130717)

“Light of the West” Cross Team-Key Laboratory Cooperative Research Project(A314021402-1912)


Climate change and human activities have profoundly altered ecosystem services in the Yellow River Basin (YRB) since the Grain for Green project was implemented, but have not been accurately revealed on a year-by-year scale. This study combined the InVEST model to reveal the year-by-year changes in the water-related ecosystem services (WRESs) in YRB during 1990-2020, including water yield, soil conservation and water purification services. The trade-off/synergy of WRESs and impacts of land management measures on WRESs were assessed fully. The results showed that from 1990 to 2020, cropland and barren land were considerably converted to forest and grassland in YRB. WRESs were continuously improved as a result of increase of water yield and reductions of soil export and nitrogen export, at rates of +1.11 mm·yr-1, -0.23 t·km-2·yr-1 and -1.01 kg·km-2·yr-1, respectively. We found that in YRB water purification service showed trade-off relationships with soil conservation and water yield services in recent decades, and water yield and soil conservation maintained a synergitic effect. Additionally, the revegetation measures showed a potential of enhancing soil conservation and water purification, but reducing water yield. This study provided a thorough understanding of WRESs dynamics and a valuable reference for the ecological restoration practices.

Cite this article

WANG Peng , XU Mingxiang . Dynamics and interactions of water-related ecosystem services in the Yellow River Basin, China[J]. Journal of Geographical Sciences, 2023 , 33(8) : 1681 -1701 . DOI: 10.1007/s11442-023-2148-0

1 Introduction

Ecosystem services (ESs) are defined as the various benefits that humans acquire directly or indirectly from natural ecosystems, bridging natural systems and human society. The Millennium Ecosystem Assessment of the United Nations reported that an estimated 60% of global ESs were in a state of continuous degradation (MA, 2005). The factors of ES degradation are very complex and are generally attributed to the impacts of climatic change, underlying surface conditions, and regional human activities, sometimes including trade-offs and synergies among ESs themselves (Karimi et al., 2021). For example, degradation of habitat quality services mostly indicates a decrease in biodiversity in an ecosystem, which will weaken the pollination role with decreases in insects and further reduce food production services, thus creating a synergy between habitat quality services and food production services; expansion of cropland through deforestation will increase food production but could cause a trade-off effect between habitat quality services and food production services. Increases in food supply services often imply reductions in water yield services and water purification services due to crop water consumption and fertilizer application, thus producing a trade-off relationship (Liang et al., 2021). Moreover, revegetation in an ecosystem will increase carbon storage and improve soil conservation capacity but accordingly will weaken water yield due to high evapotranspiration of vegetation (Gao et al., 2017). Due to the complex relationships among ESs, irrational ecological management practices may result in unexpected ES degradation. Therefore, clarifying the interactive relationships among multiple ESs and their influencing elements assists in achieving ES optimization and management for the sustainable development of ecosystems.
Currently, ES assessments commonly adopt field investigation methods and ecological models. Typical models include the ARIES model, EcoAIM model, SWAT model and InVEST model. Among them, the InVEST model is characterized by requiring few input data, having rigorous physics principles and performing rapid calculations (Sharp et al., 2016) and has been widely applied (Gao et al., 2017; Redhead et al., 2018; Pei et al., 2022). It can quickly achieve a reliable ES assessment at multiple spatial and temporal scales (regional and even national scales) (Redhead et al., 2018) and is capable of clearly quantifying the impacts of driving factors on ESs (Zhu et al., 2020). For example, the InVEST-based scenario analysis approach can effectively distinguish the relative contributions of land use and climate change to ESs (Bai et al., 2019; Gao et al., 2022). Additionally, future changes in ESs can be achieved by the InVEST model but require a combination with land prediction models such as the CA-Markov and CLUE-S models (Gashaw et al., 2018; Peng et al., 2020). Notably, model-based methods cannot fully quantify the impact of specific land use changes on ESs. The future land scenario by land prediction models produced all kinds of land type changes through spatially influenced configurations, which is incapable of explaining the impacts of individual land use change on ESs, while methods that change only a single land type actually make more sense in revealing the impact of human management (e.g., afforestation and cropland reclamation) on ESs.
The Yellow River Basin (YRB) is located in a special zone of semiarid transition to arid zones in northern China. It is a large base for energy, agriculture and animal husbandry outputs and the most important ecological barrier in the central-eastern part of China (Jiang et al., 2015). The national Grain for Green project was first implemented there. In the YRB, dry climate conditions are a major threat to the stability of local ecosystems, and areas with less than 400 mm of annual precipitation are distributed in most regions. Historically, the YRB has experienced numerous flow breaks due to low precipitation and interception of runoff by upstream dams, which resulted in widespread vegetation degradation, triggered severe soil erosion and sandstorms (Wang et al., 2006; 2016). These factors have severely constrained ecological recovery and socioeconomic development. The major ecological problems in the YRB include the shrinkage of vegetation and frozen soils, degradation of the water conservation function upstream, severe soil erosion and groundwater reduction midstream, and water pollution caused by excessive agricultural fertilization and sediment accumulation in the lower reaches. These ecological problems are closely linked to water-related ecosystem services (WRESs), i.e., water yield, soil conservation, and water purification services. Previous studies on WRESs in the YRB and at regional scales considered only a few fixed years to represent changes over decades (Bai et al., 2019; Liang et al., 2021; Yang et al., 2021), e.g., the amount of water yield in 2020 minus water yield in 2000 was regarded as the change from 2000 to 2020, which does not reveal the full dynamic of ES changes. Only continuous time-series changes in one variable can provide a more accurate understanding of variable dynamics and be more effective in exploring its driving mechanism (Pei et al., 2022; Wei et al., 2022). Currently, a detailed, complete change in the WRES time series in the YRB has not been studied at a yearly scale, greatly limiting the understanding of ecological management and ecosystem service science.
Considering a series of ecological problems in YRB, the Grain for Green project (1999-present) was first launched in YRB, whose aims are to restore degraded ecosystems and enhance ESs by increasing vegetation coverage. Chinese scholars Ouyang et al. (2016) have demonstrated the positive role of the Grain for Green project on the national ESs. Evidence shows that vegetation coverage in YRB has been increasing over the past 20 years, and anthropogenic afforestation practices have produced a significant role (Jiang et al., 2015). However, evidence also indicated ecological restoration mainly by afforestation seems to have deviated from its purpose, due to the adverse effect of excessive afforestation on YRB ecological recovery (Cao et al., 2011). Specifically, overextended vegetation largely enhances evapotranspiration processes and therefore consumes large amounts of water (Wang et al., 2019). For instance, the current vegetation restoration in China has increased evapotranspiration by 0.25 mm·yr-1 and decreased water yield by 0.16 mm·yr-1 in the whole country (Bai et al., 2020). The decrease of water yield on the Loess Plateau (located in YRB) even reached 50 mm annually caused by vegetation restoration (Sun et al., 2006); surface water storage decreased in the Mu Us Sandy Land (located in YRB) of due to intense anthropogenic tree planting (Zhao et al., 2021); plans for further vegetation expansion has been suggested to stop (Chen et al., 2015). In summary, the hydrological effects of ecological restoration projects and the impacts on aquatic ecosystems are complex and have not been well revealed, and a reliable assessment is needed to improve the understanding of ecosystem changes in YRB, which helps provide a scientific basis for future ES restoration.
Based on the above, this study combined the InVEST model and the scenario designing-based approach to assess the spatiotemporal changes and trade-off relationships of WRESs in YRB in 1990-2020, and to predict the impacts of future ecological restoration on WRESs. The objectives of this study are: (1) to reveal changes of the WRES and land cover in YRB before and after the implementation of the Grain for Green project; (2) to clarify the trade-offs/synergies relationships of WRESs; and (3) to estimate changes of WRESs under different land management measures.

