Modeling the Ecological Network in Mountainous Resource-Based Cities: Morphological Spatial Pattern Analysis Approach

Abstract

Landscape fragmentation in mountainous resource-based cities has become increasingly serious, particularly in blue-green spaces. This study aims to establish a quantitative theoretical foundation for constructing an ecological network using the integrated morphological spatial pattern analysis (MSPA)–Conefor–minimum cumulative resistance (MCR) model. It employs multiple data sets, including land use data, remote sensing images, Shuttle Radar Topography Mission (SRTM) digital elevation, vegetation coverage data, etc., to conduct the quantitative analysis. Five groups of spatial resolution datasets (i.e., 30 m, 60 m, 90 m, 150 m, and 300 m) are employed for comparison and selection through MSPA to identify and analyze core landscape types. Connectivity analysis uses Conefor 2.6 software, and ecological sources are selected accordingly. Subsequently, the MCR model is applied to construct ecological corridors. Moreover, 153 ecological corridors are delineated, comprising 78 primary and 58 secondary corridors. The results show that most ecological core patches are fragmented and dispersed, while ecological corridors are vulnerable to disruption by external interference. This study also identifies 470 ecological breakpoints, mainly concentrated in the northeast, central, and southwestern areas characterized by high corridor density and intense anthropogenic activity. Additionally, 39 biological resting points are primarily located in the central urban area, and peripheral areas show few or no such points. This suggests establishing additional biological resting points to facilitate species migration and diffusion and complement the ecological network. This research addresses a significant gap in ecological network modeling within mountainous resource-based cities by developing a blue-green ecological network model. The findings encourage ecological governance bodies and technical professionals to recognize the interdependent relationship between blue and green spaces. This study supports the formulation of targeted planning strategies and helps maintain the potential connectivity essential for ecological balance.

1. Introduction

1.1. Research Background

1.1.1. Urban Blue-Green Space Urgently Needs Reconstruction

Amid growing concerns over global climate change, ecological degradation, and resource depletion, the urban blue-green spaces fragmentation has become increasingly severe. Blue-green spaces refer to the aquatic systems (blue areas), such as rivers and lakes, and terrestrial vegetation systems (green regions), which include parks, wetlands, and green belts [1]. These natural or artificial spaces are critical in protecting and sustaining urban ecosystems, maintaining ecological stability, enhancing resilience, and improving the overall quality of the urban living environment [2]. However, urbanization and excessive development have continually reduced blue-green spaces. Current urban blue spaces face systemic challenges such as diminished resilience to floods and droughts, widespread pollution, and the deterioration of ecological functionality [3,4]. Addressing these challenges requires improving hydrological cycle efficiency by integrating blue-green-gray infrastructure to create a multifunctional and synergistic ecological security framework. Simultaneously, urban green spaces face reduced thermal regulation capacity, environmental justice disparities, and fragmented governance [5]. To tackle these problems, the ecological network must be rebuilt while incorporating socio-natural system coupling mechanisms. This approach supports the development of a resilient landscape system rooted in nature-based solutions (NBS) based on interdisciplinary, collaborative governance models. At the same time, international environmental organizations promote ecological restoration and sustainable resource use. For example, the United Nations’ 2030 Agenda for Sustainable Development (United Nations General Assembly, 2015) states the sustainable urban development objectives, such as urban blue-green spaces’ preservation and urban ecological health’s promotion.

1.1.2. Blue-Green Space Reconstruction’s Needs in Mountainous Resource-Based Cities

Mountains cover about 22% of the Earth’s land surface and house over 900 million people worldwide; they are essential to global sustainability [6]. Among all countries, China has the largest mountain areas. They comprise about sixty-seven percent of its total area, implying the potential for responsible utilization and development [7]. They offer invaluable ecosystem services, such as biodiversity conservation, habitat provision [8], climate regulation [9], and support cultural services which are essential for human well-being [10].

Yet, these areas’ complicated topography and exceptional geographical features cause challenges [11]. Natural and anthropogenic factors, including urban sprawl, flooding, and forest fires, are degrading blue-green spaces in mountainous regions. Population concentrations in mountainous, resource-based cities also often cause imbalanced supply-demand dynamics between urban and suburban areas, which worsen ecological and environmental problems, such as resource depletion, energy shortages, and pollution [12].

The natural environment in mountainous resource-based cities is intrinsically fragile and highly sensitive to human disturbances. Due to the topographical limitations, land use is restricted, and space is scarce, necessitating effective planning and management. Studies have highlighted the challenges of integrating ecosystem services into landscape planning, governance, and decision-making processes [13], offering valuable insights into the ecosystem service functions’ practical implementation [14]. Currently, academic approaches to blue-green space reconstruction include green space planning, urban forest design, wetland restoration [15], water resource management, ecological landscape design, rainwater harvesting systems [16], and urban park planning, among others [17]. Most scholars aim to study ecological restoration and protection through blue-green space reconstruction [18,19].

1.1.3. Ecological Networks Promote the Construction of Blue-Green Space

Constructing rational ecological networks improves a city’s ecological carrying capacity. Researchers commonly employ ecological networks to restore fragmented blue-green spaces [20]. Some scholars utilize morphological spatial pattern analysis (MSPA) to identify core landscape types within ecological spaces [20,21]. Connectivity analysis software is then used to determine ecological sources. A “patch-corridor matrix” model is constructed to identify ecological cores and form robust ecological networks [20].

Consequently, ecological networks have become widely accepted and effective in academic research for protecting ecological environments. They also serve as a valuable reference in planning urban green space systems under national land use planning frameworks. Currently, ecological network research is gaining increased attention [21]. While research institutions focus on ecological network sustainability, limited research has focused on constructing such networks for fragmented blue-green spaces in mountainous resource-based cities.

