Ecological Security Pattern Construction for Carbon Sink Capacity Enhancement: The Case of Chengdu Metropolitan Area

Abstract

Constructing regional ecological security patterns (ESP) and enhancing carbon sequestration are essential for achieving China’s dual-carbon goals. However, rapid urbanization has intensified landscape fragmentation, disrupted ecosystem flows, and significantly altered urban ecological networks, weakening their carbon sink functions. Using the Chengdu metropolitan area (CMA) as a case study, a time-series ESP from 2000 to 2020 was constructed. Morphological Spatial Pattern Analysis (MSPA), the Minimum Cumulative Resistance (MCR) model, the gravity model, and complex network theory were employed to assess the spatiotemporal evolution and carbon sequestration capacity of the ecological network. Results revealed continuous declines in ecological sources and corridors, an initial rise then stabilization in resistance, and weakening connectivity, particularly in central and western subregions. Reductions in modularity and topological indices reflected lower ecological stability and greater vulnerability. Forest land served as the primary carbon sink, closely associated with multiple topological metrics; grassland sequestration correlated with clustering, while water bodies were positively linked to centrality measures. Adding 10 stepping-stone nodes and 45 corridors could enhance carbon sequestration by approximately 4156 Mg C yr⁻¹, with forests contributing 94.8% by 2020. This study provides scientific support for resilient regional ESP construction and dual-carbon strategy advancement.

1. Introduction

Excessive CO₂ emissions are accelerating global warming and intensifying extreme weather events, including heatwaves and floods, thereby posing a substantial threat to the sustainable development of human societies [1]. As the world’s largest developing country and one of the major emitters, China’s efforts to reduce emissions are pivotal to global climate governance [2]. To address this challenge, the Chinese government has proposed the “dual carbon” strategy achieving peak carbon emissions by 2030 and carbon neutrality by 2060 [3]. However, urban areas have emerged as major emission centers, accounting for 85% of China’s energy consumption [4]. In particular, metropolitan regions driven by dense transportation networks and spatial agglomeration have intensified habitat fragmentation and weakened ecosystem carbon sequestration [5,6]. Under the dual-carbon context, balancing urban development with ecological protection to improve regional carbon sink potential has become an urgent priority.

Against this backdrop, the ecological security pattern (ESP) has emerged as a critical spatial approach to enhance carbon sequestration potential and mitigate urban carbon emissions [7]. ESP studies have established a relatively mature technical framework encompassing “ecological source identification—resistance surface construction—ecological corridor extraction” [8]. In the early stages, nature reserves, ecological redlines, or large ecological patches were commonly designated as ecological sources [9]. With further research advancements, methodologies such as Morphological Spatial Pattern Analysis (MSPA) and ecosystem service value assessments have been incorporated to build a more integrated ecological source system [10,11]. Ecological resistance surfaces, describing landscape resistance to species migration and dispersal, are typically constructed using expert scoring or parametric models [12]. Ecological corridors—ribbon-like regions connecting ecological sources—facilitate the flow of energy, materials, and information and are extracted using models such as the Minimum Cumulative Resistance (MCR) model, circuit theory, and gravity models [12,13]. ESP has been widely applied in constructing ecological networks for forests and watersheds [14], playing a key role in improving landscape connectivity, preserving biodiversity, and mitigating habitat fragmentation [15]. With the deepening of ecological network research, carbon emission reduction and carbon storage have become emerging focal areas [16,17]. The research focus has gradually shifted toward multi-objective optimization, aiming to enhance regional carbon storage capacity through corridor construction and spatial land optimization [18]. For example, Liang et al. [17] achieved carbon-neutral landscape reconstruction by optimizing the spatial layout of desert mining areas [19]; Qiu et al. [20] proposed strategies based on ecological network theory to enhance the carbon sequestration function of forest ecosystems; and Fang et al. [21] found a significant correlation between ecological network topology and vegetation carbon sink capacity. These findings indicate that optimizing ecological networks can not only enhance carbon sink efficiency but also improve the overall stability of ecosystems.

