A Study on the Optimization of Ecological Spatial Structure Based on Landscape Risk Assessment: A Case Study of Wensu County, Xinjiang, China

Abstract

Ecological network construction has been widely accepted and applied to guide regional ecological conservation and restoration. For arid regions, ecological networks proposed based on ecological risk assessments are better aligned with the sensitive and fragile characteristics of local ecosystems. This study assesses landscape ecological risk in Wensu County, located on the southern slope of the Tianshan Mountains in the arid region of northwestern China, and it further proposes an optimized ecological network. A multidimensional framework composed of the natural environment, human society, and landscape patterns was employed to construct an ecological risk assessment system. Spatial principal component analysis (SPCA) was applied to identify the spatial pattern of ecological risk. Morphological spatial pattern analysis (MSPA) and a minimum cumulative resistance (MCR) model integrated with circuit theory were used to extract the ecological sources and delineate the ecological corridors. The results reveal significant spatial heterogeneity in terms of ecological risk: Low-risk zones (16.26%) are concentrated in the southwestern forest and water areas. In comparison, high-risk zones (28.27%) are mainly distributed in the northern mountainous mining region. A total of 24 ecological source patches (4105.24 km²), 44 ecological corridors (313.6 km), 39 ecological pinch points, and 38 ecological barriers were identified. Following optimization, the Integral Index of Connectivity (IIC) increased by 89.04%, and the Landscape Coherence Probability (LCP) rose by 105.23%, indicating markedly enhanced ecological connectivity. The current ecological network exhibits weak connectivity in the south and fragmentation in the central region. Targeted restoration of critical nodes, optimization of corridor configurations, and expansion of ecological sources are recommended to improve landscape connectivity and promote biodiversity conservation.

1. Introduction

Amid the accelerating process of global urbanization and the intensification of anthropogenic disturbances, regional landscape patterns have experienced profound transformations, resulting in escalating ecological fragmentation that poses significant threats to biodiversity and the integrity of ecosystem functions [1]. On a global scale, intensified human interventions such as urban sprawl, agricultural expansion, and mineral exploitation have increasingly disrupted natural ecosystems, leading to the degradation of ecological sources and the severance of ecological corridors, which collectively constitute one of the foremost challenges to ecological systems [2]. In response to these pressing issues, the establishment of ecological security patterns has emerged as a pivotal strategy for harmonizing economic development with ecological preservation [3,4,5].

The construction of an ecological security pattern centers on the identification of ecological sources, resistance surface construction, and ecological corridor extraction, aiming to enhance regional ecological resilience through the optimization of spatial ecological structures. Ecological sources, as core areas for maintaining ecosystem functions, are typically identified based on indicators such as ecosystem service importance, biodiversity, or landscape connectivity [6]. Resistance surfaces integrate factors like land use types and human activity intensity to quantify the impediments to species migration or ecological processes [7]. Ecological corridors are extracted using minimum cumulative resistance models or circuit theory to optimize ecological network connectivity [8]. This methodological framework has proven effective in ecological planning in the humid regions of eastern China; however, its application in arid zones faces significant challenges. Ecosystems in arid regions not only endure intense disturbances from human activities such as population growth and urban expansion but are also rigidly constrained by natural stresses like water scarcity and extreme heat. Traditional approaches may overlook the ecological vulnerability induced by water shortages when identifying ecological sources, such as the dependence of vegetation distribution on water availability. Likewise, during resistance surface construction, it is often difficult to quantify the compound resistance imposed by high-temperature stress on species migration. These limitations undermine the scientific rigor and adaptability of ecological network design in arid environments [9,10].

Landscape ecological risk assessment offers a new perspective for improving the construction of ecological security patterns in arid regions. By analyzing how landscape patterns respond to external disturbances and their ecological consequences, this approach effectively reveals the coupling mechanism among “risk–pattern–optimization” [11,12,13]. However, current research presents notable limitations. On the one hand, traditional assessments often rely on static indicators such as landscape diversity and patch density, making it difficult to quantify the dynamic interactions between natural stressors and human activities [14,15]. On the other hand, landscape risk assessment has long been disconnected from ecological security pattern construction, resulting in risk identification outcomes that cannot be readily translated into spatial optimization practices [16]. This gap is especially pronounced in arid regions, where the spatial heterogeneity of ecological risk is not only influenced by land use fragmentation but also closely linked to uneven water distribution and extreme climatic events [17,18,19]. There is an urgent need to establish a collaborative assessment framework that integrates multiple stressors. A key challenge lies in how to deeply embed landscape ecological risk assessment into the ecological security pattern construction process [20,21]. By systematically evaluating the combined effects of human disturbances and natural stressors in arid zones, it becomes possible to design ecological security patterns that better reflect regional ecological characteristics [22]. Such integration not only enables comprehensive quantification of multi-source risk factors such as climate change and land use transformation [23] but also uses spatial risk differentiation to guide the optimized layout of ecological elements, thereby offering more scientifically grounded decision support for ecological conservation in arid regions [24].

Wensu County, located in the Aksu Prefecture of the Xinjiang Uyghur Autonomous Region and situated on the southern slopes of the Tianshan Mountains, serves as a vital ecological barrier in the region. Its ecosystems play a strategically important role in soil and water conservation, windbreak and sand fixation, and biodiversity preservation [25,26]. However, intensified anthropogenic activities such as mineral exploitation and agricultural expansion have significantly altered the original landscape patterns, leading to ecological fragmentation and habitat degradation [27,28]. Existing studies have predominantly focused on the assessment of individual ecological elements, with limited attention paid to the quantitative identification of critical nodes within ecological networks, such as pinch points and barriers, and their responses under multiple scenarios. This gap constrains the effectiveness and precision of ecological restoration strategies [29].

