The SYRB is situated in the midst of the YRB (110°14′-114°33′) and includes 11 cities and 86 counties, spanning an area of 114,600 km² (Figure 1). The SYRB is a typical loess plateau with a variety of topographies, of which mountains and hills account for more than 80%, while the rest are plains and basins in the intermountain valleys. The research region experiences a moderate continental environment with four different seasons. The average temperature over the year is 10.0°C–11.5°C, and precipitation is concentrated from July to September, totaling 350–680 mm each year. Cultivated land covers the most ground, followed by forest land and grassland. Due to fast economic growth, the study area’s natural environment is very vulnerable and rapidly influenced by human activities (Wang J. et al., 2024).

Precipitation, evaporation, and temperature data were obtained from the National Science and Technology Resource-Sharing Service platform (Table 1). The rainfall erosion factor was calculated based on the rainfall grid data. The depth of the root-restriction layer was selected from a soil depth map of 1 km in China. The soil data were collected from the Harmonized World Soil Database, which was created by the United Nations Food and Agriculture Organization and the Vienna International Institute for Applied Systems. The land use/land cover data came from the Chinese Academy of Sciences Resource and Environmental Science and Data Center. The data for the digital elevation model (DEM) were obtained from the geospatial data cloud. The gross domestic product (GDP) statistics came from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences. The distance between the road and the railway was calculated using the National Catalogue Service for Geographic Information’s basic geographic information vector data.

First, the driving factors—including the natural environment, social economy, and traffic accessibility—were chosen based on land use change between 2005 and 2020, and then the region’s land use in 2035 under an natural development scenario was predicted using the PLUS model (Figure 2). Second, the InVEST model was utilized to estimate the four typical ecosystem services for water yield, habitat quality, carbon storage, and soil conservation from 2005 to 2035. Through weighted superposition analysis, the ecosystem service evolution in the SYRB was determined. Third, the natural breakpoint method was utilized to identify the ecosystem service function, and the first three types of areas were chosen as the MSPA method’s core area to determine the SYRB’s ecological source. Then, the land use elevation, slope, and habitat quality were utilized to create the resistance surface. Finally, the circuit theory was utilized to extract ecological corridors, ecological pinch points, and ecological barrier points, construct the ESP, and determine the priority locations and measures for ecological protection and restoration.

A region that possesses excellent ecological stability, expansibility, and ideal ecological function is known as the ecological source, which may enable the ecological process to develop in a positive direction (Dai, 2022). The results of water yield, habitat quality, carbon storage, and soil conservation were normalized, and then weight was calculated using the entropy approach as follows: 0.24, 0.26, 0.26, and 0.24. Through weighted superposition analysis, the ecosystem service functions evaluation results in the SYRB were achieved. The natural breakpoint method (Gao et al., 2022) divides the ecosystem service functions into five levels, ranging from high to low: excellent, good, general, poor, and worst. The original ecological sources were determined to be the first three levels (Zhou et al., 2023).

MSPA is a quantitative approach to detecting ecological sources. It mainly identifies and classifies ecological sources through neighborhood analysis to obtain the distribution (Vogt et al., 2007; Saura and Pascual-Hortal, 2007). In this study, the original ecological sources were used as prospect data for the MSPA analysis. The SYRB yielded seven distinct landscape types: core, islet, perforation, edge, bridge, loop, and branch areas (Dai, 2022). The core area was recognized as its final ecological source in the SYRB. In the process of source identification, there are many patches with small areas and uneven distribution, which provide limited ecosystem services and have little impact on the overall ESP. As a result, the research removed those ecological patches with areas smaller than 10 km².

The resistance surface reflects the difficulties of species movement across ecological sources (Ding et al., 2022). It reflects the horizontal resistance to ecological processes. The resistance surface in this study was constructed using land use, elevation, slope, and habitat quality. The resistance value was determined by consulting the necessary literature and expert opinions, and the weights of the various resistance factors were calculated utilizing the analytic hierarchy approach (Table 4).

The ecological corridors are the channel for the movement and exchange of materials, energy, and information across ecological sources, and it is the easiest linked ecological channel with minimum resistance. The least-cost paths were determined using the Linkage Pathways Tool in Linkage Mapper based on circuit theory, with the lowest-cost path serving as the best ecological corridor.