1.1.4. A Mountainous Resource-Based City Case Study

To address this research gap, this study selects Panzhihua City, a typical mountainous resource-based city in China, as a case study. Located between 101°08′ and 102°15′ east longitude and 26°05′ north latitude, Panzhihua has a total land area of approximately 7440 km² [22,23]. The city has a subtropical climate and is the warmest area in Sichuan Province (Figure 1), with an average annual temperature of around 21 °C. In this city, forest land comprises 344,298.08 hectares, approximately 46.5% of the total land area. Water bodies and conservancy facilities cover 1.68% (12,468.14 hectares) of the land [22]. The region has complex terrain and significant elevation differences (Figure 1) exceeding 3000 m. It is predominantly mountainous, with basins accounting for only 0.16% of the land area [22,23]. Despite its ecological foundation, the city faces significant environmental protection pressures.

In recent years, rapid urbanization has led to the continued expansion of construction land. Ongoing mining activities, dominated by resource extraction and heavy industry, have an adverse impact on forest and water protection. These long-term economic activities have caused severe environmental degradation (Figure 2a,b). As shown in Figure 2a,b, the ecological patch fragmentation is becoming increasingly evident, with growing environmental problems and severely fragmented blue-green spaces requiring restoration. This paper uses ecological network theory to explore application gaps in national land use planning. It analyzes land-use classification, vegetation cover, and Landsat remote sensing data specific to this mountainous resource-based city.

1.2. Research Questions

(1) Given mountainous resource-based cities’ distinctive topography and industrial characteristics, this study emphasizes nature-based solutions (NBS). It integrates remote sensing and geographic information systems (GIS) to evaluate the distribution and function of the existing blue-green spaces. The first key research question is: How can we optimize the spatial layout in these cities with minimal intervention and enhance their ecosystem service functions?

(2) Blue-green spaces are a crucial element of the urban ecological network and significantly influence overall connectivity and environmental function. The second key research question is: How do ecological networks and blue-green spaces interact in mountainous resource-based cities?

(3) Mining activities, urban expansion, and industrial development have severely fragmented blue-green spaces in these regions. The third key research question is: How can an ecological network model be constructed to tightly couple fragmented blue-green spaces, providing a theoretical foundation for visualizing and quantitatively describing restoration and protection strategies for mountainous resource-based ecosystems?

1.3. Research Purpose

This study has the following objectives:

(1) To construct an ecological network model that tightly integrates fragmented blue-green spaces, offering theoretical support for restoring and protecting mountainous resource-based ecosystems.

(2) To provide insights to urban and rural planners and government officials for evaluating the ecological, economic, and social functions of blue-green spaces in industrial mountainous resource-based cities.

2. Literature Review

2.1. Ecological Network

An ecological network is a continuous and integrated system composed of fundamental elements such as environmental sources, corridors, and nodes [20]. Originally used to manage nature reserves, it has gradually been incorporated into urban and rural land use planning and other spatial management strategies. The ecological network has emerged as a core strategy for environmental protection. Ecological network research integrates network science tools, concepts, and models, particularly environmental studies [24]. As shown in Table 1, ecological network research covers multiple aspects.

Analyzing ecological networks of species diversity patterns usually employs various analytical methods. These methods reveal spatial distribution patterns of species diversity at different scales, offering vital insights for biodiversity conservation and management [25,26]. Some scholars have modified the minimum cumulative resistance (MCR) model by integrating network evaluation methods with night light data. For example, Miao et al. [27] analyzed the impact of road network development on fragmentation and connectivity in ecological networks, helping mitigate adverse anthropogenic effects on ecosystems and promoting harmony between humans and nature.

Table 1. The focus of ecological networks worldwide.

Country	Worldwide Research Focus
The United States	In-depth research on food webs, mutualistic relationships between plants and animals, and biodiversity maintenance [28].
Canada	The impact of ecosystem services and climate change on ecological networks.
The United Kingdom	Biodiversity conservation, ecosystem restoration, and urban ecological network planning [29].
Australia	Tropical ecosystem network structures, coral reef ecosystem functions, and conservation [30].
China	Urban landscape planning, ecological security pattern construction, urban expansion-conservation synergy, and administrative-division-oriented ecological network optimization [31,32].

Table 1. The focus of ecological networks worldwide.

Country	Worldwide Research Focus
The United States	In-depth research on food webs, mutualistic relationships between plants and animals, and biodiversity maintenance [28].
Canada	The impact of ecosystem services and climate change on ecological networks.
The United Kingdom	Biodiversity conservation, ecosystem restoration, and urban ecological network planning [29].
Australia	Tropical ecosystem network structures, coral reef ecosystem functions, and conservation [30].
China	Urban landscape planning, ecological security pattern construction, urban expansion-conservation synergy, and administrative-division-oriented ecological network optimization [31,32].

2.2. Morphological and Ecological Spatial Patterns

Scholars increasingly advocate interdisciplinary collaboration by integrating methodologies and theories from disciplines such as geography, ecology, and GIS to examine surface morphology, spatial patterns, and their influencing factors [33]. Remote sensing and GIS technologies are crucial in analyzing morphological spatial patterns using large-scale, high-resolution surface data [34]. For example, Lu et al. [35] used remote sensing and GIS technologies to study landscape morphology spatial patterns in agricultural areas, revealing the impact of human activities on their morphology. Huang et al. [34] studied the spatial distribution of land use and landscape pattern in a coastal gulf region of southeast China. Xu et al. [36] used digital elevation models to quantify the landform morphology spatial patterns in the mountainous areas, uncovering the spatial distribution patterns relevant to geological disaster risk assessments. Other studies have summarized calculation methods of various morphological indicators in urban planning and land use management, providing valuable insights for urban spatial planning [37].

Researchers such as Beroya-Eitner [38] and Xue et al. [39] have combined ecological security patterns with geological disaster sensitivity assessments to develop environmental restoration strategies that support the fragmented blue-green spaces recovery. Multi-scale morphological spatial pattern methods have also analyzed micro- to macro-level spatial patterns. These patterns often vary across scales. For example, the MSPA has demonstrated varying effects on forest fragmentation, with noticeable differences in settlement distribution and affinity propagation assessments [40]. Research on tropical rainforests shows that species diversity exhibits significant variations across spatial scales, essential for understanding these ecosystems’ structure and function [41]. A comprehensive, multi-scale analysis enhances our understanding of the complexity of surface morphology [35] and ensures more accurate assessments. However, few studies have systematically combined the MAPA–Conefor–MCR model to quantitatively construct ecological networks in mountainous resource-based cities—a vital research gap this study addresses.