Building on this foundation, this study takes landscape ecological risk (LER) assessment as a key basis for the construction of ecological security patterns (ESPs). A multidimensional evaluation framework comprising the natural environment, human society, and landscape pattern dimensions is established, and spatial principal component analysis (SPCA) is applied to reveal the spatial heterogeneity of ecological risk, which then guided the optimization of ESPs. The research objectives are as follows: (1) to develop a landscape ecological risk assessment framework for arid regions by integrating multi-source stressors, including natural environmental constraints, human disturbances, and landscape pattern dynamics; (2) to optimize the criteria for ecological source identification and resistance surface construction based on spatial risk differentiation and extract multi-level ecological corridors and nodes using Morphological Spatial Pattern Analysis (MSPA) and the Minimum Cumulative Resistance (MCR) model; and (3) to quantify the structural improvement of ecological networks before and after optimization using landscape connectivity indices, thereby validating the spatial rationality of the constructed ecological security pattern. This study integrates LER assessment with ESP construction, enhancing the scientific rigor and adaptability of ecological spatial optimization in arid regions and provides both theoretical support and practical guidance for regional ecological conservation and spatial planning.

2. Study Area and Data Sources

2.1. Study Area

Wensu County is located in the northwest of Aksu Prefecture, Xinjiang Uygur Autonomous Region, China, spanning geographic coordinates from 40°52′ N to 42°15′ N and 79°28′ E to 81°30′ E. Geographically, it lies on the southern slope of the central Tianshan Mountains and the northern fringe of the Tarim Basin, functioning as an important ecological barrier in northwestern China (Figure 1). The total area of the county is approximately 14,500 km², with a population of around 208,900. Similar to most arid zone oasis landscapes, Wensu County exhibits a distinct mountainous–oasis–desert structure. This physiography exerts an irreplaceable influence on hydrothermal conditions, resulting in pronounced elevation-driven differentiation in landscape patterns. The high-altitude zones are typically perennially snow- and ice-covered, serving as the primary water source for the entire region. As elevation decreases, temperatures rise, and meltwater from snow and ice alleviates moisture stress on vegetation, creating favorable conditions for meadows or coniferous forests. Within the lower mountain zone (1000–2000 m a.s.l.), high-temperature stress becomes increasingly evident and intensifies, driving a vegetation transition to desert steppe. With further reduction in elevation, high-temperature stress strengthens further, leading to a gradual succession towards desert vegetation dominated by species such as Alhagi sparsifolia, Haloxylon ammodendron (saxaul), and Reaumuria soongorica. However, the low-elevation areas feature relatively flat terrain and favorable soil conditions, which have enabled their large-scale conversion to cultivated farmland. Wensu’s climate is characterized by hot summers and cold winters, with summer temperatures exceeding 40 °C and winter temperatures dropping below −20 °C. Precipitation is concentrated in the summer months under monsoon influence, yielding an average annual rainfall of less than 100 mm. However, significant altitudinal gradients are evident for both temperature and precipitation.

2.2. Methodology

This study applied the SPCA method to assess landscape ecological risk in Wensu County by integrating natural, human, and landscape pattern indicators. Principal components with a cumulative contribution rate above 85% were extracted, and risk levels were classified using the Natural Breaks method and spatial overlay. An ecological resistance surface was then constructed, and ecological sources were identified using MSPA. Ecological corridors were simulated to enhance connectivity. Furthermore, Conefor 2.6 was used to evaluate landscape connectivity and ecological risk, providing a scientific basis for ecological restoration and landscape pattern optimization. The overall research process is shown in Figure 2.

2.3. Data Sources

The datasets used in this study include the following: 1. Elevation data: Digital Elevation Model (DEM) data with a spatial resolution of 12.5 m were obtained from the Alaska Satellite Facility Distributed Active Archive Center (ASF DAAC) at https://search.asf.alaska.edu/ (accessed on 4 February 2025). This dataset was used to derive terrain factors such as elevation and slope. 2. Land use/land cover data: The 2023 land use dataset was sourced from the Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, with a spatial resolution of 30 m. The dataset includes six first-level land use categories (forest, grassland, water bodies, cultivated land, construction land, and unused land) and eighteen second-level subcategories, such as urban construction land, reservoirs, and mining ponds. 3. Anthropogenic feature data: Data on roads, industrial sites, residential settlements, and water features were obtained from OpenStreet Map https://www.openstreetmap.org/ (accessed on 19 June 2025). Using ArcGIS 10.8, the Euclidean Distance Analysis tool was applied to calculate four distance-based indicators: distance to roads, industrial land, residential areas, and water bodies, respectively. 4. Landscape pattern metrics: Based on the landscape classification of Wensu County, two landscape pattern indices were calculated using Fragstats 4.2: the contagion index (CONTAG), which measures landscape connectivity and spatial aggregation, the landscape division index (DIVISION), which reflects the degree of landscape fragmentation.

2.4. Selection of Landscape Ecological Risk Indicators

Based on the multidimensional framework of natural environment, human society, and landscape pattern, eleven indicators were selected to construct a comprehensive landscape ecological risk assessment system. The selection basis for each indicator is shown in Table 1. All spatial indicators were processed and classified using the Reclassification tool in ArcGIS, and the final ecological risk levels were divided into four classes: low, medium, high, and very high.