The ecological pinch points are the high-frequency areas where ecological processes flow, serving as an alternative path in the study area for materials, energy, and organisms to flow between different ecological sources or when no other options are available. This study identified the pinch point using the Pinchpoint Mapper module in Linkage Mapper, chose the “all to one” mode, and combined with Circuits pace tools to effectively identify the pinch area through the current flow between ecological sources. Then, using the natural breakpoint approach, the identification findings were classified into five levels, with the highest level reflecting the ecological pinch points.

The ecological barrier points are the key nodes that hinder the connectivity between ecological sources, and their removal improves connectivity. This research employed the Barrier Mapper module in Linkage Mapper, selecting 500 m as the iterative radius to identify the obstacle point.

Cultivated land was the primary land use category in the SYRB, followed by forest land and grassland. Cultivated land was generally concentrated in the center and southern sections of the SYRB with ample water supplies and relatively level terrain, while forest and grassland were mostly concentrated in the Taiyue, Zhongtiao, and Lvliang mountains (Figure 5).

The areas of the six land use groups ranged from high to low: cultivated land (43,399.17 km²), forest land (34,690.24 km²), grassland (31,216.10 km²), construction land (6,048.13 km²), water area (1,006.92 km²), and unused land (39.61 km²) in 2020 (Figure 6). From 2005 to 2020, the total area transferred between land use types was 6,080.99 km², or 5.22% of the SYRB’s total area. Over the past 15 years, construction land showed a clear expansion trend, increasing by 2,837.44 km², an increase of 88.37%. Forest land also showed an upward trend, growing by 203.05 km², or 0.59%. Meanwhile, cultivated land, grassland, water area, and unused land showed a shrinking trend, decreasing by 1990.12, 864.21, 182.83, and 3.34 km², with a decrease of 4.38%, 2.69%, 15.37%, and 7.78% respectively.

The areas of the six land use groups ranged from high to low in 2035 were cultivated land (41,961.21 km²), forest land (34,764.42 km²), grassland (30,393.08 km²), construction land (8,276.51 km²), water area (968.20 km²), and unused land (36.59 km²). 4,605.10 km² of land were transferred from 2020 to 2035. Construction and forest land expanded by 2,228.37 and 74.18 km², respectively, whereas cultivated land, grassland, water area, and unused land decreased by 1,437.79, 823.02, 38.72, and 3.02 km² respectively.

From 2005 to 2035, the water yield and soil conservation in the SYRB showed a fluctuating upward trend (Figure 7), while the habitat quality and carbon storage showed a downward trend. The average water yield grew by 10.045%, from 279.030 mm in 2005 to 307.059 mm in 2035. The average soil conservation rate grew by 17.494%, from 94.712 t/ha in 2005 to 111.281 t/ha in 2035. The average habitat quality dropped from 0.035 in 2005 to 0.014 in 2035, a 60.430% reduction. The average carbon storage dropped from 101.428 t/ha in 2005 to 100.875 t/ha in 2035, a 0.545% reduction.

The general trend of the geographical distribution of water yield in the SYRB from 2005 to 2035 declined from southeast to northwest (Figure 8). In 2005, high-value water yield areas were primarily concentrated along the southeastern route, while low-value areas were mostly distributed along the northwest route and the central urban agglomeration of the region (Wang J. et al., 2024), with sporadic distribution in the southern basins. In 2010, the high-value range of water yield in the study area expanded, whereas low-value areas were less distributed, only sporadically in the central urban agglomeration of the region and southern basins. In 2015, the water yield in the majority of the study area’s middle and northern regions was low, while high-value portions were still concentrated along the southeastern edge. High-value water yield areas expanded obviously in 2020. The majority of the study region south of the center was high-value, while low-value areas were mainly scattered in the central urban agglomeration. In 2035, the central urban agglomeration and southern basins were still the primary distribution locations for low water yield. The geographical distribution of water yield is intimately linked to the distribution of precipitation and vegetation cover. Areas with heavy precipitation but low evapotranspiration have higher water yields and vice versa.