2.3. MAPA–Conefor–MCR Model

2.3.1. MSPA

As highlighted in previous studies, MSPA is a significant approach in GIS. It is used to analyze geographic elements’ morphological characteristics and reveal spatial distribution patterns and relationships of geographical phenomena [40]. MSPA offers a solid foundation for urban planning, environmental management, and resource optimization. Researchers have applied MSPA in spatial data mining, spatiotemporal analysis, multi-scale analysis, and predictive spatial simulations [42]. This study uses MSPA to analyze mountainous resource-based cities’ spatial and morphological structures. It helps identify the landscape types in ecological spaces and analyze spatial structure and interactions across scales. Subsequently, core area patches with high ecological potential are selected as representative ecological source areas [42]. Using MSPA, the land cover data, which is converted into TIFF format, is processed by segmenting and eroding pixel elements, resulting in eight distinct raster layers representing various landscape types [26]. These include cores, bridges, and other relevant landscape categories. The core area consists of large contiguous foreground features; bridges connect fragmented patches, facilitating ecological flows. The core and bridges are further analyzed for landscape connectivity for identifying primary landscape types that affect connectivity. This process extracts blue-green ecological source areas hierarchically, supporting the optimized environmental network design.

2.3.2. Patch Connectivity Analysis (Conefor)

After selecting the core patches using MSPA, the landscape connectivity of larger core patches in the study area is evaluated to determine whether they qualify as ecological sources. In landscape ecology, connectivity evaluation indicators are essential for quantifying the environmental functional connections between landscape patches and assessing their significance in the ecosystem [43]. The most commonly used evaluation indicators include the integral index of connectivity (IIC), the probability of connectivity (PC), and the delta probability index of connectivity (dPC) values [44,45]. These indicators lay a solid foundation for constructing landscape ecological networks, ecological protection planning, and corridor design. The natural breaks method is employed for classification, helping identify core patches with significant environmental importance to be designed as ecological sources [46].

In ecological source selection, researchers typically calculate each patch’s IIC, PC, and dPC values and then classify these connectivity indicators using the natural breaks method. Higher connectivity patches are selected as potential ecological sources [47,48]. These are usually large core patches with high connectivity that function as critical migration bridges in environmental networks. This approach enables researchers to design effective ecological networks that enhance species migration capacity and mitigate the negative impacts of landscape fragmentation [49]. This methodology has been widely adopted in species conservation, landscape planning, and ecosystem service assessments [50]. In global change and habitat fragmentation, these quantitative indicators are robust tools for theoretical research and provide strong support for practical ecological protection and planning.

2.3.3. MCR Model

This study integrates the core area patches’ spatial scale to identify ecological source areas. It applies the MCR model to extract minimum-cost corridors, forming a comprehensive environmental network. The MCR model evaluates landscape ecological connectivity by calculating the least-resistance path between ecological nodes [51]. In the MCR model, resistance reflects the difficulty organisms face when moving between different locations in the landscape, typically influenced by terrain, land use, and anthropogenic activity [52]. MCR thus supports ecological migration path planning, ecological corridor design, landscape optimization, and ecosystem service evaluations. It also contributes to maintaining landscape biodiversity and population connectivity while broadening the research scope within the intersecting disciplines of landscape ecology, economics, and environmental science [53].

2.4. Analytic Hierarchy Process (AHP)

The AHP solves multi-criteria decision-making problems via a quantitative approach. It breaks a complicated decision problem into a hierarchical structure at several levels. Using this structure and a pairwise comparison matrix, the relative weights of each factor can be calculated [54]. It allows us to comprehensively evaluate and make decisions by integrating subjective and objective factors, offering a more scientific and accurate approach than traditional expert scoring [55]. It applies AHP to weigh the ecological resistance factors incorporated into the MCR resistance surface [56].

3. Research Methods

3.1. Research Framework

Constructing an ecological network and identifying its shortcomings and loopholes provide a theoretical basis for restoring, improving, and maintaining blue-green spaces in mountainous resource-based cities. The research framework is illustrated in Figure 3, which outlines the steps of this study. Initially, MSPA was used to reclassify categorical land use data. Core areas with larger patch sizes were then extracted, and Conefor was applied to calculate landscape connectivity indices and identify ecological sources. Resistance values were assigned to each relevant resistance factor based on findings from the existing literature. Subsequently, using ArcGIS 10.8, the Cost Path tool and the gravity model formula were employed to develop an interaction intensity matrix between ecological sources. Finally, an environmental network optimization model was constructed to enhance ecological planning.

The ecological network construction focused on the classification of ecological elements and the design of environmental corridors. Additionally, the “patch-corridor matrix” model described the landscape structure. In this model, patches (e.g., hedgerows, settlements, and woodlands) are embedded in a dominant matrix (e.g., forests), while corridors (e.g., river networks) serve as linear connectors that enhance ecological connectivity [20,57]. The environmental network was constructed based on the natural conditions of the study area and followed the fundamental principles of the “patch-corridor matrix” concept.

Based on supervised classification and MSPA, the key ecological elements identified are the natural components of blue-green space. It primarily includes woodland, grassland, and water bodies. By integrating the connectivity index and resistance modeling, this study simulated and evaluated blue-green ecological core areas and multi-level corridors to pinpoint key nodes and weak points in the environmental network, providing optimization direction.

3.2. Data Sources

The MSPA–Conefor–MCR integrated method was used to construct ecological corridors and identify blue-green ecological source areas at different spatial scales. This study utilized data from four key sources, as shown in

Table 2. Data information about a typical mountainous resource-based city.

