The Ecological Risks in Arid Zones from a Production–Living–Ecological Space Perspective: A Case Study of the Tuha Region in Xinjiang, China

Abstract

The ecological and environmental problems of arid zones have become an urgent global concern. Current research on ecological risk is based mainly on the dominant functions of land use, with a primary focus on land use landscape projections and less consideration of potential risks to ecosystems, system resilience and interactions between nature and future sustainable development. In this study, a potential–connectivity–resilience ecological risk assessment model based on the SDGs was constructed using multisource data to spatially quantify indicators at the grid scale in the Turpan and Hami regions of Xinjiang, China. This model was used as a basis for studying ecological risk in arid zones from a production–living–ecological space (PLES) perspective. The results revealed that, during the period 2000–2020, PLESs in the Turpan and Hami regions presented significant spatial similarity, with an increasing trend in overall risk. The production space in the Turpan and Hami regions showed a parabolic trend of increasing and then decreasing, whereas the living space and ecological space in the Hami region showed continuous linear upward trends. The state of ecological security in the Turpan and Hami regions is gradually deteriorating, and comprehensive ecological protection and restoration measures are urgently needed to rationally allocate the structure and layout of the production-–living-–ecological space. The study of ecological risk from a PLES perspective not only helps in fully understanding the development trend of the arid zone; it also provides new ideas and methods for evaluating regional ecological environmental safety and scientific references for formulating regional sustainable development, ecological risk prevention and control and the rational allocation of resources.

1. Introduction

The arid zone of China is delineated by the national boundary in the northwest of the country, which extends from the Helan Mountains in the east to the Kunlun Mountains in the south. This zone comprises mainly the western region, including the Xinjiang Uygur Autonomous Region, northwestern Gansu Province, the Ningxia Hui Autonomous Region and the Inner Mongolia Autonomous Region, with an area of approximately 24.5% of the country’s land mass [1]. This region is located inland, where the influence of the humid maritime climate is minimal. As a result, the area is characterized by an arid climate, water scarcity, sparse vegetation and a fragile ecosystem. These factors make it one of the most sensitive regions in China concerning regional and climate change [2]. Contemporary social development and global warming issues pose many challenges for the region, including regional land degradation, the overconsumption of soil and water resources [3,4] and the degradation of ecosystems [5,6,7]. Therefore, scientific risk assessments of the ecological environment are important for rationally allocating regional land-scape patterns and effectively preventing and controlling environmental damage.

The concept of ecological risk originally emerged from the field of environmental impact assessment. In 1990, Hunsaker [8] proposed a regional ecological risk assessment approach, which laid a solid foundation for studying ecological risks at a geographical spatial scale. Since the release of the “Framework for Ecological Risk Assessment” in 1992 [9], the concept of ecological risk assessment has become more clearly defined and specific, providing clear guidance for the development of this field. The Guidelines for Ecological Risk Assessment, published in 1998 [10], highlight the need for a sound ecosystem risk assessment [11]. The World Health Organization subsequently modified the risk assessment framework to introduce risk management into the assessment process, creating an integrated framework for health status assessment and environmental risk assessment. Scholars worldwide have quantitatively assessed the risk level of regional ecosystems, mainly through relative risk modeling, which combines risk factors with sources, receptors and representatives of risk for each risk unit in the study area to comprehensively assess risk levels [12,13]. Fan et al. [14] conducted a study from a landscape perspective to assess the ecological risks induced by land use dynamics in the upper and middle reaches of the Heihe River, analyzing the impacts of different land use type changes on the regional ecosystem. Kanwar et al. [15] employed a relative risk model to evaluate the intensity of risk sources and their impact on receptors within the Kaipara watershed in New Zealand, conducting an ecological risk assessment and dividing the watershed into nine comprehensive risk zones to develop targeted prevention and optimization strategies. Prior to 2010, China’s development in the field of ecological risk assessment lagged behind the international level. However, with the gradual development of society and the increasing impact of human activities on ecosystems, Chinese scholars have gradually begun to pay attention to ecological risk assessment. Research has gradually shifted from the assessment of single sources of risk in small regions to the integrated study of large regions, multiple receptors and multiple sources of risk. Currently, the more commonly used evaluation methods are based on varying risks affecting the health of regional ecosystems or specific risk receptors, or they are designed to address a specific assessment purpose; these methods use different evaluation indicators to develop the corresponding assessment frameworks [16,17,18]. These approaches focus on the consideration of land use and individual ecosystem factors but fail to adequately capture ecosystem resilience and the interactions between nature and society. Therefore, in view of all these factors, a multidimensional risk evaluation based on potential, connectivity and resilience is essential for the Turpan and Hami areas.