2 Data and methods

2.1 Study area

The YRB is located at the northern Bayankara Mountain on the Qinghai-Tibet Plateau, where the Yellow River flows through provincial-level regions of Qinghai, Sichuan, Gansu, Ningxia, Inner Mongolia, Shaanxi, Shanxi, Henan, and Shandong, covering a total area of 79.5×104 km2. The climate type of the YRB is featured by arid, semi-arid and semi-humid climate. The upstream of YRB covers the region above Hekou Town in Gansu with an area of 42.8×104 km2 and an elevation above 3000 m. Regions above Longyangxia Reservoir is the headwater area of the Yellow River, with a total area of about 13.2×104 km2 and an elevation above 4000 m. The midstream of YRB are from Hekou Town to Taohuayu in Henan province with an area of 34.4×104 km2 and the elevation is generally 1000-2000 m; areas below Taohuayu are the downstream of YRB, with an area of 2.3×104 km2 and the elevation of less 100 m, where the terrain is low and flat. Over the past 20 years, vegetation coverage was increasing mainly in the midstream of YRB due to large-scale tree planting (Figure 1). Currently, the water issue is the biggest limiting factor of the “Ecological Protection and High-quality Development of the YRB”. YRB used only 2% of the total national water resource to be responsible for 13% of the country’s food production, 15% of the arable land area, more than 50 large cities of water supply (Ma et al., 2020). The national half of coal base and 70% of coal power base are located in YRB, and huge energy production would consume considerable water resources and cause industrial pollution. In addition, the soil losses and agricultural non-point source pollution in YRB are also very serious. The WRESs are highly relevant to these issues in YRB.
Figure 1 Location of the Yellow River Basin

2.2 Data source and processing

The 30-m resolution land cover data from 1990 to 2020 were from Yang and Huang (2022), and land cover types include 9 categories: cropland, forest, shrub, grassland, water, snow, barren land, impervious land, wetland; 1-km elevation data were obtained from the Resource and Environment Science Data Center of Chinese Academy of Sciences (https://www.resdc.cn/); 1-km Normalized Difference Vegetation Index (NDVI) product 1998-2019 can be downloaded in the website: http://www.vgt.vito.be. Yearly NDVI was calculated using the maximum value of NDVI datasets in a year; 1-km precipitation and reference crop evapotranspiration datasets were obtained from the National Earth System Science Data Center (http://www.geodata.cn/). The annual observed runoff data during 1999-2020 were from the Water Resources Bulletin (http://www.yrcc.gov.cn/other/hhgb/) in the study area; 1-km soil data were from the Harmonized World Soil Database product (https://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/harmonized-world-soil-database-v12/en/).

2.3 Soil conservation service

The soil retention module in the InVEST model is based on the universal soil loss equation (Wischmeier et al., 1971), which consists of two components: soil erosion reduction and sediment retention. The former demoted the difference between the potential soil erosion and the actual soil erosion, and the latter indicated the retention of the incoming sand from upstream, expressed as the product of sediment volume and sediment retention rate (Sharp et al., 2016). In this study, we used soil export to describe the soil conservation service, i.e., the more soil export, the lower soil conservation service, vice versa. The model formula is as follows:
$USL{{E}_{i}}={{R}_{i}}\times {{K}_{i}}\times L{{S}_{i}}\times {{C}_{i}}\times {{P}_{i}}$
where USLEi denotes the actual soil erosion in raster i; Rx denotes the annual rainfall erosivity factor in raster i, Ki denotes the soil erodibility factor in raster i, LSi is the slope-length factor in raster i, Ci denotes vegetation management factor in raster i, and Pi is the anthropogenic management factor in raster i.

2.4 Water purification service

The water purification module of the InVEST model is based on the principle that vegetation and soil can reduce or remove nutrients by storing and converting them and prevent nutrients into rivers, to achieve water purification service (Sharp et al., 2016). Since the calculation of nitrogen and phosphorus in the InVEST model adopted the same principle, and moreover the response of nitrogen to landscape pattern is more sensitive to land use change. Hence, in this study, the nitrogen export was used to measure level of water purification service. i.e., the more nitrogen export, the lower water purification service, vice versa. The main calculation equation is as follows:
$AL{{V}_{i}}=HS{{S}_{i}}\times po{{l}_{i}}$
where ALVx denotes the adjusted loads in raster i; HSSx denotes the hydrologic sensitivity in raster i, and polx denotes the export coefficient of in raster i.

