The Coupling Relationship Between Ecological Quality and Ecosystem Service Functions in the Sources of the Danjiangkou Reservoir

Abstract

Identifying the key drivers behind the spatiotemporal dynamics of ecosystem service functions is essential for clarifying how ecosystems respond to environmental changes. Such insights deepen our understanding of the evolution of complex ecological processes and service functions, and provide critical references for ecological governance, policy-making, and the pursuit of high-quality development pathways. In this study, the Remote Sensing Ecological Index (RSEI) was first constructed for the upstream basin of the Danjiangkou Reservoir using satellite imagery (2015 and 2024). We then employed the InVEST model to quantify six ecosystem service functions and their corresponding services: water purification (total nitrogen and total phosphorus), soil retention (soil erosion), water yield, carbon storage, and habitat provision (habitat quality). Finally, this study analyzes the driving mechanisms as well as the coupling coordination degree between the RSEI and six ecosystem service functions. From 2015 to 2024, the area classified as “excellent” in RSEI significantly expanded from 263.34 km² (3.22%) to 2566.21 km² (31.38%), reflecting a substantial enhancement in ecological quality throughout the upstream basin. There is no serious imbalance in the coupling and coordination relationship between RSEI and the value of various ecosystem service functions. Although improvements in ecosystem quality generally enhanced overall ecosystem service functions, competition among certain services was still evident in localized areas. Future ecological management should, therefore, prioritize not only the protection of ecosystem quality but also the scientific allocation of service supply and demand, the optimization of human–land relationships, and the promotion of a virtuous ecosystem cycle.

1. Introduction

The Xichuan Reservoir area of the Danjiangkou Reservoir serves as the water source for the middle route of the South-to-North Water Diversion Project (SNWDP). Its upstream basin spans the Qinba mountainous region and parts of the transitional zone of the Loess Plateau. Characterized by significant topographic relief and ecological fragility, this area functions as a major water conservation zone and ecological barrier [1]. It harbors rich biodiversity, complex mountain terrain, and critical water-conserving forests, which not only provide clean water resources but also support essential ecosystem service functions such as soil sediment, biodiversity conservation, climate regulation, and carbon storage [2,3]. With the growing population, economic development, and rapid land use changes within the basin, ecosystems of the region are increasingly exposed to multiple ecological pressures, such as agricultural non-point source pollution, vegetation degradation, and soil erosion [4]. These issues threaten not only the ecological quality of the basin itself but also its capacity to deliver key ecosystem service functions in a sustainable manner [5]. Ecological quality (EQ) is a comprehensive indicator reflecting the health status of regional ecosystems, integrating critical ecological factors such as vegetation cover, humidity, thermal conditions, and land surface exposure [6]. The evaluation of ecosystem quality involves a qualitative or quantitative analysis of a system’s favorable and unfavorable characteristics, in order to investigate the relationship between ecosystems and human activities [7]. With the implementation of the national ecological civilization strategy, achieving a balance between economic and social development and the preservation and improvement of water source ecological quality has emerged as a pivotal challenge in the management and conservation of large-scale lakes and reservoirs [8]. As the foundation for the survival and development of humankind, ecosystem service functions are fundamentally linked to ecological quality [9]. The preservation of ecosystem structure and function is a fundamental prerequisite for ensuring the long-term stability of reservoir systems [10]. Ecological quality directly affects water safety and the reliability of the water supply [11]. Ecosystem services (ESs) refer to the benefits that human beings obtain from ecosystems [12]. It focuses on the benefits that ecosystems provide to humanity, specifically emphasizing the value of ecosystem functions for societies [13]. Ecosystem service functions are the capacities of ecosystems to perform benefits, arising from their internal structure and ecological processes [14]. These functions are the natural outcomes of interactions between an ecosystem’s biological communities and its abiotic environment. It emphasizes what an ecosystem is capable of achieving and refers to the inherent ability of an ecosystem to function and operate in a natural state [15]. Ecosystem services represent the ultimate manifestation of how the specific contributions and values of ecosystem functions translate to human well-being, while ecosystem service functions themselves serve as the fundamental basis and essential prerequisite for ecosystems to supply such services. Ecosystem service functions describe the ability of an ecosystem to serve humans, including providing material, regulatory, and cultural benefits [16,17]. Ecological quality, on the other hand, shows how healthy the ecosystem is as a whole, based on its structure, processes, and functions [18]. Notably, there exists a close coupling relationship between ecosystem services and ecological quality. Therefore, investigating the coupling relationship between ecosystem services and ecological quality of the middle route of the South-to-North Water Diversion Project helps to reveal the key ecological mechanisms maintaining water quantity, water quality, and regional ecological security, thereby providing a scientific basis for safeguarding the national strategic water source area.

Following the Millennium Ecosystem Assessment, we distinguish between ecosystem functions—the biophysical processes and structures that constitute the potential of ecosystems—and ecosystem services, which are defined as the realized benefits that human populations obtain from these functions [19]. In recent years, both international and Chinese scholars have increasingly acknowledged that examining a single ecosystem service is inadequate for fully capturing changes in overall ecosystem functionality [20]. Instead, it is necessary to analyze multiple services together to reveal the mechanisms of their interactions [21]. Many studies have assessed ecological quality through indicators such as vegetation cover [22], biodiversity indices [23], and soil erosion intensity [24], or have focused on quantifying specific services such as water conservation [25], soil sediment [26], and water purification [27]. Some have examined the effects of human activities, such as land use change, engineering projects, or climate change, on either ecological quality or individual services. However, studies that systematically integrate the multidimensional concept of “ecological quality” with that of “ecosystem services functions” remain relatively scarce. In particular, the existing body of literature exhibits several limitations: (1) Certain studies define ecological quality by means of ecosystem services [28]. This approach conflates the system’s intrinsic state with the benefits it delivers to external entities, thereby creating obstacles to establishing valid causal or correlational relationships between them [29]. This reduces the independence and comparability of ecological quality and ecosystem services, hindering a clear understanding of their interaction and obscuring the underlying linkages among ecosystem structure, function, and services; (2) In the scale of large-scale major water source basins, significant research gaps remain in the quantitative assessment of the dynamic correlation between ecological quality and ecosystem service capacity [30]. Few studies have systematically evaluated the degree of interdependence and constraint among different systems. Without understanding this coupling relationship, changes in ecosystem services are thus simplistically attributed to changes in ecological quality [31]. This misattribution not only obscures the causal relationship between the ecosystem services and ecological quality but also fails to uncover their dynamic coupling patterns and long-term evolutionary trends; (3) Moreover, the key driving mechanisms underlying the coupling between ecological quality and ecosystem service functions have not been thoroughly investigated. In particular, the roles of natural conditions and socio-economic drivers, as well as their interactions, remain insufficiently explored, which limits the scientific basis for precise watershed ecological management [32]. Without disentangling the relative roles and interactions of these natural and socio-economic drivers, policymakers cannot prioritize the most effective regulations or investments, hindering the implementation of precise and cost-effective watershed ecological management strategies.

