Scenario Simulation of Ecosystem Services Based on Land Use/Land Cover Change in the Bailong River Basin, in China

Abstract

Land use/land cover changes (LUCCs) significantly reshape ecosystem services (ESs) within the framework of climate change. Studying LUCC and its impact on ESs is crucial for a comprehensive understanding of the impact of human activities on ecosystems. The InVEST model coupled with the predicted land use data were used to analyze the spatiotemporal characteristics of four ESs (soil conservation (SC), water yield (WY), carbon storage (CS), and habitat quality (HQ)) under three scenarios from 2040 to 2100 and quantified trade-offs/synergies and bundles of these ESs within the Bailong River Basin (BRB). The results indicated that (1) under the SSP1-2.6 scenario, there is an anticipated increase in forestland, a concurrent decrease in grassland, farmland, and built-up land, and an enhancement in four ESs from 2040 to 2100. The forestland and farmland in the SSP2-4.5 scenario showed a gradual decrease, with an expansion of grassland and built-up land. Except for HQ, the other three ESs were reduced. Both forestland and grassland decreased. Built-up land and farmland increased, and ESs decreased significantly under the SSP5-8.5 scenario. (2) Synergistic effects were identified among the ESs, with the most pronounced synergy observed between CS and HQ. Spatially, six pairs of ESs under the SSP1-2.6 scenario showed synergistic effects. Under the SSP2-4.5 and SSP5-8.5 scenarios, most of the ESs present trade-off effects. (3) The characterization of ES bundles revealed that the balanced enhancement of the four ESs predominantly occurred in the southern region of the basin. Among the scenarios, SSP1-2.6 had the highest representation, followed by the SSP2-4.5, while the SSP5-8.5 had the lowest proportion. The findings facilitate the sustainable and balanced development of diverse ESs and offer theoretical and technical insights for devising spatial regulation policies and ecosystem-based management strategies.

1. Introduction

The ES framework acts as an essential bridge between ecosystems and socioeconomic systems, playing a critical role in safeguarding human well-being [1,2]. In the last century, factors such as global climate change [3], environmental pollution [4], ecological destruction [5], resource depletion [6], and LUCC [7] have profoundly affected ecosystem functioning, resulting in considerable modifications to ESs. According to the MEA report, over 60% of the world’s ecosystems are either degrading or have already degraded. The ensuing degradation poses a significant threat to the integrity of regional ecosystems and jeopardizes the feasibility of sustainable socioeconomic development [8,9]. This includes a decline in vegetation productivity, habitat fragmentation, water scarcity [10], and soil and water erosion [11]. Therefore, assessing and identifying the spatial patterns of ESs can offer significant support for their management and inform decision-making processes, thereby helping to mitigate the adverse effects of the threats to these services [12].

In recent years, progress in technologies has catalyzed a significant shift toward quantitative approaches in ecological and environmental studies [13,14]. The availability of diverse datasets, enabled by advanced technologies, lays a solid groundwork for the quantitative assessment of ESs [15,16]. Additionally, the evolution of diverse evaluation techniques has empowered the scientific community to harness existing knowledge for assessing ESs across a range of spatial scales and within multiple planning sectors [17]. The InVEST model, celebrated for its sophisticated capabilities, is particularly adept at visualizing and quantifying [18]. Arunyawat et al. (2016) analyzed the impact of LUCC on ESs based on the InVEST model. Utilizing this tool, they were able to map and quantify various ES functions in Northern Thailand, including SC, WY, CS, and HQ [19]. Yang et al. (2021) applied the InVEST model to assess HQ and analyzed the effects of LUCC on HQ in the Taihang Mountains [20]. Additionally, Li et al. (2021) employed the InVEST model to assess WY in the Danjiang River Basin, thereby assessing the basin’s capacity for water conservation [21]. In a subsequent study, Yang et al. (2023) used the InVEST model to analyze the impact of ecological restoration projects in the Yellow River Delta on changes in HQ and CS [22]. Huang et al. (2024) leveraged the InVEST model to project the spatiotemporal dynamics of WY, HQ, and SC within oases and urban clusters in Xinjiang, China [23]. The InVEST model has been widely recognized for evaluating WY, CS, HQ, and SC. Water resources, being a critical component, are essential in the mitigation of water scarcity and pollution, as well as in maintaining the vitality of aquatic ecosystems. SC is vital for preserving land productivity and fertility, preventing erosion and desertification, and maintaining the balance of soil and water, as well as the surface hydrological cycle. HQ contributes to the healthy development of the ecosystem environment. CS plays an important role in mitigating climate change and protecting biodiversity. These ESs form the foundation of environmental sustainability and societal well-being [24]. Therefore, we quantitatively evaluated the four key ESs (WY, SC, HQ, and CS) of BRB using the InVEST model.

Human activities are a major cause of ecosystem degradation. Among the myriad factors that influence ESs, LUCC emerges as the most dynamic and dominant, with the power to directly alter ecosystems, thereby affecting the sustainability and change in ESs [25,26]. Therefore, the potential negative impact of LUCC on ESs has attracted considerable attention [27,28]. Research does indicate that LUCC is affected by climate change and socioeconomic development [29,30]. The most recent CMIP6 offers a comprehensive framework for exploring various future development scenarios. This is achieved by integrating shared socioeconomic pathways (SSPs) with representative concentration pathways (RCPs) [2,31]. An escalating trend in research is adopting SSP-RCP scenarios to foresee shifts in future land use requirements. In a prominent study, Dong et al. (2018) employed the GCAM to project future land use demands across China and to simulate the spatial distribution of these demands [32]. Recently, many new perspectives have emerged to evaluate the spatiotemporal changes of ESs in SSP-RCP scenarios. Qiu et al. (2023) analyzed the spatiotemporal distribution of LUCC and ESs in Central Asia under different SSP-RCP scenarios using the LUH2 dataset [33]. Guo et al. (2024) [34] and Li et al. (2023) [35] assessed CS in SSP-RCP scenarios using LUH2 data Nonetheless, the majority of these investigations concentrate on LUCC at the national or global scales. In addition, some scholars have predicted the ecosystem of Shenzhen in 2030 based on LUCC [36]. Wu et al. (2022) utilized the FLUS model to assess HQ, integrating a framework that considers climate change and development patterns [37].