Despite notable research progress, several gaps and deficiencies persist. First, carbon sink functionality has not been systematically integrated into the indicator systems for identifying ecological sources and corridors, leading to an overemphasis on structural connectivity while overlooking spatial variations in carbon benefits [22]. Second, research on ecological networks in regions with high human activity intensity, such as metropolitan areas and urban agglomerations, remains relatively limited, and there is a lack of optimization pathways tailored to the complexities of urban spatial structures and land use conflicts [17,23]. Third, most current ESP constructions fail to adequately address multi-objective synergies, particularly the trade-off mechanisms between ecological connectivity and carbon sink capacity [24]. Moreover, redundant corridors, broken nodes, and insufficient connectivity are common issues within ecological networks, significantly restricting the realization of carbon sink functions [18,25]. Therefore, there is an urgent need to establish an ecological network optimization system that integrates “structure-function-carbon benefits” to enhance the carbon sequestration potential and sustainability of megacity ecosystems [26,27]. To overcome these challenges, complex network theory provides new perspectives on the topological optimization of ecological security models [28]. By incorporating topological indicators such as degree centrality, betweenness centrality, and clustering coefficients into network analysis [14,29], it becomes possible to quantitatively evaluate the ecological cycling capacity of key nodes, offering structural support to enhance carbon sequestration [30]. In topologically optimized ecological networks, species migration becomes more efficient, vegetation restoration potential increases, and carbon storage capacity is significantly improved [31,32]. Therefore, investigating how metropolitan ecological networks affect carbon sequestration capacity can offer a scientific basis for network optimization, supporting carbon neutrality targets and advancing the sustainable development of regional ecosystems.

This study focuses on the CMA. Ecological sources were identified through a combination of Morphological Spatial Pattern Analysis (MSPA) and Conefor, incorporating ecological importance and ecological redline data. Potential ecological corridors were initially delineated using the Minimum Cumulative Resistance (MCR) model and further refined with the gravity model to construct the ecological security pattern. The resulting ecological network was abstracted into a topological structure, with ecological sources represented as nodes and corridors as edges. Carbon sequestration capacity was assessed using land use-specific carbon coefficients and analyzed relative to network topology. Key patches with degraded ecological function or low carbon sink capacity were identified, and an optimization strategy involving the addition of new corridors and stepping-stone nodes was proposed and verified through comparative analysis. This research aims to address three core scientific questions.

(1)

What were the spatial and temporal characteristics of the ecological network and carbon sink capacity in the CMA from 2000 to 2020?

(2)

Does a significant relationship exist between the topological structure of the ecological network and its carbon sink capacity?

(3)

What strategies can be developed to optimize the ecological network structure and improve carbon sink potential in metropolitan regions?

2. Materials and Methods

2.1. Study Area

The CMA is located in the western Sichuan Basin (102°49′–105°27′ E, 29°15′–31°42′ N), covering 33,114 km² and encompassing four core cities: Chengdu, Deyang, Meishan, and Ziyang (Figure 1). As a densely populated and economically active region in western China, the CMA plays a pivotal role in the Yangtze River Economic Belt and the Belt and Road Initiative [33]. By 2022, it hosted over 31.2 million residents, contributed more than 38% of Sichuan Province’s GDP, and accounted for over 85% of its total carbon emissions [34,35]. The CMA also serves as an ecological barrier in the upper Yangtze River, encompassing key ecological zones such as the Longmen and Qionglai Mountains and Longquanshan Urban Forest Park [36]. However, rapid urbanization and transportation expansion have intensified landscape fragmentation and ecological degradation, substantially increasing carbon emissions. Thus, constructing a well-connected ecological network is vital for harmonizing urban development with ecological sustainability.

2.2. Data Source and Processing

This study employed datasets spanning 2000 to 2020, all standardized to a 30 m × 30 m resolution to ensure consistency and comparability. The data included land use and land cover change (LUCC), Digital Elevation Model (DEM), slope, topographic relief, Normalized Difference Vegetation Index (NDVI), water bodies, roads, and population (Figure 2). LUCC data were derived from Landsat imagery and categorized into six classes. Slope and topographic relief were extracted from DEM data, while NDVI was calculated using spectral bands. Vector data for roads and water bodies were rasterized, and population data were interpolated. All datasets were resampled to match the resolution of land use data, ensuring spatial accuracy and analytical reliability (

Table 1. The list of data used in this study.

Category	Data	Resolution	Data Source
Environment	LUCC	30 m	Globe Land global surface coverage data (https://www.webmap.cn/commres.do?method=globeIndex, accessed on 30 January 2025)
	DEM, slope, topographic relief	30 m	Geo spatial data cloud (https://www.gscloud.cn/, accessed on 30 January 2025)
	NDVI	30 m	Resource and Environmental Science Data Center of the Chinese Academy of Sciences (https://www.resdc.cn/, accessed on 30 January 2025)
Socioeconomic	Road network	-	Open Street Map (https://www.openhistoricalmap.org/, accessed on 30 January 2025)
	Water net
	Demographic data	-	Data from World Pop website (https://hub.worldpop.org, accessed on 30 January 2025)

).