No.	Types of Data	Time Range	Precision	Processing Method	Data Source
1	The current status of land use in Panzhihua City	2023	30 m resolution	The data underwent manual visual interpretation correction, topological verification, and reclassification to ensure analytical rigor	Panzhihua City Natural Resources and Planning Bureau http://zgj.panzhihua.gov.cn/
2	Remote sensing images of Panzhihua City	2023	Achieving a positional accuracy of <0.5 pixel	The data are processed with atmospheric correction and geometric registration	http://www.gscloud.cn/
3	SRTM digital elevation products	2023	30 m resolution	The data underwent void filling and projection transformation	http://www.gscloud.cn/
4	Vegetation coverage in Panzhihua City	2023	MODIS NDVI 250 m resolution	The data are processed by using the Maximum Value Composition (MVC) method to generate monthly composite datasets	http://www.gscloud.cn/

. GuidosToolbox (2.8) is a geospatial data analysis and visualization toolkit used to perform MSPA on reclassified land use data. Conefor software calculates landscape connectivity indicators and is a decision-support tool for landscape planning and habitat conservation. It identifies key landscape elements by quantifying connectivity using the IIC, PC, and dPC [58]. Conefor supports the computation of different dispersal distances to determine optimal connectivity thresholds [58].

4. Data Analysis

4.1. Landscape Type Identification Based on MSPA Technology

Land use, water bodies, green spaces, and other relevant factors in the study area were compiled to delineate habitat patches composed of blue-green space elements. This study categorized forest land, grassland, water bodies, and wetlands within the city into highly sensitive overlapping areas. The primary research targets included the following landscape elements: forest land, grassland, water bodies, and wetlands. Using MSPA, the core areas’ landscape categories were identified by extracting the foreground elements. The spatial extent and distribution of forests and grasslands in the city are vast. However, after spatial pattern analysis at 30 m resolution, it became difficult to distinguish landscape types other than core areas. Smaller landscape types with strong connectivity but limited areas could not be accurately identified. Therefore, a control group was established using the following five spatial scales for the foreground feature file (Figure 4): 30 m, 60 m, 90 m, 150 m, and 300 m [26,49,59,60,61]. The analysis was conducted using Guido’s software (2.8).

(1) 30 m

A scale of 30 m is typically used for high-detail landscape analysis, capable of detecting small green patches and water bodies. At this scale (Figure 4a), the core area appears mainly contiguous, with no distinguishable secondary patches.

(2) 60 m

The 60 m scale balances spatial detail and data processing efficiency. Most of the core areas remain connected at this scale (Figure 4b). However, extracting large contiguous zones as ecological sources is still impossible.

(3) 90 m

A 90 m scale is suitable for regional-scale ecological network analyses and can identify large patches and major corridors. At this scale (Figure 4c), the core areas remain mostly connected, and a few additional zones appear, but the distinction remains unclear.

(4)150 m and 300 m

Furthermore, 150 m and 300 m scales are used for macro-scale ecological network analysis, focusing on broader environmental processes and patterns [49,62]. Compared with the 30 m, 60 m, and 90 m results, the 150 m scale shows significantly lower fragmentation (Figure 4d). The 60 m and 90 m scales did not effectively identify medium-sized rivers and lakes or anthropogenically modified areas. Compared to the 300 m scale (Figure 4e), it better distinguishes human agglomerations, such as the center.

In conclusion, the 150 m scale offers an optimal balance: it enhances the details’ visibility without excessive fragmentation or overgeneralization, improves the medium and small rivers and lakes’ identification, and supports the core area patches selection with potential as ecological sources. This study refers to the MSPA scale effect theory proposed by Soille and Vogt [59] and Saura’s scale selection recommendations for landscape connectivity analysis [49]. Additionally, the MSPA-Conefor study by Peng in southwest China’s terrain is similar to Panzhihua City [60]. Based on these references, 150 m was ultimately chosen as the optimal research scale. This scale effectively identifies landscape core areas and ecological corridors while avoiding the small plot fragmentation illusion at higher resolutions and the loss of key corridors at lower resolutions.

Table 3. Landscape type classification statistics.

Type	Area (km2)	Occupancy in Foreground Elements	Occupancy in Study Area
Core area	4872.00	77.81%	65.72%
Islet	19.18	0.31%	0.26%
Perforation	322.46	5.15%	4.35%
Edge	315.31	5.04%	4.25%
Bridge	260.96	4.17%	3.52%
Loop	401.23	6.41%	5.41%
Branch line	70.42	1.12%	0.95%
Total	6261.56

and Figure 5 present the main landscape types and their corresponding statistics.

4.2. Landscape Connectivity Evaluation Based on Conefor

Using Conefor software, the IIC, PC, and dPC indices were calculated for the core patches identified as having potential ecological sources within the study area. The evaluation results are summarized as follows: (1), (2), and (3) [26,63]. Based on these indices, the importance of each patch was assessed. Several core patches were selected as ecological sources and ranked by importance using the natural breaks classification method [26] (

Table 4. Landscape connectivity index based on Conefor.

Rank	Connectivity Index (dIIC)	Probability Index of Connectivity (dPC)	Numbers
1	95.49	95.20	17
2	17.13	22.02	29
3	6.68	9.07	26
4	3.36	5.67	7
5	1.70	4.25	14
6	1.87	3.04	15
7	0.45	0.67	20
8	0.39	0.66	24
9	0.32	0.48	8
10	0.16	0.25	23
11	0.15	0.23	4
12	0.12	0.21	2
13	0.09	0.15	28
14	0.09	0.15	13
15	0.09	0.14	6
16	0.08	0.14	11
17	0.07	0.13	16
18	0.069	0.11	5

and Figure 6).

The dIIC and dPC indices are the major measurement systems for revealing ecosystems’ structural and functional characteristics. The dIIC results depicts the relationship between species diversity and spatial connectivity. A higher value suggests stronger species interactions in the landscape. The patch importance index (often depicted as dPC values) shows the degree of connectivity between ecosystem fragments. A higher dPC suggests stronger patch connectivity , leading to a more mixed and efficient ecological network.

$$\mathrm{I}\mathrm{I}\mathrm{C}={\displaystyle \frac{{\sum }_{i=1}^n{\sum }_{j=1}^n\frac{a_i{a}_j}{1+{nl}_{ij}}}{A_L^2}}$$

(1)

$$\mathrm{P}\mathrm{C}={\displaystyle \frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{a}_i\times a_j\times p_{ij}}{A_L^2}}$$

(2)

$$\mathrm{d}\mathrm{P}\mathrm{C}={\displaystyle \frac{PC-{PC}_{remove}}{PC}}\times 100\mathrm{\%}$$

(3)

$n$ is the total number of patches in the landscape; $a_i$ and $a_j$ are the areas of patch i and patch j respectively; ${nl}_{ij}$ is the number of connections between patch i and patch j; $A_L$ is the area of the background landscape. $p_{ij}$ is the maximum possibility of direct diffusion of species between patches i and j. ${PC}_{remove}$ is the overall index value of the remaining patches after removing a single patch [63,64,65].