2.4.1. Natural Environment Indicators

Five indicators were chosen to represent natural ecological attributes: elevation, slope, gross primary productivity (GPP), leaf area index (LAI), and distance to water bodies. Elevation and slope are critical topographic factors influencing the likelihood of soil erosion and geological hazards. Higher values indicate increased ecological vulnerability. According to [30], slope gradients of 0–8°, 8–15°, 15–25°, and >25° correspond to mild, moderate, strong, and severe erosion intensity, respectively. Water availability is a key limiting factor in arid ecosystems. In Wensu County, proximity to water bodies is essential in maintaining water supply, habitat quality, and ecological recreation [31]. Areas closer to water bodies tend to have lower ecological risk due to stronger ecosystem support functions.

Table 1. Landscape ecological risk indicators and their classification criteria in Wensu County.

Dimension	Indicator	Risk Level	Classification Criteria	Reference
Natural Environment	DEM (m)	1	<1200	(Chen et al., 2023) [32]
		2	1200–1900
		3	1900–2000
		4	>2000
	Slope (°)	1	<8	(Tang and Song, 2006) [33]
		2	8–15
		3	15–25
		4	>25
	Gross Primary Productivity	4	0–2	(Xie et al., 2019) [34]
		3	2–6
		2	6–8
		1	>8
	Leaf Area Index (LAI)	1	0–7	(Tian et al., 2023) [35]
		4	7–15
		3	15–24
		2	>24
Human Society	Distance to Water Bodies (m)	4	0–1000	(Wang et al., 2016) [31]
		3	1000–2000
		2	2000–3000
		1	>3000
	Distance to Industrial Sites	4	0–3500	(Zhang et al., 2021) [36]
		3	3500–7000
		2	7000–10,000
		1	>10000
	Distance to Roads	4	0–500	(Gao et al., 2023) [37]
		3	500–1500
		2	1500–2500
		1	>2500
	Distance to Residential Areas	4	0–3078	(Gao et al., 2023) [37]
		3	3078–6003
		2	6003–17,706
		1	>17,706
Landscape Pattern	Land Use Type	1	Forest, shrubland, sparse forest, other forest, high-cover grassland, lakes	(Wang et al.,2023) [38]
		2	Paddy fields, medium-cover grassland, rivers, marshland, reservoirs/ponds
		3	Dry farmland, low-cover grassland, beaches, bare land
		4	Other construction land, urban land, rural settlements
	Contagion Index (CONTAG)	4	0–17.45	(Jian et al., 2014) [39]
		3	17.45–45.90
		2	45.90–69.80
		1	69.80–96.74
	Landscape Division Index	1	0–0.15	(Gao et al., 2023) [37]
		2	0.15–0.35
		3	0.35–0.61
		4	>0.61

Table 1. Landscape ecological risk indicators and their classification criteria in Wensu County.

Dimension	Indicator	Risk Level	Classification Criteria	Reference
Natural Environment	DEM (m)	1	<1200	(Chen et al., 2023) [32]
		2	1200–1900
		3	1900–2000
		4	>2000
	Slope (°)	1	<8	(Tang and Song, 2006) [33]
		2	8–15
		3	15–25
		4	>25
	Gross Primary Productivity	4	0–2	(Xie et al., 2019) [34]
		3	2–6
		2	6–8
		1	>8
	Leaf Area Index (LAI)	1	0–7	(Tian et al., 2023) [35]
		4	7–15
		3	15–24
		2	>24
Human Society	Distance to Water Bodies (m)	4	0–1000	(Wang et al., 2016) [31]
		3	1000–2000
		2	2000–3000
		1	>3000
	Distance to Industrial Sites	4	0–3500	(Zhang et al., 2021) [36]
		3	3500–7000
		2	7000–10,000
		1	>10000
	Distance to Roads	4	0–500	(Gao et al., 2023) [37]
		3	500–1500
		2	1500–2500
		1	>2500
	Distance to Residential Areas	4	0–3078	(Gao et al., 2023) [37]
		3	3078–6003
		2	6003–17,706
		1	>17,706
Landscape Pattern	Land Use Type	1	Forest, shrubland, sparse forest, other forest, high-cover grassland, lakes	(Wang et al.,2023) [38]
		2	Paddy fields, medium-cover grassland, rivers, marshland, reservoirs/ponds
		3	Dry farmland, low-cover grassland, beaches, bare land
		4	Other construction land, urban land, rural settlements
	Contagion Index (CONTAG)	4	0–17.45	(Jian et al., 2014) [39]
		3	17.45–45.90
		2	45.90–69.80
		1	69.80–96.74
	Landscape Division Index	1	0–0.15	(Gao et al., 2023) [37]
		2	0.15–0.35
		3	0.35–0.61
		4	>0.61

2.4.2. Human Society Indicators

Two indicators were selected to reflect the intensity and spatial distribution of anthropogenic disturbance: distance to residential areas and distance to industrial sites. These features represent concentrated zones of human activity and are direct sources of disturbance to natural landscapes. The closer a location is to roads, settlements, or industrial zones, the higher the likelihood of ecological degradation, fragmentation, and risk.

2.4.3. Landscape Pattern Indicators

To characterize landscape spatial structure and its influence on ecosystem resilience, three landscape metrics were adopted: land use type, the contagion index (CONTAG), and the landscape division index (DIVISION). Land use types reflect spatial heterogeneity in surface cover and are fundamental to ecological structure and function. Different land use categories are associated with varying levels of ecological risk based on their species composition, biomass productivity, and ecological functionality. The landscape division index measures the spatial separation of patches within the same land type. Higher values indicate greater patch isolation and higher landscape fragmentation, which correlates with increased ecological risk. The contagion index assesses the continuity and connectivity of dominant patch types. Higher CONTAG values suggest better connectivity and lower ecological risk, whereas lower values indicate severe fragmentation, reduced connectivity, and elevated risk.