2.3. Methods

The research framework (Figure 3) begins by identifying ecological sources through MSPA and Conefor, integrating ecological importance and redline data. Potential corridors were delineated using the MCR model and further refined with the gravity model to construct the ecological security pattern. The network was then abstracted into a topological structure, with sources and corridors represented as nodes and edges, respectively. Carbon sequestration was estimated based on sequestration coefficients and correlated with network topology. Finally, key nodes exhibiting degraded ecological function or low carbon capacity were identified, and an optimization strategy involving edge augmentation and stepping-stone insertion was proposed.

2.3.1. Ecological Security Pattern Construction

(1)

Identification of ecological sources based on MSPA

The Morphological Spatial Pattern Analysis (MSPA) method identifies ecologically significant patches within a study area using morphological algorithms [12]. Binary classification was performed in ArcGIS [5], and the data were processed in Guidos Toolbox 2.8, with forests, grasslands, and watersheds set as foreground, and cropland, unused land, and built-up areas as a background [37]. Landscape elements were classified into seven structural categories, including core, bridge, edge, and island. Given the extensive study area and high fragmentation, core patches larger than 2.5 km² were selected as candidate ecological sources using the eight-neighborhood method. MSPA results were supplemented by calculating potential connectivity (PC) and patch importance index (dPC) in Conefor 2.6 [38], with a 1000 m threshold and 0.5 connectivity probability. Patches with dPC > 0.20 were intersected with ecological redline zones to identify final ecological sources, classified into forest, grassland, and water categories. The formula is as follows [39].

$$PC={\displaystyle \frac{\sum _{i=1}^n\sum _{j=1}^n{a}_i{a}_j{p}_{ij}^{\ast }}{A_L^2}}$$

(1)

$$dPC={\displaystyle \frac{PC-P{C}_{remove}}{PC}}\times 100$$

(2)

where n is the number of patches, aᵢ and aⱼ represent the areas of patches i and j, Pᵢⱼ denotes the maximum dispersal probability between them, A_L is the total landscape area, and PC_remove is the connectivity index after patch removal.

(2)

Constructing resistance surfaces

Resistance surfaces quantify the barriers to species and energy flows during horizontal migration [40]. Considering the specific characteristics of the Chengdu Metropolitan Area and prior studies, eight resistance factors were selected: DEM, slope, topographic relief, NDVI, LUCC, population density, water network density, and road network density. Each factor was classified and assigned values based on ecological sensitivity, combining expert judgment with literature review. The Analytic Hierarchy Process (AHP) was employed to determine the relative weights of these factors [18]. Subsequently, a weighted overlay analysis in ArcGIS 10.2 was performed to generate the composite ecological resistance surface (

Table 2. Assignment of resistance factor.

Resistance Factor	Weight	1	10	20	40	70	100
DEM/m	0.15	216–553	553–936	936–1539	1539–2235	2235–3064	3064–7100
Slope/°	0.11	0–6.63	6.634–13.62	13.62–22.70	22.70–34.57	34.57–54.83	54.83–89.40
Degree of topographic relief	0.06	0.57–1	0.43–0.57	0.33–0.43	0.26–0.33	0.20–0.26	0–0.20
NDVI	0.10	201–254	186–201	169–186	145–169	109–145	0–109
LUCC	0.37	Forest	Grassland	Water	Farmland	Unutilized land	Construction land
Water network density	0.09	0–0.015	0.015–0.04	0.041–0.068	0.068–0.099	0.099–0.143	0.143–0.203
Road density	0.07	0–0.38	0.38–1.18	1.18–2.45	2.45–4.28	4.28–7.39	7.39–12.05
Population density	0.05	0–19.56	19.56–88.01	88.01–244.46	244.46–586.70	586.701–1271.19	271.19–2503.27

).