4.3. Selection of Ecological Source Areas

Landscape patterns and use were considered during the selection process. Zhou et al. [66] demonstrated that lower thresholds (e.g., dPC > 0.1) can effectively enhance connectivity and link more dispersed ecological patches. Based on this criterion, 18 ecological patches with a dPC index > 0.1 and potential as environmental sources were identified as core areas for further analysis (Table 4 and Figure 7). Darker colors indicate higher dIIC and dPC values, representing better ecological network connectivity.

The study area exhibits a strong ecological foundation, as shown in Table 3 and Figure 5. Its core area covers 4871.997 km², accounting for 65.72% of the total area. Large patches within the core area are primarily located in the northern region. In contrast, the southern region, characterized by higher altitudes and multiple river systems, is less influenced by human activity. The central urban area is situated along the Jinsha River and is dominated by construction land. Due to the intense impact of human activity, the core area in this region is fragmented, forming a “fault zone” that disrupts connectivity from south to north. The edge zone comprises 5.04% of the total area and is a buffer between the core and peripheral zones. This buffer plays an essential protective role and must remain stable. The presence of small, isolated patches of islets indicates the degree of ecological fragmentation. Their proportion is about 0.31%, suggesting a relatively intact environmental condition. The bridge and loop are connecting channels. Branch lines provide good connectivity, and perforations, i.e., hollow areas, are within the core area. The four areas constitute 4.17%, 6.41%, 1.12%, and 5.15% of the area, respectively (Table 3). Overall, core patches exhibit low independence and are connected in contiguous formations. However, numerous perforations render their morphology unsaturated and susceptible to external environmental influences. A comparison of landscape distribution data uncovers that the ecological sources include national forest parks, ecological tourism areas, and nature reserves. Larger ecological sources are mainly located in the northern part, indicating the best connectivity.

4.4. Construction of Ecological Resistance Surface

The terrain of this mountainous resource-based city is characterized by steep undulations, high elevations, and numerous valleys. Due to these characteristics, relatively flat areas were selected for constructing ecological corridors. Given Panzhihua’s unique river valley topography and water system distribution, the environmental corridor direction aligns with river valleys to connect blue-green spaces. This helps improve the urban microclimate, enhances the water resources’ efficient use, and contributes to strengthening urban ecosystem services. Consequently, the construction of ecological corridors corresponds with the water network layout. On top of using current land use data, the distance between environmental corridors and water bodies was included as a resistance factor. The ecological resistance surface was constructed using ArcGIS 10.8 software through overlay and raster calculation, incorporating variables such as land use, slope, elevation, NDVI, and distance from water bodies. Resistance surface values are assigned using the expert scoring method, which may introduce subjectivity based on personal research backgrounds and experiences. It also may not objectively reflect the degree of obstruction of different land uses to specific species. To minimize the bias, the AHP [55,67] was employed to determine the weight of each factor. The following five key resistance factors were identified: NDVI, land use, elevation, slope, and distance from water. Drawing on the findings of previous researchers, the assignment of ecological resistance values of the five factors is as follows: “Weight (land use) > Weight (land cover) > Weight (distance from water) > Weight (elevation) > Weight (slope)” [39,68,69,70]. Five domain experts were consulted, and the Saaty 1–9 scale [71,72] (

Table 5. Saaty 1–9 scale [73].

Scale	1	3	5	7	9	2, 4, 6, 8
Meaning	Equal	Moderate	Strong	Very strong	Extremely strong	The median value of the adjacent scale

and Appendix A) was used to compare the relative importance of factors. A judgment matrix was constructed (

Table 6. A judgment matrix.

Factors	Land Use	Elevation	Slope	NDVI	Distance from Water
Land use	1	5	6	2	3
Elevation	1/5	1	2	¼	1/2
Slope	1/6	1/2	1	1/5	1/3
NDVI	1/2	4	5	1	2
Distance from water	1/3	2	3	1/2	1

), and the geometric mean method was applied to combine individual expert scores into a comprehensive matrix. This matrix was used to derive the weight of each factor (Figure 8). The calculation process is shown in the following equations: (4)–(6). Where n is order of the pairwise comparison matrix; $A$ is the pairwise comparison matrix, $W$ is the normalized weight vector; λmax is the maximum eigenvalue [73,74,75].

$${\lambda }_{max}={\displaystyle \frac{1}{n}}{\sum }_{i=1}^n{\left(AW\right)}_t/W_i$$

(4)

$$\mathrm{C}\mathrm{I}={\displaystyle \frac{({\lambda }_{max}-n)}{n-1}}$$

(5)

$$\mathrm{C}\mathrm{R}={\displaystyle \frac{CI}{RI}}$$

(6)

Weights were calculated using the eigenvector method, and a consistency test was performed. The consistency ratio (CR = 0.047) confirmed that the matrix satisfied the consistency requirements (i.e., CR = CI/RI < 0.1) [71,73]. If the CR had exceeded 0.1, adjustments would have been necessary. The maximum eigenvalue ${\lambda }_{max}$ was calculated using the judgement matrix and weight vector (W) [73]. To reduce errors, the CI was compared with the average random index (RI) [73]. As shown in Figure 8, the results indicate that land use type had the highest weight (0.4081), followed by NDVI (0.2373). Elevation (0.0993) and slope (0.0733) contributed less. A sensitivity analysis (coefficient of variation < 5%) validated the weight results [76].

GIS-based natural breaks were used to classify each resistance factor into five categories. Original values across these five intervals were obtained (

Table 7. Basic resistance values of each indicator.