However, the existing body of research has predominantly concentrated on the alterations in ESs as a result of LUCC under projected future climate scenarios. However, future changes in multiple ecological services under different LUCCs and climate scenarios, especially in ecologically fragile areas in China, are still unknown. Moreover, assessing ESs in isolation across various scenarios is insufficient to offer tangible directives for environmental management. It is imperative to integrate the interrelationships among ESs into spatial planning and to bolster the nexus between ESs and ecological governance [38]. ESs are not discrete entities. Rather, they are interconnected, with alterations in one ES potentially diminishing or augmenting the capacity of another [39,40]. These interlinkages are characterized as trade-offs and synergies [41]. Scholars acknowledge the significance of ecosystem interdependencies, with extensive research dedicated to their quantitative assessment [42]. However, integrating this analysis into land use planning poses an ongoing challenge. Neglecting the interconnectivity of ESs can impede the enhancement of ecological benefits and may lead to the degradation of ecosystems within land use practices [43,44]. Therefore, clarifying the complex relationships between ecosystems under the background of land use change can promote the healthy development of ecosystems.

To address the above limitations, this research opts for the BRB as the focal study area. The BRB is a pivotal region for water conservation and acts as an essential ecological barrier, with significant water conservation services and high HQ. The basin features highly varied terrain, severe soil erosion, and improvements in SC services that contribute to the ecological security and stability of the watershed. However, the basin is characterized by rugged mountains, deep valleys, and limited land resources. Land reclamation and policy implementations, such as the “grain for green”, may also affect changes in ESs and their interrelationships under the broader context of regional climate change. To this end, considering the ecological and environmental attributes of the research area, as well as the features of the InVEST model, four typical services, namely CS, SC, WY, and HQ, were selected for in-depth analysis. The ES functions under future land use scenarios in the BRB were assessed, and their trade-offs/synergies and bundling were discussed, providing scientific support for regional environmental management, decision-making, and sustainable development. Section 2 offers a detailed description of the study area, the data and their sources, the method used for land use simulation, the method for assessing ESs, and the methods for analyzing ES trade-offs/synergies and identifying ES bundles. Section 3 presents the research findings. Section 4 provides a discussion on the contributions, policy implications, research limitations, and future prospects. Section 5 summarizes the key conclusions. Our objectives are 1) to forecast the forthcoming LUCC trajectories in the BRB from 2040 to 2100; 2) to evaluate the variations in ESs across diverse scenarios; 3) to assess the interplay among ecosystems within the context of various scenarios; 4) to identify ES bundles and provide relevant recommendations for ES management.

2. Materials and Methods

2.1. Study Area

The BRB, defined by the geographic coordinates 32°36′ N to 34°24′ N and 103°00′ E to 105°30′ E, covers an area of 17,756 km². The BRB primarily encompasses Diebu County, Zhouqu County, Wen County, Tanchang County, and Wudu District (Figure 1). The basin’s topography is marked by elevated terrain in the northwest and lower elevations in the southeast, ranging in altitude from 602 to 5280 m. The predominant landform types within the basin include plateau mountains, river valleys, and loess landforms. There is an obvious vertical change trend in the domain. The valley is warm and rainless, and the high mountain area is cold and rainy. Due to the influence of tectonic movement, regional stratum characteristics, and climate, the soil layer in the watershed is shallow, the site conditions are poor, and the water and soil losses on both sides of the river are serious [45]. With the development of the regional society and economy, forestry resources have been developed and utilized in large quantities in basins, grasslands have been overgrazed, and the regional ecological environment has been seriously degraded.

2.2. Data Source and Pre-Processing

This study collected multisource datasets to accommodate the diverse data demands of various models, encompassing spreadsheet data, vector data, and raster data across a spectrum of spatial and temporal resolutions. (

Table 1. Overview of the Key Datasets.

Data Type	Application	Data Source	Data Format	Spatial Resolution
Land use/land cover	Simulation of land use, HQ, WY, SC, CS	The land use data are the second (2010) and third (2020) national land use survey data collected from the Natural Resources Bureau of the five counties and districts	Shpfile	/
Precipitation	WY, SC	National Earth System Science Data Center, National Science and Technology Infrastructure of China	Raster	1 km
Temperature	WY	National Earth System Science Data Center, National Science and Technology Infrastructure of China	Raster	1 km
Evapotranspiration	WY	National Ecosystem Science Data Center	Raster	1 km
Root depth, Soil depth, and soil texture	WY, SC, Simulation of land use	China soil map-based harmonized world soil database (HWSD) (v1.1)	Raster	1 km
Digital elevation model(DEM)	WY, SC	Geospatial Data Cloud	Raster	30 m
Slope	Simulation of land use	Slope data is calculated using DEM data.	Raster	30 m
Carbon density	CS	A dataset of carbon density in Chinese terrestrial ecosystems (2010s)	Spreadsheet	/
Residential point and road data	Simulation of land use, HQ	National Basic Geographic Information Center	Raster	Shpfile
Population density (POP)	Simulation of land use, HQ	Resource and Environmental Science Data Center	Raster	1 km
Gross domestic product (GDP)	Simulation of land use	Resource and Environmental Science Data Center	Raster	1 km
Soil type	Simulation of land use	Resource and Environment Science and Data Center	Raster	1 km
Lithology	Simulation of land use	1:200,000 regional geological map	Shpfile	/
Water system data	Simulation of land use	National Basic Geographic Information Center	Shpfile	/
Nighttime light data	Simulation of land use	Earth observation group	Raster	500 m
Precipitation and Temperature Data of BCC-CSM2-MR Climate Model	Simulation of land use	CMIP6	Raster	110 km
GDP and POP under different Shared Socioeconomic Pathways (SSPs)	Simulation of land use	Scientific Data Bank	Raster	55 km

). To address the discrepancies in land use data classifications between the two surveys, we performed a reclassification of the datasets from both periods to align their categorizations. All the data underwent format conversion, clipping, projection, and resampling using Python. The sub-basins were delineated using DEM data with the ArcSWAT hydrological analysis tool (Figure 2).