2.4.4. Calculation of Landscape Ecological Risk Values

The landscape ecological risk value is a quantitative measure of ecological risk. Based on the raster data of landscape ecological risk indicators, it is calculated using the spatial principal component analysis (SPCA) method [40] as follows:

$$\mathrm{R}={\sum }_{i=1}^m{\sum }_{i=1}^n(A_{ij}\times C_j)$$

(1)

where R is the risk value; Aij is the j th principal component on the i th raster in the region; and Cj is the eigenvalue contribution rate of the j th principal component. Principal components with a cumulative contribution rate exceeding 85% are considered statistically significant.

According to the evaluation results, to extract the contribution rates of the principal components, a fishnet was generated using the Create Fishnet tool in ArcGIS, and principal component analysis (PCA) was conducted in SPSS (https://www.ibm.com/products/spss (accessed on 19 June 2025)). The results indicate that the evaluation factors are positively correlated with the contribution rates—the larger the value, the higher the contribution.

2.5. Landscape Pattern Optimization

2.5.1. Identifying and Optimizing Ecological Source Areas

In landscape pattern optimization aimed at enhancing ecosystem connectivity and stability while reducing landscape ecological risks, “ecological sources” refer to areas with high environmental stability and strong expansion potential.

This study initially selected forest, wetland, and water body patches larger than 5 km² as potential ecological sources and imported them into Conefor_Input for further analysis. To address the issue of small and fragmented patches within the study area, the selection process included merging or eliminating these patches, thereby creating a more cohesive and continuous ecological source area. This optimization was achieved using the “Aggregate Polygons” and “Eliminate Polygon Part” tools in ArcGIS. Following this, landscape connectivity was assessed using four indices: Landscape Coincidence Probability (LCP), Integral Index of Connectivity (IIC), Probability of Connectivity (PC), and Delta of Importance (DI). A threshold distance of 5000 m was set, and the connectivity probability was defined as 0.5. Based on these calculations, the core areas were evaluated, and patches with a core importance value (DI) greater than 0.5 were identified as the final ecological sources.

In this study, the commonly used software Conefor 2.6 was employed to evaluate the effectiveness of landscape pattern optimization. By calculating and comparing the IIC and LCP indices before and after optimization, the level of improvement in landscape connectivity was quantitatively assessed.

Among these, LCP and IIC have demonstrated strong performance in representing ecological connectivity; therefore, they were selected as the primary indicators in this study. The calculation formulas are as follows [41]:

$$IIC={\displaystyle \frac{\sum _{i=1}^n\sum _{j=1}^n\frac{a_i\times a_j}{1+{NL}_{ij}}}{{\mathrm{B}}_L^2}}$$

(2)

$$\mathrm{L}\mathrm{C}\mathrm{P}={\sum }_{\mathrm{i}=1}^{{\mathrm{N}}_{\mathrm{c}}}{\left({\displaystyle \frac{{\mathrm{C}}_{\mathrm{i}}}{{\mathrm{A}}_{\mathrm{L}}}}\right)}^2$$

(3)

where Ai and Aj are the areas of patches i and j in the basin’s landscape pattern, respectively; BL is the total area of all patches in the basin; NLij is the number of connections between patches i and j. The IIC value ranges between 0 and 1 (0 < IIC < 1), with values closer to 1 indicating better connectivity. Nc represents the number of key ecological elements in the landscape pattern of the study area, and Ci is the sum of their areas. The LCP value also ranges between 0 and 1 (0 < LCP < 1), representing the probability that a randomly placed point falls within a connected patch; the closer the probability is to 1, the better the landscape connectivity.

Landscape pattern connectivity can quantitatively reflect changes in connectivity. The rate of change in landscape connectivity is defined as:

$$R={\displaystyle \frac{b-a}{a}}\times 100\%$$

(4)

where R represents the rate of change in the LCP or IIC index; a is the value of the LCP or IIC index before optimization; and b is the value after optimization.

2.5.2. Construction of the Integrated Ecological Resistance Surface

Spatial heterogeneity is a key characteristic that distinguishes landscape patterns from other inherent attributes. This heterogeneity causes material, energy, and ecological flows to encounter varying levels of resistance across different spatial areas [42,43].

In this study, we constructed the resistance surface model using the minimum cumulative resistance (MCR) model. The minimum cumulative resistance (MCR) model was first proposed by Knaapen in 1992 to study the migration cost that species face when moving from a “source” to a “sink” during their migration process. This model quantifies the resistance that species must overcome along their migration paths, thereby reflecting the integrated resistance of heterogeneous landscape elements to ecological flows between ecological sources. Therefore, identifying the resistance of landscape patterns is a key step in optimizing overall landscape configuration. The MCR model is expressed by the following formula [44]:

$$MCR=f_{min}\sum _{i=1}^m\sum _{j=1}^n\left(D_{ij}{F}_i\right)$$

(5)

where MCR represents the minimum cumulative resistance from ecological source i to j; Dij is the spatial distance from ecological source patch i to ecological surface j; Fi is the resistance value of the i-th factor impeding ecological flow across the landscape surface [45].

Within this study, anthropogenic disturbances, notably roads, constitute key factors impeding connectivity between core ecological source areas. Furthermore, situated in an arid zone, Wensu County exhibits extensive bare ground devoid of vegetation due to water scarcity; this condition exerts a greater influence on the connectivity of ecological sources in this region. Consequently, the primary objective in delineating ecological corridors is to establish connectivity through vegetation between source areas, rather than facilitating gene flow or migration for fauna. This rationale underpinned the selection of distance from water sources as a critical parameter in constructing the resistance surface. To classify the resistance levels, we applied the Natural Breaks classification method, categorizing the surface into four levels: low, medium, high, and very high resistance (

Table 2. Classification criteria for cumulative landscape resistance in Wensu County.