(3)

Extraction of potential ecological corridors

The minimum cumulative resistance (MCR) model, originally proposed by Dutch ecologist Knaapen [41], quantifies the least-cost path required for material or species flow between ecological sources. It is widely applied in ecological corridor design to enhance connectivity and ecosystem functionality [42]. In this study, ecological sources and resistance surfaces were utilized to calculate minimum resistance paths, thereby identifying potential corridors within the Chengdu Metropolitan Area. These calculations were conducted using the Cost Path tool in ArcGIS 10.2 [25]. The MCR formula is presented as follows [43]:

$$V_{MCR}=f_{min}\sum _{j=n}^{t=m}\left(D_{tj}\times R_t\right)$$

(3)

where f represents a positive function indicating the cumulative resistance from a spatial unit to all ecologically favorable source patches. D_tj is the distance between landscape unit t and source patch j, R_t denotes the resistance of unit t, and min is the minimum cumulative resistance to any ecological source patch.

(4)

Screening ecological corridors

The Gravity Model (GM) is widely utilized to identify ecological corridors by quantifying the interaction strength between landscape patches [44]. Higher gravity values indicate lower ecological resistance and greater connectivity, underscoring the importance of corridors linking ecological sources [45]. By applying the GM, key corridors can be prioritized to maintain ecological integrity and enhance landscape-scale connectivity. The GM formula is presented as follows [46]:

$$G_{ab}={\displaystyle \frac{N_a{N}_b}{D_{ab}^2}}={\displaystyle \frac{L_{max}^{2}ln{S}_{a}ln{S}_b}{L_{ab}^2{P}_a{P}_b}}$$

(4)

G_ab denotes the interaction force between patches a and b. N_a and N_b represent the weight coefficients of these patches. D_ab represents the standardized resistance of the ecological corridor between them, while L_max denotes the highest resistance observed in the study area. S_a, S_b, P_a, and P_b represent the area and resistance of patches a and b, respectively. L_ab indicates cumulative corridor resistance.

2.3.2. Ecospatial Network Topology

The ecological spatial network integrates ecological network and complex network theories, abstracting ecological sources and corridors as nodes and edges [47]. Network modeling was conducted using Gephi 0.10.0, a Java-based visualization tool. The landscape was transformed into an undirected ecological network and analyzed through six topological indicators: degree, betweenness centrality, closeness centrality, clustering coefficient, eigenvector centrality, and PageRank [39]. These metrics reveal interactions between ecological sources and network nodes, supporting spatial structure characterization [48]. Moreover, by linking network topology with carbon sequestration capacity, this method facilitates the identification of optimization strategies for ecological restoration and carbon enhancement (Table S1).

2.3.3. Calculation of Carbon Sequestration Capacity

Different land use types exhibit varying carbon sequestration potentials within the carbon cycle (

Table 3. Carbon sequestration coefficients for different land use types.

Land Use Type	Carbon Sequestration Factor (t/hm2a)	Bibliography
Forest	283.9	[52,53]
Grassland	143.8	[54]
Watershed	67.1	[55,56]

). Natural ecosystems such as forests, shrublands, grasslands, lakes, and wetlands are widely recognized as major carbon sinks [20,49]. In this study, the carbon sink capacity of ecological sources was estimated using carbon sequestration coefficients derived from land use classifications and previous studies conducted in China [50]. The primary land types in this study: forests, grasslands, and watersheds play a vital role in regulating the regional carbon cycle. To quantify their contributions, carbon sequestration coefficients for each land type were calculated based on established methodologies [51].

$$C_t=\sum _{i=1}^n{A}_i{S}_i$$

(5)

where C_t is the total amount of sequestered carbon; i is the land use type; A is the land area; and S is the sequestration factor.

3. Results

3.1. Ecological Security Pattern Construction and Analysis

3.1.1. Identification and Analysis of Ecological Source Site

The core area, identified as potential ecological source land through MSPA, covered 6762.18 km² in 2000, 6780.32 km² in 2010, and 6740.11 km² in 2020, reflecting a slight increase followed by a decline over the 20-year period. This trend suggests localized degradation of ecological patterns under continued urbanization. These core ecological sources are primarily located in the Longmen, Qionglai, and Longquan mountain ranges, as well as the Minjiang and Tuojiang river basins. They substantially overlap with ecological redline areas and function as key zones for maintaining regional ecological security and supporting carbon sequestration. In contrast, ecological patches in the central and eastern CMA are highly fragmented. Some ecological sources fall outside the redline boundaries and are mainly situated on the urban fringe, where severe patch fragmentation forms potential blind spots in ecological protection. Rapid urban expansion has disrupted ecological corridors, weakened landscape connectivity, constrained ecosystem carbon sink capacity, and created spatial ecological gaps, resulting in stronger ecological performance in the west and weaker in the east (Figure 4).