2.5 Water yield service

The InVEST water yield module defines water yield as the amount of water remaining after subtracting actual evapotranspiration from precipitation based on Budyko hydrothermal coupling hypothesis (Budyko, 1974), and it denotes a total amount of water volume including surface and subsurface runoffs. The water yield module parameters consist of vegetation type, evapotranspiration coefficient and soil depth. The main algorithms of the model are as follows:
${{Y}_{mn}}=\left( 1-\frac{AE{{T}_{mn}}}{{{P}_{m}}} \right)\times {{P}_{m}}$
where Ymn is the annual water yield in raster m and land cover type n; Pm is the annual precipitation of raster m; AETmn is the annual evapotranspiration in raster m and land cover type n.
$\frac{AE{{T}_{mn}}}{{{P}_{m}}}=\frac{1+{{\text{ }\!\!\omega\!\!\text{ }}_{m}}{{U}_{mn}}}{1+{{\text{ }\!\!\omega\!\!\text{ }}_{m}}{{U}_{mn}}+1/{{U}_{mn}}}$
where Umn is the dryness index of grid m on land cover type n, dimensionless, indicating the ratio of potential evapotranspiration to precipitation, calculated by equation (5); ωm is the ratio of annual available water to precipitation for modified vegetation, dimensionless
${{U}_{mn}}=\frac{K{{c}_{n}}\times E{{T}_{0,m}}}{{{P}_{m}}}$
where Kcn is evapotranspiration coefficient of land cover type n, determined by land cover types; ET0,m annual reference crop evapotranspiration in raster m.
${{\text{ }\!\!\omega\!\!\text{ }}_{m}}=Z\cdot \frac{AW{{C}_{m}}}{{{P}_{m}}}$
where Z is the Zhang coefficient, a constant relevant to the seasonal characteristics of precipitation (Zhang et al., 2001). We compared the modeled water yield and observed water yield, finding that the two have high consistency while Z=7.7; AWCm is the effective available water for vegetation.

2.6 Linear trend analysis

The change of a variable time series can be expressed by a one-dimensional linear regression equation. This study used the linear trend to showed the change of WRESs in YRB during 1990-2020 at a raster scale. A positive trend value indicates an increase of variable and a negative value indicate decrease of variable. The equation is the following:
$\theta=\frac{n \sum_{i=1}^n i y_i-\sum_{i=1}^n i \sum_{i=1}^n y_i}{n \sum_{i=1}^n i^2-\sum_{i=1}^n i^2}$
where θ is the linear trend of inter-annual WRESs; n is the number of years in the study time period; yi is the WRESs value in the i-th year. The significance level of the linear trend was judged at the p<0.05.

2.7 Trade-off and synergy analysis

Correlation analysis is an effective method for determining ESs trade-off and synergistic relationship (Gong et al., 2020). This study selected the correlation analysis to judge trade-off and synergy among WRESs. That the correlation coefficient between WRESs more/less than 0 represents synergy/trade-off. The calculation of correlation coefficients is as follows:
where Rxy is the correlation coefficient of variable x and variable y with values ranging from –1 to 1; k is the length of time series; xv and yv are the values of variable x and variable y in raster v, respectively; $\bar{x}$ and $\bar{y}$ are the raster average values of variable x and y, respectively.

2.8 Spatial hotpot and cluster analyses

The Getis-Ord G* statistic is an indicator to analyze the degree of spatial aggregation (Sandeep and Ranjan, 2022). When the Getis-Ord G* statistic is significantly positive/negative, the high values of WRESs are considered to be clustered in space, i.e., the hot/cold spot region; Getis-Ord G* statistic is calculated as follows:
$G_i^*=\frac{\sum_{j=1}^n W_{i j} X_i-\bar{X} \sum_{j=2}^n X_j}{\sqrt{\frac{\left[n \sum_{j=1}^n W_{i j}^2-\left(\sum_{j=1}^n W_{i j}\right)^2\right]}{n-1}}}$
where Gi* is the Getis-Ord G* statistic, when Gi* is greater than 1.96, the region represents a hot spot, i.e. the higher WRESs in the region shows aggregation in space, when Gi* is less than -1.96, the region represents a cold spot, i.e. the lower WRESs in the region shows aggregation in space; mathematical expectation Xj is the WRESs value of the j-th unit; Wij is the spatial weight matrix component in the i-th unit and j-th unit, respectively; the adjacent denotes 1 in Wij and non-adjacent denotes 0 in Wij; n is the number of total units; S is the standard deviation of the sample; Xi and Xj denote the value of WRESs in the i-th unit and j-th unit, respectively.
The local spatial autocorrelation LISA index can be used for describing the degree of local spatial aggregation of the extreme values of spatial variables, which consist of four cluster types: low-low cluster, high-high cluster, low-high cluster (low-value region is wrapped high-value region), and high-low cluster (high-value region is wrapped low-value region). Its local spatial autocorrelation features can be effectively identified as the following calculation:
where LISAv denotes the local spatial auto-correlation index in unit v; fvz is the spatial weight matrix in unit v and unit z, respectively; Bv and Bz are the spatial attribute values of variable B in adjacent paired cells; h is the number of total units.

2.9 Impacts of land management measures on WRESs

We selected the land cover in 2020 as base scenario, based on which the three land management scenarios were established by referring to the land management policies in YRB and historical change of YRB land ecosystem. These scenarios include the river protection (RP) scenario, green development (GD) scenario and a food supply (FS) scenario.
(1) In the base scenario, land cover data in 2020 and 31-year average climate data as InVEST model inputs. Our purpose is to assess the impacts of different land management measures on WRESs, and therefore a basic scenario needs to be used as a reference to reflect change of WRESs in different scenarios relative to that in the base scenario. As WRESs are also affected by climate factor, we therefore did not directly use the climate data in 2020 because it may be an unusual climate condition. Therefore, the average of multi-year climate data was used in climate input of every scenario to avoid overestimating or underestimating the impact of these scenarios.
(2) For GD scenario, afforestation is a long-term ecological practice that YRB has been continuously starting and will continue in the future. It focuses on establishing forest on land with high slopes to reduce soil erosion. We have converted all cultivated land with a slope of >15° into forest land to reduce soil erosion and improve ecological stability. Finally, the area of forest increased by 20,868 km2.
(3) For RP scenario, the Yellow River has experienced many flow breaks, and the application of agricultural fertilizers has increased river pollutants. The planting of protective forests around the rivers can effectively improve the water purification and water conservation functions of ecosystems. 1-km grid around the water body grid in the whole YRB was adjusted to the forest grid, thus leading to an increase of 30,672 km2 of forest.
(4) For FS scenario, with the increasing population in YRB, the food demands will grow quickly, which also requires more cropland to support food production. Lands with low slopes are less prone to soil erosion, but are suitable for plantation of crop. Consequently, barren land with slopes of <6° were converted into cropland to increase the cultivated area for higher food supply. Ultimately, the area of cropland increased by 22,805 km2.