Resistance Level	Cumulative Resistance Value
1	0~391
2	391~1192
3	1192~2295
4	2295~4537

).

2.5.3. Construction of Ecological Corridors

Ecological corridors are virtual pathways that facilitate species migration and dispersal between different ecological sources, playing a crucial role in maintaining ecological functions. These corridors not only enhance spatial connectivity for species but also promote gene flow, stabilize populations, and strengthen the resilience of ecosystems to external disturbances. The Linkage Mapper tool under circuit theory was employed to extract the ecological corridors. Moreover, ecological corridors were categorized into three levels based on their centrality, from high to low: key corridors, important corridors, and general corridors. This classification was carried out to prioritize the most critical corridors for conservation and management efforts, ensuring the protection of the most essential pathways that maintain ecosystem connectivity.

In this study, ecological sources, corridors, and ecological resistance surfaces were used and processed through the sf and terra packages of the R language. First of all, all spatial data were converted into a unified coordinate system. Then, the length of each corridor was calculated, and its average resistance value on the resistance surface was extracted. The missing values were handled to ensure data integrity. Furthermore, the altitude information was extracted using the centroid of the corridor, and the tortuosity was calculated. Finally, the width of the corridor was predicted through the XGBoost regression model. The input features of the model included corridor length, average drag value, altitude value, and tortuosity. During the training process, random noise processing was carried out on the data to increase diversity. To ensure the rationality of the prediction results, the width value was limited to between 30 and 300 m. Finally, the buffer zone of the corridor was generated based on the predicted width, and the corresponding area was calculated.

2.5.4. Identification of Ecological Nodes

Ecological nodes are key points within an ecological network structure and represent the most vulnerable links in an ecosystem. They are typically located at the intersections or critical positions of ecological corridors and act as “stepping stones” connecting different ecological sources. In this study, ecological nodes in Wensu County were defined as the intersection points between “ridge lines” and “valley lines” on the resistance surface. Ridge lines were extracted based on areas with maximum resistance values, while valley lines were derived from areas with minimum resistance values. Ridge lines typically represent high-resistance areas, such as mountainous regions or zones with intensive human activity, whereas valley lines indicate low-resistance areas, such as rivers or valleys. Through spatial overlay analysis, the intersections of ridge and valley lines were identified as potential ecological nodes. These nodes serve not only as critical junctions of ecological corridors but also as important habitats or stopover sites during species migration.

To ensure the accuracy and reliability of the identified ecological nodes, field surveys and ecological model validation were conducted. Species distribution data and ecological process simulations were used to verify the ecological functions of the nodes. Additionally, practical conservation measures for ecological nodes were proposed, including restricting human activity, restoring vegetation, or constructing ecological bridges. These measures can significantly enhance the connectivity and stability of ecological nodes, thereby improving the overall resilience of the ecological network.

3. Results and Analysis

3.1. Landscape Ecological Risk Assessment in Wensu County

3.1.1. Three-Dimensional Landscape Ecological Risk Assessment: Natural Environment–Human Society–Landscape Pattern

The results of the principal component analysis (PCA) indicate significant differences in the contribution rates of the 11 landscape ecological risk factors, with natural environmental factors playing a dominant role (

Table 3. Eigenvalues and contribution of each principal component.

Principal Component	Eigenvalue	Variance Contribution (%)	Cumulative Contribution (%)
1	3.62	32.88	32.88
2	1.73	15.76	48.63
3	1.21	11	59.63
4	1.03	9.34	68.98
5	0.8	7.25	76.23
6	0.72	6.5	82.73
7	0.64	5.82	88.55
8	0.55	5	93.54
9	0.34	3.08	96.63
10	0.2	1.84	98.47
11	0.17	1.53	100

). The first principal component (PC1) has an eigenvalue of 3.62. It explains 32.88% of the total variance, primarily driven by elevation (DEM), suggesting that topographic factors exert a decisive influence on the spatial distribution of ecological risk. The second principal component (PC2), with an eigenvalue of 1.73 and a contribution rate of 15.76%, brings the cumulative variance to 48.63%. This component is mainly influenced by slope and gross primary productivity (GPP), reflecting the synergistic effect of terrain variability and vegetation cover on ecological risk. The third principal component (PC3), with an eigenvalue of 1.21 and a contribution rate of 11.00%, incorporates the leaf area index (LAI), raising the cumulative contribution to 59.63%. This highlights the critical regulatory role of vegetation conditions in shaping risk patterns in the midstream and downstream areas.

The fourth principal component (PC4) has an eigenvalue of 1.03 and accounts for 9.34% of the variance, bringing the cumulative contribution to 68.98%. It is primarily driven by the distance to water bodies (Water), indicating that hydrological factors significantly affect the spatial differentiation of ecological risks. PC5 (7.25%) mainly reflects the influence of industrial and mining activities (INDIST), implying that anthropogenic industrial disturbances have become a non-negligible source of risk. PC6 (6.50%) further reinforces the importance of proximity to water bodies; PC7 (5.82%) emphasizes the influence of distance to roads (ROAD); and PC8 (5.00%) is dominated by the distance to residential areas (RA). The remaining components, PC9 through PC11 (3.08% to 1.53%), capture the effects of landscape pattern characteristics.

Overall, the PCA results reveal a multi-dimensional driving mechanism behind ecological risk in the study area: natural geographic elements constitute the fundamental influence, human activities introduce secondary disturbances, and landscape pattern characteristics play a fine-tuning role.