Using Conefor 2.6, woodlands, grasslands, and water bodies larger than 0.25 km² were selected, resulting in 16, 15, and 14 key patches (dPC > 0.2) across the three periods. These patches are primarily located in the southwestern part of the metropolitan area and serve as core ecological nodes that make the greatest contributions to landscape connectivity and carbon sink functionality. Based on dPC values and spatial characteristics, ecological sources were categorized into three levels: important, general, and secondary.

3.1.2. Analysis of Resistance Surfaces

The ecological resistance surface of the CMA for 2000, 2010, and 2020 was modeled using eight factors, including LUCC, DEM, and slope (Figure 5). Results indicated that overall ecological resistance increased between 2000 and 2010 due to rapid urban expansion and intensified human activities. From 2010 to 2020, resistance changes stabilized, suggesting a saturation of development intensity. Ecological resistance exhibited a distinct “high-center, low-edge” spatial pattern. High ecological resistance is primarily observed in the urban core of Chengdu and adjacent cities, including southern Deyang and central Ziyang. In these areas, high population density, extensive transportation infrastructure, and concentrated industrial activities collectively contribute to elevated resistance levels. Medium-resistance areas were located in urban–rural transition zones, including northeastern Deyang and southeastern Meishan, where development and natural landscapes are interwoven, and ecological risk remains uncertain. Low-resistance zones were primarily found in the forested Longmen, Qionglai, and Longquan Mountains, and in wetland and river valley regions around the Minjiang and Tuojiang Rivers, where ecological connectivity remains strong.

3.1.3. Ecological Corridor Analysis

Using the cost-path model and the iterative tool in ArcGIS, 120 potential ecological corridors were identified in 2000, 105 in 2010, and 91 in 2020, indicating a significant downward trend (Figure 6). This decline is primarily attributed to the reduction in ecological source areas and increased landscape fragmentation driven by urban expansion. Ecological corridors are primarily distributed in the Longmen and Qionglai Mountains to the west, and in Longquanshan Forest Park to the east, forming a spatial pattern characterized by edge concentration and central sparsity. The central city of Chengdu, eastern Deyang, Ziyang, and southeastern Meishan exhibited sparse corridor distributions and fragmented networks with weak connectivity. The significance of each ecological corridor was assessed using the GM. Corridors were categorized into primary, secondary, and general levels (Tables S2–S4). The number of primary corridors remained relatively stable (8–10), while secondary and general corridors decreased, reflecting a trend of increasing simplification and reduced redundancy. Priority should be given to restoring ruptured areas and establishing “stepping-stone” nodes to enhance network connectivity and carbon sink capacity.

3.2. Topological Characteristics of the Ecospatial Network

3.2.1. Topological Network Structure Overview

Ecological sources identified between 2000 and 2020 were represented as nodes, while ecological corridors connecting these sources were modeled as edges, thereby constructing the topological structure of the ecological network. The resulting network topology was analyzed and visualized using Gephi 0.10.0. The results revealed that the ecological network in the study area experienced significant structural transformations over the 20-year period (Figure 7). At a resolution of 1.0, the network consisted of three ecological communities in both 2000 and 2010, which decreased to two in 2020, indicating a decline in landscape heterogeneity and a weakening of ecological connectivity. This change is largely attributed to the loss of ecological corridors due to expanding construction land, which has diminished the region’s carbon sink capacity. Further analysis showed that the spatial distribution of ecological communities aligns with the ecological resilience pattern of the Chengdu metropolitan area: the network remained relatively stable in the central and western regions, whereas ecological degradation in the eastern and southeastern areas disrupted energy and material flows, reducing the potential for ecological restoration.

3.2.2. Degree and Betweenness Centrality

Degree centrality reflects the number of direct connections a node has with surrounding nodes, while betweenness centrality indicates its role as a bridge within the network. Analysis conducted using Gephi 0.10.0 revealed a consistent decrease in the average values of both metrics from 2000 to 2020 (

Table 4. Ecological network topology indicators.

Indicators/Years	2000	2010	2020
Degree	7.5	7.067	6.286
Pagerank	0.063	0.066	0.071
Closeness centrality	0.676	0.679	0.658
Betweenness centrality	3.75	3.51	3.50
Eigenvector centrality	0.601	0.671	0.662
Clustering coefficient	0.579	0.563	0.50

), indicating weakened connectivity, diminished structural redundancy, and increased ecological vulnerability (Figure 8a). Nodes 5, 10, and 14 in 2000, and nodes 9 and 12 in 2020, showed high values in both metrics, serving as crucial bridges for ecological flow and carbon transfer. The degradation of these nodes could diminish the carbon sequestration capacity of adjacent ecological sources.