2.3. Method

Our research framework is shown in Figure 3. Initially, based on socioeconomic and climate change data in the SSP-RCP scenario, the PLUS model was used to predict land use from 2040 to 2100. Next, the InVEST model is employed to calculate four types of ESs using land use and climate data. Then, the spatiotemporal distribution characteristics of ES trade-offs/synergies from 2040 to 2100 were revealed. Finally, the SOM method was used to determine the ES bundles. The findings offer significant guidance for ecosystem management and the formulation of implementation plans.

2.3.1. Simulation of Future Land Use

PLUS Model

We used the PLUS model to simulate land use in the BRB. The PLUS model is a patch-based land use change simulation model [46]. Liang et al. (2022) explained the basic principles and formulas of the PLUS model [46].

Model Validation

Land use is categorized into six types, namely farmland, forestland, grassland, water bodies, built-up land, and unused land, along with 13 subcategories in the BRB. Nineteen driving factors influencing land use change were selected, including socioeconomic, DEM, transportation, and environmental factors. Subsequently, based on the land use data from 2010, a simulation of the land use pattern in 2020 was conducted using transition matrices, domain weights, and land demand. Then, the Kappa statistical method was used to validate the simulation results:

$$\mathrm{Kappa}={\displaystyle \frac{{\mathrm{P}}_0-{\mathrm{P}}_{\mathrm{c}}}{{\mathrm{P}}_{\mathrm{p}}-{\mathrm{P}}_{\mathrm{c}}}}$$

(1)

where ${\mathrm{P}}_0$ denotes the proportion of correct simulations, ${\mathrm{P}}_{\mathrm{c}}$ denotes the correct proportion of the model under random conditions, and ${\mathrm{P}}_{\mathrm{p}}$ denotes the proportion of correct simulations under ideal classification conditions. The verification results indicate that the Kappa coefficient is 0.8881. Consequently, the simulation results are deemed reliable and can be utilized for land use projections in the years 2040, 2060, 2080, and 2100 [47,48].

Multi-Scenario Setting

We identified three representative SSP-RCP scenarios within the CMIP6 framework and integrated them with plausible future developmental trajectories to formulate three distinct scenarios: (1) SSP1-2.6 and the ecological protection scenario (SSP126-EP), where ecological conservation is paramount, with socioeconomic development predicated on ecological preservation, a reduction in fossil fuel usage, and the attainment of sustainable green growth; (2) SSP2-4.5 and natural development scenario (SSP245-ND), based on the original land use development evolution; and (3) SSP5-8.5 and economic growth scenario (SSP585-EG), which is dominated by economic development and has the main goals of putting a large amount of fossil fuels into use and economic improvement. We used four driving factors (GDP, POP, temperature, and precipitation) to estimate LUCC under the SSP-RCP scenario [49].

2.3.2. ESs Assessment

SC

SC is calculated through the sediment delivery ratio module in the InVEST model [50]. The specific calculation of the model is as follows:

$$\mathrm{S}\mathrm{E}\mathrm{D}\mathrm{R}\mathrm{E}\mathrm{T}=\mathrm{P}\mathrm{K}\mathrm{L}\mathrm{S}-\mathrm{U}\mathrm{S}\mathrm{L}\mathrm{E}$$

(2)

$$\mathrm{P}\mathrm{K}\mathrm{L}\mathrm{S}=\mathrm{R}·\mathrm{K}·\mathrm{L}\mathrm{S}$$

(3)

$$\mathrm{U}\mathrm{S}\mathrm{L}\mathrm{E}=\mathrm{R}·\mathrm{K}·\mathrm{L}\mathrm{S}·\mathrm{C}·\mathrm{P}$$

(4)

SEDRET denotes the amount of soil conservation. PKLS denotes the potential soil erosion amount in tons (t). USLE denotes the actual soil erosion amount in tons (t). LS denotes the slope length and slope factor. R and K represent the precipitation and soil erosion factors, respectively. The magnitudes of the C and P factors are between 0 and 1, and the values are assigned by consulting the relevant literature in similar research areas and combining the special conditions of the research area [38] (

Table 2. C and P values for different land use types.

	Farmland	Forestland	Grassland	Water Bodies	Built-Up Land	Unused Land
C	0.20	0.05	0.30	0.00	0.00	1.00
P	0.15	1.00	1.00	0.00	1.00	1.00

).

WY

We employed the InVEST model to calculate WY. The calculation formula within the model is presented below:

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

(5)

$${\displaystyle \frac{AET\left(x\right)}{P\left(x\right)}}=1+{\displaystyle \frac{PET\left(x\right)}{P\left(x\right)}}-{⌊{1+{\displaystyle \frac{PET\left(x\right)}{P\left(x\right)}}}^{\omega \left(x\right)}⌋}^{1/\omega \left(x\right)}$$

(6)

$$\omega \left(x\right)=\mathrm{Z}{\displaystyle \frac{AWC\left(x\right)}{P\left(x\right)}}+1.25$$

(7)

Y(x) denotes the water production (mm). AET(x) represents the actual evapotranspiration. PET(x) represents the potential evapotranspiration. P(x) denotes the average monthly precipitation. ω(x) is the physical parameters of natural climate and soil properties. AWC(x) represents derived from the plant-available water capacity (PAWC), the maximum rooting depth of the soil, and the minimum value of the vegetation root depth. The Z value ranges from 1 to 10 and denotes the distribution of precipitation in the watershed, as well as other hydrological and geological characteristics, without any units.

$$\begin{array}{c}\mathrm{P}\mathrm{W}\mathrm{A}\mathrm{C}=54.509-0.132\times Sand\%-0.003\times {\left(Sand\%\right)}^2-0.055\times Silt\%-0.006\times {(Silt\%)}^2\\ -0.738\times Clay\%+0.007\times {(Clay\%)}^2-2.688\times \mathrm{C}\mathrm{\%}+0.501\times {(C\%)}^2\end{array}$$

(8)

Specifically, Sand%, Silt%, Clay%, and C% represent the respective percentage compositions of sand, silt, clay, and organic matter in the soil.

