Towards Smarter Urban Green Space Allocation: Investigating Scale-Dependent Impacts on Multiple Ecosystem Services

Abstract

Urban green space (UGS) is crucial for enhancing ecosystem services (ESs), offering both ecological and social benefits. The multifunctional and synergistic development of UGS is essential for addressing ecological security challenges and meeting the demand for high-quality urban living. In densely urbanized areas, optimizing green space scale is essential for maximizing its multifunctionality. This study focuses on the Taihu Lake region in China, assessing six ESs. A self-organizing map (SOM) was employed to identify five distinct ecosystem service bundles (ESBs), while redundancy analysis (RDA) explored how green space scale characteristics influence ESs within each bundle. The results indicate that ESs exhibit significant spatial heterogeneity, with the ESBs showing two typical patterns in terms of synergistic-tradeoff relationships. The green ratio (GR) is the primary driver, with largest patch index (LPI) acting as the secondary factor, while other indicators’ effects vary across ESBs. This study systematically examines the pathways through which UGS scale characteristics influence ESs under multiple scenarios, adopting the ESB perspective. It proposes a tiered UGS scale regulation framework aimed at achieving synergistic, multi-value outcomes. Such a framework has strong potential to enhance both the ecological performance and spatial efficiency of UGS allocation. The findings contribute a novel approach to resolving multifunctional integration challenges in high-density urban settings and providing valuable insights for landscape planning and management.

1. Introduction

With the ongoing acceleration of global urbanization, urban green space (UGS), as a crucial element in maintaining urban ecosystem functions and promoting human well-being, are facing unprecedented challenges, particularly concerning the scientific basis of their scale and configuration. According to the United Nations’ “World Urbanization Prospects” report [1], by 2050, the global urban population is projected to reach 68%, with China’s high-density urbanization characteristics being especially prominent. This trend is expected to exacerbate the fragmentation and functional degradation of urban ecological spaces [2,3], which in turn will affect the comprehensive capacity of green space ecosystem services (ESs) [4,5]. In this context, optimizing the scale and configuration of UGS to maximize the comprehensive benefits of ESs has become a key issue in sustainable urban development.

Through the interaction between humans and the green environment, UGS play multiple ecological roles, providing provisioning, supporting, regulating, and cultural services [6,7,8]. Previous studies have made significant strides in understanding how the scale of UGS affects ESs, with metrics such as green space area, green space ratio, and three-dimensional green volume being commonly used in research [9,10,11]. Based on the theory of “green infrastructure,” numerous scholars have confirmed that large, contiguous green spaces offer significant advantages in climate regulation, biodiversity conservation, and carbon sequestration [4,12]. However, studies in high-density urban areas suggest that small, dispersed green spaces are more effective in enhancing community green equity and improving residents’ mental health [13,14]. In recent years, an increasing number of studies have focused on the nonlinear relationship between green space scale and ESs, highlighting the differentiated benefits of ESs as the size of green spaces increases in different development scenarios. For instance, in central urban areas, the size of green spaces in high-density, hard-surfaced communities can significantly enhance their cooling benefits [15,16]; in forested regions, carbon sequestration responds significantly to changes in forest cover [17].

There is a growing consensus that promoting the synergistic optimization of multiple values is vital for UGS planning. Ecosystem services often exhibit significant trade-offs or synergies. For example, carbon sequestration services tend to synergize with water quality purification, water conservation, and biodiversity, while recreational services and habitat provision may be in conflict due to human disturbance [18,19,20]. However, existing research still faces significant limitations. Static analyses of ESs often fail to capture the differentiated impacts of multiple scenarios, such as urban expansion and land-use transformation, on the synergistic or trade-off relationships among ESs [21,22,23]. Furthermore, a one-size-fits-all approach to green space size adjustments may trigger conflicts in ESs combinations within specific zones (e.g., increasing large green spaces may reduce cultural service provision) [24,25].

Therefore, differentiated green space management strategies, based on the characteristics of green spaces and the specific ES requirements, are necessary to enhance dominant services in each region and respond precisely to functional needs [26]. For example, mountainous areas should prioritize the protection of large, continuous green spaces to maximize their support and regulation services [15,27], while compact cities can optimize certain small green spaces to enhance cooling and cultural service efficiency [9,16,28] to better serve residents. Ecosystem service bundles (ESBs) serve as a key tool for analyzing the relationships between ESs, offering new insights [20,23,29]. Using clustering algorithms such as K-means and self-organizing map (SOM), this approach can identify functional units with ecological integrity [30]. This provides a methodological foundation for identifying UGS development needs, quantifying the comprehensive benefits of ESs, and guiding smart zoning management under multiple scenarios [3,31].

In summary, amidst the shift in urbanization from extensive spatial expansion to intensive quality enhancement, this study focuses on the Taihu Lake region of China, a representative urbanized area. It aims to identify clustering patterns and spatial heterogeneity of ESs, reveal the influence mechanism of green space scale on the multidimensional comprehensive benefits of ESs, and clarify key scale characteristics under different development scenarios. This research seeks to provide more precise guidance for zoning strategies of UGS. The findings will provide a scientific foundation for the precise management of green resources, promoting sustainable urban development and offering practical guidance for the “efficiency -oriented” approach to high-density urban green space planning.

2. Materials and Methods

2.1. Study Area

The study area is located in the Taihu Lake region (30°36′–31°53′ N, 119°18′–120°54′ E) of the Yangtze River Delta in eastern China (Figure 1). It corresponds to the Taihu Lake Strategic Coordination Zone, as defined in the “Shanghai Metropolitan Area Spatial Collaborative Planning”, encompassing the four cities of Suzhou, Wuxi, Changzhou, and Huzhou. This region is rich in water and green ecological resources and is home to Lake Taihu, one of the most important freshwater bodies in the Yangtze River basin. The area plays a crucial role as a key supplier of ESs.

Amid rapid urbanization and intensive industrialization, the tension between development and ecological conservation in the region has become increasingly evident [32]. This underscores the need to incorporate the synergistic enhancement of multiple ecosystem values into the broader spatial planning framework. Additionally, the green space development strategies of the four cities vary due to factors such as resource endowments and differing development models. As such, the Taihu Lake region serves as a representative case, offering valuable insights into multi-scenario-oriented planning and control strategies aimed at optimizing the synergistic benefits of UGS.