The geographical distribution of soil conservation in the research region changed slightly between 2005 and 2035 (Figure 8). As time went by, the high-value regions of soil conservation increased slowly and fluctuated, while the low-value parts decreased slowly in a fluctuating manner. The high-value regions were mainly located in the western Lvliang Mountains, the central Taiyue Mountains, and the southern Wangwu and Zhongtiao Mountains, being forest lands and grasslands with high vegetation coverage. The low-value regions were primarily concentrated in basins with considerable human activity and limited vegetation covering.

From 2005 to 2035, the geographical distribution patterns of carbon storage, habitat quality, and soil conservation in the study area were basically consistent (Figure 8). High-value locations were distributed in the research area’s mountainous regions on the east and west sides, while low-value areas were located in regions with a high level of urbanization. The research region did not show an overall decline in carbon storage or habitat quality.

In 2005, the area classified as providing outstanding ecosystem services amounted for 8.12% of the total area, increasing to 8.22% in 2010 and 8.88% in 2015 (Figure 9). However, by 2020, there was a decrease to 7.04% of the total area, followed by an increase to 12.92% by 2035. Excellent-grade areas were mostly located in the forest land on the research area’s east and west edges, where the vegetation was better. The area graded as having good ecosystem services accounted for 17.44% of the total area in 2005, 19.99% in 2010, 20.09% in 2015, 20.76% in 2020, and 21.31% in 2035, showing a continuous expansion trend. This grade was mainly located on the edge of the excellent grade of ecological sources. The areas of general grade made up 34.23% of the total area in 2005, 32.56% in 2010, 34.61% in 2015, 33.59% in 2020, and 33.85% in 2035. This grade showed fluctuating changes, mainly extending along the edges of the areas of good and excellent grades. The ecosystem services in the poor and worst grades were mainly dispersed in urban built-up regions and cultivated land, providing fewer ecosystem services.

From 2005 to 2035, the core area showed a fluctuating development trend, with the majority of it concentrated in locations with significant biodiversity and superior ecological conditions, such as the Lvliang, Taiyue, Zhongtiao, and Taihang mountains (Figure 10). The core area was generally located in the research area’s eastern and western regions, with a weak connection between these two parts. The bridge area was less than that of the core area, which connects different core area patches as a structural corridor, and it has significant implications for biological migration and landscape connectedness. The edge and perforation sections served as a transition between the core area and non-green landscape patches. From 2005 to 2035, the edge area continued to increase while the perforation area decreased continuously. The edge region was situated on the outside of the core area, and its ability to protect the core area from external disturbances gradually increased. The perforation region was situated on the inner border of the core area, and its ability to maintain ecological stability within the core area declined. The branch area expanded and subsequently declined, indicating that the connectivity of ecological corridors within the study area is not particularly tight. Islet and loop areas had the smallest proportions, with the islet area decreasing, indicating that the fragmented patches that can be used as stepping stones for biota gradually decreased, which is detrimental to the overall connection of ecological patches. The fluctuating growth of loop areas indicates the instability of support for migration and movement of organisms within patches, and the region was very small.

The final ecological sources were obtained by choosing patches bigger than 10 km², which MSPA designated as ecological core regions. In 2005, there were 81 ecological sources covering 35,767 km², or 30.73% of the research area (Figure 11). In 2010, there were 64 ecological sources covering 33,225 km², or 28.54% of the overall area. In 2015, there were 80 ecological sources covering 36,171 km², or 31.07% of the area. In 2020, there were 114 ecological sources covering 31,062 km², or 26.69% of the area. In 2035, there were 110 ecological sources covering a total of 39,931 km², or 34.30% of the area. The ecological sources increased from 2005 to 2015 but decreased from 2015 to 2020. The overall area of ecological sources increased from 2020 to 2035, but the number of source regions remained stable, showing a good trend in the ecological environment. However, the phenomenon of ecological fragmentation still existed.

The research area’s ecological sources were mainly found in the southeastern Taihang, Zhongtiao, and Taiyue mountains, as well as the center and western Lvliang Mountains. These places are mainly covered by forests, with high habitat quality and relatively intact ecosystem services functions, making them priority regions for ecological development. The ecosystem services functions along the western corridor were relatively high, but they were distributed in a scattered manner, with small areas that were easily influenced by human activity. As a result, they were more prone to fragmentation or disappearance, with fewer ecological sources spread over the area. In the central basin, cultivated and construction land predominated, with more anthropogenic interventions, leading to the dispersion of important areas for ecosystem service functions.