CS

This study used the InVEST model to calculate CS. The calculation formula is:

$$C=C_{\mathit{above}}+C_{\mathit{below}}+C_{\mathit{soil}}+C_{\mathit{dead}}$$

(9)

C_above denotes the aboveground biomass carbon storage in tons (t). C_below signifies the underground biomass carbon storage in tons (t). C_soil represents the soil organic carbon storage in tons (t). C_dead indicates the carbon storage from dead organic matter in tons (t). C is regional carbon storage (t). The carbon density data come from the collection and arrangement of the results, such as [51] and the reference database [52].

HQ

The InVEST model assesses the status of biodiversity through the magnitude of HQ. The calculation formula is:

$$Q_{\mathrm{x}\mathrm{j}}=H_j\left[1-\left({\displaystyle \frac{D_{xj}^z}{D_{xj}^z+k^z}}\right)\right]$$

(10)

In the formula, Q_xj is the HQ (dimensionless) of grid unit x in land type j; H_j denotes the habitat attribute or suitability of land type j; D_xj signifies the degree of habitat degradation for grid unit x in land type j; k is the half-saturation coefficient; and z is the default normalization constant of the model (z = 2.5).

$$D_{xj}={\sum }_{r=1}^R{\sum }_{y=1}^{Y_r}\left({\displaystyle \frac{w_r}{\sum _{r=1}^R{w}_r}}\right)r_y{i}_{rxy}{\beta }_x{S}_{jr}$$

(11)

where $w_r$ is the weight of each threat factor, $r_y$ is the intensity of the threat factor, ${\beta }_x$ is the anti-interference level of the habitat, and $S_{jr}$ is the relative sensitivity of different habitats to different threat factors. The habitat degradation degree ranges from 0 to 1, with $D_{xj}$ values close to 1 indicating a high habitat degradation degree [53]. The parameters are detailed in

Table 3. Parameters of threat factors.

Threat Factor	MAX_DIST(km)	Weight	Decay Type
National and Provincial Highways	2.5	0.6	Linear
County and township roads	0.5	0.5	Linear
Residential area	2.5	0.4	Exponential
Urban	6.0	0.8	Exponential
Farmland	1.5	0.6	Exponential
Population density	3.5	0.3	Exponential

and

Table 4. Sensitivity parameters of various land types to habitat threat factors.

Land Use Type	Habitat Suitability	National and Provincial Highways	County and Township Roads	Residential Area	Urban	Farmland	Population Density
Farmland	0.30	0.50	0.60	0.40	0.50	0.00	0.80
Forestland	0.90	0.90	0.70	0.50	0.60	0.30	0.70
Grassland	0.60	0.70	0.50	0.20	0.30	0.50	0.50
Water bodies	0.90	0.75	0.65	0.70	0.80	0.10	0.50
Built-up land	0.00	0.00	0.00	0.00	0.00	0.00	0.85
Unused land	0.01	0.20	0.20	0.10	0.10	0.10	0.30

.

2.3.3. Trade-Offs/Synergies

This article uses a Spearman correlation analysis method to determine the trade-offs/synergies between ESs. The negative correlation between the ESs represents a trade-off, while the positive correlation represents a synergistic effect. We conducted a Spearman correlation analysis using R 4.4.1 software for the period from 2040 to 2100.

2.3.4. Identification of Bundle

“Bundle” refers to the spatially recurrent aggregation of ESs [54], which can identify various patterns of ES aggregation. We implemented self-organizing maps (SOM) to determine ES bundles at the sub-basin scale. This method assigns each sub-basin to an ES bundle based on the spatial similarity of the four ESs.

3. Results

3.1. LUCC Characteristics in the BRB from 2040 to 2100

Our predicted land use in the BRB from 2040 to 2100 is shown in Figure 4, Figure 5 and Figure 6. From 2040 to 2100, the proportion of forest land will always be the largest (>60%), mainly distributed in the western and southern regions of the basin, with the widest distribution in Diebu County. Grassland (accounting for 12.39% and 17.81% respectively) was distributed on the slopes on both sides of the ravine. Both farmland and built-up land (including industrial land, common land, urban land, village land, township land, and traffic land) are concentrated in flat areas on both sides of the Bailong River and its secondary tributaries. It is worth noting that Wudu District has the largest built-up land area, while Tunchang County has the widest distribution of farmland.

Under the SSP126-EP scenario, forestland increased by 3.83% from 2040 to 2100, while grassland, farmland, and built-up land decreased by 14.47%, 13.41%, and 23.10%, respectively. Similar to the SSP126-EP scenario, farmland decreases in the SSP245-ND scenario, whereas grassland and built-up land show an increasing trend. Forestland shows a trend of gradual increase and then sudden decrease. Specifically, farmland and forestland decreased by 13.41% and 1.32%, respectively, while grassland and built-up land increased by 14.45% and 24.38%, respectively. In 2040–2100, the forestland area in the SSP585-EG scenario is significantly lower than that in the SSP126-EP and SSP245-ND scenarios. The forestland and grassland decreased by 5.03% and 5.47%, respectively. However, farmland and built-up land increased by 22.42% and 25.81%, respectively.