3 Results

3.1 Changes in land cover transfer

The land cover composition in YRB is dominated by grassland and cropland (Table 1). The cropland showed a consistent decreasing trend from 1990 to 2020, with the largest decrease of about 14,000 km2 between 2000 and 2010. The forest has increased by 17,048 km2 in the last 30 years, mainly between 2010 and 2020, accompanied by an increase of 7513 km2. The grassland increased in 1990-2010 and decreased in 2010-2020, with an increase of 7589 km2 from 1990-2020. Barren land has been showing a reducing trend, with a reduction of 4665 km2 in 1990-2000, 8213 km2 in 2000-2010 and 317 km2 in 2010-2020, respectively. Impervious area increased from 8797 km2 in 1990 to 22,071 km2 in 2020, denoting an increase of 151% of urbanization expansion. Other land cover types are few with insignificant changes. Overall, the land cover change in YRB is characterized by increases in vegetation and impervious surface, and significant decreases in cropland and barren land.
Table 1 Areas of land cover from 1990 to 2020 (km2)
Types 1990 2000 2010 2020
Cropland 207,040 202,112 188,083 183,597
Forest 75,197 79,173 84,732 92,245
Shrub 5500 5235 4394 3811
Grassland 456,039 458,927 470,648 463,628
Water 5030 4600 5537 5880
Snow 264 289 558 220
Barren land 37,143 32,478 24,265 23,948
Impervious land 8797 12,608 17,486 22,071
Wetland 916 504 223 526
In terms of land cover (Figure 2), the cropland is mainly concentrated at the end of the upstream of YRB and at the junction of the middle and lower reaches, and the forest is mainly distributed around the midstream cropland. The grassland is the main land cover type, spreading over almost the entire study area. The barren land is mainly observed in the northern part of YRB, around the Yellow River course. The impervious surface is almost all in the cultivated area. In terms of spatial distribution, the increase in impervious area is near the region of cropland, indicating that rapid urbanization is all derived from the occupation for cropland. The increase of grassland is very high, mainly in the midstream of YRB. Areas with considerable increase of cropland is observed in the latter section of the upstream headwaters. Overall, the land cover transfer in YRB is mainly between grassland, impervious area, forest and cropland. In the upstream headwater of YRB, land cover changes are not evident due to high elevation and few human activities there.
Figure 2 Land cover change in the Yellow River Basin from 1990 to 2020

3.2 Inter-annual changes in precipitation and WRESs

Figure 3 shows a comparison between the modeled and observed water yield, and the results showed a strong correlation between them with a high correlation coefficient of 0.83, validating that the modeled water yield is highly accurate. Figure 4 shows the precipitation, soil conservation service (in terms of soil export), water purification service (in terms of nitrogen export) and water yield service during 1990-2020. The precipitation at YRB increased at a rate of +2.52 mm·yr-1 from 1990 to 2020, with minimum (437 mm) and maximum (588 mm) values occurring in 1997 and 2003, respectively. The water yield service of YRB is maintaining an increasing trend of +1.11 mm·yr-1, with the lower amount of about 40 mm in 1991 and 1997 and the highest yield (129 mm) in 2003. In addition, the soil conservation service in YRB is at a moderate increasing trend in the last 30 years as a result of a decreasing rate of -0.23 t·km-2·yr-1 in soil export. Between 2002 and 2012, the soil conservation service was enhancing due to the decrease of soil export, and soil conservation service in the post-2002 period is better than that in the pre-2002. The water purification service is also enhancing overall, due to a significant decrease in nitrogen export at the rate of -1.01 kg·km-2·yr-1, with the highest and lowest nitrogen export occurring in 1990 and 2019, respectively. Overall, the soil conservation, water purification and water yield services have been increasing entirely in YRB over the last 30 years, and the enhancement of WRESs has been mainly after 2002, which was attributed to positive land management and climate change.
Figure 3 Comparison of modeled water yield and observed water yield from 1999 to 2020
Figure 4 Precipitation, water yield, soil export and nitrogen export from 1990 to 2020
Figure 5 shows the multi-year average and linear trends for the three WRESs. Areas with high water yield are mainly in the headwater area of YRB upstream, where high elevation and low temperature result in low evapotranspiration and very high water yield service. The significant increase of water yield service is also mainly distributed in the headwater area of YRB upstream, where the increase trend of water yield is significant (p<0.05). There is a slight decrease in water yield service in the latter part of the upstream and downstream areas, while there is little significant change in water yield service in the midstream YRB. The areas with low water purification service are mainly distributed near the cropland area in the midstream and downstream, where the large amount of fertilizer application significantly increased the amount of nutrients and produced high nitrogen export into rivers; the areas with high water purification service are observed in and around the forest area in the latter section of the upstream, where the nitrogen export was very low. The areas with weakened water quality purification service are mainly in the headwater area of upstream and near the midstream of the Yellow River, where the nitrogen export maintained an increasing trend. The higher soil export in the headwater area of upstream implies a low soil conservation service there, and the soil export output is also low in the cropland area of the midstream, where the terrain is flat and not prone to suffer soil erosion. The trend of increasing soil export is mainly in the headwater area of upstream, implying a decrease of soil conservation service, and decreasing soil export is primarily distributed in the midstream, denoting an increase in soil conservation service there.
Figure 5 Annual average water yield (a) and its trend (b), nitrogen export (c) and its trend (d), and soil export (e), and its trend (f)