NDVI
Original value	−0.26–0.11	0.11–0.24	0.24–0.33	0.33–0.39	0.39–1.00
Reclassified value	100.00	70.00	50.00	20.00	1.00
Land use
Original value	Water area	Forest land	Wetland	Grassland	Cultivated land	Construction land	Unused land
	4.00	2.00	5.00	3.00	1.00	6.00	7.00
Reclassified value	1.00	1.00	5.00	5.00	40.00	60.00	100.00
Elevation
Original value	838.00–1439.00	1439.00–1803.00	1803.00–2192.00	2192.00–2680.00	2680.00–4143.00
Reclassified value	1.00	20.00	50.00	70.00	100.00
Slope
Original value	0.00– 11.33	11.33– 19.66	19.66– 27.65	27.65– 36.98	36.98– 84.96
Reclassified value	1.00	10.00	50.00	70.00	100.00
Distance from water
Original value	0.00– 7212.93	7212.93– 16,006.42	16,006.42– 26,175.90	26,175.90– 38,959.54	38,959.54– 62,381.30
Reclassified value	1.00	20.00	50.00	70.00	100.00

). Reclassification was conducted using a 1–100 scale [39], where 1 indicates the lowest resistance and 100 is the highest, to improve the accuracy of the resistance surface and enhance the visual distinction between resistance values.

Each factor was divided into five levels based on the specific conditions of the study area, and corresponding resistance values were assigned (Table 7 and Figure 9). Finally, raster calculations were conducted using GIS, and the AHP-derived weights (NDVI = 0.24, land use = 0.41, elevation = 0.10, slope = 0.07, and distance from water = 0.18) were applied. The resulting ecological resistance surface is shown in Figure 10.

4.5. Ecological Network Construction Based on the MCR Model

This study established an MCR model using land use type as the basic element, starting from the blue-green space within a mountainous resource-based city. The focus was on selecting ecological sources and constructing resistance surfaces. Eighteen patches with dPC > 0.1 were identified as environmental sources using landscape connectivity and patch importance indices as indicators. Concurrently, the land use data for Panzhihua City was rasterized at a grid size of 30 m × 30 m to generate the cost data for the MCR model [26,68,77]. The cost-distance module in ArcGIS 10.8 was then used to combine the raster data of each ecological source with the cost data to calculate the MCR for each environmental source. Finally, an MCR model was constructed for the 18 ecological sources. In the cost path module of ArcGIS, these 18 ecological sources were treated as both sources and targets. Cost paths were calculated using the cost distance and backlink raster data, generating 153 potential ecological corridors (Figure 11).

4.6. Analysis of Ecological Network Structure Characteristics

This study integrated the data with the generated ecological resistance surface (Figure 10). The analysis revealed that resistance values are highest in the center and northeastern regions of Panzhihua, while lower resistance values are mainly found in the north, northwest, and southwest. As a result, this study identified 153 ecological corridors based on the MCR model, from which redundant corridors were eliminated to yield 136 effective environmental corridors. The higher the interaction intensity among ecological source patches, the more valuable and beneficial it is to establish and maintain corridors that facilitate the materials and energy exchange between species.

The interaction intensity matrix between source areas was established using the gravity model Formula (7) (

Table 8. The interaction intensity matrix between ecological source areas and environmental corridors is based on the gravity model.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
1		98.07	94.76	3.07	2.70	2.76	29.50	10.25	9.63	5.03	4.11	3.36	5.73	2.53	4.72	4.62	1.17	1.66
2			57.03	4.07	3.50	3.49	67.04	18.91	15.56	7.25	5.55	4.34	8.34	3.25	6.48	6.30	1.49	2.00
3				1.76	219.20	63.13	5.20	2.16	3.99	9.57	20.22	13.34	4.79	7.04	3.32	4.37	1.27	1.73
4					1.67	128.20	4.72	2.01	3.57	10.15	23.68	15.26	4.62	7.76	3.03	4.72	1.37	1.86
5						1.79	19.60	9.31	5.55	3.08	2.60	2.18	3.48	1.65	2.96	2.91	0.77	1.12
6							4.70	2.06	3.47	7.75	15.20	11.34	4.16	6.36	3.03	4.33	1.33	1.92
7								189.40	37.02	11.01	8.15	6.24	13.19	4.65	10.75	9.58	2.17	2.83
8									10.51	9.75	5.93	4.35	27.53	3.50	19.47	14.19	2.58	2.57
9										4.52	3.64	2.96	5.31	2.22	4.61	4.20	1.05	1.46
10											40.99	101.40	11.24	44.73	5.16	11.10	2.72	3.13
11												20.40	7.36	119.50	5.27	11.13	2.87	3.51
12													44.03	15.71	12.75	27.13	4.10	4.43
13														8.26	32.47	121.40	6.76	5.01
14															5.63	49.44	10.16	5.71
15																14.83	18.12	8.79
16																	3.71	4.46
17																		10.47
18

). The interaction intensity was classified into 78 primary ecological corridors with a value of 5 or more, 58 secondary corridors ranging from 2 to 5, and 17 general corridors with less than 2. The process established an ecological network model for Panzhihua City (Figure 12).

$$G_{ab}={\displaystyle \frac{N_a{N}_b}{D_{ab}^2}}={\displaystyle \frac{\left[\frac{1}{P_a}\times \ln S_a\right][\frac{1}{P_b}\times \ln S_b]}{{\left(\frac{L_{ab}}{L_{max}}\right)}^2}}={\displaystyle \frac{L_{max}^2\ln S_a\ln S_b}{L_{ab}^2{P}_a{P}_b}}$$

(7)

Based on Huang et al. (2019) [78], $G_{ab}$ is the interaction force between ecological source patches a and b; $N_a$ and $N_b$ are the weights of ecological source patches a and b respectively; $D_{ab}$ is the standardized value of the potential ecological corridor resistance between ecological source patches a and b; $P_a$ is the resistance value of ecological source patch a; $S_a$ is the area of ecological source patch a; $L_{ab}$ is the cumulative resistance value of the potential ecological corridor between ecological source patches a and b; $L_{max}$ is the maximum cumulative resistance values of the potential ecological corridors between all ecological source patches.