Therefore, optimizing the ecological network should emphasize the preservation of these structurally vital nodes and incorporate them into Chengdu’s “Carbonwise Tianfu” carbon credit mechanism through targeted compensation and incentive policies to channel resources toward high carbon sink potential areas [57].

3.2.3. Proximity Centrality and Clustering Coefficients

Closeness centrality measures the average shortest path from a node to all other nodes, reflecting its spatial accessibility, while the clustering coefficient indicates the degree of local aggregation around a node. From 2000 to 2020, both indicators showed an overall declining trend (Table 4), suggesting increased distances between ecological sources and a more fragmented local structure (Figure 8b). Nodes 5, 11, and 13 (2000); 7, 9, and 15 (2010); and 6 and 10 (2020) exhibited low clustering coefficients, indicating weak local connectivity, limited carbon sequestration efficiency, and reduced system stability.

It is recommended to prioritize corridor construction in these areas to enhance local connectivity and designate them as priority restoration zones under Chengdu’s “Carbonwise Tianfu” mechanism, promoting resource investment through market-based carbon incentives.

3.2.4. Eigenvector Centrality and PageRank

Eigenvector centrality measures a node’s connection to other influential nodes, while PageRank indicates its importance in a random walk process. Both indicators peaked in 2010 and had slightly declined by 2020 (Table 4), suggesting a weakening of the network’s structural hierarchy under external disturbances (Figure 8c). Nodes 5, 10, 13, and 15 in 2000, and nodes 5, 9, 11, and 12 in 2020, had high PageRank values, functioning as key components for maintaining structural integrity and carbon linkages. In contrast, peripheral nodes 8 and 14 in 2020 showed low values, indicating limited contributions to carbon sinks.

Therefore, conservation efforts should prioritize high-impact nodes while also enhancing the functional role of low-centrality nodes. It is recommended to incorporate this approach into Chengdu’s “Carbonwise Tianfu” initiative by establishing a mechanism that couples carbon value with network topological influence.

3.3. Correlation Analysis Between Carbon Sequestration Calculations and Ecological Network Structure

3.3.1. Calculation of Carbon Sequestration

Ecological source areas were categorized into forest, grassland, and water bodies based on their corresponding land use types. Carbon sequestration was estimated for 16, 15, and 14 source sites within the CMA in the years 2000, 2010, and 2020, respectively, by applying land use-specific carbon coefficients (Table 2), with corresponding proportions shown in (

Table 5. Calculation of carbon sequestration for each ecological source.

Particular Year	Main Land Use Types	Area (ha)	Total Source Area (ha)	Proportion of Total Ecological Resources	Carbon Fixed (Mg C yr−1)	Proportion of Total Carbon Sequestration	Total Carbon Sequestration (Mg C yr−1)
2000	Forest	411,951.8	541,959.5	76.01%	116,953,116.6	87.17%	134,174,061.8
	Grass	110,787.8		20.44%	15,931,289.34	11.87%
	Water	19,219.91		3.60%	1,289,655.84	0.96%
2010	Forest	414,045.6	541,757.5	76.43%	117,547,537.1	87.53%	134,296,416.5
	Grass	106,641.6		19.68%	15,335,060.05	11.42%
	Water	21,070.33		3.89%	1,413,819.28	1.05%
2020	Forest	410,771.3	536,894.7	76.51%	116,617,965	87.51%	133,264,170
	Grass	106,692.6		19.87%	15,342,399.48	11.51%
	Water	19,430.78		3.62%	1,303,805.522	0.98%

). The results indicate a decline in both the total number of ecological source sites and the overall carbon sequestration capacity over the past two decades, primarily due to the reduction in source area numbers. Despite some variation in the carbon uptake of grasslands and aquatic systems, forests consistently contributed more than 87% of total sequestration, maintaining a dominant role. Spatially, ecological assets were concentrated in forested and grassland-covered zones such as the Longmen Mountains, Qionglai Range, Longquanshan Forest Park, and the Tuojiang River Basin. In contrast, areas such as central Chengdu, southeastern Deyang, and Ziyang exhibited limited ecological land, thus constraining regional sequestration potential. Strengthening both the quantity and quality of ecological source areas is critical to enhancing the carbon sink capacity of the metropolitan area.