3.3 Hot/cold spot and spatial cluster of WRESs

Figure 6 shows the hot/cold spots and spatial cluster characteristics of the three WRESs. The extremely significant hot spots of water yield service are in sub-basins 8 and 15 in the upstream, and the very significant hot spots are in sub-basin 9 in the downstream. The significant cold spots and region with low-low cluster of water yield service are distributed in the latter section of the upstream, including sub-basins 4, 11, 23 and 25, which implies that areas with low water yield are also surrounded by areas with low water yield. Areas with high-high cluster of water yield service are only distributed in region 15, which implies that high water yield are clustered in this sub-basin. Areas with significantly high soil export are mainly in sub-basins 6, 10, 20 and 26, where soil conservation services are deficit. Regions with lowsoil export cluster are in sub-basins 4 and 28, indicating soil conservation services are very abundant there; the sub-basin 10 is surrounded by areas with high soil export, suggesting very low soil conservation capacity. Areas with significant hot spots of nitrogen export are mainly in and near the YRB downstream, where high sediment accumulation are concentrated, suggesting very low water purification service. Areas with significant cold spots and low-low clusters of nitrogen export are observed in sub-basins 4, 23 and 25, where the water purification service are very high. The sub-basin 13 has low nitrogen export and is surrounded by areas with high nitrogen export. The sub-basin 28 is areas with high nitrogen export, suggesting lacks of water purification service.
Figure 6 (a) Hot/cold spots of annual average water yield (WY) and (b) its spatial cluster, (c) hot/cold spots of annual average nitrogen export (NE) and (d) its spatial cluster, and (e) hot/cold spots of annual average soil export (SE) and (f) spatial cluster. *, ** and *** indicated that significance at the p < 0.10, p < 0.05 and p < 0.01 level, respectively.

3.4 Trade-offs and synergies between WRESs

Figure 7 illustrates the trade-off and synergy relationships among the three WRESs. In the headwater area of YRB upstream, the soil conservation and water purification services showed significantly strong trade-off relationship. Areas with high trade-off occupy 16.73% of the whole YRB (Table 2), and areas with low trade-off relationships are observed in many areas, accounting for 53.68% of YRB. The water yield and soil conservation services maintained synergistic relationships in most areas with the negative correlation between water yield and soil export, and areas with synergistic relationships were distributed in 68% of YRB. The significantly high synergies were mainly in the midstream of YRB, where both water yield and soil conservation services showed enhanced trends. The trade-offs between water yield and water purification services are distributed throughout the entire study area, and 86% of YRB showed the positive correlation between water yield and soil export. Increases in water yield will strengthen the transport of runoff nutrients, which will lead to more water quality pollution.
Figure 7 Trade-offs and synergies among water yield (WY), soil export (SE) and nitrogen export (NE). * indicated that significance at the p < 0.05.
Table 2 Areas of the various degrees of trade-offs/synergies (km2)
Relationship SE vs NE WY vs SE WY vs NE
High trade-off 136,365 79,246 8,700
Low trade-off 437,605 554,880 63,209
Unrelated 43,828 47,823 41,917
Low synergism 148,423 116,248 175,460
High synergism 49,012 17,036 525,945

3.5 Future land management of impacts on WRESs

Table 3 presents the impacts of the three land management measures on the three WRESs. In the base scenario, the multi-year average water yield in YRB is 68.82 mm. Compared to the average, RP, GD and FS scenarios reduce water yield by 1.08, 1.39 and 0.64 mm, respectively, equivalent to of 0.86 billion m3, 11.10 billion m3 and 0.51 billion m3 water losses in YRB. The negative impact of afforestation on water yield is greater than that of the agricultural development scenario. The multi-year average soil export for YRB is 94.13 t·km2 in the base scenario. Compared with the baseline level, the soil export output is reduced by 3.17 t·km-2 in the FS scenario, followed by the GD (-2.44 t·km-2) and RP (-0.63 t·km-2) scenarios. The multi-year average soil export of YRB reached about 304.22 kg·km-2. Compared to the baseline level, RP and GD scenarios reduce the nitrogen export by 35.37 kg·km-2 and 48.03 kg·km-2, respectively, both effectively improving water purification service. However, FS scenario weakened water purification service with an increase of 5.55 kg·km-2 in nitrogen export compared to the multi-year average level. Overall, the three scenarios effectively decreased water yield, soil export, and nitrogen export, except for nitrogen export in FS scenario.
Table 3 Changes in water yield, nitrogen export and soil export under different scenarios
Scenario WY (mm) SE (t·km-2) NE (kg·km-2)
Baseline 68.82 94.13 304.22
RP 67.74 93.50 268.85
GD 67.43 91.69 256.19
FS 68.18 90.96 309.77
Figure 8 illustrates the changing trade-offs and synergistic relationships of WRESs under the three scenarios. In RP scenario, the correlation between nitrogen export and soil export increases compared to the base scenario, implying an increased synergy between soil conservation and water purification services. There were no significant changes in trade-offs and synergies between the other WRESs. In GD scenario, the synergy between nitrogen export and soil export decreased compared to the base scenario, and the correlation between nitrogen export and water yield increased significantly, implying an increase in the trade-off between water yield and water purification services. In FS scenario, the synergy between soil conservation service and water purification service was slightly weakened, while the trade-off between water yield and water purification services reduced. These land management measures failed to significantly change the relationship between soil conservation and water yield services.
Figure 8 Trade-offs and synergies in different land management scenarios. WY, NE and SE denote water yield, nitrogen export and soil export, respectively; RP denotes riparian forest buffer scenario; GD denotes the Grain for Green project scenario; FS denotes the agricultural expansion scenario.