To address these limitations, the research takes the upstream basin of the Danjiangkou Reservoir as the research area, integrating remote sensing data, geospatial datasets, and the InVEST model to build a comprehensive analytical framework. To effectively manage these interdependent relationships, we propose a framework that combines ecological quality assessment, ecosystem service quantification, synergy/trade-off analysis, and coupling coordination evaluation. This study aims first to delineate the spatiotemporal patterns of ecological quality and multiple ecosystem service functions within the basin. It then aims to assess the dynamics of the relation between RSEI and ecosystem service functions. And finally, it aims to analyze the coupling coordination between ecological quality and ecosystem service functions and their driving factors. The findings are expected to offer scientific insights and robust decision-making support for similar large-scale water source basins with comparable ecological and hydrological characteristics; their direct generalizability to global contexts remains to be verified by further cross-basin comparative studies.

2. Study Area and Methods

2.1. Overview of the Study Area

The Danjiangkou Reservoir, with a total area of 1022.75 km², located in the middle reaches of the Han River, serves as the primary water source for the South-to-North Water Diversion Project. The reservoir spans Henan and Hubei provinces, with the Henan section mainly encompassing Xichuan. With an area of 546 km², the Xichuan section represents 52% of the reservoir’s total area. It is a typical large-scale lake-reservoir type water source area (Figure 1). In December 2014, Danjiangkou Reservoir officially began to supply water. As of April 2023, the reservoir had transferred over 55 billion cubic meters of water, directly benefiting a population surpassing 85 million people. The reservoir’s source lies at the transitional zone between the Qinba Mountains and the western Henan mountainous region, characterized by low-to medium-elevation mountains and hills and interspersed with river valley plains. The climate is a transition between the northern subtropical and warm temperate zones, with four distinct seasons and an annual average precipitation of approximately 800–1000 mm, providing relatively abundant water resources. The region is rich in virgin forest resources, and forest coverage is approximately 34%, with exceptional water conservation capacity and abundant rare flora and fauna, making it a vital ecological repository in China. The distinctive natural landscape of the water source has given rise to a dense hydrological network dominated by the Han River. The region contains 21 rivers with a basin area exceeding 1000 km² and around 220 rivers exceeding 100 km². The water source is characterized by excellent quality, and the water quality maintains full compliance with at least Class II standards, with Class I attainment recorded at specific monitoring sections. Since the region has the dual attributes of being a strategic water source for the South-to-North Water Diversion Project and a local economic development area, land use change profoundly influences the functions of ecosystem service functions. Consequently, this area serves as a typical and representative ecosystem for investigating ecosystem services under land use change.

2.2. Research Methods

2.2.1. Research Framework

This study first employs remote sensing imagery to construct the Remote Sensing Ecological Index (RSEI) for the upstream watershed of the Danjiangkou Reservoir, the water source of the middle route of the South-to-North Water Diversion Project. Then, five modules from the InVEST model—water purification, sediment delivery ratio, carbon storage, water yield, and habitat quality—are used to calculate six ecosystem service functions: total nitrogen, total phosphorus, total carbon storage, soil erosion, water yield, and habitat quality. After normalizing the RSEI and ecosystem service outputs separately, we analyze their driving mechanisms, synergistic and trade-off relationships, and the degree of coupling coordination.

2.2.2. Data Sources and Processing

To mitigate the confounding effects of seasonal vegetation dynamics, climatic phenological variations, and disparate data acquisition conditions, and to ensure that the analysis results can accurately reflect the real changes in ecological quality, rather than those caused by natural seasonal fluctuations or data errors, Level 2 Collection 2 (C2L2) Landsat 8 imagery from May and June of 2015 and 2024 with a spatial resolution of 30 m were obtained from the U.S. Geological Survey (USGS) website (https://eros.usgs.gov/ (accessed on 14 April 2014 and 16 May 2024)). NDVI serves as an effective indicator of vegetation condition, reflecting cover and photosynthetic activity to provide a reliable proxy for ecosystem productivity and ecological quality in large-scale watershed studies. To improve the accuracy of NDVI calculations, images with minimal cloud cover and high quality were selected from these months. The data were pre-processed with geometric correction and radiometric calibration, followed by atmospheric correction, mosaicking, and clipping using ENVI 5.6.

To ensure spatial comparability, all geographic datasets were reprojected to the CGCS2000 coordinate system and resampled to a uniform 30 m resolution. To avoid bias due to differences in units and scales, all data were normalized using the min–max normalization method. The Modified Normalized Difference Water Index (MNDWI) was applied to extract water bodies in the study area and mask them out to reduce the influence of water features on principal component load distribution. Atmospheric correction was performed using the FLAASH module in ENVI, which accounts for atmospheric absorption and scattering based on MODTRAN radiative transfer modeling. The accuracy of atmospheric correction was evaluated by comparing surface reflectance values across overlapping image scenes and verifying their spectral consistency with standard surface reflectance products, ensuring physically realistic reflectance values and minimal residual atmospheric effects.

2.2.3. Construction of the Remote Sensing Ecological Index (RSEI)

(1) Greenness Index—NDVI

The Normalized Difference Vegetation Index (NDVI) is used to reflect vegetation growth conditions and vegetation cover dynamics in the study area [33]. It is calculated as follows:

$$NDVI=(NIR-Red)/(NIR+Red)$$

(1)

where Red is the red band; NIR is the near-infrared band.