Although different ecological risk evaluation index systems have expanded the scope and content of ecological risk research, with the Beautiful China initiative and green and sustainable development, people are no longer satisfied with evaluation systems that consider only a single ecosystem factor but rather expect them to reveal the impacts of interactions between nature and future sustainable development on ecological risk. The United Nations Sustainable Development Goals (SDGs) are initiatives that address social, economic and environmental challenges in a holistic manner through an integrated strategy; they emphasize the need for coordinated development in all regions to lead humanity toward sustainable development [19]. This approach is closely related to ecological risk assessment; for example, the assessment of ecosystem risk can contribute to the SDG15 objective of “protecting, restoring and promoting the sustainable use of terrestrial ecosystems” by analyzing issues such as global warming, biological habitat quality and land degradation; moreover, the assessment of relevant water resources allows for the identification and management of water use efficiency issues, thus contributing to the SDG6 goal of “clean water and water ecosystems”. In view of the impacts of natural and human activities, urban-related risk indicators are included to achieve SDG11, “safe and sustainable cities and human communities”, by assessing social development, urban expansion and population increase in the region during the study period. Therefore, an ecological risk assessment model based on potential–connectivity–resilience and the SDGs can be used to systematically identify and analyze multiple risks existing in the environment and enhance the comprehensiveness and reliability of ecosystem health evaluations. Then, targeted measures can be formulated to mitigate the impacts of these risks on the ecosystem.

During the spatial development of the national territory, an imbalance in the proportion of the land use structure has gradually become prominent. Furthermore, the 18th National Congress explicitly proposed the construction of a production-–living-–ecological space, namely, “intensive and efficient production space, livable and moderate living space, and clear and beautiful ecological space”, which highlights the direction for the optimization of national land space and principles [20]. This proposal led to the concept of the production–living–ecological space (abbreviated as PLES) [21]. The intertransformation of the PLESs are a key driving factor for changes in the state of the ecological environment, and an in-depth understanding and mastery of its transformation process can provide an important foundation for effectively preventing ecological risk. Thus, the theory of production–living–ecological spaces has received widespread attention in academia, and scholars have conducted extensive and in-depth research on PLESs that covers definitional interpretation [22,23], rational planning and demarcation [24,25,26,27,28], carrying capacity [29,30], quantitative analysis by function [31], the optimized layout of village and town spatial structure under PLESs [32,33], landscape ecological risk analyses and other aspects. Classifying and evaluating PLESs in a scientific and rational manner has become a basic prerequisite for planning the spatial layout of PLESs and reducing regional ecological risk. However, existing studies rarely combine PLESs with multi-indicator ecological risk assessments; they have focused mainly on economically developed regions in the central and eastern parts of China, and there is a lack of research on arid regions [34,35]. Therefore, it is necessary to conduct a study on ecological risk from the perspective of PLESs in the Turpan and Hami regions.