3.2. Spatiotemporal Distribution of ESs

The distribution of four types of ESs in the BRB shows significant spatial differences. Specifically, the areas with high SC values are mainly located in the Wen County and southeastern regions of the Wudu District. These areas mostly belong to stony mountainous areas or nature reserves with less surface disturbance. The areas with low SC values are distributed on both sides of the Bailong River in the Zhouqu–Wudu–Wen County section where human activities are relatively frequent and industry and agriculture are relatively developed. That is, the low-value SC areas are consistent with areas with high amounts of human activity and areas prone to landslides and debris flows. In the SSP1-2.6 scenario, the basin SC shows an upward trend. Its growth area is mainly concentrated on both sides of the upper reaches of the Bailong River, and the total SC volume increased by 10.38%. In contrast, SC shows a downward trend under the SSP2-4.5 and SSP5-8.5 scenarios, and the declining areas are mainly concentrated between Wudu District and Wen County (Figure 7).

Spatially, the distribution pattern of HQ does not change much. Areas with a higher HQ are mainly distributed in nature reserves and forestry management reserves (Figure 8). These include the Baishuijiang National Nature Reserve in Wen County and the Dala, Axia, and Doer Nature Reserves and forestry development zones in Diebu County. The areas with a low HQ are mainly concentrated in river valleys, urban and rural settlements, farming areas, bare land, and low-cover grasslands, where human activities are relatively frequent. The HQ under the SSP1-2.6 scenario is the most elevated, demonstrating a positive trend of escalation, with the most significant increase (3.27%). The HQ of the SSP2-4.5 scenario comes second, also showing increasing change characteristics. The SSP5-8.5 scenario has the worst HQ and shows a significant decreasing trend (−3.19%).

The spatial distribution of the CS across the three scenarios within the BRB exhibits minimal variation, but there are still local changes. Regions exhibiting significantly elevated CS are primarily located in the forested mountainous areas, including the southern mountainous forests of Tanchang County, the high mountainous regions on the southern bank of the Baishui River in Wen County, and the upper to middle reaches of the Boyu River and Gongba River. Low CS value areas are located in agricultural planting areas, towns, high-altitude mountainous areas in Northern Diebu, major landslide areas, and alpine bare-rock areas (Figure 9). This result indicates that forests are the main vegetation type for carbon sequestration, and vegetation coverage has a significant impact on the carbon sequestration function of ecosystems. Within the framework of the SSP1-2.6 scenario, CS within the basin exhibited a significant upward trajectory from 2040 to 2100, with an increase of 3.27 × 10⁶ t, an increase of 12.67%, and the total CS in the entire basin reached 26.06 × 10⁷ t in 2100. The regions experiencing the most significant growth are predominantly located within the Baishui River Basin and the southwestern sector of Wudu District. Under scenarios SSP2-4.5 and SSP5-8.5, CS within the basin exhibited a declining trend over the period from 2040 to 2100, decreasing by 1.64 × 10⁶ t (6.43%) and 5.73 × 10⁶ t (22.86%) respectively. By 2100, the total CS in the entire basin reached 25.33 × 10⁷ t and 24.50 × 10⁷ t, respectively. The areas with reduced CS are primarily concentrated in the northern regions of Wen County and Wudu District.

The spatial distribution of WY is shown in Figure 10. Under the SSP1-2.6 scenario, the WY is the smallest and shows increasing change characteristics. In the SSP2-4.5 scenario, WY is lower compared to the SSP5-8.5 scenario, with a downward trend observed in both. Under the SSP1-2.6 scenario, the implementation of ecological conservation and restoration projects has successfully mitigated ecological degradation and led to a partial recovery, thereby enhancing the basin’s water production capacity. Within the SSP5-8.5 scenario, extensive regions of grasslands and forests are repurposed as agricultural and urban land, leading to persistent ecosystem degradation, and WY gradually decreases. The spatiotemporal distribution of WY within the basin is intricately linked to precipitation patterns and land use alterations, as evidenced by various studies.

3.3. Spatiotemporal Distribution of Trade-Offs/Synergies Between ES Pairs

We performed a Pearson correlation using SPSS20.0 software, and the results are shown in

Table 5. Pearson correlation in the SSP1-2.6 scenario.

	HQ126_2040	HQ126_2060	HQ126_2080	HQ126_2100
SC126_2040	0.449 **	0.444 **	0.439 **	0.438 **
SC126_2060	0.446 **	0.440 **	0.435 **	0.433 **
SC126_2080	0.442 **	0.439 **	0.434 **	0.433 **
SC126_2100	0.442 **	0.440 **	0.435 **	0.434 **
CS126_2040	0.972 **	0.963 **	0.955 **	0.954 **
CS126_2060	0.964 **	0.970 **	0.966 **	0.966 **
CS126_2080	0.957 **	0.967 **	0.968 **	0.968 **
CS126_2100	0.955 **	0.966 **	0.967 **	0.968 **
WY126_2040	0.348 **	0.343 **	0.346 **	0.345 **
WY126_2060	0.343 **	0.340 **	0.342 **	0.341 **
WY126_2080	0.355 **	0.352 **	0.353 **	0.353 **
WY126_2100	0.360 **	0.361 **	0.362 **	0.362 **
	SC126_2040	SC126_2060	SC126_2080	SC126_2100
CS126_2040	0.406 **	0.407 **	0.403 **	0.408 **
CS126_2060	0.396 **	0.396 **	0.395 **	0.401 **
CS126_2080	0.389 **	0.389 **	0.388 **	0.394 **
CS126_2100	0.387 **	0.386 **	0.385 **	0.391 **
WY126_2040	0.316 **	0.303 **	0.312 **	0.318 **
WY126_2060	0.334 **	0.324 **	0.335 **	0.348 **
WY126_2080	0.340 **	0.335 **	0.349 **	0.371 **
WY126_2100	0.350 **	0.352 **	0.372 **	0.415 **
	CS126_2040	CS126_2060	CS126_2080	CS126_2100
WY126_2040	0.257 **	0.244 **	0.247 **	0.245 **
WY126_2060	0.256 **	0.243 **	0.245 **	0.243 **
WY126_2080	0.283 **	0.271 **	0.271 **	0.270 **
WY126_21000	0.309 **	0.302 **	0.301 **	0.300 **

Note: ** Significant correlation at the 0.01 level (two-sided).