4 Discussion

High-intensity soil erosion was the most serious eco-environmental problem in the YRB. Since the 1970s, local managers have conducted a series of ecological restoration projects in the YRB, such as dam construction and sloping land remediation projects, watershed management projects, soil and water conservation projects, and reforestation and grass restoration projects (Fu, 2022). In this study, we revealed the changes in land cover and vegetation coverage in the YRB from 1990 to 2020 and found continuous reductions in cropland and barren land and continuous increases in grassland and forest. The land cover change confirmed that ecological restoration has achieved great effects on vegetation restoration in the YRB (Yang et al., 2021). The significant increase in vegetation cover was in the YRB midstream (Figure 1), the most severely eroded region, which reflected the effects of positive human interventions on revegetation. The midstream area is the priority area for vegetation restoration projects (Jia et al., 2014). In addition, we found an increase in barren land and vegetation degradation in the headwater area of the YRB upstream, which warrants attention from local environmental authorities. As a result, the core region of ecological restoration needs to be shifted upstream to improve the ecological stability in the headwater area of the YRB upstream, which is critical for sustainable water supply for residents in the midstream and downstream. Impervious areas in the downstream were expanding significantly, suggesting increasing land use intensity and a rising demand for ESs from downstream residents.
Precipitation in the YRB has been increasing over the last 30 years in the YRB, reflecting an overall increase in the atmospheric water inputs of the whole study area. The modeled water yield is very close to the observed water yield from the Water Resources Bulletin of the YRB (Figure 3), which supports the reliability of the results. It is noteworthy that the simulated water yield in 2003 is much higher than the observed water yield. Excessive precipitation allowed the hydraulic facilities (i.e., dam) to fully intercept and store water from upstream, ultimately leading to low runoff downstream (i.e., water yield). The water yield service in the YRB improved from 1990 to 2020, which was mainly determined by the increase in precipitation. Although forest and grassland in the YRB were expanding and decreased in runoff due to increased evapotranspiration caused by afforestation, these would partially offset such negative consequences by positive vegetation-climate feedback on reduced water yield (Li et al., 2018; Zhang et al., 2022). Hence, the increased amplitude of water yield resulting from increased precipitation indeed exceeded the additional water losses by evapotranspiration. The water purification service in the YRB has been weakening due to decreasing nitrogen export, which is mainly influenced by positive human management of agricultural land. The decrease in cropland means less fertilizer is imported into ecosystems, and vegetation often strengthens nutrient interception (Gao et al., 2017; Liang et al., 2021). Hence, the water purification service of the YRB is enhanced mainly due to positive human interventions. The soil conservation service in the YRB showed a fluctuating trend in the last 30 years, with a significant enhancement period occurring from 2002-2012, which experienced the most significant land cover change (Table 1) and indicated that decreases in soil loss were led by positive anthropogenic land management. Overall, the change in water yield services is very consistent with changes in precipitation, reflecting the decisive role of climatic factors (Yang et al., 2019; Pei et al., 2022), while soil conservation and water purification services are associated with changes in land use (Gao et al., 2017; Liang et al., 2021; Liu et al., 2021).
The headwater area of the YRB upstream is a hotspot region of water yield and soil export, indicating high water supply and soil loss there. Because of the poor management upstream, vegetation degradation was significant here, thereby leading to increased soil losses to some extent. Increased precipitation in the upstream mountain region has led to a total increase in the water yield service in the YRB. The latter section of the upstream is the cold spot area of water yield and nitrogen export and is also the transition area of arid and semiarid areas with annual precipitation below 400 mm, thus causing a precipitation deficit and low-intensity agricultural activities. The downstream YRB is a highly significant hotspot of nitrogen export, suggesting a cold spot for water purification services. The cropland in the YRB is mainly distributed downstream, where the nutrient interception roles of the ecosystem are very limited due to sparse vegetation and high fertilizer inputs, resulting in weak water purification services and very high nitrogen export and aggregation. Spatially, areas with high soil conservation services spread mainly in the midstream, along with a statistically significant decreasing trend of soil export, as large-scale vegetation restoration measures were implemented in the midstream. Significant soil losses were observed in the upstream mountain area, where the water purification service was reduced, which demonstrated that the upstream of the YRB suffered ecological degradation and ES decline. Immediate action must be taken to protect and restore ecological functions in the Yellow River headwater, such as establishing natural reserves and strengthening vegetation restoration, which are both important for improving the well-being of residents downstream.
In the midstream of the YRB, further afforestation was reported to be a poor solution for ecological restoration (Chen et al., 2015). Due to the pursuit of economic and selection preferences for single tree species in afforestation practices (Cao et al., 2011) and the inadaptability of planted trees to the local arid climate, further vegetation restoration has exacerbated the trade-offs between ESs (Chen et al., 2015; Feng et al., 2016). Due to economic pursuits, large numbers of single species are planted in afforestation practices, reducing biodiversity in small ecosystems. Eventually, trees are cut down for trade along with the harvesting of other crops, increasing the amount of virtual water transfer removed from the local ecosystems (Zhuo et al., 2022), and ultimately impairing the carbon sequestration and water conservation functions. As revegetation-driven carbon sequestration levels increase and reach a certain threshold, further enhancement of carbon sequestration through tree planting can lead to unsustainable water resources (Feng et al., 2016), including negative roles in groundwater, runoff, and soil water (Jia et al., 2017; Liang et al., 2018; Li et al., 2020). These results are supported by our results, which found that vegetation restoration measures could cause a reduction in water yield and increases in soil conservation and water purification services, with a trade-off effect among WRESs. The trade-off implies that vegetation restoration should be implemented in some specific areas with adequate precipitation to ensure that the vegetation-induced reduction in water resources does not conflict with total water requirements and can achieve a win-win outcome. We found that different vegetation restoration measures had different effects on the trade-offs and synergies of WRESs. Tree planting near rivers further promoted the synergy between soil conservation and water purification services, while tree planting on steep slopes reduced the synergistic effect, which indicated a complex driving mechanism among WRESs. The GD scenario significantly increased the trade-off between water yield and water purification services, as the conversion of cropland to forest improved nutrient interception and reduced nitrogen transport.
WRESs in the whole YRB showed a clear spatial heterogeneity of trade-off/synergy relationships, which can provide wise decision-making for ecological conservation and management measures. For example, the trade-off between water purification and soil conservation services is strong in the mountain region of the upstream YRB, where increasing soil export reduces nitrogen export; additionally, grassland is the main land cover type in this region, and it is distributed with a complex topography. Therefore, upstream ecological management should focus on grassland protection in the reserve, limiting human activities and coordinating the relationship between water purification and soil conservation services. Among the WRESs, the water yield and soil conservation services maintained a synergistic relationship; however, areas with low water yield spread over most of the YRB. Because of the large water demand for tree growth, vegetation restoration practices in the YRB need to consider tree species with lower water consumption and should moderate the competition between water use sectors. In particular, fertilizer input from cropland should be strictly limited to improve water purification.
Overall, the driving mechanisms of WRESs are complex and directly influenced by climate change and human-induced land use changes. Studies have shown that the contribution of rainfall changes to water yield reached 84% in YRB (Yang et al., 2021), 88% (Pei et al., 2022) and 85% (Zhao et al., 2019) in other regions of northern China, respectively. Hence, it can be concluded that water yield is mainly dominated by precipitation. Historically, the channel of Yellow River has undergone several diversions, and water yield varied in different runoff confluence processes due to topographic changes; Anthropogenic reservoirs could directly regulate runoff and sand transport, ensuring irrigated water of agriculture but increasing more nutrient output from fertilizer application, and reducing flood hazards also providing more food supply. Agricultural activities could increase the vulnerability of soils to erosion, due to the natural protective layer in natural soils being broken and eroding during cultivation activities (Cao et al., 2011). A wetter climate would increase rainfall erosion, and increased runoff exports more sediment downstream (Yang et al., 2022). In addition to the effects of rainfall erosion forces, studies have shown that soil erosion is primarily related to the height of topographic slopes (Bao et al., 2022), with steeper slopes being more susceptible to erosion. Additionally, climate warming increases the atmospheric evapotranspiration requirements and increases the percentage of terrestrial moisture partitioned into the atmosphere, causing unexpected extreme climate events. Extreme precipitation and drought will increase rain erosion and wind erosion of soils, respectively, severely decreasing the soil conservation service of ecosystem. On the whole, the driving mechanism of WRESs still needed to be deeply explored and more accurately understood, so as to better serve scientific management of ESs and response with future risks.
In addition, uncertainties of the results arise from the sensitivity of the parameters used in the InVEST model. For example, the result of soil export depends highly on the parameter setting of vegetation factor and anthropogenic management factor, and moreover the output of nitrogen export is closely linked to the nitrogen load of each land cover type (Sanchez-Canales et al., 2012). Due to the lack of field observations, sensitivity analyses are worth conducting in the future. Despite these uncertainties, this study constructively provided an understanding of year-by-year dynamic of ESs from the perspective of WERSs. The three WRESs are the indicators most important indicators related to ecological problems in YRB, including soil losses, water quality pollution, and water availability. At present, YRB ecological management should urgently address problems on degraded vegetation and water retention function in the upstream, soil loss in the midstream, and sediment accumulation and non-point pollution in the downstream. Our study released a clear understanding and an applicable approach on revealing these ecological issues. Notably, many studies on assessing ESs only considered a few years (Gao et al., 2017; Liang et al., 2021; Ge et al., 2022), which greatly limits the understanding of ESs dynamic. This paper complements the gap in studies on year-by-year scale and provides a valuable reference.