The study focuses on UGS, which are classified according to “Urban Green Space Classification Standards of China CJJ/T85-2017” [33]. The specific types selected for this study include park green spaces (G1), protective green spaces (G2), square land (G3), scenic and recreational green spaces (EG1), ecological conservation green spaces (EG2), regional facility protective green spaces (EG3), and production green spaces (EG4). UGS in the study area exhibit significant spatial differentiation: large clusters are found along Taihu Lake and the surrounding mountains, with a clear urban–rural gradient. In urban areas, small, scattered park green spaces are more prevalent.

The study adopted the 1.2 km × 1.2 km square grid as the study unit, dividing the research area into square grid units with 1.2 km sides. This unit size is chosen based on the following considerations: (1) to encompass the full distribution range of dominant green patches within each grid as much as possible, thereby minimizing artificial fragmentation of continuous ecological processes; (2) to align closely with the average scale of basic management units defined in urban control detailed plans—the core statutory framework for urban planning and management in China—thereby enhancing the practical relevance of the research findings; (3) to ensure an adequate number of study units to meet statistical requirements, thereby improving the robustness and reliability of subsequent analyses. In total, there are 7621 research units, of which 2606 are dedicated to UGS.

2.2. Data Processing

Based on data acquisition feasibility and actual conditions, and striving to eliminate the impact of temporal differences, the study collected and processed multi-source data for the Taihu Lake region in 2020. The data mainly includes land use, green vegetation, built environment, and meteorological data, as shown in

Table 1. Data description.

Type	Format	Period	Data sources
Administrative boundaries	Vector	2020	National Catalog Service For Geographic Information https://www.webmap.cn (accessed on 1 October 2024).
Remote Sensing Imagery	Raster (1 km × 1 km)	August 2020	MODIS MOD021KM dataset (LAADS DAAC https://ladsweb.modaps.eosdis.nasa.gov/ (accessed on 1 October 2024))
Land cover data	Raster (1 m × 1 m)	2020	SinoLC1 Dataset [34]
Digital Elevation Model (DEM)	Raster (30 m × 30 m)	2020	ALOS PALSAR dataset (United States Geological Survey, https://glovis.usgs.gov/ (accessed on 1 October 2024))
UGS data	Vector	2020	Open Street Map https://www.openstreetmap.org/ (accessed on 1 October 2024), supplemented by manual interpretation based on remote sensing imagery.
Building and Road network	Vector	2020	Open Street Map https://www.openstreetmap.org/ (accessed on 1 October 2024)
Mean annual precipitation (PRE)	Raster (1 km × 1 km)	2020	National Tibetan Plateau/Third Pole Environment Data Center http://data.tpdc.ac.cn (accessed on 1 October 2024)
Mean annual evapotranspiration (ET)	Raster (1 km × 1 km)	2020	National Tibetan Plateau/Third Pole Environment Data Center http://data.tpdc.ac.cn (accessed on 1 October 2024)
Soil attribute data	Raster (1 km × 1 km)	2020	National Tibetan Plateau/Third Pole Environment Data Center http://data.tpdc.ac.cn (accessed on 1 October 2024); HYSOGs250m dataset (ORNL DAAC https://daac.ornl.gov/ (accessed on 1 October 2024))
Vegetation canopy height data	Raster (10 m × 10 m)	2020	The ETH Global Canopy Height 2020 product [35]

.

Due to variations in raster formats across data sources, we resampled the data to match the smallest data pixel size (i.e., land-use data at 1 m × 1 m). Subsequently, we performed zonal statistics by study unit for variable characterization.

2.3. Ecosystem Service Measurement

Based on the regional characteristics of the Taihu Lake area and in alignment with the development objectives outlined in the master plan, six key ESs were selected for analysis (Table 2): climate regulation (CR), stormwater regulation (SR), water conservation (WC), water purification (WP), habitat quality (HQ), and outdoor recreation (OR). These six ESs comprehensively consider the core ecological characteristics and needs of the Taihu Lake region, drawing on the existing relevant literature (see Table 2). The selection criteria primarily include the following: (1) critical importance to the sustainable development and human well-being of the study area; (2) close association with the scale characteristics of UGS; (3) availability and quantifiability of relevant data; (4) alignment with commonly used ESs assessment frameworks in similar studies to enhance comparability and applicability of results.

The study employed ArcGIS 10.7, ENVI 5.3, and the InVEST 3.14.0 model to conduct quantitative assessments of these services. Specifically, the CR service was estimated using ENVI to infer surface temperature [36,37,38], while SR [39], WC [40,41], WP [42,43], and HQ [44,45] were quantified through the Urban Flood Risk Mitigation, Water Yield, Nutrient Transport Ratio, and Habitat Quality modules within the InVEST model [42,46,47]. OR services were assessed based on the service supply capacity and recreational accessibility of green spaces [48,49]. Detailed calculations for each ecosystem service are provided in the Supplementary Materials.

Given the differences in measurement units among the evaluation indicators for various service types, the study standardized the original data using minimum–maximum normalization. It is important to note that, while the calculations adhered strictly to official standards and standardized coefficients, the complexity of the ecological environment in the Taihu Lake region resulted in some deviation in the actual calculations. However, the errors in the relative relationships among ecosystem service units are considered to be within a controllable range.

Table 2. Key ESs in the Taihu Lake region.