(2) Wetness Index—Wet

The Tasseled Cap Transformation is a multispectral image-enhancement technique based on matrix operations that rotates the multidimensional spectral space to separate information on vegetation and soil, yielding components such as brightness, greenness, and wetness [34]. The third component derived from the Tasseled Cap Transformation of Landsat 8 imagery represents the wetness index, calculated as follows:

$$\begin{array}{cc}\hfill Wet=& 0.1511Blue+0.1973Green+0.3283Red\hfill \\ \hfill & +0.3407NIR-0.7117SWIR1-0.4559SWIR2\hfill \end{array}$$

(2)

where Blue, Green, Red, NIR, SWIR1, and SWIR2 denote the reflectance values in the blue, green, red, near-infrared, and shortwave infrared 1 and 2 bands, respectively.

(3) Dryness Index—NDBSI

The Normalized Difference Built-up and Soil Index (NDBSI) is used as an indicator of surface dryness. This composite index is computed by averaging two constituent indices that represent surface characteristics: the Soil Index (SI) and the Index-based Built-up Index (IBI) [35]. The formula is as follows:

$$NDBSI={\displaystyle \frac{IBI+SI}{2}}$$

(3)

where

$$IBI={\displaystyle \frac{\frac{2NIR}{NIR+Red}-\left(\frac{Red}{Green+Red}+\frac{Blue}{Blue+NIR}\right)}{\frac{2NIR}{NIR+Red}+\left(\frac{Red}{Green+Red}+\frac{Blue}{Blue+NIR}\right)}}$$

(4)

$$SI={\displaystyle \frac{\left(NIR+Green\right)-\left(Red+Coastal\right)}{\left(NIR+Green\right)+\left(Red+Coastal\right)}}$$

(5)

In these formulas: SI represents the Soil Index; IBI represents the Index-based Built-up Index; NDBSI is the Normalized Difference Built-up and Soil Index, used as an indicator of impervious surface and dryness; coastal refers to the coastal/aerosol band; spectral bands including Blue, Green, Red, NIR, SWIR1, SWIR2 are consistent with those applied in the Wetness index calculation.

(4) Heat Index—LST

The Land Surface Temperature (LST) derived from the thermal infrared band of remote sensing imagery is used as the heat index [36]. The calculation is as follows:

$$\mathrm{LST}=T/[1+(\lambda T/\rho )\ln \epsilon ]$$

(6)

where T is the sensor temperature; λ is the central wavelength of the thermal infrared band; ρ = 0.01438 (a constant derived from Planck’s law); ε is the land surface emissivity; L is the spectral radiance.

(5) RSEI Construction

Since the four indices used to compute RSEI differ significantly in scale and units, the range normalization method is used to standardize all indicators, reducing the influence of unit discrepancies on the evaluation results [37]. The normalization formula is as follows:

$$N={\displaystyle \frac{X-X_{\min }}{X_{\max }-X_{\min }}}$$

(7)

where X is the original value of the indicator; Xmax and Xmin are the maximum and minimum values of the indicator, respectively.

After normalization, principal component analysis (PCA) is applied, and the first principal component (PC1) is extracted to represent the initial remote sensing ecological index (RSEI₀):

$$RSE{I}_0=1-\left\{PC1[f(NDVI,Wet,LST,NDBSI)]\right\}$$

(8)

The initial RSEI is then normalized again to obtain the final RSEI:

$$RSEI={\displaystyle \frac{(RSE{I}_0-RSE{I}_{0_{\min }})}{(RSE{I}_{0_{\max }}-RSE{I}_{0_{\min }})}}$$

(9)

where RSEI₀ is the initial ecological index; RSEI_0min and RSEI_0max are the minimum and maximum values of RSEI₀, respectively.

The final RSEI ranges from 0 to 1, with higher values indicating better ecological environmental quality. The RSEI results are then classified into five categories using the equal-interval method: Poor, Fairly Poor, Moderate, Good, and Excellent, to reflect different levels of ecological quality across the region [38].

2.2.4. InVEST Model

The InVEST model was designed to assess the spatial dynamics and interactions of multiple ecosystem service functions, supporting integrated assessment across diverse geographic and socio-economic contexts [39]. This study utilizes five InVEST modules: water purification, sediment delivery ratio, carbon storage, water yield, and habitat quality for analysis.

(1) Water Purification

The Nutrient Delivery Ratio (NDR) serves as a functional indicator of nutrient pathways, reflecting both aquatic transport and sediment release processes to reveal the sources, movement, and ultimate fate of nutrients. The average and total runoff per sub-watershed were derived from input reference data and water yield values. Then, these values are used to estimate the pollutant loads exported from each area [40]. The calculation formulas are as follows:

$$AL{V}_x=HS{S}_x\times po{l}_x$$

(10)

$$HS{S}_x={\displaystyle \frac{{\gamma }_x}{\overline{{\gamma }_W}}}$$

(11)

$${\gamma }_x=\log \left({\displaystyle \sum ^u{Y}_u}\right)$$

(12)

where $AL{V}_x$ denotes the load of water quality elements; $HS{S}_x$ is the hydrological sensitivity; $po{l}_x$ is the export coefficient for the relevant water quality element; runoff indices correspond to runoff volume; the total water yield upstream of the target area is considered in calculations.

(2) Soil Erosion

Soil erosion, a key ecosystem service, contributes significantly to curbing land degradation and alleviating hazardous events like floods and landslides. The Sediment Delivery Ratio (SDR) module, which is based on the Revised Universal Soil Loss Equation (RUSLE), is widely used to quantify soil erosion [41]. Its effectiveness stems from the integration of critical driving factors: topography, climate, soil characteristics, vegetation conditions, and land use patterns [42]. The calculation formula is as follows:

$$RUSLE=R\times K\times LS\times C\times P$$

(13)

where RUSLE (t/ha/year) denotes the annual soil loss, R [MJ∙mm/(ha∙h∙a)] denotes the rainfall erosivity factor, K [(t∙ha∙h/(ha∙MJ∙mm)] denotes the soil erodibility factor, LS denotes the slope length and gradient factor, C denotes the vegetation cover and management factor, and P denotes the soil and water conservation measure factor.

a. R factor

It was calculated by a simple algorithm model for estimating erosion using data from 71 representative meteorological stations throughout China [43]. The 71 representative meteorological stations were selected based on China’s agricultural climatic regionalization, with daily rainfall and daily maximum 10 min rainfall intensity data collected from each station’s establishment up to 1984. Stations located in the mid-tropical and south-tropical zones of Qiong’nan, the Xisha, Zhongsha, and Dongsha Islands, as well as the Nansha Islands, were not included due to their extremely small proportion of China’s total land area. Similarly, no stations were selected from the arid southern temperate zone of the Tarim-Hami Basin, where annual precipitation is extremely low, or from the alpine frigid and sub-frigid zones, where precipitation is scarce and mainly falls as snow. Although these three zones cover large areas, they are sparsely populated, have very few meteorological stations, and often contain incomplete records. Among the 71 selected stations, 52 have data records spanning 25–29 years, 12 have records for 20–24 years, 4 have records for 10–20 years, and the remaining 3 have 6–9 years of data. The lowest annual precipitation was recorded at Altay in Xinjiang (137 mm), while the highest was observed at Guilin in Guangxi (1667 mm).