The Turpan and Hami Basin is not only an important part of the national strategic base but also an arid region with a fragile ecological environment. The diverse topographic and climatic conditions have led to land covers that are predominantly sand, Gobi and desert. The 74 townships in the six districts and counties in the region are suffering from the hazards of land desertification; the degree of land degradation is becoming increasingly serious; resources and energy are continuously being developed; and the contradiction between social development and the ecological and environmental protection of land is prominent. On this basis, in this study, the actual situation of the Turpan and Hami regions is taken, and land use is divided based on PLES functions. The aims of this study are as follows: (1) to analyze the functions of different land use subjects and study the spatial and temporal evolution of PLESs in the Turpan and Hami regions over the past 20 years, and then use the land use type matrix and the land center of the gravity migration model to assess the specific factors of PLES transformation; (2) to use the potential–connectivity–resilience model based on the SDGs to objectively analyze the dynamic changes in the ecological risk in the region, thus revealing the pattern of change in the ecological risks and clarifying the deficiencies of the ecological environment in the arid zone; and (3) based on the aforementioned research, an ecological risk assessment from the perspective of PLESs is conducted to provide scientific guidance and recommendations for promoting sustainable development in the Turpan–Hami region. This assessment aims to mitigate regional ecological risks and foster the coordinated development of PLESs in arid urban areas.

2. Materials and Methods

2.1. Study Area

The Turpan–Hami region (41°18′~43°43′N, 86°40′~96°04′E), east of the Tianshan Mountains in Xinjiang, is the collective name for the cities of Turpan and Hami, and serves as a crucial transportation hub linking Xinjiang with inland regions of China (Figure 1). With a total area of approximately 21,000 km², the region is situated at the heart of the Asia–Europe continent and is deeply landlocked, with hot and arid seasons, scarce precipitation and rapid evaporation. Furthermore, it has an extremely uneven distribution of water resources, resulting in a severe lack of water resources, sparse vegetation, severe desertification and extremely fragile ecosystems [36,37].

2.2. Data

In this study, land use data, normalized difference vegetation index (NDVI) data, digital elevation model (DEM) data and soil type data were used; the detailed information is shown in

Table 1. Data sources.

Data	Time	Resolution	Data Sources
Land use data	2000~2020	30 m	Resource Sciences and Satellites of the Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 7 December 2023))
NDVI	2000~2020	1 km	Resource Sciences and Satellites of the Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 7 December 2023))
DEM	2020	30 m	geospatial data clouds (https://www.gscloud.cn/ (accessed on 22 November 2023))
Mean annual temperature data	2000~2020	1 km	National Earth System Science Data Centre (https://www.geodata.cn/ (accessed on 22 November 2023))
Mean annual precipitation data	2000~2020	1 km	National Earth System Science Data Centre (https://www.geodata.cn/ (accessed on 22 November 2023))
Population density data	2000~2020	1 km	National Earth System Science Data Centre (https://www.geodata.cn/ (accessed on 22 November 2023))
Road data	2020	30 m	Digital Globe Open Platform (https://open.geovisearth.com/ (accessed on 7 December 2023))
Soil data	2020	1 km	National Glacial Tundra Desert Science Data Centre (https://www.ncdc.ac.cn/ (accessed on 20 January 2024))
Potential evapotranspiration data	2000~2020	0.1 mm	National Tibetan Plateau Science Data Centre (https://data.tpdc.ac.cn/ (accessed 22 December 2023))
Social statistical data	2000~2020	\	https://tjj.xinjiang.gov.cn/ accessed on 20 March 2024

.

(1)

Land use data came from the Centre for Resource Sciences and Satellites of the Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 7 December 2023). It was published every five years from 2000 to 2020 and used Landsat series of remote sensing image data. The land resources are classified into 25 land types based on their natural attributes.

(2)

NDVI data came from the Centre for Resource Sciences and Satellites of the Chinese Academy of Sciences (https://www.resdc.cn/ (accessed on 7 December 2023)); the spatial resolution is 1 km.

(3)

DEM data came from geospatial data clouds (https://www.gscloud.cn/ (accessed on 22 November 2023)); the spatial resolution is 30 m.

(4)

Mean annual temperature, annual precipitation and population density data came from the National Earth System Science Data Centre (https://www.geodata.cn/ (accessed on 22 November 2023)); the spatial resolution is 1 km.