5 Conclusions

This study assessed the year-by-year changes in land cover and WRESs in YRB from 1990 to 2020 based on the InVEST model. We revealed the 30-year spatial trends and trade-offs/synergy relationship of WRESs, and estimated WRESs changes under land management scenario. The whole findings can be summarized as follows:
(1) During 1990-2020, the land cover change of YRB is dominated by the reductions of cropland and wasteland as well as increases of forest, grassland and impervious surface. The significant recovery of vegetation coverage is concentrated mainly in the midstream of YRB, and the degraded trends of the vegetation occurred in the headwater area of upstream.
(2) The WRESs in YRB are improving in the last 30 years, with water yield increasing at a rate of +1.11 mm·yr-1 and nitrogen export and soil export decreasing at rates of -1.01 kg·km-2·yr-1 and -0.23 t·km-2·yr-1, respectively. Spatially, areas with the increased water yield were in the headwater area of YRB upstream, yet where the water quality purification and soil conservation services were decreasing.
(3) The water yield, nitrogen export and soil export cold spots are clustered in the latter section of YRB upstream, suggesting lower water yield service and higher soil conservation and water purification services. The soil conservation and water purification services in YRB maintained a trade-off relationship, the water yield sustained a synergistic relationship with soil conservation service but a trade-off relationship with water purification service.
(4) According to the scenario analysis, the afforestation-related measures could reduce the water yield but increase the soil conservation and water purification services. Measures to expand cropland showed the potential of reducing soil loss but increasing pollutant export.
Bai P, Liu X, Zhang Y et al., 2020. Assessing the impacts of vegetation greenness change on evapotranspiration and water yield in China. Water Resources Research, 56(10): e2019WR027019.

Bai Y, Ochuodho T O, Yang J, 2019. Impact of land use and climate change on water-related ecosystem services in Kentucky, USA. Ecological Indicators, 102: 51-64.


Bao H, Tang M, Lan H et al., 2022. Soil erosion and its causes in high-filling body: A case study of a valley area on the Loess Plateau, China. Journal of Mountain Science, 20(1): 182-196.


Budyko M I, 1974. Climate and Life. San Diego: Academic Press.

Cao S, Chen L, Shankman D et al., 2011. Excessive reliance on afforestation in China’s arid and semi-arid regions: Lessons in ecological restoration. Earth-Science Reviews, 104(4): 240-245.


Chen Y, Wang K, Lin Y et al., 2015. Balancing green and grain trade. Nature Geoscience, 8(10): 739-741.


Feng X, Fu B, Piao S et al., 2016. Revegetation in China’s Loess Plateau is approaching sustainable water resource limits. Nature Climate Change, 6(11): 1019-1022.


Fu Bojie, 2022. Ecological and environmental effects of land-use changes in the Loess Plateau of China. Chinese Science Bulletin, 67(32): 3769-3779, 3768. (in Chinese)

Gao J, Bian H, Zhu C et al., 2022. The response of key ecosystem services to land use and climate change in Chongqing: Time, space, and altitude. Journal of Geographical Sciences, 32(2): 317-332.


Gao J, Li F, Gao H et al., 2017. The impact of land-use change on water-related ecosystem services: A study of the Guishui River Basin, Beijing, China. Journal of Cleaner Production, 163: S145-S155

Gashaw T, Tulu T, Argaw M et al., 2018. Estimating the impacts of land use/land cover changes on ecosystem service values: The case of the Andassa watershed in the Upper Blue Nile Basin of Ethiopia. Ecosystem Services, 31: 219-228.


Ge G, Zhang J, Chen X et al., 2022. Effects of land use and land cover change on ecosystem services in an arid desert-oasis ecotone along the Yellow River of China. Ecological Engineering, 176: 106512.


Gong J, Liu D Q, Gao B L et al., 2020. Tradeoffs and synergies of ecosystem services in western mountainous China: A case study of the Bailongiang watershed in Gansu, China. Chinese Journal of Applied Ecology, 31(4): 1278-1288. (in Chinese)

Jia X, Fu B, Feng X et al., 2014. The tradeoff and synergy between ecosystem services in the Grain-for-Green areas in northern Shaanxi, China. Ecological Indicators, 43: 103-113.