This method has been widely applied across most regions of China, and its applicability and calculation accuracy have been fully verified in existing studies [44]. The calculation formula is as follows:

$$R=\alpha {{\displaystyle \sum _{j=1}^k\left(D_j\right)}}^{\beta }$$

(14)

$$\alpha =21.586{\beta }^{-7.189}$$

(15)

$$\beta =0.8363+{\displaystyle \frac{18.177}{P_{d12}}}+{\displaystyle \frac{24.455}{P_{y12}}}$$

(16)

where k is the annual number of erosivity rainfall days, D_j represents the daily rainfall (≥12 mm) of the j day, P_d12 is the average daily rainfall of all days with daily rainfall ≥ 12 mm, P_y12 is the average annual rainfall with daily rainfall ≥ 12 mm, and α and β are the coefficients to be determined.

b. K factor

The K factor is the soil erodibility calculated using the Erosion Productivity Impact Calculator (EPIC) model, in which data on soil properties are derived from soil texture, organic matter, structure, and permeability. It represents the susceptibility of soil particles to detachment and transport by rainfall and surface runoff [45]. The calculation formula is as follows:

$$\begin{array}{c}K=\left\{0.2+0.3exp\left[0.0256SAN\left(1-{\displaystyle \frac{SIL}{100}}\right)\right]\right\}{\left({\displaystyle \frac{SIL}{CLA+SIL}}\right)}^{0.3}\\ \left[1.0-{\displaystyle \frac{0.25C}{C+exp(3.72-2.95C)}}\right]\left[1.0-{\displaystyle \frac{0.7SN}{SN+exp(-5.51+22.9SN)}}\right]\end{array}$$

(17)

where K is the soil erodibility (t∙hm²∙h∙m² MJ⁻¹∙mm⁻¹ h), SAN, SIL, and CLA (%) are the content value (%) of sand, silt, and clay, respectively. SN = 1 − SAN/100.

c. LS factor

The topographic factor is one of the most important and spatially variable components in the RUSLE. It represents how slope length (L) and slope steepness (S) together influence soil erosion [46]. The calculation formula is as follows:

$$\begin{array}{c}\mathrm{I}S=S\cdot \left[{\left(A_{in}+D^2\right)}^{m+1}-A_{in}^{m+1}\right]/\left(D^{m+2}\cdot x_i^m\cdot {22.13}^m\right)\\ S=\left.\left\{\begin{array}{c}10.8\sin \theta +0.03,\mathrm{where}\theta <9\%\\ 16.8\sin \theta -0.50,\mathrm{where}\theta \ge 9\%\end{array}\right.\right\}\\ m=\left.\left\{\begin{array}{c}0.2,\mathrm{where}\theta \le 1\%\\ 0.3,\mathrm{where}1\%<\theta \le 3.5\%\\ 0.4,\mathrm{where}3.5\%<\theta \le 5\%\\ 0.5,\mathrm{where}5\%<\theta \le 9\%\\ \lambda /(1+\lambda ),\mathrm{where}\theta >9\%\end{array}\right.\right\}\\ \lambda =(\sin \theta /0.0896)/\left(3\sin {\theta }^{0.8}+0.56\right)\end{array}$$

(18)

where L is the slope length factor; S is the gradient factor; A_in(m²) is the area of the sediment yield area above the pixel entrance, calculated using the D8 flow direction algorithm; D (m) is the pixel size; x_i = |sinα_i| + |cosα_i|, α_i is the sediment export direction of the pixel i; m is the slope length index; θ is the slope; λ is the intermediate variable.

d. C and P factor

The vegetation cover and management factor (C) quantifies the ratio of soil loss from land under specific vegetation cover or management practices to that from continuously bare soil subjected to conventional tillage. The C factor ranges from 0 to 1 and is determined according to relevant studies. The soil and water conservation practice factor (P) reflects the effectiveness of soil and water conservation measures in reducing soil erosion. It is defined as the ratio of soil loss from land with conservation practices to that from land without such measures on sloping arable terrain. The P factor ranges from 0 to 1 and is assigned based on relevant studies [47] (

Table 1. C&P values for each land use type in the upstream basin of Danjiangkou Reservoir.

Land Use Type	A	F	G	W	C	U
C	1	0.51	0.33	0	0.28	1
P	1	0.15	0.3	0	0.1	1

Note: Land use types were categorized into six classes: cultivated land (A), forest land (F), grassland (G), water (W), construction land (C), and non-used land (U).

).

(3) Carbon Storage

This module was designed to evaluate the carbon sequestration capacity of the study area. Total carbon storage, which is closely linked to regional land use conditions, is quantified as the sum of four distinct carbon pools: aboveground biomass carbon density, belowground root carbon density, soil organic carbon storage, and dead organic matter carbon storage [48]. The formula is as follows:

$$C_{Total}=C_a+C_b+C_s+C_d$$

(19)

where $C_{Total}$ is the total carbon storage; $C_a$ is the aboveground biomass carbon storage; $C_b$ is the belowground root carbon density; $C_s$ is the soil carbon storage; $C_d$ is the dead organic matter carbon storage.

(4) Habitat Quality

Habitat quality serves as a key indicator of biodiversity. As such, this module assesses it by quantifying the extent and degree of degradation of various habitat or vegetation types within a region [49]. The formula is as follows:

$$HQ=H_i\times \left[1-{\displaystyle \frac{D_i^z}{D_i^z+k^z}}\right]$$

(20)

where HQ denotes a given land use habitat quality value, ranging from 0 to 1. The HQ value closer to 1 indicates superior habitat quality, whereas the HQ value closer to 0 signifies inferior quality. $H_i$ denotes the habitat suitability of a certain land use; $D_i^z$ denote the degree of habitat degradation corresponding to a specific land use type in the study area; k denotes half-fullness and the coefficient.