ESs Type	Description	Importance Within the Study Area	References
Climate Regulation (CR)	Green spaces reduce local temperatures and mitigate extreme heat events through evapotranspiration, shade, and reflection of long-wave radiation.	The built-up areas surrounding Taihu Lake experience elevated summer temperatures, exacerbated by industrial heat emissions that worsen the regional thermal environment. The plan proposes the creation of an ecological barrier around the lake to enhance carbon sequestration and mitigate heat island effects.	[38,50]
Stormwater Regulation (SR)	Retain and store rainfall runoff through vegetation, soil, and terrain to reduce peak flood flows and delay peak onset times.	Climate change has intensified the uneven distribution of rainfall both temporally and spatially, while rapid urbanization has led to a decline in surface permeability. The plan emphasizes the importance of utilizing the composite system of polders, river networks, and lakes to strengthen flood retention capacities.	[39,51]
Water Conservation (WC)	Green spaces promote rainwater infiltration, replenish groundwater, and maintain base flow and drinking water sources.	Taihu Lake serves as a vital water source for cities such as Shanghai and Suzhou city, with upstream areas like the Yili Mountain designated as key conservation zones for water sources.	[52,53,54]
Water Purification (WP)	Improve water quality by filtering, adsorbing, and decomposing pollutants through vegetation, soil microorganisms, and substrate.	To address issues like non-point source pollution and eutrophication, the plan outlines measures for developing lakeside wetlands and fostering regional collaboration in managing water systems. It also mandates the construction of an ecological interception system.	[20,53]
Habitat Quality (HQ)	Green spaces provide food, breeding grounds, and shelter for local flora and fauna, maintaining urban biodiversity.	Habitat fragmentation disrupts species migration and environmental pollution contributes to the decline of sensitive species populations. The plan calls for the protection of Taihu Lake’s critical ecological spaces to preserve regional biodiversity.	[53,55]
Outdoor recreation (OR)	Offering natural landscapes and ecological spaces addresses residents’ needs for physical and mental well-being, as well as the preservation of cultural heritage.	Intensive development has reduced public green spaces, and natural recreational resources are unevenly distributed. In response, the plan adopts an “Ecology+” approach to foster a green economy.	[56,57,58]

Table 2. Key ESs in the Taihu Lake region.

ESs Type	Description	Importance Within the Study Area	References
Climate Regulation (CR)	Green spaces reduce local temperatures and mitigate extreme heat events through evapotranspiration, shade, and reflection of long-wave radiation.	The built-up areas surrounding Taihu Lake experience elevated summer temperatures, exacerbated by industrial heat emissions that worsen the regional thermal environment. The plan proposes the creation of an ecological barrier around the lake to enhance carbon sequestration and mitigate heat island effects.	[38,50]
Stormwater Regulation (SR)	Retain and store rainfall runoff through vegetation, soil, and terrain to reduce peak flood flows and delay peak onset times.	Climate change has intensified the uneven distribution of rainfall both temporally and spatially, while rapid urbanization has led to a decline in surface permeability. The plan emphasizes the importance of utilizing the composite system of polders, river networks, and lakes to strengthen flood retention capacities.	[39,51]
Water Conservation (WC)	Green spaces promote rainwater infiltration, replenish groundwater, and maintain base flow and drinking water sources.	Taihu Lake serves as a vital water source for cities such as Shanghai and Suzhou city, with upstream areas like the Yili Mountain designated as key conservation zones for water sources.	[52,53,54]
Water Purification (WP)	Improve water quality by filtering, adsorbing, and decomposing pollutants through vegetation, soil microorganisms, and substrate.	To address issues like non-point source pollution and eutrophication, the plan outlines measures for developing lakeside wetlands and fostering regional collaboration in managing water systems. It also mandates the construction of an ecological interception system.	[20,53]
Habitat Quality (HQ)	Green spaces provide food, breeding grounds, and shelter for local flora and fauna, maintaining urban biodiversity.	Habitat fragmentation disrupts species migration and environmental pollution contributes to the decline of sensitive species populations. The plan calls for the protection of Taihu Lake’s critical ecological spaces to preserve regional biodiversity.	[53,55]
Outdoor recreation (OR)	Offering natural landscapes and ecological spaces addresses residents’ needs for physical and mental well-being, as well as the preservation of cultural heritage.	Intensive development has reduced public green spaces, and natural recreational resources are unevenly distributed. In response, the plan adopts an “Ecology+” approach to foster a green economy.	[56,57,58]

2.4. ESBs Identification

The study employed the SOM model to identify ESBs. The SOM algorithm, first proposed by Kohonen [59], simulates topological relationships in data through neural networks. It effectively retains the key features of the input data, captures spatial co-occurrence patterns of ESs, and provides a visual output. Compared to the K-means algorithm, SOM is better suited for revealing the spatial distribution characteristics of large sample sizes and complex combinations of ESs. Additionally, pre-comparison of clustering results from both algorithms showed that SOM offers stronger spatial continuity in clusters, aligning more effectively with the spatial associative characteristics of ecological processes. This further confirms that the SOM clustering results are robust and interpretable [60].

For the analysis, the study utilized the “kohonen” package 3.0.12 in R, with all ecosystem service metrics normalized. SOM clustering followed a standard analytical framework, and the final number of clusters was determined based on the specific needs of this study. On one hand, it aimed to maximize intra-cluster homogeneity and inter-cluster heterogeneity, comparing various clustering scenarios to ensure that the ecological service characteristics of each unit were consistently similar within ESBs while maintaining distinctiveness between clusters. On the other hand, considering ecological spatial management needs, the study also evaluated whether the ecosystem service combinations within each ESBs reflected typical regional characteristics and whether their multidimensional attributes aligned with the cognitive logic of management departments for ecological functional zoning. This ensured the feasibility of spatial decision-making.

2.5. Measurement of ES Comprehensive Benefits

2.5.1. Correlation Analysis and Selection of Typical ES Pairs

The study uses correlation analysis to examine the relationships between the values of various ESs within each ESB, select typical service pairs, and further explore the correlation between comprehensive ES benefits and scale characteristic indicators. Specifically, the correlation coefficient (r) was calculated within the range of [−1, 1], where positive values indicate synergistic relationships, negative values indicate trade-off relationships, and the absolute value reflects the strength of the relationship. Significant correlations were identified using significance tests (p value). Based on this, a significant correlation network was constructed for each ESB following the “dominant ES-statistical correlation-functional value” framework. This allowed for the assessment of the following three comprehensive benefit indicators.