3.3.2. Analysis of the Relationship Between Ecological Network Structure and Carbon Sequestration Capacity

To examine the relationship between carbon sink performance and ecological network structure, this study utilized Spearman’s rank correlation to assess how total and per-unit-area carbon sequestration correlate with six topological metrics across different ecological land types (Figure 9). In 2000, water-related ecological sources showed a positive correlation with the clustering coefficient (p < 0.05, r = 0.50), while other indicators displayed negative correlations. Per-unit carbon sequestration in forested areas was positively associated with node degree, closeness centrality, and eigenvector centrality (p < 0.05), with a mean r value of 0.51. By 2010, clustering coefficients remained significantly positively correlated with total carbon in both forest (p < 0.05, r = 0.62) and grassland (p < 0.01, r = 0.70) areas. However, forest carbon showed negative correlations with betweenness centrality, closeness centrality, PageRank, and eigenvector centrality. Grassland per-unit carbon was positively correlated with the clustering coefficient (p < 0.05, r = 0.63), while its negative correlations with betweenness and PageRank were not statistically significant. In water ecosystems, per-unit carbon was positively associated with degree, betweenness, closeness, and PageRank (p < 0.05), although the average r value of 0.54 indicated moderate rather than strong correlations. By 2020, forest areas continued to exhibit a positive correlation with the clustering coefficient (p < 0.05, r = 0.49), and their per-unit carbon sequestration also showed a positive association with eigenvector centrality (p < 0.05, r = 0.48).

Forest ecosystems are the primary contributors to carbon sequestration and exhibit strong correlations with various topological indicators of the ecological network. In contrast, grassland ecosystems show strong positive correlations mainly with clustering coefficients, while their negative associations with centrality indicators suggest a relatively decentralized role in the network. Aquatic ecosystems, on the other hand, display positive correlations between carbon sequestration intensity and several centrality metrics, highlighting their localized importance in the network structure. These findings underscore that carbon sequestration capacity varies significantly with ecological network topology. Accordingly, improving network structure—especially by increasing the number and quality of corridors connected to low-efficiency sources such as specific grasslands and aquatic zones near expanding urban areas—can enhance both connectivity and carbon sink performance. Strengthening corridor linkages between key ecological sources can also generate positive externalities that support adjacent economic development. This integrated approach is essential for enhancing carbon sink functions in metropolitan regions and provides a sound foundation for land use planning and unlocking the value of regional land resources.

4. Discussion

4.1. Optimizing Regional Ecological Security Patterns to Enhance Regional Carbon Sequestration

This study reveals the key ecological challenges facing the Chengdu metropolitan area amid rapid urban expansion between 2000 and 2020 (Figure 10), including fragmentation of ecological source sites, reduction in ecological corridors, declining spatial connectivity, and weakening carbon sink capacity. These transformations have undermined the structural integrity and ecological functionality of regional systems, thereby constraining the realization of China’s national “dual-carbon” strategy. Empirical evidence shows a significant positive relationship between ecological network topology and carbon sequestration performance, indicating that higher spatial connectivity enhances both ecological energy flow and carbon sink efficiency. Therefore, constructing a well-connected ecological spatial network is essential for improving carbon sequestration potential.

Using the 2020 optimization scenario as a case study and drawing on the framework proposed by Qiu et al. [50] network enhancement can be achieved by increasing ecological corridors and inserting “stepping-stone” nodes to shorten paths between low-efficiency nodes. These stepping stones serve as relay habitats, reducing migration risks and enhancing species movement success rates. Based on this approach, this study optimized structurally weak nodes (e.g., nodes 6, 10, 11, and 15), resulting in the addition of 10 new ecological nodes and 45 corridors, significantly improving network accessibility and topological balance. Compared to the baseline scenario, total carbon sequestration increased by approximately 4.16 million tons, with marked improvements in both structural stability and operational efficiency of the ecological network. Structural analysis based on carbon sink contributions revealed that forest land accounted for 94.8% of the increase, confirming its dominant role. Grassland and water bodies contributed 1.0% and 4.2%, respectively, further underscoring the critical role of forests in enhancing connectivity and carbon sequestration performance.