(5)

Road data came from the Digital Globe Open Platform (https://open.geovisearth.com/ (accessed on 7 December 2023)); the spatial resolution is 30 m. There are six categories of road types: highways, national highways, provincial highways, railways, county roads and rural roads.

(6)

Soil data came from the National Glacial Tundra Desert Science Data Centre (https://www.ncdc.ac.cn/ (accessed on 20 January 2024)); a wide range of elemental contents, including sand content, silt content, clay content and organic carbon content, were included at a resolution of 1 km.

(7)

Potential evapotranspiration data came from the National Tibetan Plateau Science Data Centre (https://data.tpdc.ac.cn/ (accessed 22 December 2023)); they are in NetCDF (.nc) format, with a spatial resolution of 0.1 mm, and required conversion to raster data for use in ArcGIS 10.8.

All the data mentioned above were processed using ArcGIS software, with the resolution standardized to 30 m and projected to the WGS_1984_UTM_zone_46N coordinate system.

2.3. Methods

2.3.1. Categorization of Production, Living and Ecological Functions

Land is a multifunctional complex that integrates production, living and ecological functions, which interact and are related to each other; different land use patterns lead to the differentiation of the production, living and ecological functions of each land use [25]. Specifically, production space refers to land used for agro-industrial and other economic production activities; living space includes land for residential use and related infrastructure; and ecological space includes areas such as nature reserves, wetlands and forests, which are essential for maintaining ecological balance and environmental protection. PLESs reflect a dynamic process of reallocation and spatial repositioning of different land cover resources [38]; therefore, identifying the core functions of each land use is a key step in rationally allocating the PLES layout. Drawing on relevant research results, this study adopts a three-level scoring standard (maximum of 5 points, medium of 3 points, minimum of 1 point, and 0 points for missing functions) [21,39] to assess the strength and completeness of the production, living and ecological functions of the current land use status of the secondary land use category in the Turpan–Hami Basin area; the scoring results are detailed in

Table 2. Land use classification system and PLES function scoring in the Turpan and Hami areas.

Category I		Category II		Production Space	Living Space	Ecological Space
Code	Name	Code	Name
1	Cropland	11	Paddy field	5	0	1
		12	Dryland	5	0	1
2	Woodland	21	Woodland	5	0	5
		22	Shrubland	0	0	5
		23	Sparse woodland	0	0	5
		24	Other forestland	5	0	1
3	Grassland	31	High-coverage grassland	0	0	5
		32	Medium-coverage grassland	0	0	5
		33	Low-coverage grassland	0	0	3
4	Waters	41	Rivers and canals	1	0	3
		42	Lakes	0	0	5
		43	Reservoirs and ditches	3	0	3
		44	Permanent glaciers and snowfields	0	0	5
		46	Shoals	0	0	5
5	Urban, rural and industrial residential land	51	Town land	1	5	0
		52	Rural settlements	1	5	0
		53	Other constructed land	5	1	0
6	Unused land	61	Sandy land	0	0	3
		62	Gobi	0	0	3
		63	Saline soil	0	0	3
		64	Marshland	0	0	3
		65	Bare ground	0	0	3
		66	Bare rock texture	0	0	3

.

Cluster analysis is a crucial tool for quantitative research in the classification of geographical entities and regional zoning, as it enables the analysis and expression of regional differences through the calculation of multi-factor similarities [40,41]. In this study, a hierarchical clustering method is employed to identify the thematic functions of land use types, allowing for a deeper understanding of the spatial distribution characteristics and similarity patterns of various land use types. This approach effectively reveals the functional differences between regions, providing a scientific basis for land use management and regional planning. The formula is presented as follows:

$$D_{ij}=\sqrt{{\displaystyle \frac{1}{n}}{\sum }_{k=1}^n{\left(X_{ik}-X_{jk}\right)}^2}$$

(1)

In the formula, $D_{ij}$ represents the similarity coefficient of the “production-–living-–ecological” (abbreviated as PLE) functions between the ith and jth land use types. Here, $X_{ik}$ and $X_{jk}$ denote the scores of the ith and jth land use types on the kth PLE function, respectively. The variable $n$ indicates the total number of land use types in the Turpan and Hami regions.