2.5.2. Comprehensive Indicator Measurement

(1)

MESLI

The multiple ecosystem services landscape index (MESLI) is used to assess the extent to which a region maintains the provision of multiple ESs simultaneously [61,62]. This index helps to identify and understand the overall status of ESBs. MESLI characterizes the comprehensive supply benefits of multidimensional services within a region by summing the standardized values of each indicator. The formula is as follows:

$$\mathrm{MESLI}={{\displaystyle \sum }}_{\mathrm{i}=1}^{6}E{S}_i$$

(1)

where $E{S}_i$ represents the normalized standard value of the i-th ES type.

(2)

Coupling coordination degree

This study focuses on the synergistic development mechanisms of three types of service combinations. The first type consists of dominant services within each ESB, along with service combinations that exhibit the most significant synergistic effects. The second type involves strongly positively correlated service combinations with high value and concentration. The third type is when multiple services exhibit similar degrees of correlation, priority is given to combinations where both values are high and the ecological functions within the cluster are complementary.

To quantitatively measure these interrelationships, a coupling coordination model is introduced. This model, based on system coupling theory from physics, originally described interactions between systems through material flow, energy conversion, and information transmission [63]. In ES research, each service is treated as a system with independent assessment indicators. The interactions between them reflect nonlinear dynamic coupling processes [64]. This model simultaneously reflects the coupling degree (strength of association) and coordination degree (level of positive interaction) between key ecosystem service pairs, which is essential for guiding balanced development of the comprehensive benefits of green space ESs [65]. The calculation formula is as follows:

$$C=\sqrt[2]{{\displaystyle \frac{V_1\times V_2}{{\left(V_1+V_2\right)}^2}}}$$

(2)

$$T=\alpha V_1+\beta V_2$$

(3)

$$D=\sqrt{C\times T}$$

(4)

where C represents the coupling degree, indicating the extent to which two ESs enhance each other’s interactions, with a value range of [0, 1]. When C approaches 0, it signifies low interconnectivity between system elements and insufficient synchronization in their development. Conversely, when C approaches 1, it reflects a highly synergistic relationship between the two systems. V₁, V₂ quantifies the assessment value of the service pair. T denotes the coordination degree, which measures the overall efficiency of the different services. α and β are the weighting coefficients for the ESs, with all services being considered equally important, thus both are set to 0.5. D is the coupling coordination degree, which integrates the C and T parameters to characterize the level of synergistic development between the two services, with a value range of [0, 1]. A higher D value indicates a stronger synergistic relationship.

(3)

Trade-off Strength

The trade-off strength among ESs can represent conflicting demands or interests among different stakeholder groups. The trade-offs between multiple land management objectives need to be carefully considered [66] in order to achieve UGS sustainability. Therefore, the study focuses on three types of service combinations: first, the dominant service within the region and the service type with the most significant conflict; second, two strongly negatively correlated service combinations with high value and concentration; third, when multiple services exhibit similar correlation strengths, prioritize combinations with significant numerical differences but prominent trade-off effects. The corresponding root mean square error (RMSE) values are then calculated.

The RMSE measures the deviation between the actual and expected values of each ecosystem service, reflecting the strength of trade-off relationships between services. The RMSE ranges from [0, 1]. When the value approaches 0, it indicates weaker trade-off relationships between services; as the value increases, it indicates intensified resource competition between services, corresponding to enhanced trade-off effects. The calculation formula is as follows:

$$RMSE=\sqrt{{\displaystyle \frac{1}{n}}\times {\displaystyle \sum }_{i=1}^n{\left(E{S}_i-\overline{ES}\right)}^2}$$

(5)

where $\overline{ES}$ represents the average value of the ESs included in the calculation; $E{S}_i$ represents the normalized standard value of the i-th ES type.

2.6. Redundancy Analysis

Redundancy analysis (RDA) is a direct gradient analysis method primarily used to uncover correlation patterns among multiple sets of variable data [67,68]. The process begins by integrating multiple explanatory variables into more representative composite indicators. These indicators are then tested for their explanatory power in relation to the response variable, ultimately identifying key variables that are statistically significant. To ensure the reliability of the results, explanatory variables with a p-value less than 0.05 are typically selected.

In this study, an RDA model was constructed for each ESB, with five UGS scale characteristics indicators as explanatory variables and three ES comprehensive benefit characteristics as response variables (

Table 3. Measurement methods for UGS scale characteristics indicators.

Scale Characteristic Indicators	Calculation Formula
Green Ratio (GR)	${\mathrm{GR}}_i={\displaystyle \frac{A_{gi}}{A_i}}\times 100\%$ where Agi is the green space area within unit i; Ai is the area of unit i.
Tridimensional Green Volume (TGV)	${\mathrm{V}}_i=\sum (C{H}_{xy}\times R_x\times R_y)$ $TG{V}_i={\displaystyle \frac{V_i}{A_i}}$ where CH is the canopy height of the grid cell; Rx, Ry are the resolutions of the grid cell, with data of 10 m; Vi is the total three-dimensional green volume of unit i.
Largest Patch Index (LPI)	$LP{I}_i={\displaystyle \frac{\max \left(a_{ij}\right)}{A_i}}\times 100\%$ where aij is the area of green space patch j in unit i; max(aij) is the maximum area of green space patch j in unit i; Ai is the total area of unit i.
Mean Patch Index (MPI)	$MP{I}_i={\displaystyle \frac{{{\displaystyle \sum }}_{j=1}^n{a}_{ij}}{n{A}_i}}\times 100\%$ where aij is the area of green space patch j in unit i; n is the number of green space patches in unit i; Ai is the total area of unit i.
Patch Area Coefficient of Variation (PACV)	$PAC{V}_i={\displaystyle \frac{\sqrt{\frac{1}{n-1}\times {{\displaystyle \sum }}_{i=1}^n{\left(a_{ij}-\overline{a}\right)}^2}}{\overline{a}}}\times 100\%$ $\overline{a}={\displaystyle \frac{{{\displaystyle \sum }}_{j=1}^n{a}_{ij}}{n}}$ where aij is the area of green space patch j in unit i; n is the number of green space patches in unit i; Ai is the average area of green space patches in unit i.