,

Table 6. Pearson correlation in the SSP2-4.5 scenario.

	HQ245_2040	HQ245_2060	HQ245_2080	HQ245_2100
SC245_2040	0.451 **	0.447 **	0.445 **	0.438 **
SC245_2060	0.453 **	0.450 **	0.450 **	0.439 **
SC245_2080	0.444 **	0.441 **	0.440 **	0.430 **
SC245_2100	0.445 **	0.440 **	0.438 **	0.431 **
CS245_2040	0.959 **	0.970 **	0.972 **	0.931 **
CS245_2060	0.973 **	0.971 **	0.968 **	0.939 **
CS245_2080	0.948 **	0.963 **	0.969 **	0.923 **
CS245_2100	0.939 **	0.956 **	0.963 **	0.915 **
WY245_2040	0.344 **	0.325 **	0.314 **	0.345 **
WY245_2060	0.365 **	0.354 **	0.349 **	0.368 **
WY245_2080	0.327 **	0.314 **	0.308 **	0.328 **
WY245_2100	0.324 **	0.303 **	0.292 **	0.324 **
	SC245_2040	SC245_2060	SC245_2080	SC245_2100
CS245_2040	0.408 **	0.416 **	0.406 **	0.402 **
CS245_2060	0.417 **	0.424 **	0.414 **	0.411 **
CS245_2080	0.402 **	0.411 **	0.401 **	0.396 **
CS245_2100	0.397 **	0.408 **	0.397 **	0.391 **
WY245_2040	0.316 **	0.324 **	0.310 **	0.306 **
WY245_2060	0.360 **	0.385 **	0.364 **	0.348 **
WY245_2080	0.323 **	0.343 **	0.327 **	0.313 **
WY245_2100	0.311 **	0.316 **	0.304 **	0.302 **
	CS245_2040	CS245_2060	CS245_2080	CS245_2100
WY245_2040	0.233 **	0.260 **	0.216 **	0.206 **
WY245_2060	0.277 **	0.296 **	0.267 **	0.262 **
WY245_2080	0.236 **	0.257 **	0.224 **	0.218 **
WY245_2100	0.212 **	0.240 **	0.194 **	0.183 **

Note: ** Significant correlation at the 0.01 level (two-sided).

and

Table 7. Pearson correlation in the SSP5-8.5 scenario.

	HQ585_2040	HQ585_2060	HQ585_2080	HQ585_2100
SC585_2040	0.425 **	0.426 **	0.423 **	0.421 **
SC585_2060	0.415 **	0.416 **	0.414 **	0.411 **
SC585_2080	0.421 **	0.421 **	0.420 **	0.417 **
SC585_2100	0.434 **	0.434 **	0.432 **	0.428 **
CS585_2040	0.968 **	0.976 **	0.958 **	0.948 **
CS585_2060	0.975 **	0.976 **	0.970 **	0.964 **
CS585_2080	0.976 **	0.971 **	0.975 **	0.971 **
CS585_2100	0.973 **	0.965 **	0.975 **	0.974 **
WY585_2040	0.364 **	0.371 **	0.359 **	0.357 **
WY585_2060	0.367 **	0.376 **	0.360 **	0.356 **
WY585_2080	0.333 **	0.341 **	0.327 **	0.325 **
WY585_2100	0.345 **	0.354 **	0.340 **	0.338 **
	SC585_2040	SC585_2060	SC585_2080	SC585_2100
CS585_2040	0.412 **	0.402 **	0.406 **	0.415 **
CS585_2060	0.415 **	0.406 **	0.410 **	0.419 **
CS585_2080	0.416 **	0.408 **	0.412 **	0.420 **
CS585_2100	0.416 **	0.408 **	0.411 **	0.419 **
WY585_2040	0.410 **	0.390 **	0.375 **	0.366 **
WY585_2060	0.297 **	0.280 **	0.277 **	0.287 **
WY585_2080	0.359 **	0.342 **	0.328 **	0.321 **
WY585_2100	0.326 **	0.309 **	0.301 **	0.300 **
	CS585_2040	CS585_2060	CS585_2080	CS585_2100
WY126_2040	0.329 **	0.320 **	0.315 **	0.313 **
WY126_2060	0.319 **	0.306 **	0.298 **	0.293 **
WY126_2080	0.300 **	0.289 **	0.283 **	0.280 **
WY126_2100	0.303 **	0.292 **	0.285 **	0.282 **

Note: ** Significant correlation at the 0.01 level (two-sided).

. In the three scenarios, there are synergistic effects among the four ESs. The strongest correlation was observed between CS and HQ, with correlation coefficients exceeding 0.95 (p < 0.01). The SC also showed a significant correlation with HQ, while the correlation between WY and HQ was the weakest, with coefficients below 0.30. However, the changing trends of the correlation coefficients are different over time. Among them, under the SSP1-2.6 scenario, except for the three pairs of ESs related to WY, the correlations of other ESs show a decreasing trend over time. Under the SSP2-4.5 scenario, the correlation between SC-HQ showed a decreasing trend, and the correlation between other ESs exhibited an initial increasing phase followed by a subsequent decline. Under the SSP5-8.5 scenario, the correlation change trend between ESs is more complex. The correlation between CS-HQ shows an increasing trend over time. The correlation between CS-HQ and WY-HQ exhibits a pattern of increasing correlation initially, followed by a subsequent decline, and the correlation between WY-CS shows a decreasing trend. Other ES pairs exhibit an initial negative correlation, which is subsequently followed by a positive correlation over time. Ecosystem service management should prioritize ES pairs characterized by trade-offs, and also consider ES pairs with weakened synergies that imply increased trade-offs [55].