2.3.2. Evolution of Spatial and Temporal Patterns in PLESs

Single Land Use Attitudes

The attitude of single land use dynamics reflects the degree and speed of change in land types during the study period and can clearly reveal the magnitude of change in different land use cover types in the study area; thus, it is used to represent the fluctuation and change in a certain land cover type within the study interval [42]. The expression is:

$$K={\displaystyle \frac{U_b-U_a}{U_a}}\times {\displaystyle \frac{1}{T}}\times 100\mathrm{\%}$$

(2)

where $K$ represents the attitude of movement of a particular landform during the study period; $U_a$ and $U_b$ represent the areas of a particular landform at the beginning and end of the study period, respectively; and $T$ represents the length of the study.

Degree of Integrated Land Use Dynamics

The degree of integrated land use dynamics can express the overall changes in all land cover types in the study area over the study period and can be used to study regional differences in land use dynamics [43]. The specific expression for this is as follows:

$$M=\left[{\displaystyle \frac{\sum _{i=1}^n∆U_{i-j}}{2{\sum }_{i=1}^n{U}_i}}\right]\times {\displaystyle \frac{1}{T}}\times 100\mathrm{\%}$$

(3)

where $M$ represents the degree of integrated land use dynamics; $U_i$ represents the area of land class i in the initial period; $∆U_{i-j}$ represents the absolute value of the area of land class $i$converted to other land classes$j$ during the study period; and $T$ represents the length of the study.

2.3.3. Construction of Ecological Risk Evaluation Indicators

Potential–Connectivity–Resilience Evaluation System Based on SDGs

In this study, on the basis of the exposure and disturbance effects attributable to the risk sources and the special geomorphology and local policies of the arid zone, 17 evaluation indices were finally selected, and an ecological risk evaluation system based on the potential–connectivity–resilience of the SDGs was constructed, as shown in

Table 3. Evaluation system of “potential–connectivity–resilience” based on the SDGs.

Normative Layer	Risk Layer	Targets	Connotation	SDG	Nature of the Indicator	Weights
Potential	Exposure	Slope	Land degradation risk	/	+	0.125
		Vegetation covered	Spatial distribution patterns of plant communities	15.1.2	-	0.210
		Land use risk	Human activity patterns	15.3	+	0.128
	Disturbance	Average annual temperature	Potential desertification risks	15.3	+	0.105
		Water yield		6.6	-	0.168
		Population density	Regional social development situation	11.3.1	+	0.264
Connectivity	Exposure	Intensity index	Ecological connectivity and fragmentation	/	+	0.093
		Patch density			-	0.050
		Shannon Diversity Index	Diversity and heterogeneity of landscape structure and function	/	-	0.166
		Contagion index	Degree of plaque aggregation	/	-	0.052
	Disturbance	Distance to plowland	Ecological impacts of farming	/	-	0.299
		Distance from construction site	Ecological impacts of human development activities	/	-	0.215
		Road network density		11.2	-	0.125
Resilience	Exposure	Ecosystem resilience index	Resilience of ecosystems	15.5	-	0.181
		Habitat quality index	Ecosystem resilience to risk	15.5	-	0.301
		Ecosystem services value index	Ecosystems provide services and value to humans	15.1.2	-	0.349
	Disturbance	Drought risk	Probability of risk of drought on land	15.3	+	0.169

.