(5) Water Yield

This module was constructed based on the Budyko water–energy balance framework. The core principle of this framework is that the net water yield of any grid cell within a watershed is quantified as the difference between precipitation and actual evapotranspiration [50]. The formula is as follows:

$$Y\left(x\right)=\left(1-{\displaystyle \frac{AET(x)}{P(x)}}\right)\times P(x)$$

(21)

where Y(x) is the annual water yield of land use type at grid x; AET(x) is the actual annual evapotranspiration at grid x; P(x) is the annual precipitation at grid x.

2.2.5. Redundancy Analysis (RDA)

Redundancy Analysis (RDA) is an ordination method that combines regression analysis with principal component analysis (PCA), allowing for the clarification of relationships between explanatory and response variables [51]. In this study, RDA is employed to quantitatively explore the mutual influence between ecological environmental quality and ecosystem service functions in the Danjiangkou Reservoir area.

Six ecosystem service indicators are designated as response variables, while five remote sensing indices are set as explanatory variables. Values for both sets of variables are extracted using a 1 km grid resolution. The RDA is conducted using R Studio 4.4.3.

In the RDA results plot, a perpendicular angle between two arrows indicates that the corresponding variables are relatively independent and uncorrelated. The angle between arrows represents the degree of correlation: Acute angles suggest a positive correlation, with smaller angles indicating stronger relationships. Obtuse angles suggest a negative correlation, with larger angles indicating stronger negative relationships. The cosine of the angle between explanatory and response variable arrows reflects the overall contribution of explanatory variables to the variance in response variables [52].

In this study, R was used to analyze the response mechanism between RSEI and ecosystem service functions. The response variables include RSEI, NDVI, Wetness (WET), NDBSI, and LST, while the explanatory variables include six ecosystem service functions: total nitrogen, total phosphorus, soil erosion, water yield, total carbon storage, and habitat quality. Before performing RDA, all variables (both RSEI and ecosystem service functions) are standardized, and then the RDA is conducted. A permutation test was employed to evaluate the statistical significance of the observed results.

2.2.6. Coupling Coordination Degree Model

The coupling coordination degree (CCD) model is used to evaluate the degree of interaction and coordinated development between different systems, elements, or variables [53]. A higher CCD value indicates a stronger level of mutual promotion and coordination between the elements [54]. The commonly used formulas are as follows:

$$CD={\displaystyle \frac{2\sqrt{U_1\times U_2}}{U_1+U_2}}$$

(22)

$$CCD=\sqrt{C\times (\alpha U_1+\beta U_2)}$$

(23)

where CD is the coupling degree; CCD is the coupling coordination degree; U₁ represents the RSEI value; U₂ represents the value of an ecosystem service function; referring to relevant studies, the contributions of the two factors to coordination are assumed to be equal, and the weights are set as α = β = 0.5.

To more accurately evaluate the results of coupling coordination, the CCD values are classified into different levels based on defined numerical intervals. Existing research typically uses the equal-interval method, dividing the CCD into 3 to 10 types [55]. A5-level classification is considered to offer a good balance between differentiation and interpretability, effectively reflecting the coordination status between different factors. In this study, based on the existing literature and the ecological characteristics of the Danjiangkou Reservoir region, the coupling coordination degree is classified into five types: high coordination, good coordination, basic coordination, mild disorder, and severe disorder. For the specific classification thresholds, see

Table 2. Coupling coordination degree classification.

Type	Characteristics	Value Range
High Coordination	Benign interaction is strengthened; relationship is well-coordinated and orderly	(0.8, 1.0]
Good Coordination	Mutual balance and cooperation; relationship is relatively coordinated	(0.6, 0.8]
Basic Coordination	Interaction is basically coordinated; normal development can be sustained	(0.4, 0.6]
Mild Disorder	Competitive interaction; coordination deteriorates	(0.2, 0.4]
Severe Disorder	Obvious conflict; development becomes disordered and uncoordinated	(0.0, 0.2]

.

3. Results and Analysis

3.1. RSEI Temporal Variation

Since the four indicators (NDVI, WET, NDBSI, and LST) that were used to calculate the RSEI have different units and scales, unequal weighting may occur during the principal component analysis (PCA). Meanwhile, to mitigate the influence of extensive water bodies on the principal component loadings, the MNDWI was employed to extract the water extent for masking. Subsequently, all four indicators were normalized to a uniform [0, 1] range. Finally, principal component analysis was performed on the integrated imagery comprising these four indicators. The eigenvalues and respective contribution rates of all indicators to the principal components were derived for the years 2015 and 2024. The results are shown in

Table 3. Principal component analysis results.

Year	Indicator	PC1	PC2	PC3	PC4
2015	NDVI	0.341	0.389	0.482	0.707
	WET	0.341	0.389	0.482	−0.707
	NDBSI	−0.474	0.491	−0.731	0.000
	LST	−0.737	−0.676	0.023	0.000
	Eigenvalue	0.184	0.059	0.004	0.000
	Variance (%)	74.33	23.91	1.76	0.00
2024	NDVI	0.423	0.792	−0.149	−0.414
	WET	0.045	0.275	0.930	0.239
	NDBSI	−0.270	0.253	−0.313	0.875
	LST	−0.863	−0.482	0.122	−0.083
	Eigenvalue	0.189	0.036	0.008	0.001
	Variance (%)	80.97	15.25	3.55	0.24

.

The contribution rates of the first principal component in 2015 and 2024 were 74.33% and 80.97%, respectively, both exceeding 55%. This indicates that the characteristic information of the NDVI, WET, NDBSI, and LST is largely concentrated in the first principal component. In the first principal component, the loadings of the four indicators exhibit a clear pattern: NDVI and WET have positive loadings, indicating that they contribute positively to regional ecological environmental quality. NDBSI and LST have negative loadings, suggesting that they have a degrading effect on ecological quality. These PCA results are consistent with former research, confirming that the selected indicators accurately reflect the ecological environment of the research area [56]. Therefore, the four indicators NDVI, WET, NDBSI, and LST can be effectively used to construct the Remote Sensing Ecological Index (RSEI) for the Danjiangkou Reservoir area.