). For the Taihu Lake region, a multidimensional evaluation framework for scale characteristics was developed through literature reviews and case studies. Five indicators were chosen: GR (planar dimension), TGV (three-dimension), and three landscape pattern indicators—LPI, MPI, and PACV [69,70,71,72,73].

A preliminary correlation analysis was first conducted to assess whether the five green space scale characteristics influenced the three comprehensive benefit indicators. Subsequently, the RDA model was applied to further investigate the combined impact of these scale characteristics on the ES comprehensive benefits, shedding light on how these factors collectively influence the overall structure of the ESBs.

3. Results

3.1. Classification of Ecosystem Service Bundles

To uncover the spatial structure and functional differentiation of ESs, this study begins by identifying ESBs through SOM clustering and analyzes their distribution and dominant characteristics.

Based on the assessment of six ESs, 7621 valid spatial units were subjected to SOM clustering. Using the “kohonen” package in R, six distinct ESBs were delineated after five iterations of optimization. The clustering results were evaluated using the silhouette coefficient [65,66,67], a metric that assesses clustering performance by quantifying both the cohesion of samples within clusters and their separation from neighboring clusters. The coefficient ranges from −1 to 1, with higher values indicating better internal compactness and clearer separation between clusters. In this study, the silhouette coefficient was 0.44, suggesting that the SOM algorithm effectively distinguished ESBs [67] and met the required level of clustering accuracy [29,67,68]. The spatial distribution map is shown in Figure 2, where ESB6 was excluded from subsequent statistical analyses due to its limited sample size.

Each ESB displays unique dominant functional characteristics and spatial differentiation. ESB1 consists of 981 spatial units, primarily located in the southwestern regions of the Tianmu Mountains, Xisi Mountain, and Xishan. This cluster is particularly notable for its dominance in four Ess—WC, HQ, OR, and CR—reflecting a diversified functional synergy and forming a core hub within the regional ecosystem. ESB2 is found in the suburban transition zone, primarily between Suzhou City and Wuxi City, as well as between Changzhou City and Huzhou City, encompassing 2091 spatial units. The contributions of various ESs to this cluster are relatively balanced. ESB3 is mainly located in the southwestern farmland areas, comprising 1656 spatial units. These regions are predominantly characterized by WP service, representing typical agricultural ES. ESB4 is distributed around lakes in the central and northeastern regions, adjacent to medium-to-large water bodies such as Taihu Lake, Chenghu Lake, Yangcheng Lake, Dianhu Lake, and Changdang Lake. This cluster includes 922 spatial units and relies heavily on blue-green spaces, with CR as the dominant ecosystem service function. ESB5 is located in urban areas and towns near mountainous regions, comprising 1848 spatial units. The cluster is primarily characterized by WC capability, supporting the ecological development needs of urban areas. ESB6 is scattered in suburban areas near Huzhou City in the southern region, comprising 33 spatial units. This cluster primarily provides SR and WC services. However, due to the limited sample size, it was difficult to support statistically significant analyses of correlations and impact mechanisms. Therefore, it was excluded from subsequent research discussions.

3.2. Identification of ES Values and Related Relationships

A detailed analysis was performed on the five ESBs. By dissecting the internal correlations within ESs, we conduct an in-depth study of the value of each ESB feature within ESs and their interrelated characteristics. Combining the numerical features of ESs (see Figure 3a) with the correlation coefficients and significance levels between two ESs (see Figure 3b), we focus on identifying typical synergistic service pairs and trade-off service pairs.

The results of the K-S normality test revealed that the ecosystem service capacities within these five clusters do not follow a normal distribution. Consequently, Spearman’s rank correlation analysis was applied to the ecosystem service capacities across all five ESBs. The numerical characteristics of each ES and the analysis of trade-off and synergistic relationships are shown in Figure 3.

ESB1 highlights the dominant roles of CR, WC, HQ, and OR services, with OR exhibiting the highest average value (0.828). OR shows significant synergistic relationships with CR, WC, and HQ, with the strongest synergy observed with HQ. However, OR also exhibits a significant trade-off relationship with WP. Therefore, “OR-HQ” is identified as a typical synergistic service pair, while “OR-WP” is a typical trade-off service pair.

ESB2 features a relatively balanced demand for various services, with WC, WP, and OR showing the highest values (mean values of 0.383, 0.233, and 0.389, respectively). Notably, WP and OR have a strong significant negative correlation, indicating that “WP-OR” is a typical trade-off service pair. In contrast, CR and HQ exhibit the strongest positive synergistic correlation, making “CR-HQ” a typical synergistic service pair.

ESB3 is dominated by WP, which has the highest mean value (0.486), followed by WC (0.418). WP shows significant positive correlations with SR, WC, and HQ. Given that the value of WC is significantly higher than that of SR and HQ, “WC-WP” is chosen as the typical synergistic service pair within this cluster. Moreover, the four ES components, other than CR and WP, exhibit significant negative correlations, with CR being the key trade-off point. Based on the comparison of ES values, “WC-CR” is selected as the typical trade-off service pair.

ESB4 is characterized by CR, which holds the highest mean value (0.416), followed by OR (0.394), WC (0.278), HQ (0.245), and WP (0.176), with the latter showing no significant differences among them. CR services exhibit a strong synergistic relationship with WC and HQ, with the highest correlation coefficient found with HQ. Therefore, “CR-HQ” is identified as a typical synergistic service pair. In addition, WP and HQ exhibit the strongest negative correlation, making “WP-HQ” the typical trade-off service pair.

ESB5 is primarily dominated by WC, which has the highest mean value (0.653), followed by OR (0.423). WC exhibits strong trade-off relationships with CR, WP, and HQ. Among these, the strongest negative correlation is observed between “WC-HQ,” making it a typical trade-off service pair. OR shows a strong significant trade-off with WP and a strong synergy with both CR and HQ, with the highest correlation coefficient found with HQ. Hence, “OR-HQ” is identified as a typical synergistic service pair.