4.7. Analysis of the Current Situation of Blue-Green Space Coupling

Panzhihua City faces significant challenges in blue-green space construction. Shortages, low water levels, poor quality, and distinct seasonal variations between floods and droughts characterize the water system [22].

According to a 2024 poll conducted by the local government [22,23], the average annual precipitation is 932.00 mm, surface water resources total 2381.1272 million m³, groundwater resources amount to 846.0836 million m³, and total water consumption is 717.7379 million m³, with average annual surface water resources at 4.128 billion m³. The water bodies exhibit poor quality, distinct seasonal fluctuations, and difficulties in water use during dry periods. Meanwhile, the forest area covers 344,298.08 hectares, yielding a forest coverage rate of 46.5%. Based on the ecological sources and ecological network derived from the MSPA–Conefor–MCR model, it was found that the city’s blue-green space is highly fragmented. The average area of environmental patches is small, their distribution is discrete, and the degree of fragmentation is high. Moreover, densely populated urban areas are characterized by gaps and fragmentation, with a notable lack of ecological corridors connecting the patches.

5. Results

5.1. Ecological Network Optimization: New Ecological Sources and Ecological Resting Points

Based on the current conditions of the study area and the results of the ecological network construction, this study introduces new ecological sources, ecological nodes, and ecological resting points to enhance the restoration of environmental breakpoints (Figure 7 and Figure 13). The spatial framework of the ecological network in mountainous, resource-based cities is optimized accordingly.

5.1.1. Ecological Sources, Nodes, and Resting Points

Each ecological source grants an ecosystem services’ range. This study aims to enhance the overall environmental benefits and establish extra ecological source points by identifying areas falling short of coverage in the existing environmental sources’ radiation range.

The service range of ecological sources is quantified using the MCR model. The average radiation radius of the 18 identified sources (Figure 7) is approximately 14.1 km (ranging from 12.3 km to 16.8 km). As shown in Figure 11, the network includes 78 primary corridors (57.4%), 58 secondary corridors (42.6%), and 17 general corridors, which are primarily located in the western fault zone (accounting for 64.7% of all corridors in that region). Moreover, 136 of the 153 potential ecological corridors have been retained after screening, resulting in a corridor density of 0.40 km/km². After calculating the gravity model formula (7) and Table 8, the final environmental network model for the city is illustrated in Figure 12. This study identifies and establishes four new ecological sources, primarily in the northwest and northern parts of the city, using 15 km and 20 km radiation buffer zones around the geometric center points of the primary and secondary ecological sources. Furthermore, based on existing ecological sources, corridors, and the constructed ecological network, 30 new ecological corridors are added through selection and optimization to strengthen network robustness further. With these new sources, the total environmental source area increases by approximately 23.1%.

Given the large area occupied by ecological sources in the study region (Figure 7), the ecological network requires high structural stability. To this end, a comprehensive resistance surface is employed (Figure 10). Since ecological processes in these areas are vulnerable to anthropogenic disturbance and ecological corridors are prone to fragmentation, constructing ecological resting points to serve as stepping stones is essential [26]. These small ecological patches significantly enhance the corridors’ stability and the network. Such resting points are particularly vital in large-scale corridors where disconnection from the nearby ecological sources is common, for example, when corridors transverse the central urban area of Panzhihua or fragmented spaces in the city’s west. This study conducts a buffering analysis of ecological source areas and overlays ecological corridors with the source buffers. By integrating these components, intersection points are identified and planned. Using ArcGIS technology, 39 ecological resting points are established at least, strategically located in the central, eastern, and western parts of the study area.

5.1.2. Strengthening Ecological Breakpoint and Integration with Road Networks

Ecological networks in densely populated areas may have conflict with human infrastructure, increasing resistance to environmental processes and compromising the network’s landscape functions. Based on road data for Panzhihua City, 470 ecological breakpoints are identified in the road system (Figure 13). It is necessary to repair these breakpoints and prevent direct conflicts between anthropogenic activity and ecological functions. Potential solutions include constructing overpasses or culverts that meet biological movement, ensuring the environmental corridors’ continuity. Additionally, establishing zones around existing green spaces can help mitigate the impacts of anthropogenic activities. Minimizing disruptions and expanding buffer coverage wherever possible is recommended to maximize ecological stability.

5.2. Causes and Potential Impacts of the Blue-Green Space Fragmentation in Panzhihua City

This study concludes that the blue-green space in Panzhihua City is highly fragmented, necessitating the ecological network construction. The ecological patches are small, sparsely distributed, and highly fragmented. Urban areas, incredibly densely populated regions, expose significant ecological voids, falling short of environmental corridors connecting these patches. A few reasons cause this fragmentation. First, the rapid urbanization of industrial mountainous resource-based cities has led to intensive land development and construction, resulting in the blue-green spaces’ degradation and disconnection. Second, as industrial cities expand, large pieces of land are requisitioned, replacing the natural blue-green areas with industrial land and infrastructure. Third, urban planning often overlooks the protection and responsible blue-green space usage, further exacerbating fragmentation. Infrastructure development, such as urban roads and railways, disconnects the blue-green spaces, creating ecological breakpoints and intensifying spatial discontinuity. Lastly, other anthropogenic activities degrade these environmental resources.

Severe fragmentation of blue-green space in industrial, mountainous, resource-based cities negatively affects the urban ecological environment, public health, and sustainable urban development. Therefore, measures are needed to protect and restore blue-green spaces and to achieve coordinated development between urban ecology and the economy.

6. Discussion

6.1. Scientific Basis and Innovations in Ecological Network Construction

The negative effects of blue-green space fragmentation on ecosystem services, such as biodiversity conservation and hydrological regulation, are well documented [79]. The results of this study further support this while highlighting the unique challenges mountainous resource-based cities face. Despite an increase in ecological core areas, local fragmentation remains high. This suggests that traditional “scale first” protection strategies are insufficient to address the spatial degradation driven by industrialization [26,60]. Unlike conventional ecological network studies that often “prioritize network construction over management” [26,60,80], this study emphasizes addressing ecological gaps and weaknesses. This approach offers a more practical foundation for optimizing ecological environments, conserving natural resources, and restoring degraded landscapes.