Based on Landsat 8 imagery from 2015 and 2024, the values of NDVI, WET, NDBSI, and LST for the upstream watershed of the Danjiangkou Reservoir were calculated. PCA was then used to derive RSEI for both years. Since RSEI is continuous (0–1), it must be divided into discrete ecological quality levels. The most objective way of classification is the natural breaks (Jenks) method. The RSEI proportion changes in the area in 2015 and 2024 were presented in

Table 4. RSEI proportion changes in 2015 and 2024.

RSEI Level	2015		2024
	Area (km2)	Proportion %	Area (km2)	Proportion %
Poor	0.40	0.00	16.91	0.21
Fairly Poor	114.32	1.40	376.65	4.61
Moderate	2001.55	24.48	2340.14	28.62
Good	5827.67	71.26	2877.58	35.19
Excellent	263.34	3.22	2566.21	31.38

.

The spatial distribution pattern of RSEI in the upstream basin of Danjiangkou Reservoir in 2015 and 2024 is shown in Figure 2. In 2015, the “Good” level had the largest area and proportion across the entire watershed, reaching 5827.67 km² and 71.26%, respectively. This was followed by the “Moderate” level, with an area of 2001.55 km² and a proportion of 24.48%. The remaining three levels occupied relatively smaller portions of the watershed. In 2024, the “Good” level still covered the largest area, but both its area and proportion dropped significantly to 2877.58 km² and 35.19%, marking a decrease of 2950.09 km² and 36.07 percentage points. In contrast, the “Excellent” level showed a substantial increase, reaching 2566.21 km² and 31.38%, which is an increase of 2302.87 km² and 28.16 percentage points, respectively. The remaining three levels (“Poor,” “Fairly Poor,” and “Moderate”) also experienced varying degrees of increase in area and proportion. Overall, from 2015 to 2024, the area classified as “Excellent” in terms of RSEI increased significantly in the upstream watershed of the Danjiangkou Reservoir.

Spatial analysis of RSEI in Figure 2 reveals that the “Poor” and “Fairly Poor” quality classes were primarily clustered in the northeastern part of the watershed in 2015. “Moderate” RSEI levels were primarily distributed across the northern mountainous areas, while regions with “Good” and “Excellent” RSEI values were mainly located in the central part of the study area. By 2024, the spatial extent of “Poor” and “Fairly Poor” levels expanded in the northern high-altitude mountainous areas, with small-scale increases also observed in the southern region. The area covered by the “Moderate” level also increased notably in the northern watershed. Most significantly, areas with “Good” and “Excellent” ecological quality expanded markedly in the central region. In 2015, the study area was predominantly characterized by moderate and good RSEI levels, with excellent areas scattered mainly in the southern and central regions. By 2024, there is a marked spatial agglomeration of “excellent” levels, forming large continuous patches—especially in the southern, central, and eastern regions—indicating a significant enhancement in ecological quality. The northern mountainous and riparian areas remain dominated by fairly poor-to-moderate RSEI levels. The agglomeration of high RSEI levels (“Excellent” and “Good”) strengthens over time. The spatial pattern transitions from a relatively uniform “Good” agglomeration to more differentiated clusters of “Excellent,” indicating enhanced spatial heterogeneity in ecological quality, with high-quality areas becoming more concentrated. The shift reflects a spatial clustering trend of improving ecological quality.

3.2. Spatial Changes in Ecosystem Service Functions

Over the decade period from 2015 to 2024, ecosystem service functions in the research area have undergone varying degrees of change in both scope and intensity (Figure 3).

The two maximum values of the water quality purification indicators, total nitrogen increased from 2.478 kg·hm⁻²·a⁻¹ in 2015 to 2.507 kg·hm⁻²·a⁻¹ in 2024, and total phosphorus decreased from 0.283 kg·hm⁻²·a⁻¹ in 2015 to 0.265 kg·hm⁻²·a⁻¹ in 2024, respectively. Under certain conditions, the sediment of a reservoir will release nitrogen, and factors such as temperature rise and water disturbance will accelerate the release of nitrogen from the sediment. The cumulative amount of nitrogen released from the overlying water is proportional to the disturbance rate. With the operation of the Danjiangkou Reservoir, nitrogen release from sediment may increase due to changes in water level and water flow velocity, leading to an increase in total nitrogen output. Due to the implementation of extremely strict soil erosion control and ecological protection projects such as returning farmland to forests, afforestation, and slope conversion in the Danjiangkou reservoir area and upstream areas, the sediment content in the incoming rivers has been greatly reduced. As soil particles with attached phosphorus are effectively intercepted, the phosphorus flux entering the water through surface runoff naturally decreases significantly. This is the most critical reason for the decrease in total phosphorus during the research period. During the research period, water yield and soil erosion exhibited the largest changes among all ecosystem services. Water yield increased from 579.45 m³ to 787.768 m³. Spatially, the northern part of the watershed showed an expanding low water yield area in 2024, corresponding to higher elevations, while the central region, influenced by farmland to forest and grassland conversion projects, became predominantly a high-water yield area. Soil erosion decreased from 318.707 t to 241.589 t. The decrease in soil erosion is the result of the combined effects of human governance, policy control, and natural factors. On the one hand, by implementing slope and channel management projects, vegetation restoration and ecological construction, optimizing agricultural production methods, and promoting comprehensive management of small watersheds, the soil and water conservation capacity has been significantly improved; on the other hand, the implementation of ecological protection policies and the strengthening of supervision and law enforcement have constrained human activities that damage the ecology, while natural factors such as changes in precipitation patterns may also assist in reducing erosion. Among them, proactive human intervention is the dominant factor leading to a significant decrease in erosion in the short term. Total carbon storage increased from 20.595 × 10⁹ t·km⁻² to 23.684 × 10⁹ t·km⁻², and the carbon sequestration capacity of ecosystems has been enhanced. The increase in vegetation coverage has enhanced the carbon storage of plant biomass; the soil carbon pool accumulates due to the input of dead branches and leaves, protective tillage, and soil and water conservation measures. The optimization of land use has enhanced carbon sequestration capacity, and these factors are the main reasons for the increase in total carbon storage. Habitat quality across the watershed improved significantly during the study period.