3.3. Comprehensive Impact of UGS Scale Characteristics on ESs

To understand the drivers behind ESs variations, we further quantified the influence of UGS scale characteristics on ESs performance using correlation and redundancy analysis. This study conducted a preliminary correlation analysis, which confirmed that five UGS scale characteristics indicators influence three comprehensive ecosystem service benefit indicators. Redundancy analysis was then applied to quantify the explanatory power of each scale characteristic indicator on the multidimensional comprehensive benefits of ESBs (

Table 4. Explanatory power, variance contribution, and significance of UGS scale characteristics indicators on the comprehensive benefits of ESs.

Scale Characteristics Indicators		GR	TGV	LPI	MPI	PACV
ESB1	Explains %	49.7	1	0.5	0.7	<0.1
	Contribution %	95.7	2	0.9	1.4	<0.1
	P	0.002	0.002	0.012	0.002	0.354
ESB2	Explains %	14.3	0.2	0.8	0.1	3.3
	Contribution %	76.2	1.3	4.5	0.6	17.3
	P	0.002	0.25	0.016	0.532	0.002
ESB3	Explains %	13.1	8.1	2.3	<0.1	1.1
	Contribution %	53.3	32.9	9.4	0.1	4.3
	P	0.002	0.002	0.006	0.944	0.144
ESB4	Explains %	21	1	1.4	<0.1	2.7
	Contribution %	80.1	3.9	5.3	0.4	10.3
	P	0.002	0.056	0.032	0.684	0.006
ESB5	Explains %	45.5	0.7	1.9	0.4	4.9
	Contribution %	85.1	1.4	3.6	0.7	9.2
	P	0.002	0.002	0.002	0.006	0.002

Note 1: Green ratio (GR), tridimensional green volume (TGV), largest patch index (LPI), mean patch index (MPI), and patch area coefficient of variation (PACV). Note 2: Explains represents the percentage of total variance in the comprehensive benefits of ESs explained by individual UGS scale indicator, reflecting their explanatory power. A higher value indicates stronger explanatory power for that indicator. Contribution represents the contribution of individual UGS scale indicator to the construction of the RDA ordination axis, reflecting their relative importance. A higher value indicates greater relative importance for that indicator. p-value obtained through permutation testing to assess the statistical significance of explanatory effects; p < 0.05 indicates significant results, shown in green in the table.

, Figure 4). In Figure 4, the length and angle of the arrows directly reflect the magnitude and direction of the explanatory power that the explanatory variables have on the variation in the response variables.

Regarding the importance of explanatory variables (Table 4), First, the GR ranked first across all five ESB categories, serving as the core influencing indicator. Notably, in ESB1, GR demonstrated the highest explanatory power and was statistically significant. Second, LPI showed some explanatory capacity for the comprehensive benefits of ESs across different ESB categories. However, its independent explanatory power and relative contribution were generally low, indicating a more limited influence. The TGV displayed significant differences in its influence. In ESB3, TGV had an independent explanatory power of 8.1% and a relative contribution of 32.9%, signaling a strong influence. However, in other ESB categories, its explanatory power was generally weak, with no significant effect in ESB2 and ESB4. This variation may be attributed to the different types and distribution characteristics of UGS in the study areas. The MPI exhibited weak explanatory power, with significant effects only in ESB1 and ESB5. Meanwhile, PACV was significant in ESB2, ESB4, and ESB5, but not in the other categories. This discrepancy might be linked to the different mechanisms through which the heterogeneity of UGS (as reflected by PACV) affects various ecosystems.

In terms of the relationships between explanatory variables and response variables (Figure 4), GR, TGV, LPI, and MPI all show a positive correlation with both the D and MESLI. Regarding the RMSE, GR and TGV exhibit a moderate inhibitory effect in ESB5, while LPI and MPI demonstrate a similar effect in ESB3 and ESB5, suggesting that they help mitigate competition among ESs. Overall, increasing the green space ratio, enhancing green volume, and preserving large natural patches can significantly enhance the comprehensive value of ESs. The PACV presents a more complex pattern. It shows a positive correlation with D and MESLI in ESB2 to ESB5, but a negative correlation in ESB1. Similarly, PACV is negatively correlated with RMSE in ESB1 and ESB5, and it also exhibits negative correlations in ESB2 to ESB4. These results reflect the variable mechanisms through which spatial heterogeneity affects ecosystem functioning in different urban contexts.

In summary, the scale characteristics of UGS exert a multidimensional influence on ESs. Among these, GR emerges as the most critical factor in enhancing the overall benefits of ESs. Although other scale characteristics—such as TGV, LPI, MPI, and PACV—vary in their influence, they also contribute to regulating the synergies and trade-offs among ESs across different spatial and ecological contexts.

4. Discussion

Based on the results, the study further analyzes the interactions between dominant ESs and interactive relationships across different ESBs—encompassing both stable synergistic effects and context-dependent trade-offs. The differential role of UGS scale characteristics underscores the importance of implementing refined UGS management strategies. Building upon these scale-dependent effects, this study proposes a multi-tiered UGS management framework aimed at enhancing ESs provision through strategic, targeted, and precise interventions to promote integrated and adaptive governance.

4.1. Spatial Differentiation and Synergistic–Trade-Off Mechanisms of ESBs

Through the identification and analysis of regional ESBs, this study reveals the pronounced spatial differentiation of the ecological functions of UGS. Under the influence of ecological resources and population density, different ESBs exhibit unique combinations of ESs and interaction patterns, including stable synergistic effects and spatially sensitive trade-offs.

The results indicate that ESs in the study area follow a distinct spatial gradient: mountain forests → farmland → lakes and marshes → suburban areas → urban areas. Within this gradient, ESB1 (southwestern mountain forest area) and ESB4 (central lakes and marshes area) serve as core functional zones for terrestrial and aquatic ecosystems, respectively. ESB2 (suburban transition zone) and ESB5 (urban built-up area) reflect the integrated and coordinated interplay between human activity and natural ecosystems. This spatial differentiation pattern vividly reflects the interaction between natural geographic conditions and anthropogenic influences [74]. In multifunctional, synergistic forested and lake–marsh areas (ESB1 and ESB4), low construction intensity helps enrich blue-green resources, enabling them to synergistically deliver ecological benefits primarily centered on natural attributes such as climate regulation, water conservation, and habitat quality. This positions them as critical elements within the regional ecological security framework. This aligns with the blue-green space protection strategy outlined in the Shanghai Metropolitan Area Spatial Coordination Plan. In medium-to-low density peri-urban areas (ESB2) and farmland regions (ESB3), all ESs develop relatively evenly at generally lower-to-medium levels, with water conservation and purification—critical functions for sustaining production and living conditions—maintained at a certain baseline. In high-density urban areas (ESB5), water conservation and outdoor recreation services are the primary ESs. These can effectively deliver high-level benefits by leveraging small-scale, highly connected green spaces, thereby ensuring residents’ quality of life. This finding is consistent with the ecological spatial development strategy proposed in the Shanghai Metropolitan Area Spatial Coordination Plan and further confirms that ESB-based analysis provides a robust scientific foundation for the designation of ecological functional zones and the development of differentiated management strategies in regional land-use planning [20].