Among the SDGs, SDG6.6 emphasizes the protection and restoration of water-related ecosystems; ecosystems such as forestland and grassland promote water infiltration through vegetation and effectively contain groundwater and surface water, thus reducing water loss and improving water use efficiency. Therefore, water yield was chosen to reflect this indicator. The rational planning and construction of road networks can significantly improve the sustainable, accessible and safe transport system proposed in SDG11.2; thus, we chose road network density to reflect this indicator. SDG11.3.1 advocates more efficient land use and urban planning in cities with high population density to ensure sustainable development and quality of life; thus, we chose population density to reflect this indicator. SDG 15.1.2 emphasizes the conservation of terrestrial and freshwater biodiversity; furthermore, areas of high vegetation cover support higher biodiversity by providing diverse habitats for a wide range of flora and fauna. Thus, we chose vegetation cover to represent this indicator. SDG 15.3 states that desertification should be prevented and that the goal of zero growth in land degradation should be pursued. Furthermore, irrational land use practices such as overcultivation, overgrazing and over-abstraction of groundwater can exacerbate the risk of drought on land and persistent drought, and rising temperatures can further exacerbate land degradation. Thus, land use risk, mean annual temperature and drought risk were chosen as the key indicators for preventing desertification. SDG15.5 emphasizes significant action to reduce the degradation of natural habitats. Both habitat quality, an important tool for assessing the health of biological habitats, and the ecosystem resilience index, an important measure of the resistance and resilience of ecosystems, are closely related to the state of health of biological habitats. Thus, they were selected as key indicators for reducing habitat degradation. In addition, on the basis of the ecological policy documents recently released by the national government and local governments, as well as the characteristics of the ecosystem itself, we also included indicators such as the spreading degree index, the Shannon diversity index and the distance from construction sites. By integrating the links between the SDGs and these evaluation indicators, a comprehensive ecological protection strategy can be formulated to ensure the coordinated development of ecosystems, thus effectively promoting the realization of regional sustainable development goals.

Subjective–Objective Combination Empowerment Method

Hierarchical analysis is a systematic decision-making method for addressing complex multicriteria problems. It acheives this by decomposing the problem into different levels of objectives, criteria and scenarios; constructing judgment matrices for comparison; calculating eigenvalues and eigenvectors to determine the subjective weighting values of each indicator W_i; and ultimately carrying out a consistency test to ensure its reasonableness.

The entropy weight method integrates and quantifies the information of each indicator and finally determines the objective weight by measuring the information entropy of the indicator. Its specific formula is as follows [44]:

$$P_{ij}=X_{ij}/{\sum }_{i=1}^k{X}_{ij}$$

(4)

$$M_j=-k{\sum }_{i=1}^n{P}_{ij}\mathit{ln}{P}_{ij},k=1/\mathit{ln}n$$

(5)

$$W_j=\left(1-M_j\right)/{\sum }_{j=1}^m\left(1-M_j\right),\left(j=\mathrm{1,2}\dots ,m\right)$$

(6)

where ${ X}_{ij}$ is the normalized value of the indicator; $P_{ij}$ is the weight of the indicator accounted for by the ith sample under the jth indicator; $M_j$ is the value of the information entropy of the jth indicator; $W_j$ is the value of the weight of the entropy-weighted objective method; and $n$ is the number of samples.

The final weight value $W$ is calculated via the following formula:

$$W=\sqrt{W_i{W}_j}/{\sum }_{i, j=1}^n\sqrt{W_i{W}_j}$$

(7)

3. Results

3.1. Main Functional Division of Land Use Types in the Turpan–Hami Region

According to the results of the cluster analysis (Figure 2), when the group spacing is set to 18, the secondary class of land use status can be divided into three clustering groups with clear trituration functions. The first group, whose main function is ecological, includes land use types such as bare ground, marshland, sandy land and lakes; the second group, whose main function is production, includes land use types such as dryland, paddy land and other constructed land; and the third group, whose main function is living, includes land use types for town land and rural settlements. The final result of the classification of the PLE function of the land use types in the Turpan and Hami regions was obtained; the detailed information is shown in

Table 4. Land use status types in the Turpan and Hami areas: PLE function divisions.