Overall, total nitrogen, soil erosion, total carbon, and water yield exhibited an increasing trend, while total phosphorus exhibited a declining trend. The changes in total nitrogen output, total phosphorus output, and habitat quality were relatively small, whereas the changes in water yield, soil erosion, and total carbon storage were more pronounced.

3.3. Analysis of Driving Mechanisms Between RSEI and Ecosystem Service Functions

Redundancy Analysis (RDA) and regression analysis were conducted on the 2015 and 2024 data for the entire watershed to examine the relationship between the Remote Sensing Ecological Index (RSEI) and six ecosystem service functions. These analyses aimed to evaluate the extent to which ecosystem service functions influence ecological environmental quality. The results are presented in Figure 4.

The RDA results shown in Figure 4 indicate that in 2015, the correlation coefficient between the RSEI and habitat quality is 0.62, suggesting that there is a significant positive correlation. In contrast, RSEI showed significant negative correlations with water yield, total nitrogen, total phosphorus, soil erosion, and total carbon storage, as reflected by the correlation coefficients are −0.55, −0.70, −0.68, −0.60, and −0.52, respectively.

Compared to 2015, in 2024, the correlation coefficient in RSEI and habitat quality, and water yield is 0.68, 0.42, exhibiting significant positive correlations. While the correlation coefficient in RSEI and total nitrogen, total phosphorus, soil erosion, and total carbon storage, respectively, are −0.72, −0.69, −0.65, and −0.58, maintaining significant negative correlations. From 2015 to 2024, good ecosystem service functions have been able to maintain and improve ecological quality, while a high-quality ecosystem has continuously provided sustained and effective ecosystem service functions. Specifically, RSEI consistently showed a positive correlation with habitat quality throughout this period. Ecosystem service functions in the upstream watershed of the Danjiangkou Reservoir, such as increased biodiversity, enhanced water purification capacity, and improved habitat quality, have contributed to the improvement of regional ecological quality. Beginning in early 2015, with the commencement of water diversion through the South-to-North Water Transfer Middle Route Project, water yield across the watershed gradually increased. This enhancement enriched biodiversity, strengthened water purification capacity, and reduced soil erosion, leading to a significant positive correlation between RSEI and water yield.

However, after the Danjiangkou Reservoir began diverting water northward, upstream watershed management authorities implemented ecological protection measures such as farmland conversion to forest and ecological water replenishment, which improved the water conservation capacity of the ecosystem. Nevertheless, after rainfall, pollutants such as nitrogen and phosphorus from fertilizers and pesticides increased in runoff entering water bodies, affecting river water quality. As a result, the trade-off relationships among water yield, total nitrogen, and total phosphorus ecosystem service functions have intensified.

3.4. Analysis of Coupling Mechanisms Between RSEI and Ecosystem Service Functions

The calculated coupling degree (CD) and coupling coordination degree (CCD) values between the RSEI and the five selected ecosystem service functions are shown in Figure 5.

As shown in Figure 5, during the study period, the coupling coordination degree between the RSEI and all ecosystem service functions across the entire region generally exhibited no instances of severe disorder. Among the six ecosystem service functions, total nitrogen output, water yield, and soil erosion showed significant changes, while the coupling coordination degrees of the other three ecosystem service functions improved to some extent, albeit with relatively smaller variations. Specifically, the area with a “good coordination” level for total nitrogen output expanded throughout the study region. The area with a “high coordination” level for water yield noticeably decreased in the eastern part of the region, whereas the area with a “basic coordination” level in the northern part of the study area increased substantially. Vegetation, as the predominant land use type in the northern region, performs essential regulatory functions in soil and water conservation as well as water quality purification. Correspondingly, the area with a “basic coordination” level for soil erosion increased markedly in the southeastern and northern parts of the study area.

4. Discussion

Ecological environmental quality is influenced not only by natural factors within a region but is also closely related to the interaction with human activities [57]. In the upstream watershed of the Danjiangkou Reservoir—a critical water source area for the South-to-North Water Diversion Project—this interaction is particularly pronounced. Water diversion requirements exert significant pressure on local ecosystems, influencing both water quality and quantity, while regional afforestation and soil conservation policies actively shape land cover and ecosystem stability. Therefore, the lower the disturbance from such integrated natural and anthropogenic factors, the higher the value of ecosystem service functions and the better the ecological quality [58]. This study analyzes the spatiotemporal variation characteristics of the regional RSEI and six key ecosystem service functions, aiming to clarify the driving mechanisms and coupling coordination relationships between ecosystem service functions and the RSEI at the watershed scale. By doing so, it addresses the limitation of relying on a single indicator to comprehensively and dynamically assess ecological benefits. The results offer practical guidance for ecological management and contribute to promoting the coordinated development of society, economy, and the environment in ecologically sensitive water source areas.

The value range of RSEI results is reported as 0.49 to 0.69 in regions surrounding large river basins; over 0.63 in areas with high vegetation coverage; and close to 0.80 in concentrated agricultural land such as farmland. In the upstream watershed of the Danjiangkou Reservoir, the main land use type in the northern part is forestland, while the middle and southern parts are dominated by farmland. The RSEI results in this study mainly range between 0.40 and 0.80, which is consistent with the value ranges in previous studies [59]. The central and southern parts of the watershed clearly show higher ecological quality than the northern regions. This suggests that slope and channel management projects, vegetation restoration and ecological construction, optimizing agricultural production methods, and promoting comprehensive management of small watersheds in the Danjiangkou Reservoir area have been most effective in improving ecological conditions in the central and southern areas.