Jia X, Shao M, Zhu Y et al., 2017. Soil moisture decline due to afforestation across the Loess Plateau, China. Journal of Hydrology, 546: 113-122


Jiang W, Yuan L, Wang W et al., 2015. Spatio-temporal analysis of vegetation variation in the Yellow River Basin. Ecological Indicators, 51: 117-126.


Karimi J D, Corstanje R, Harris J A, 2021. Bundling ecosystem services at a high resolution in the UK: Trade-offs and synergies in urban landscapes. Landscape Ecology, 36(6): 1817-1835.


Li C, Zhang Y, Shen Y et al., 2020. Decadal water storage decrease driven by vegetation changes in the Yellow River Basin. Science Bulletin, 65(22): 1859-1861.


Li Y, Piao S, Li Z et al., 2018. Divergent hydrological response to large-scale afforestation and vegetation greening in China. Science Advances, 4(5): eaar4182.


Liang H, Xue Y, Li Z et al., 2018. Soil moisture decline following the plantation of Robinia pseudoacacia forests: Evidence from the Loess Plateau. Forest Ecology and Management, 412: 62-69.


Liang J, Li S, Li X et al., 2021. Trade-off analyses and optimization of water-related ecosystem services (WRESs) based on land use change in a typical agricultural watershed, southern China. Journal of Cleaner Production, 279: 12385.

Liu M, Min L, Zhao J et al., 2021. The impact of land use change on water-related ecosystem services in the Bashang area of Hebei province, China. Sustainability, 13(2): 716.


Ma Z G, Fu C B, Zhou T J et al., 2020. Status and ponder of climate and hydrology changes in the Yellow River Basin. Bulletin of Chinese Academy of Sciences, 35(1): 52-60. (in Chinese)

Millennium Ecosystem Assessment MA, 2005. Millennium Ecosystem Assessment Synthesis Report. Island Press, Washington, D.C.

Ouyang Z, Zheng H, Xiao Y et al., 2016. Improvements in ecosystem services from investments in natural capital. Science, 352(6292): 1455-1459.


Pei H, Liu M, Shen Y et al., 2022. Quantifying impacts of climate dynamics and land-use changes on water yield service in the agro-pastoral ecotone of northern China. Science of the Total Environment, 809: 151153.


Peng K, Jiang W, Deng Y et al., 2020. Simulating wetland changes under different scenarios based on integrating the random forest and CLUE-S models: A case study of Wuhan Urban Agglomeration. Ecological Indicators, 117: 106671.


Redhead J W, May L, Oliver T H et al., 2018. National scale evaluation of the InVEST nutrient retention model in the United Kingdom. Science of the Total Environment, 610-616: 666-677.

Sanchez-Canales M, Benito A L, Passuello A et al., 2012. Sensitivity analysis of ecosystem service valuation in a Mediterranean watershed. Science of the Total Environment, 440: 140-153.


Sandeep K, Ranjan P B, 2022. Hydroponic farming hotspot analysis using the Getis-Ord Gi* statistic and high-resolution satellite data of Majuli Island, India. Remote Sensing Letters, 12(4): 408-418.


Sharp R, Tallis H T, Ricketts T et al., 2016. InVEST+VERSION+User’s Guide. Stanford: Stanford University Press.

Sun G, Zhou G, Zhang Z et al., 2006. Potential water yield reduction due to forestation across China. Journal of Hydrology, 328(3/4): 548-558.


Wang H, Yang Z, Saito Y et al., 2006. Interannual and seasonal variation of the Huanghe (Yellow River) water discharge over the past 50 years: Connections to impacts from ENSO events and dams. Global and Planetary Change, 50(3/4): 212-225.


Wang S, Fu B, Piao S et al., 2016. Reduced sediment transport in the Yellow River due to anthropogenic changes. Nature Geoscience, 9(1): 38-41.


Wang Y, Li X, Shi F et al., 2019. The Grain for Green project intensifies evapotranspiration in the revegetation area of the Loess Plateau in China. Chinese Science Bulletin, 64(5/6): 588-599.

Wei P J, Wu M H, Jia Y L et al., 2022. Spatiotemporal variation of water yield in the upstream regions of the Shule River Basin using the InVEST model. Acta Ecologica Sinica, 42(15): 6418-6429. (in Chinese)

Wischmeier W H, Johnson C B, Cross B V, 1971. A soil erodibility monograph for farmland construction sites. Journal of Soil and Water Conservation, 6(5): 189-193.

Yang D, Liu W, Tang Y et al., 2019. Estimation of water provision service for monsoon catchments of South China: Applicability of the InVEST model. Landscape and Urban Planning, 182: 133-143.


Yang J, Huang X, 2022. The 30 m annual land cover datasets and its dynamics in China from 1990 to 2021. Earth System Science Data, 13(8): 3907-3925.


Yang J, Xie B, Zhang D et al., 2021. Climate and land use change impacts on water yield ecosystem service in the Yellow River Basin, China. Environmental Earth Sciences, 80(3): 72.


Yang P, Li R, Gu Z et al., 2022. Runoff sediment characteristics affected by erosive rainfall patterns in a small watershed in karst areas of Southwest China. Catena, 219: 106591.


Zhang B, Tian L, Yang Y et al., 2022. Revegetation does not decrease water yield in the Loess Plateau of China. Geophysical Research Letters, 49(9): e2022GL098025.

Zhang L, Dawes W R, Walker G R, 2001. Response of mean annual evapotranspiration to vegetation changes at catchment scale. Water Resources Research, 37(3): 701-708.


Zhao M, Geruo A, Zhang J et al., 2021. Ecological restoration impact on total terrestrial water storage. Nature Sustainability, 4(1): 56-62.


Zhao Y R, Zhou J J, Lei L et al., 2019. Identification of drivers for water yield in the upstream of Shiyang River based on InVEST model. Chinese Journal of Ecology, 38(12):3789-3799. (in Chinese)

Zhu W, Zhang J, Cui Y et al., 2020. Ecosystem carbon storage under different scenarios of land use change in Qihe catchment, China. Journal of Geographical Sciences, 30(9): 1507-1522.


Zhuo L, Li M, Zhang G et al., 2022. Volume versus value of crop-related water footprints and virtual water flows: A case study for the Yellow River Basin. Journal of Hydrology, 608: 127674.