Analysis of the principal component loading matrix reveals the functional characteristics of various factors across different dimensions and elucidates the underlying mechanisms of ecological risk formation (Figure 3). In the PC1 dimension, DEM (−0.83), slope (−0.70), GPP (−0.72), and LAI (−0.64) exhibit significant negative loadings, while RA (0.83) and ROAD (0.73) display strong positive loadings. This marked polarity of loadings highlights an important spatial differentiation pattern: areas with high elevation, steep slopes, and dense vegetation cover typically experience minimal human disturbance, thus forming low-risk zones dominated by natural factors. In contrast, low-lying and flat areas tend to be more heavily populated and developed with road infrastructure, leading to higher ecological risks.

The PC2 dimension presents a more complex loading structure: LUCC (−0.66) shows a strong negative loading, while slope (0.39) and INDIST (0.47) have positive loadings. This suggests the presence of two distinct risk formation pathways. On the one hand, specific land use types such as built-up land are closely associated with high ecological risks. On the other hand, topographic conditions may indirectly shape risk patterns by influencing the spatial distribution of industrial activities. This dual mechanism helps explain the unique spatial characteristics observed in PC2.

Notably, the distance to water bodies consistently shows high loadings across multiple components, such as PC4 (0.46), PC5 (0.33), and PC6 (0.59), indicating that its impact on ecological risk operates across multiple spatial scales. This distinct pattern of loadings may be attributed to the differentiated ecological effects of water bodies: at a local scale, they offer positive ecosystem services such as water supply, while at a regional scale, they may indirectly influence ecological risk by shaping patterns of human settlement.

The loading characteristics of landscape pattern metrics also provide critical insights. In PC3, the strong negative loadings of DIVISION (−0.77) and CONTAG (−0.61) suggest an inverse relationship between landscape fragmentation and ecological risk. Highly fragmented areas, often concentrated in zones of intensive human activity, may exhibit lower levels of natural ecological risk not due to intact ecosystem function but rather due to artificial regulation and management that mitigates natural hazards despite ecosystem degradation.

3.1.2. Landscape Ecological Risk Evaluation

The landscape ecological risk assessment is illustrated in Figure 4. Among the five natural environmental factors (Figure 5a–e), elevation and slope exhibit similar spatial patterns, both showing a general decrease from north to south and from the southern to western parts of the study area. The leaf area index (LAI) presents a risk distribution centered in the northern region, with risks gradually decreasing outward. Areas with greater distances from water bodies exhibit the highest proportion of ecological risk.

The three anthropogenic factors (Figure 5f–h) are mainly concentrated in the central part of the region. Risk associated with proximity to roads demonstrates a distinct linear spatial distribution. The spatial distribution of risk related to distance from settlements is largely opposite to that of distance from water bodies. The influence of industrial and mining activity (INDIST) expands from the mountainous north outward, with high-risk zones also concentrated in the central area.

For the land use type and landscape pattern metrics within the natural environment and landscape structure dimensions (Figure 5i–k), the risk distributions are relatively consistent. The contagion index (CONTAG) and division index (DIVISION) exhibit similar patterns of landscape ecological risk, with notable spatial heterogeneity. Using the Natural Breaks (Jenks) method in ArcGIS, landscape ecological risk levels were classified into four categories: low, moderate, moderately high, and high (Figure 4).

The low-risk zones are primarily located in the southwest, covering 208.76 km² or 16.26% of the total area. These zones consist mainly of forest, water bodies, and grasslands, characterized by minimal human disturbance, good vegetation cover, and a relatively intact ecological environment. The moderate-risk zones are primarily distributed in the central region, with an area of 422.79 km², accounting for 32.92% of the total area, and they are mainly dominated by forest land. The moderately high-risk zones are located on the northern mountains, covering 363.05 km² or 28.27% of the total area, and they represent the smallest proportion of high-risk areas. These zones are frequently subjected to human disturbances. The high-risk zones are mainly distributed in mountainous areas, covering 289.63 km² or 22.55% of the total area. These zones are dominated by industrial and mining land use, contributing to elevated levels of landscape ecological risk.

3.2. Landscape Pattern Optimization in Wensu County

3.2.1. Identification of Ecological Source Areas in Wensu County

Using Morphological Spatial Pattern Analysis (MSPA), seven types of landscape components were identified in Wensu County, with a total area of 3792.47 km², accounting for 26.44% of the study area (

Table 4. MSPA landscape classification statistics in Wensu County.

MSPA	Area (km2)	Percentage/%
Core	2777.5	19.38
Bridge	105.38	0.73
Isiet	153.58	1.07
Edge	417.14	2.91
Perforation	119.49	0.83
Branch	165.07	1.15
Loop	54.31	0.37

). Among them, the core area covers 2777.5 km², representing 19.38% of the county’s total area, while the bridge area covers 105.38 km², accounting for 0.73%. In addition, the edge, islet, perforation, branch, and loop areas collectively account for 5.5% of the total area. As shown in Figure 6a, the core areas exhibit a clear spatial pattern, mainly distributed across the alluvial fan plains in central Wensu County, forming linear zones stretching from west to east across the region. These core patches are relatively large and morphologically intact, providing essential habitats for species to survive and reproduce.

Within the core areas, abundant vegetation offers food sources as well as adequate space for shelter and breeding. These ecosystems possess strong self-regulating capacity, helping maintain biodiversity and supporting stable, healthy ecological development. Particularly in the southern grassland regions, the lush vegetation and fertile soils provide favorable conditions for diverse biological communities to thrive. The relative dispersion of these ecological units creates critical spaces for species migration and reproduction, making them vital components of the regional ecological network.