During the study period, the area classified as having an “excellent” RSEI grade expanded substantially, demonstrating the effectiveness of ecological restoration efforts and land use management policies implemented in the region. Overall, the ecological environment has shown significant improvement, reflecting positive progress in ecosystem conservation and governance. Our results join a growing consensus from studies in critical water source and ecologically vulnerable regions, collectively underscoring that policy-driven ecological projects can significantly enhance ecological quality [60,61]. Similar improvements have been observed in other major water source basins such as the Miyun Reservoir, Shiyang River Basin, and Yangtze River, where vegetation restoration projects have also led to substantial increases in RSEI and vegetation coverage, alongside improved soil conservation capacity [62,63,64]. These consistent patterns across different basins indicate that the mechanisms and policy effects identified in this study have strong regional applicability and universality, demonstrating that integrated watershed management and ecological restoration can effectively improve ecological quality and ecosystem service functions at the watershed scale. However, our results also reveal that the areas categorized as “poor” and “relatively poor” have increased to some extent, indicating persistent ecological stress in localized zones. Similar phenomena have been observed in other watershed studies, where ecological governance improved overall conditions but failed to fully resolve problems such as vegetation degradation, soil erosion, and rocky desertification [65]. Such issues continue to weaken ecosystem service functions and pose risks to long-term ecological stability. Compared with earlier findings, this study highlights a stronger spatial heterogeneity of ecological quality, suggesting that broad-scale ecological governance measures are effective but insufficient on their own [66]. Therefore, targeted interventions—particularly in upstream watershed management, soil erosion control, and vegetation restoration—are essential for consolidating ecological achievements, narrowing regional disparities, and enhancing ecosystem resilience.

In 2015, the ecosystem within the study area was in a relatively degraded state, with a low vegetation coverage rate. Under this context, regions with higher RSEI values tended to overlap with areas characterized by superior vegetation conditions and lower soil erosion intensity. Consequently, a negative correlation was observed between RSEI and soil erosion during this period. By 2024, large-scale implementation of ecological restoration projects (e.g., reforestation and farmland-to-forest conversion) in mountainous regions of the study area had led to a significant increase in vegetation coverage. Notably, these vegetation-improved areas not only maintained higher RSEI values but also exhibited elevated simulated soil erosion intensity. This phenomenon can be attributed to two key factors: the inherent steep topographic features of mountainous areas and the enhanced rainfall infiltration capacity induced by vegetation restoration—both of which collectively promote the redistribution of surface runoff. Thus, the positive correlation between RSEI and soil erosion detected in 2024 does not indicate ecological degradation; instead, it reflects a critical shift in the dominant controlling factors of soil erosion—from being constrained primarily by vegetation scarcity (in 2015) to being regulated by topographic and hydrological processes (in 2024). Simultaneously, the water production capacity of areas with high RSEI values showed a decreasing trend. This reduction is not a sign of ecological deterioration but rather a direct consequence of improved vegetation and soil properties: the enhanced water retention capacity of the ecosystem (driven by vegetation root systems and improved soil structure) effectively reduces the volume of simulated surface runoff, thereby lowering the overall water production.

From the spatial distribution pattern of RSEI, slope and channel management projects, vegetation restoration and ecological construction, optimizing agricultural production methods were implemented in the central and southern parts of the research area have significantly improved the regional ecological environment quality. Meanwhile, these projects increased vegetation coverage, enhanced regional carbon sequestration capacity, delayed surface runoff, effectively suppressed soil erosion, and reduced soil erosion quantity, which further lowered total nitrogen and phosphorus output. The coupling coordination degree improvement area of soil erosion mainly lies in the central region, matching well with the coupling coordination improvement areas of total nitrogen and phosphorus input.

After reservoir water diversion, watershed management departments implemented ecological regulation and water replenishment measures according to water resource demands and water quality conditions [67]. These measures improved the reasonable flow and self-purification capacity of water bodies, helping reduce total nitrogen output, which explains the significant expansion of the “good coordination” level area. Additionally, increased vegetation coverage significantly enhances evapotranspiration; higher plant water demand reduces soil moisture content, which would typically cause a reduction in water yield [68]. However, during the study period, water yield increased in the research area, while the coupling coordination degree between RSEI and water yield decreased. The water diversion project of the South-to-North Water Transfer increased overall regional water yield but caused uneven spatial distribution-water yield rose in some areas and declined in others-negatively impacting the coordination degree between RSEI and water yield.

Coupling coordination analysis results indicate that improvements in ecosystem quality have, to some extent, promoted the overall enhancement of ecosystem service functions but have also revealed competitive relationships among service functions in certain local areas. Future ecological management must prioritize the protection of ecological quality while scientifically balancing the supply and demand of ecosystem service functions. This requires optimizing human–land interactions and fostering a virtuous cycle within the ecosystem to ensure long-term sustainability.

5. Conclusions

This study focuses on the Xichuan Reservoir area of the Danjiangkou Reservoir, a crucial water source for the middle route of the South-to-North Water Transfer Project. By employing a framework combining the RSEI and six ecosystem service functions, this research investigates the underlying drivers, synergistic-trade-off relationships, and coupling coordination degree among them. The main conclusions are as follows:

During the study period, the proportion of the watershed classified as “excellent” in RSEI grades increased significantly in the upstream area of the Danjiangkou Reservoir. This improvement can be attributed to a series of ecological management projects in the central and southern regions of the basin, including agricultural non-point source pollution control, aquatic ecological restoration, and the conversion of cropland to forest. These measures enhanced vegetation coverage, strengthened regional carbon sequestration capacity, mitigated surface runoff, and effectively curbed soil erosion. As a result, total nitrogen and phosphorus inputs were further reduced, leading to a substantial improvement in ecological environmental quality.

Among the six ecosystem service functions studied, total nitrogen, total phosphorus, soil erosion and carbon storage displayed significant negative correlations with RSEI, whereas habitat quality showed a distinct positive relationship. Water yield changed from a negative correlation with RSEI in 2015 to a positive correlation in 2024. The water diversion from the South-to-North Water Transfer Project increased water yield in the watershed, which enriched biodiversity, enhanced water purification capacity, and reduced soil erosion occurrence. Synergistic relationships among total carbon storage, habitat quality, and soil erosion were strengthened, while trade-off relationships among water yield, total nitrogen, and total phosphorus became more prominent.

Throughout the research period, the coupling coordination degree between RSEI and all ecosystem service function values exhibited no severe imbalance state across the entire region. Notable changes were observed in total nitrogen, soil erosion, and water yield. Improvements in water self-purification capacity, enhanced surface rainfall interception, and better soil and water conservation contributed to the increase in coupling coordination degree between RSEI and total nitrogen output. The uneven spatial distribution of water resources was the primary factor driving the decline in the coupling coordination degree between RSEI and water yield. Increased vegetation coverage delayed surface runoff and effectively suppressed soil erosion, leading to the rise in coupling coordination degree between RSEI and soil erosion.