The spatial correlations among ESs are illustrated in Figure 11. The spatial correlation of the three scenarios shows significant differences. Among them, under the SSP1-2.6 scenario, most ESs show synergistic effects, and the proportion of catchments showing synergistic effects reaches more than 82%. SC-WY had the largest proportion of synergistic effects (99.45%), followed by HQ-CS (90.48%). The region exhibiting the highest intensity of synergistic effects is predominantly situated in the central part of the basin. Under the SSP2-4.5 scenario, the proportion of channels showing synergistic effects between SC and WY is 97.98%. The proportion of areas exhibiting synergistic effects between HQ and CS constitutes 32.24%, mainly distributed in Wudu District. The proportions of channels showing good effects for HQ-SC and HQ-WY were 43.58% and 43.48% respectively, concentrated in Wen County and Wudu District in the southern part of the basin. Under the SSP5-8.5 scenario, the channels showing synergistic effects between HQ and CS account for the largest proportion (85.39%). The proportions of channels showing synergistic effects of HQ-SC and SC-CS accounted for 70.43% and 75.47%, respectively, with these areas predominantly found across the central and southern regions of the basin. The proportion of channels showing a trade-off effect for HQ-WY is as high as 87.66%, which is distributed throughout the basin. The proportion of channels showing trade-off effects for SC-WY and CS-WY is as high as 85.14% and 71.32%, respectively, and the correlation gradually increases from southwest to northeast. Consequently, it is imperative for managers to be cognizant of these dynamics and to implement suitable interventions [49]. Promote ES improvement by avoiding unnecessary ES correlations.

3.4. Spatial Distribution Characteristics of ES Bundles

The SOM identified six ES bundles at the sub-basin scale (Figure 12). The ES bundle refers to a collection of different ES aggregation distributions in space or time [54]. The spatial distribution patterns of ES bundles exhibit substantial variability across distinct scenarios. For Cluster 1, this cluster is situated along the banks of the Bailong River within the urban boundaries of the Wudu District. Under the SSP1-2.6 scenario, SSP2-4.5 scenario, and SSP5-8.5 scenario, they account for 20.34%, 12.29%, and 13.39% of the basin, respectively. This ES bundle is distinguished by the spatial synergy between HQ and CS, featuring elevated levels of both services. For Cluster 2, this bundle mainly covers the northern area of Tanchang County, covering 6.04%, 7.55%, and 10.78% of the basin area in the SSP1-2.6 scenario, SSP2-4.5 scenario and SPP5-8.5 scenario, respectively. This bundle provides weak Ess, except for WY. For Cluster 3, this cluster is predominantly situated along the left bank of the Bailong River within Zhouqu County and spans the northern region of Diebu County. Under the SSP1-2.6 scenario, SSP2-4.5 scenario, and SPP5-8.5 scenario, they account for 9.11%, 12.84%, and 14.10% of the basin area, respectively. This ES bundle demonstrates spatial synergies, characterized by high levels of CS, HQ, and WY. For Cluster 4, this cluster is predominantly situated in the regions with more favorable ecological conditions located in the southern part of the basin, with the largest proportion under the SSP1-2.6 scenario (20.29%), followed by the SSP2-4.5 scenario (18.33%). The SSP5-8.5 scenario accounts for the smallest proportion (18.13%). This bundle mainly focuses on the balanced development of four ESs, especially in the SSP1-2.6 scenario. For Cluster 5, this cluster is primarily distributed along the banks of the Bailong River, spanning both Diebu County and Zhouqu County. The three scenarios account for 25.08%, 26.44%, and 23.31% of the basin, respectively. This bundle is distinguished by the spatial synergy among high CS, high HQ, and elevated SC. For Cluster 6, this cluster comprises sub-basins situated in the eastern and southern hilly regions. Under the three scenarios, this cluster accounts for 19.13%, 22.56%, and 19.76% of the watershed, respectively. This bundle is characterized by the balanced development of the other three ESs except SC (Figure 13).

4. Discussion

4.1. Analysis of the Impact of Land Use on ES

The spatial distribution of land use in the BRB exhibits significant differences under different climate models and development strategies. The changes in the area of built-up land and farmland differ across the three scenarios. In the SSP126-EP scenario, built-up land decreases, while forest area increases to achieve ecological sustainability [56]. In the SSP585-EG scenario, the amount of land available for construction increases significantly due to socioeconomic factors. This trend is similar to the historical patterns observed in the SSP245-ND scenario [57]. By comparison, it is shown that the expansion of built-up land is more influenced by economic development, while the increase in farmland area is mainly affected by population growth [58,59]. In addition, by comparing the changes in forestland area, lower economic growth rates and more stable climate environments are conducive to the expansion of the forestland area, and forests often extend to areas with better ecological environments [60], which is similar to our results. Therefore, to ensure the stability of regional ecosystems and achieve sustainable development, policies must be formulated to strictly regulate the unchecked expansion of built-up land, strengthening the protection of ecological lands (water bodies, forestland, and grassland) [61].