Moreover, the study reveals that different ESBs exhibit distinct patterns of ecosystem service (ES) synergies and trade-offs. Two typical interaction patterns are identified: (1) stable benchmark relationships which remain consistent regardless of spatial context. For instance, the synergistic relationship between habitat quality and outdoor recreation is observed across all ESBs. In ESBs dominated by natural ecosystems, multi-service synergies are more pronounced. A notable example is the strong and stable synergy among climate regulation, water conservation, and habitat quality in ESB1, which aligns with previous findings by Li et al. and Lin et al. [20,75]. In urban core areas, the concentrated protection of small habitat patches and the establishment of green networks not only help preserve biodiversity but also provide ample recreational spaces for people [72,73]. (2) Spatially sensitive conversion relationships: These occur when the relationship between services shifts depending on spatial context. For example, water conservation and habitat quality are synergistic in ESB1 but exhibit a trade-off relationship in other clusters. Water purification and habitat quality are synergistic in ESB3 and ESB5, but trade-offs emerge in other clusters. Flood regulation shows extensive synergies in ESB2 and ESB3 but weakens or reverses in ESB4 and ESB5. Water conservation, acting as a synergistic hub in ESB1, becomes a trade-off node in ESB5. These patterns underscore the spatial heterogeneity and contextual dependence of ES interactions, emphasizing the need for tailored management approaches across different urban ecological zones.

4.2. Differentiated Driving Mechanisms of UGS Scale Characteristics

Research indicates that the impact of UGS scale characteristics on ESs varies significantly across different ESBs, pointing toward the need for diversified UGS management measures.

Redundancy analysis reveals that the GR consistently exhibits the strongest positive influence across ESBs, serving as the most stable and dominant indicator for enhancing the comprehensive benefits of ESs, particularly with respect to D and MESLI. This finding supports the “quantity-first” management principle [76], which emphasizes prioritizing green space coverage in urban planning. Especially in high-density urban areas, practical implementation may involve strategies such as green space replacement (e.g., demolition and redevelopment for greening) [77] and vertical greening (e.g., rooftop greening) [78] to effectively increase the available green area in densely built urban environments.

Notably, the impact of TGV exhibits clear spatial heterogeneity. In ESB3, TGV shows a high contribution rate, consistent with the ecological function of farmland shelterbelts in enhancing water purification services. This suggests that ecological engineering, particularly tree planting in agricultural landscapes, should be reinforced in these areas. However, in ESB2 and ESB4, the influence of TGV is statistically insignificant (p > 0.05), indicating that the marginal benefits of vertical greening vary by context. These results highlight the need for careful scenario-based assessment to avoid overreliance on vertical greening, which may lead to resource inefficiency or neglect of foundational horizontal green infrastructure.

Regarding spatial configuration indicators, the maximum patch area index (LPI) exerts a significant influence on ES enhancement, whereas MPI demonstrates a weaker effect, reaching significance only in specific ESBs. This highlights the differential impact of urban population density and construction intensity on ESs, indicating that patch-scale management should avoid a one-size-fits-all approach: Natural areas (ESB1) should prioritize the protection and expansion of continuous, concentrated green patches; Transitional and semi-urban zones (ESB2–4) should emphasize the development of large-scale, concentrated green spaces; Dense urban cores (ESB5) should adopt a balanced distribution of medium- and small-sized patches to maximize spatial efficiency. Special consideration should also be given to the dual effect of PACV. In ESB1, PACV is negatively correlated with service outcomes, indicating that landscape homogeneity is vital for functional stability in natural ecosystems. Conversely, in more artificial landscapes (e.g., ESB2–5), PACV exhibits a positive effect, implying that moderate heterogeneity among green patches can enhance the functional diversity of these systems.

In the context of ecosystem service trade-offs, all ESBs—except for ESB5, characterized by high-density development and limited green space—should implement scale constraints to prevent excessive green expansion from inadvertently suppressing the supply of other critical ESs. In ESB3, for instance, scale heterogeneity should be implemented [77].

The research results clearly show that in most situations, the “amount of green space” is more decisive than the “spatial layout,” but in specific ecosystems, such as high-density urban areas, limited land resources and high population density coupled with high demand for ESs necessitate particular attention to the rational layout of green space scale and its threshold effects. This ensures that limited green space areas deliver more critical ESs (such as habitat quality, outdoor recreation, water conservation, etc.) to precisely meet residents’ needs for natural spaces [79,80,81,82]. Therefore, regional landscape management needs to adopt differentiated strategies based on the protection of natural ecosystems, combined with land use and spatial allocation, to further refine spatial allocation plans, in line with China’s goals of compact and protective urban forms [77].

4.3. Smart Hierarchical Design of UGS Management Strategies

Drawing from the differentiated effects of UGS scale characteristics, this section proposes a multi-level UGS management framework aimed at maximizing ESs benefits through strategic, targeted, and precise interventions. The framework adopts a hierarchical approach, addressing strategic control, targeted implementation, and precise regulation.