The geographical distribution of resistance surface in the SYRB indicated that the low-resistance areas were mostly found in regions with more precipitation and concentrated forest areas (Figure 12). High-resistance zones were found in densely populated regions of cities, where the ecological environment was relatively fragile and posed greater resistance to ecological processes. The comprehensive resistance values in 2005, 2010, 2015, 2020, and 2035 were 3.389, 3.402, 3.405, 3.407, and 3.426, respectively. Resistance values increased year by year, further hindering biological migration and exchange.

In 2005, there were 191 ecological corridors totaling 2,792.239 km. In 2010, there were 137 ecological corridors spanning 2,945.837 km (Figure 13). In 2015, there were 188 ecological corridors totaling 3,499.462 km. In 2020, there were 274 ecological corridors totaling 4,061.342 km. In 2035, there were 268 ecological corridors totaling 3,553.176 km. The increasing length of ecological corridors indicates that the fragmentation of ecological source areas increased the cost of resistance to biological migration. Areas with concentrated ecological source areas had a higher number of ecological corridors but shorter lengths. The distance between the source areas in the east and west was relatively far, the ecological corridors were long, and the quantity was limited. The southern basins were greatly influenced by human activity, resulting in unstable ecological sources and a rise in the number of ecological corridors due to source area growth. There were no ecological source sites or ecological corridors in the research area’s northern end.

In 2005, 27 ecological pinch points were identified in the research region (Figure 13). The number increased to 28 in 2010 and 2015 and reached 36 in 2020. By 2035, there were 40 ecological pinch points. This shows a rise in the number of ecological pinch points, with their distribution expanding from a relatively concentrated central area to the study area’s periphery. Ecological pinch points were typically found in the middle of ecological corridors or in areas intersecting with ecological source areas. These areas often face higher ecological risks as a result of the high ecological resistance in the surrounding areas, highlighting the urgent need to protect ecological pinch-point areas in order to sustain the connectivity of the ecological landscape.

In 2005, a total of 21 ecological barrier points were identified, increasing to 22 in 2010, 23 in 2015, 24 in 2020, and 28 in 2035 (Figure 13). This minor rise in the number of barriers suggests that the overall ESP remained unstable. Overlaying ecological pinch points and ecological barrier points revealed obstacles along important migration corridors for wildlife, making these areas a priority for protection and restoration.

To improve the integrity and efficiency of the ecological network, this research combined the “Ecological Protection and High-Quality Development Plan of the Yellow River Basin in Shanxi Province,” the “14th Five-Year Plan for the Conservation and Utilization of Natural Resources in Shanxi Province,” and the ecological protection and restoration project of the “Two Mountains, Seven Rivers, and One Basin” in Shanxi Province to construct an ESP of the SYRB, namely, “one axis, two zones, four belts, and multiple points.” “One axis” describes the development axis along the northeast-southwest direction of the Fen River (Figure 14). The “two zones” allude to two ecological essential protected areas: the Lvliang Mountains and the Taihang, Zhongtiao, and Taiyue mountains, as well as their surrounding areas. The “four corridors” are important bridges that connect ecological sources in 2020 depending on the distribution and extension directions of significant ecological corridors in 2020. “Multiple points” refer to multiple ecological barrier points and ecological pinch points.

According to the land use prediction results of the SYRB in 2035, an ESP model of “one circle, two zones, one axis, three belts, and multiple points” was constructed. Compared to the ESP in 2020, by 2035, many small patches had formed between the two key ecological protection areas, increasing the connectivity between the two regions and forming a “circle,” namely, the important ecological development circle of the SYRB. In the central and southern basins, as well as the southeast, multiple ecological pinch points formed, and the number of pinches increased. The distribution was relatively concentrated in the center and gradually spread to the study area’s neighboring areas, with the distribution gradually expanding.