Previous studies have indicated that vegetation has significant impacts on ESs, such as SC, WY, CS, and HQ [62,63]. Wang et al. found that the trade-off/synergy between NPP, SR, and WY varies under different climate and land use conditions in Shaanxi Province [64]. Research has shown that ecological restoration projects have an improving effect on WY, SR, and CS in the Loess Plateau of Northern Shaanxi and can also affect changes in ES bundles. Moreover, the significant synergistic enhancement in SR and CS suggests ecological amelioration [65]. The BRB is situated in a transitional ecological zone and frequently experiences geological disasters, such as landslides, mudslides, and floods. Land use is inefficient, and land degradation is severe. Our predictions align with the trends observed in other related studies. The SSP1-2.6 scenario is a relatively idealized scenario that provides the most dense and efficient ESs, suppresses rapid urban expansion, increases ecological land use, and enhances all four ESs to varying degrees. However, in the SSP2-4.5 and SSP5-8.5 scenarios, it is expected that climate change and land use changes will be more severe, resulting in greater ES losses under higher development pressures [34]. These results indicate that implementing ecological protection measures can reduce land use changes caused by human activities and increase ecosystem functions [66]. Variations in population size and agricultural development are the principal drivers of LUCC. Within the SSP5-8.5 scenario, population expansion and economic progress have resulted in a notable escalation of urban and built-up land areas, which are concurrent with a reduction in forestland, and these LUCCs have exerted adverse effects on ESs [67]. This conclusion is consistent with the research results of Zhang et al. [68].

4.2. Suggestions on the Management of ES in the BRB

The ultimate goal of ESs is their practical application, providing a theoretical basis and technical support for the management of regional ESs and environmental protection [69]. The maintenance of ESs is vital to the future of humanity. However, ecosystem degradation is common and often triggers further disasters, reducing the resilience of natural and socioeconomic systems to climate change and irrational human activities. The current research on protecting the natural capital of ecosystems is still in its infancy [70]. The pressing challenge is to translate ideas into action through social policy and decision-making [71]. The goal of the current research is to implement ecological environment management within different ES functional areas and enhance human activities that are crucial to the ecological environment through land use policy formulation and governance.

Mountainous regions are deemed optimal for studying the effects of LUCC on ESs. On one hand, their unique topography enables mountain regions to effectively store and filter diverse material and energy flows [72]. With unique geological, geomorphological, climatic, hydrological, and other environmental features, mountains nurture diverse ecosystems and provide a variety of ESs to local residents [73]. On the other hand, mountain areas are vulnerable regions and frequently experience disasters, such as landslides and mudflows. The population in mountainous areas of China is relatively large, and these regions are undergoing rapid industrialization and urbanization. So, the relationship between LUCC and ES seems more complex in mountainous areas. Meanwhile, a series of policies are being implemented in these areas, such as “Reform and Opening-up” and “Western Development”, aimed at promoting local economic growth, as well as “Grain for Green” policies designed to strengthen ecological protection. The above policies have all promoted regional land use change. In future development, mountain areas should focus on coordinating ecological protection with economic development. Regional economic development has shifted from “high-speed” development to “high-quality” development to enhance the value of ES in the research area.

The BRB is a fragile mountainous area prone to natural disasters. We combine the ES spatiotemporal trade-off/synergy status and ecological zoning to put forward relevant suggestions and measures with the goal of protecting people’s lives and property, reducing disaster hazards, and promoting regional sustainable development. Therefore, in regional land use planning, priority must be given to WY. The WY in some areas on both sides of the Bailong River is showing an increasing trend, which will increase the risk of landslides and debris flow disasters (C3). In these areas, relevant departments should strengthen the construction of disaster prevention and early-warning projects [74]. In the southern region where forestland was restored, there were relatively more areas with increased ESs (C4). Therefore, the implementation of extensive plantations in the south should be considered to improve the capacity of ESs such as SC and CS. In addition, in ecologically fragile gully areas (C5), afforestation projects should weigh the pros and cons, and the expansion of construction land should be controlled within an appropriate range. We suggest focusing on the extreme aggregation of high ESs in future development.

4.3. Shortcomings and Prospects

Our study encounters certain limitations. First, the complexity of LUCC brings a certain degree of uncertainty to land use simulation. The changes in land use are not only influenced by spatially expressive factors such as economy and population but also by policy frameworks including farmland protection and ecological protection regulations [75,76,77]. Secondly, climate and land use changes have different impacts on different ecosystems. In this study, we investigated four typical ES: SC, WY, CS, and HQ. We did not assess other services, including cultural services, due to constraints on data availability and simulation methodologies. Considering LUCC and climate change, the inclusion of a broader range of ESs could facilitate a more holistic investigation into the mechanisms of ES changes in future studies. We should choose more indicators to simulate important ESs. Finally, we simulated multiple ESs under the combined effects of LUCC and climate change, which helps to better understand ES dynamics. We incorporate climate change into the LUCC simulations, which makes the assessment of future ESs more realistic. It is imperative to delineate the independent impacts of these two primary drivers in future studies to ascertain their relative strength and dominance across various ESs [78]. Such discernment is vital for the effective integration of planning and policy-making [79].

5. Conclusions

This study coupled the PLUS and InVEST models to simulate LUCC and ESs and their mutual relationships under different SSP-RCP scenarios. Under the SSP126-EP scenario, the forestland expanded rapidly, and the grassland, farmland, and built-up land will decrease. In the SSP245-ND scenario, the land use change pattern is similar to that in the SSP126-EP scenario, but the growth of forest land is slow. Under the SSP585-EG scenario, built-up land and farmland expanded rapidly. From 2040 to 2100, all four ESs increase under the SSP1-2.6 scenario. Under the SSP2-4.5 and SSP5-8.5 scenarios, ESs show a decreasing trend. There are synergistic effects between ES pairs, and attention should be paid to ES pairs with reduced synergistic effects. The cluster with a balanced development of the four ESs is situated in the southern region of the basin, and this cluster accounts for the highest proportion in the SSP1-2.6 scenario. Clusters with weak ESs are mainly situated in the northern area of Tanchang County, and this cluster has the highest proportion in the SSP5-8.5 scenario. We recommend the development of spatial planning and management strategies that are informed by the trade-offs/synergies among ES, as well as the specific bundles of ESs, to promote the sustainable development of regional ecosystems.