However, there are notable imbalances in the spatial distribution of ecological patches within Wensu County. As indicated in Figure 6a, ecological units such as bridges, islets, and loop areas are sparsely distributed, resulting in poor environmental connectivity and significant challenges for species migration and material exchange across regions. Although the number of patches is relatively high, most are small and fragmented, limiting their effectiveness as ecological corridors. To address this issue and construct a more cohesive ecological source area, specifically, the “Aggregate Polygons” and “Eliminate Polygon Part” tools in ArcGIS were employed to optimize the spatial distribution of environmental patches, resulting in a more complete and continuous map of ecological sources (Figure 6b).

The Conefor 2.6 software was used to assess ecological connectivity. A connection threshold distance of 5000 m and a connection probability of 0.5 were set based on the environmental characteristics of Wensu County. A connectivity analysis was then conducted on the core ecological patches. The patches with high connectivity and strong ecological functions were identified as the final ecological sources. The results indicate that 24 ecological source areas were ultimately selected, with a total area of 4105.24 km² for patches with dPC > 2.5. These patches range in size from 14.86 km² to 3051.46 km².

From a spatial perspective, the ecological sources are mainly concentrated in the northern region, forming a relatively compact cluster. In contrast, source areas in the western part are sparse, while the central and southern regions show significant gaps. This spatial discontinuity weakens ecological linkages across the region. Therefore, further environmental restoration and conservation efforts, especially in the central and south gap areas, will be essential for enhancing the overall ecological connectivity of Wensu County and improving the stability and sustainability of its environmental network.

3.2.2. Construction of Ecological Corridors in Wensu County

Based on the detailed analysis of ecological source areas and the established landscape ecological risk map, 44 ecological corridors were extracted using the Linkage Mapper tool under circuit theory (Figure 7). These corridors totaled 313.6 km, with the longest corridor reaching 4 km and the shortest measuring 0.07 km. According to the dPC (delta Probability of Connectivity) values, ecological sources were classified into three levels: core sources (dPC > 4), which account for 92.95% of the total source area; important sources (2.5 < dPC < 4); and general sources (dPC < 2.5).

Corridors were also categorized into three levels based on centrality from high to low: key corridors, important corridors, and general corridors. Among them, 28 key corridors are primarily distributed in the central region. These corridors mainly traverse forested areas and connect core sources. They are relatively short in length but densely distributed, with a total length of 6.77 km. The central region features high habitat quality and low landscape ecological risk, with many core source patches that are large in size and closely spaced. This creates favorable spatial channels for species migration and material flow among core sources.

In the ecological corridor distribution map (Figure 7), longer corridors in the southern region exhibit a north–south orientation. They are aligned with the Aksu and Talan rivers’ hydrological systems. Shorter corridors are mostly situated above grasslands and forested areas. The spatial layout of ecological corridors closely aligns with the distribution of water bodies, grasslands, and forests, highlighting the critical role of natural features in maintaining environmental connectivity.

Water bodies, grasslands, and forests provide essential habitats for wildlife and effective linkages among ecological source areas. They form the basis for species dispersal and material exchange, making such connectivity essential for maintaining ecosystem health and stability.

3.2.3. Identification of Ecological Pinch Points and Barriers in Wensu County

Ecological pinch points and barriers are critical nodes within the environmental system that connect ecological sources and corridors. They directly influence species migration and genetic exchange, affecting regional biodiversity and ecosystem stability. In this study, the “all-to-one” mode was employed to identify multiple ecological pinch points and barriers characterized by high current density, and their spatial variations under different scenarios were analyzed.

As shown in Figure 8, 39 ecological pinch points with high current density were identified. Among them, 22% cover an area greater than 1 km² and are primarily located at both ends of ecological corridors and near ecological source areas. These pinch points are found in regions where corridors intersect and overlap, forming critical nodes for species flow. Notably, the presence of pinch points in the central segments of long corridors indicates that these areas also serve important functions in facilitating biological movement. Restoration of these pinch points can be achieved through relatively simple ecological interventions such as vegetation rehabilitation, soil improvement, and water resource conservation. Since most pinch points are located in grasslands or unused lands, their ecological restoration is comparatively feasible and cost-effective, contributing significantly to enhancing the ecological network and the long-term sustainability of local species populations.

In addition, the results show that the average area of the corridors is 16.15 km², with the smallest area being 0.046 km² and the largest area being 101.76 km², indicating significant spatial variation among the corridors. This variation may have a crucial impact on ecological connectivity and species migration. Wider corridors are more beneficial for the movement of large animal populations, while narrower corridors may limit species dispersion. Figure 8 demonstrates the spatial distribution of these corridors. As illustrated in the figure, significant spatial variation exists in terms of corridor width across the study area. Corridors located in the northern and eastern regions tend to be considerably wider, with those in the eastern region exhibiting broader spans and forming relatively continuous ecological linkages. In contrast, corridors in the central and southern regions are markedly narrower and display a more elongated, fragmented spatial configuration. Such heterogeneity in corridor width may substantially influence the effectiveness of landscape connectivity, potentially affecting ecosystem stability and posing challenges to regional biodiversity conservation initiatives.

In contrast, ecological barriers represent areas where species movement is hindered. Using the Linkage Mapper tool, 38 ecological barriers with high current density were identified, ranging in size from 1 to 8 km². These barriers are mostly situated at the junctions between ecological corridors and source patches or directly along corridor paths. These barriers suggest spatial constraints that obstruct species migration and reduce genetic exchange and biodiversity flow. Restoration of ecological barriers is equally important, particularly those small barriers located near source patches, which are typically less ecologically stable and urgently require intervention. Measures such as reestablishing native vegetation, improving water quality, and minimizing human disturbances can effectively reduce the negative impact of these barriers, enhance connectivity between habitat patches, and support the conservation of biodiversity and the sustainable functioning of the ecosystem.