In the ESP constructed for the SYRB in 2020, there are 114 ecological sources covering a total of 31,062 km², along with 274 ecological corridors that extend a total length of 4,061.342 km. Utilizing MSPA and Remote Sensing Ecological Index for identification (Wang B. et al., 2024), 108 ecological sources and 243 ecological corridors were identified within the SYRB, encompassing a total area of 34,157.42 km² and a total length of 3,259.44 km. By calculating various Multi-ecosystem Service Landscape Indices (Wang et al., 2023), the comprehensive capacity of ecosystem services was quantified, identifying 16 ecological sources in Shanxi Province, with a total area of 3,094 km² and 15 primary ecological corridors totaling 1,000 km in length. Our results were consistent with those of Wang J. et al. (2024), but demonstrated superior performance compared to Wang et al. (2023). This study first conducts a comprehensive assessment of ecosystem services, and then integrates the MSPA model to identify regions with high ecosystem services value as ecological sources. This approach enhances the scientific basis for the selection of ecological sources and aligns more closely with the actual conditions of the region. In the construction of the ESP for 2035, there are a total of 110 ecological sources, encompassing an area of 39,931 km², while the total length of 268 ecological corridors is 3,553.176 km. Compared to 2020, the number of ecological sources has decreased, while their total area has increased. Conversely, the number of ecological corridors has increased, although their total length has decreased, indicating an improvement in the ecological environment.

In the YRB, the Ningxia region’s ecological sources are primarily located in the southern portion. Ecological risks are higher at the border of the northern desert (Jiang et al., 2024). The ecological source areas in the Henan section of the YRB are mostly located in the forested area in the southwest of Luoyang City (Wei et al., 2023). The ecological source areas in the Gansu section of the YRB are mostly distributed on the eastern and southern edges, with uneven and fragmented spatial distribution (Xu et al., 2023). The ecological sources in the “Jiziwan” region of the YRB are scattered and fragmented (Wei et al., 2022). However, apart from the YRB’s source area, the ecological sources in Shanxi Province are the most densely distributed, forming a unique ESP (Yan et al., 2024). The high-density forest cover in the eastern and western mountainous areas of the research area, coupled with the high number of cities and frequent human activity in the central basin area, has hindered ecological exchanges between the eastern and western regions. In addition, Shanxi Province has abundant coal resources and is a significant coal resource province in China. Long-term, large-scale, and intensive coal resource exploitation has led to land subsidence and excavation damage on the surface, which can accelerate soil erosion in local areas, exacerbate vegetation destruction, and damage the ecosystem (Wu et al., 2021). By comparing the coal mines distribution map in Shanxi Province (https://zrzyt.shanxi.gov.cn), it can be found that areas with a concentrated distribution of coal mines in the research region overlap significantly with regions of the lowest ecological environment quality grade.

Compared to 2020, the ESP in 2035, based on anticipated climate change, showed a decrease in the average values of the four ecosystem services. Among them, the water yield and soil conservation in the western region of the study area decreased most significantly. Therefore, there was a noticeable shrinkage in the identification of important areas for ecosystem services and ecological sources. Research has shown that climate change leads to varying degrees of change in the frequency and severity of regional extreme weather events (floods, storms, etc.), thereby further impacting the quality of the natural environment (Fan et al., 2022). The increase in extreme weather events caused by global warming in 2030 will pose a bigger danger to the SYRB’s natural environment quality (Fu et al., 2024).

(1) For ecological resources, it is essential to establish key conservation areas, adjust forest structures, enhance biodiversity to increase vegetation coverage, maintain ecological functions, and ensure that these resources remain undisturbed by human activities. In addition, create buffer zones at the edges of ecological sources, increase vegetation coverage, and mitigate the effects of human activities.

This study used the PLUS, InVEST, and MSPA models, together with circuit theory, to identify ESPs for 2020 and 2035. However, due to the uncertainties of future climate data and future benefits, there is a certain lag in improving ecosystem services and ESPs through ecological protection and restoration measures. Long-term observation and continuous revision of ecological protection strategies are still needed. In addition, the research model selects four representative ecosystem services: water yield, habitat quality, carbon storage, and soil conservation. The exclusion of other significant ecosystem services may lead to certain discrepancies in the construction of the ESPs, which will be discussed in greater depth in subsequent research endeavors. It is essential to take into account the variations in future ESPs under different scenarios, in order to propose more precise strategies for ecological restoration and development. Additionally, data sources with varying spatial resolutions can significantly influence the results. Higher spatial resolution tends to yield more accurate outcomes. Future researches should therefore give adequate consideration to the acquisition and application of high-resolution data.