Methodologically, the use of a multi-scale MSPA method (30 m, 60 m, 90 m, 150 m, and 300 m resolutions) aligns with Wang and Chen [81] in urban spatial form analysis in Changsha. Furthermore, the MCR model incorporates proximity to water bodies by introducing a weighted coefficient (0.18), enhancing model accuracy (Figure 8). Therefore, the ecological network derived from this comprehensive data framework is precise. This study also demonstrates the effectiveness of establishing 39 ecological resting points at key intersections across the city’s central, eastern, and western areas to reduce human-induced environmental disruption. This finding aligns with Haddad et al. [82] in Science, which confirms that microhabitats can connect fragmented landscapes. It also underscores restoring ecological networks’ complexity in mountainous resource-based cities.

6.2. Mechanism Analysis of the Causes of Blue-Green Space Fragmentation

The spatial conflict model of “ecological source vs. mining land” in Panzhihua City offers useful insights for mountainous resource-based cities. Future policymakers may further receive inspiration from the “brownfield-blue-green corridor” transformation strategy used in the Ruhr area of Germany [83,84] for designing integrated mechanisms for mine reclamation and biological migration corridors. The Panzhihua City National Land Space Master Plan (2021–2035) Text 0108 (final), promulgated in 2023, includes a structural system of urban green space and water bodies and outlines a preliminary plan for building a blue-green network.

The findings in Section 5.2 identify the main drivers of blue-green space fragmentation in mountainous resource-based cities as industrial land encroachment, infrastructure development, and insufficient planning and protection. This is consistent with existing research on mountainous resource-based cities [85,86]. Additionally, Panzhihua exhibits a distinct feature of “fragmentation due to hydrological stress in the dry season” [22,23], which confirms the seasonal vulnerability of ecological networks in such regions. Therefore, planners and policymakers could consider dynamic adjustments such as incorporating temporary wetland resting points during the rainy season to minimize the problem.

6.3. Limitations and Directions for Further Research

First, there is no universally accepted industry standard for determining the width of ecological corridors [26]. This study references a range of scales from previous research [49,59,60], ultimately selecting the corridor width commonly used in municipal ecological network planning. Second, the AHP method employed in this study has limitations, such as the potential subjectivity involved in constructing the pairwise comparison matrix. Further research should focus on improving the ecological sources’ identification, developing integrated models, and refining the resistance value index system by integrating multidisciplinary knowledge.

7. Conclusions

7.1. Theoretical Implications

Based on the MSPA–Conefor–MCR model, this study constructs a blue-green space ecological network in a mountainous resource-based city. It explores this within the city’s overall land-use planning to improve the ecological network and propose construction strategies for blue-green and water spaces. Using multi-scale spatial analysis and considering the importance of ecological patches to connectivity, this study sets up five comparison groups with grid scales of 30 m, 60 m, 90 m, 150 m, and 300 m for spatial pattern analysis. It ultimately determines that a 150 m grid scale is the most appropriate for this type of research. Compared to other scales, the 150 m grid avoids fragmentation caused by excessive detail, enhances water bodies’ identification, and results in more representative core ecological patches. This provides a reliable spatial foundation for constructing ecological networks. The gravity model is used to quantify interactions between ecological sources, which allows for a more refined classification of ecological corridors. This study also verifies the ecological buffer zones’ contributions and biological stepping stones to the stability of the ecological network, offering a new theoretical perspective for ecological network research in mountainous resource-based cities.

(1) The ecological core area accounts for 65.72% (4871.997 km²) of the study area. It is primarily composed of large patches in the primeval forested mountain regions. Smaller patches are concentrated in river valleys and urban areas characterized by low terrain and intensive anthropogenic activity. The spatial fragmentation is high, landscape connectivity is low, and the area is vulnerable to anthropogenic disturbances.

(2) There are 18 ecological core patches in the urban area. Larger patches exhibit strong landscape connectivity. The primary ecological core areas are located in the primeval forests of the city’s north, northwest, and southwest parts. These zones feature excellent environmental conditions and minimal human interference. In contrast, secondary ecological core patches are distributed in the northeastern valleys and central parts, which are closer to suburban areas and subject to frequent anthropogenic activity. These patches are fragmented and scattered. Therefore, buffer zones should be established along their boundaries to mitigate the impact of human disturbances.

(3) Regarding ecological corridors, 153 were initially identified. An interaction strength matrix was developed using the gravity model to classify the connections among ecological sources. After eliminating redundant and non-essential corridors, 78 primary ecological corridors and 58 secondary ecological corridors were obtained. The northern part of the city, consisting of fragmented ecological source areas connected by patches of primeval forest, exhibits low resistance values. However, corridor density remains low despite a large, sizable ecological core area. Therefore, key ecological protections and network enhancements are needed to complete the ecological network and improve landscape connectivity. Given the low correlation among fragmented ecological cores, these zones remain susceptible to external interference that could disrupt ecological corridors. The construction of stepping stones and corridor restoration measures should be increased, particularly in the central and northeastern valley regions.

(4) The current ecological network includes 470 ecological breakpoints, mainly located in the northeastern and central areas with high corridor density and in the southwestern valleys where anthropogenic activity is frequent. It is essential to incorporate the concept of ecological breakpoints into blue-green space integration strategies. Planning for biological bridges, culverts, and migration structures for other species should be prioritized at these points. Additionally, there are currently 39 biological resting points, primarily located in the central urban area. However, more resting points are needed to serve as material carriers for species migration and biological diffusion.

7.2. Practical Implications

The results offer insights to urban planning efforts for governments and urban and rural planners. It uses a systematic framework comprising ecological source identification, corridor extraction, and environmental network construction. It stresses water and green spaces’ interconnectedness and advocates a targeted ecological network strategy to address weaknesses and gaps in the existing system. This approach intends to boost landscape connectivity and corridor stability.

This study offers valuable insights into ecological protection and sustainable development in mountainous resource-based cities as follows: (1) ecological bridges and culverts should be constructed in areas with lots of ecological breakpoints to prevent corridor fragmentation; (2) additional ecological resting points should be established in regions with weak connectivity to improve species migration; and (3) optimizing the ecological network based on corridor density helps maintain strong landscape connectivity.