3.2.4. Evaluation of Landscape Optimization Effectiveness

A comparison of the Integral Index of Connectivity (IIC) and the Least-Cost Path (LCP) index before and after landscape pattern optimization in Wensu County (

Table 5. Evaluation of landscape pattern optimization effect.

	Before Optimization	After Optimization	Change Rate
Least-Cost Path (LCP)	0.0382	0.0785	105.23
Integral Index of Connectivity (IIC)	0.0365	0.0690	89.04

) reveals notable improvements. Prior to optimization, the IIC and LCP indices were 0.0365 and 0.0382, respectively. After optimization, these indices increased to 0.069 and 0.0785, representing change rates of 89.04% and 105.23%, respectively.

Landscape ecological studies have demonstrated that areas with higher connectivity indices generally exhibit more stable and resilient ecological landscape patterns. The significant increases in both the IIC and LCP values in this study indicate that the optimization measures effectively enhanced ecological connectivity across the region, thereby improving the integrity and sustainability of the overall landscape structure.

4. Discussion

Landscape pattern optimization is a current research focus in ecological environment management [13,46], with the identification of ecological sources and extraction of ecological corridors being central to this process [5]. Approaches to effectively and accurately identify ecological sources and extract corridors vary across studies. Initially, researchers often employed qualitative methods, directly designating structurally complex ecosystems or land use types with high biodiversity (e.g., forests) as ecological sources and subsequently deriving corridors [47,48]. Later, to enable quantitative identification of ecological source and extraction of ecological corridor, numerous studies incorporated metrics such as ecosystem service value [49], ecological environment quality (e.g., the Remote Sensing Ecological Index, RSEI) [50], or habitat suitability/potential distribution for key protected species [10] into analytical frameworks. The aim was to enhance landscape connectivity to ensure ecosystem integrity and stability.

However, Wensu County, located in the arid zone of Central Asia on the northern edge of the Taklimakan Desert, exhibits a typical mountain–oasis–desert structure. Its landscape pattern displays a distinct vertical zonation, ranging from glaciers, alpine meadows, and coniferous forests in the northern mountains to desert in the south. Consequently, ecosystem diversity, complexity, and functions (e.g., provisioning and supporting services) exhibit significant hierarchical gradients. The ecological service output and environmental quality of the southern desert are substantially lower than those of the northern mountains. Relying solely on high ecosystem service value or superior environmental quality for ecological source identification would therefore largely exclude the southern desert. From an ecosystem integrity perspective, however, the southern desert constitutes the basal zone of Wensu’s vertical ecological zonation and is an essential component of the complete ecological structure. Furthermore, under intensifying anthropogenic pressures like global warming and cropland expansion, Central Asia’s arid regions face escalating threats from rising temperatures, glacial retreat, and water scarcity. Effectively managing ecological risks and utilizing limited water resources to maximize ecosystem structural stability and functional integrity are primary goals for ecological conservation and management in the arid zone. This study incorporated quantified water resource accessibility into ecological risk assessment, which subsequently informed the identification of ecological sources and extraction of corridors. Integrating ecological risk into the landscape pattern optimization process thus enables multi-objective optimization. This approach aligns with practical ecological conservation needs, and the resulting optimized landscape pattern reflects the characteristics of arid oasis ecosystems. The identification of large desert areas in the southern study region as ecological sources (Figure 8) validates the scientific rationale of this methodology.

Additionally, the southern desert of Wensu County, characterized by flat terrain, fertile soil, and readily accessible groundwater, has historically been a prime area for land reclamation. Consequently, natural ecosystems face increasing pressure from cropland expansion. Protecting ecological sources, corridors, pinch points, and mitigating barrier points is crucial. Water resources are fundamental to safeguarding these conservation targets. Firstly, ecological sources themselves form under relatively favorable water supply conditions. Secondly, surface and subsurface runoff originating from the northern mountains serve as natural linkages connecting Wensu’s vertical ecological zones; hydrological connectivity is paramount for maintaining connectivity between ecological sources and corridors. Based on this, to ensure structural stability, future water resource development should minimize interference with hydrological connectivity. This includes avoiding river fragmentation and limiting uncontrolled extraction of river water and groundwater to guarantee ecological base flows. Similarly, in future ecological restoration efforts, alongside restoring sufficiently large ecological sources, reestablishing hydrological connectivity represents a highly practical and effective measure for achieving structural ecological connectivity.

5. Conclusions

The results indicate pronounced spatial heterogeneity in landscape ecological risk across Wensu County. Based on the risk assessment, the area was classified into four levels: low-risk zones are primarily located in the southwest (16.26%), moderate-risk zones are in the central region (32.92%), and high-risk zones are in the northern mountainous areas (28.27%). These spatial patterns are closely linked to topographic variation and human disturbance, with elevation and proximity to settlements and roads identified as key driving factors.

Regarding ecological network configuration, ecological source areas are concentrated in the northern and central parts of the county. At the same time, the southern region exhibits poor connectivity and requires urgent restoration. A total of 44 ecological corridors were identified, with those in the central area playing a crucial role in maintaining species migration and ecological flows. Following landscape optimization, ecological connectivity improved substantially, with connectivity indices increasing by 89.04% and 105.13%, respectively. These findings underscore the effectiveness of targeted restoration strategies and provide a scientific basis for ecological conservation and spatial planning.