Newly added corridors were concentrated in areas previously lacking ecological sources, such as eastern Deyang, northwestern Ziyang, and southern Meishan. The optimization process not only enhanced ecological integrity but also unlocked considerable spatial potential for land value transformation. This paves the way for developing a synergistic model integrating ecology, carbon sinks, and economic growth. These findings align with similar studies conducted in Yunnan and the Jiaodong Peninsula, validating both the theoretical underpinnings and regional adaptability of the methodology [26,27]. Overall, the optimized ecological security pattern in the Chengdu metropolitan area not only improves carbon sink capacity and spatial connectivity but also provides a strategic interface between ecological restoration and the pursuit of carbon neutrality.

4.2. Strengths and Recommendations

This study constructs the ecological security pattern of the CMA from a time-series perspective (2000–2020), addressing the previous lack of dynamic analyses on ecosystem evolution in metropolitan regions. Comparative analysis across different periods reveals the interrelated evolution of ecological source fragmentation, declining network connectivity, and the degradation of carbon sink capacity. Methodologically, this study adopts complex network topology analysis to quantitatively assess the relationship between ecological network structure and regional carbon sequestration capacity through the application of relevant topological indicators [30].

Moreover, ecological source areas were overlaid with ecological redlines to identify both high-value ecological patches and conservation gaps, thereby providing a scientific basis for formulating differentiated protection and restoration strategies. It is recommended that areas already designated within the ecological redline be strictly protected; buffer zones should be established for source areas located outside the redline but possessing high ecological value; and regions that are highly fragmented yet exhibit significant restoration potential should be designated as priority zones for ecological restoration [58]. At the practical level, the study suggests implementing carbon compensation mechanisms in ecologically fragmented, low-centrality marginal areas through platforms such as “CarbonSmart Tianfu” to incentivize societal participation in restoration. It further recommends developing a carbon sink evaluation framework based on network optimization to foster synergies among policies related to carbon sink evaluation, ecological conservation, and carbon trading. Overall, this research provides scientific support for optimizing ecological security patterns and offers theoretical guidance for institutional design and policy innovation toward carbon neutrality in future metropolitan areas.

4.3. Limitations and Challenges

This study employs Spearman correlation analysis to explore the relationship between ecological network topology and carbon sink capacity. As a non-parametric method, it is suitable for detecting monotonic relationships and handling non-normal data. However, compared to Pearson or partial correlation, it offers limited precision in quantifying network effects on carbon stocks. Future research should consider integrating multiple correlation methods for validation. Additionally, land use classification in this study is relatively coarse; although forests, grasslands, and water bodies are included, ecosystems like wetlands, shrublands, and agroforestry also store significant carbon. A more detailed classification could improve assessments of ecosystem contributions to carbon balance. Moreover, future work should simulate carbon sink dynamics under different ecological security scenarios—such as urban growth, restoration, and policy changes—by combining spatial optimization models (e.g., PLUS or FLUS) with carbon accounting frameworks to support adaptive carbon neutrality strategies.

5. Conclusions

This study explores how optimizing the ecological spatial structure of the CMA influences carbon sink enhancement. It also proposes an ecological network optimization framework based on the ESP. Based on the analysis of the structure, function, and dynamic characteristics of the ecological network from 2000 to 2020, key topological indicators were employed to quantify their relationship with carbon sink capacity, and optimization strategies were derived through Spearman correlation analysis. The main conclusions are as follows:

(1)

Between 2000 and 2020, the area of ecological source sites first expanded and then declined, indicating an increase in regional ecological fragmentation. Ecological resistance displayed a spatial pattern of “high in the center and low at the edges”, with ecological networks concentrated in the west, while ecological foundations in the central and eastern areas remained weak and urgently require enhancement.

(2)

The ecological spatial network exhibited signs of degradation, with network modules decreasing from three to two. Although the core area remained stable, the southeastern region became marginalized due to corridor disruptions, and key topological indicators declined, reflecting diminished connectivity.

(3)

Forests consistently served as the primary carbon sink (>70%) and exhibited significant correlations with multiple network metrics. Grassland sinks were influenced by clustering coefficients, while water bodies showed positive associations with centrality, highlighting their role in localized connectivity.

(4)

The addition of 10 new stepping stones and 45 ecological corridors in 2020 increased the annual carbon sink capacity by 4.16 million C yr⁻¹, with forest ecosystems contributing up to 94.8%, significantly enhancing network connectivity and supporting regional carbon neutrality goals.

Although this study focuses on the CMA as a case study, many metropolitan regions and urban agglomerations face similar challenges. By emphasizing ecological restoration, carbon sink enhancement, and the optimization of ecological security patterns, this research provides both theoretical foundations and practical guidance for the formulation of carbon sequestration policies.