At the strategic level, the GR should be established as the core control indicator, with differentiated standards tailored to specific ESBs. This foundational guarantee must be aligned with the current status and developmental potential of green space allocation across zones. Moreover, rigid constraints should be enforced through statutory land-use planning to ensure the preservation and gradual enhancement of green coverage. Specifically, ESB1 should maintain its current GR level of ≥62.6%, while other zones should show continuous improvement. Drawing on China’s Urban Green Space Classification Standards CJJ/T85-2017 [33] and relevant ecological protection regulations, recommended baseline of GR is ESB3 ≥ 40%, ESB5 ≥ 30%. At the implementation level, targeted actions should be developed to optimize green space scale and spatial configuration based on the unique ecological and spatial features of each ESB: For ESB1, large, continuous green patches must be strictly protected to prevent fragmentation from construction activities. In ESB3, priority should be given to establishing a network of tree-lined protective forests—with widths of 15–30 m and a minimum coverage rate of 10%—to strengthen three-dimensional green density and ecological resilience [83,84]. For ESB5, under existing urban development constraints, optimization of patch size distribution is key. Xu et al. recommend that individual patch areas account for 3.00% to 62.50% of total green space, with 20.00% identified as the optimal benchmark [85]. Sheng and Wang [86] further advocate for scale-specific configurations based on construction density to ensure a sustainable and diversified supply of ESs [12]. At the precise regulation level, the focus shifts to identifying and managing key ecological nodes, especially threshold areas where service relationships undergo significant functional transformations. These include the following: the negative inflection point in the patch area coefficient of variation within ESB1 and the balancing thresholds between water conservation and other ESs in built-up areas. In such areas, targeted ecological restoration projects should be implemented, combined with further quantitative and qualitative research, to achieve refined and adaptive control of green functions.

In summary, this progressive three-tiered management framework of “basic assurance-spatial optimization-node regulation” not only ensures the foundational role of green space quantity but also enhances ecological benefits through fine-tuned spatial adjustments.

4.4. Research Limitations and Future Prospects

To enhance the contextual relevance of research findings, studies must acknowledge current limitations in data, methodology, and scope while outlining future research directions. The aim is to improve the precision and applicability of ecosystem services-based planning approaches.

Despite its contributions, this study has several limitations: (1) Data Limitations: The analysis relies on static spatial data, lacking insight into the dynamic coupling between green space evolution and service function responses. The long-term effects of drivers such as climate change and urban expansion remain underexplored. (2) Methodological Constraints: While redundancy analysis quantifies indicator contributions, the multi-scale interactions (e.g., regional–urban–neighborhood) are insufficiently addressed, limiting the understanding of cross-scale transmission mechanisms. Meanwhile, this study has not sufficiently explored the scale effects of ESs. The analysis of the spatial differentiation of ESBs and their relationship with green space scale was conducted solely at a single spatial resolution (1.2 km grid). A systematic evaluation of how different scale parameters might influence the inference of underlying mechanisms has not been undertaken, which to some extent limits the applicability of the findings. (3) Regional Bias: The study area is concentrated in China’s Yangtze River Delta, which may limit the generalizability of findings to arid, cold, or high-latitude regions.

To build upon these findings, future studies should implement the following measures: (1) Integrate remote sensing time-series data with ecosystem process models to construct dynamic coupling models, capturing lag effects and thresholds in green space evolution and ES responses. (2) Establish multi-level spatial sizes to further deepen the exploration and comparative analysis of dynamic differences in ESB identification and green space scale influence mechanisms across various unit scales. By integrating the analysis of cascading effects across different hierarchical levels—from blocks to watersheds—this approach aims to better guide the smart allocation and management of UGS across multiple scales and dimensions. (3) Develop multi-scale analysis frameworks to examine cascading effects from micro-level design (e.g., plant configurations) to macro-level spatial patterns, supporting a comprehensive optimization theory from site design to regional planning. Conduct cross-regional comparative studies to explore the variable roles of green space scale characteristics across climate zones and urbanization stages and establish regionally adaptive guidance systems. Investigate socio-ecological synergy mechanisms by integrating public preferences, policy frameworks, and management practices into the ES analysis, aiming to generate operational “nature–society” win–win strategies. These research directions will contribute to a more scientific and precise basis for green space governance and accelerate the transition from empirical to data-driven decision-making in urban ecological planning.

5. Conclusions

This study innovatively integrates the ESB theoretical framework, through the use of multidimensional scale characteristic indicators, reveals the differentiated impact of scale characteristics on the comprehensive benefits of multiple ESs. The findings offer not only a novel theoretical perspective on the spatial organization of UGS and their coupling relationships with ESs, but more importantly, they propose a green space scale classification and control framework that supports multi-objective, synergistic optimization. This framework provides strategic guidance for the precise adaptation and efficient management of UGS, particularly under the constraints of stock-based urban development.

The results reveal pronounced spatial heterogeneity in ESs, with the identified ESBs exhibiting two distinct patterns of synergistic and trade-off relationships. Among the scale-related variables, the green ratio emerged as the dominant driver of ESs performance, while the largest patch index functioned as a secondary but consistent influencing factor. The influence of other spatial metrics varied across different ESBs, underscoring the context-specific nature of scale–ES interactions.

By adopting an ESB-based analytical framework, this study provides a systematic examination of how UGS scale characteristics affect multiple ESs provision across multiple landscape scenarios. Building on these insights, we advocate shifting research from an “ecological perspective” to a “human–land coupling perspective,” integrating ES interrelationships with specific urban population demands. We propose a hierarchical UGS scale regulation framework—“basic assurance-spatial optimization-node regulation”—aimed at optimizing multifunctional benefits and promoting ecosystem synergistic effects. This framework demonstrates strong potential for enhancing the ecological effectiveness and spatial allocation efficiency of UGS, particularly within the constraints of high-density urban environments.

Overall, the findings offer a novel perspective for integrating multifunctionality into green space planning and governance. They contribute both a theoretical foundation and a practical strategy for addressing the complex challenges of balancing ecological functions and spatial demands in contemporary urban landscape management. Nevertheless, this study has certain limitations. These include a relatively uniform spatial granularity and scale, as well as limited consideration of the evolutionary dynamics of UGS patterns and the corresponding response mechanisms of ESs. Future research should pursue cross-regional and multi-scale comparative analyses, incorporating emerging technologies such as big data analytics and artificial intelligence to further the precision and adaptability of UGS planning.