Production, Living and Ecological Functions	Clustering Group	Land Use Type
Ecological function	1	Bare ground, bare rock texture, marshland, saline soil, Gobi, sandy land, low-coverage grassland, rivers and canals, permanent glaciers and snowfields, shoals, lakes, medium-coverage grassland, high-coverage grassland, sparse woodland, shrubland
Production function	2	Dryland, other forestland, paddy land, other constructed land, woodland, reservoirs and ditches
Living function	3	Town land, rural settlements

.

3.2. Spatial and Temporal Evolution of PLES in the Turpan and Hami Regions, 2000–2020

Based on the analysis of PLES changes in the Turpan and Hami regions in 2000, 2010 and 2020, the total areas of the Turpan and Hami regions are approximately 74,000 km² and 137,200 km², respectively; furthermore, the area shares of each category of PLES are similar. Over the past 20 years, with the rapid development of the Turpan and Hami regions, there have been significant changes in land use types within the area. Ecological space remains the predominant land use type, accounting for 97% of the total area. However, its area has gradually decreased from 208,253 km² in 2000 to 206,908 km² in 2020, with a total reduction of 1344 km². In contrast, production and living spaces have shown a continuous growth trend during this period. The area of production space has increased by 1215 km², expanding from 2851 km² to 4067 km², representing 2% of the total area. Meanwhile, living space has increased by 142 km², from 190 km² to 333 km², accounting for 1% of the total area. These changes reflect the dynamic adjustment and transformation of land use patterns against the backdrop of rapid regional economic development. This is mainly due to the rapid development of industrialization and urbanization as well as the expansion of agricultural production, which has resulted in the large-scale removal of natural vegetation and the encroachment of ecological space (Figure 3). Figure 4 shows the change in PLES land use motivation attitudes in the Turpan and Hami regions from 2000 to 2020. The results of the study revealed that the motivations for the change trends of single land use in the Turpan and Hami regions over 20 years were the same. The value of production space was the largest value during 2000–2010, indicating that the change in production space was greater in this period; the value of living space increased significantly in the latter 10 years, and the change was more obvious. However, the value of ecological space changed by only 0.04, mainly because 97% of the area in this region was ecological space, the base was larger, and the change was not obvious. In terms of the attitudes toward integrated land use dynamics, the values for the Turpan and Hami regions from 2000 to 2010 were 0.07% and 0.06%, respectively, and both decreased to 0.02% from 2010 to 2020, with a low degree of integrated land use. This situation occurred because the geomorphological types in the study area are dominated by sandy and bare rocky terrain that is unsuitable for human habitation, whereas oases and plains are small in size, resulting in significant land use conflicts.

3.3. PLES Center-of-Gravity Migration in Turpan and Hami, 2000–2020

Between 2000 and 2020, the center of gravity of the production space in the Turpan area was distributed mainly in the northern part of Gaochang District (Figure 5). The production space was more dispersed in the east-west direction in 2000, while it was more dispersed in the north-south direction in both 2010 and 2020, showing a gradual trend of spreading outward. During the first 10 years, the center of gravity shifted to the southeast by 3.51 km and then continued to move in that direction for the next 10 years by 1.53 km. This phenomenon is due mainly to the implementation of the West-to-East Natural Gas Pipeline Project, the route of which passes through the Turpan Basin, which has led to many laborers moving to major towns and cities, thus causing the center of gravity to shift to the east. The center of gravity of the production space in the Hami region during this period was located in the northern part of the Yizhou district, close to the junction of Yiwu and Barkun Kazakh Autonomous Counties (Figure 5). The standard deviation ellipse analysis indicates that the production space in the Hami region during the period of this study was dispersed mainly in the northwest–southeast direction, gradually diffusing outward. During this period, several major projects were carried out in the Hami region, including the Hami 13th Chamber Wind Power Base, the Hami 50 MW Molten Salt Tower Photovoltaic Power Plant, and the Hami-Dunhuang and Hami-South-Zhengzhou power transmission corridors; these projects are intended to export frontier electricity. The implementation of these projects resulted in the relocation of the center of gravity of the production space in the region by 11.04 km in the southeast direction during the study period.

During the period 2000–2020, the center of gravity of living space in the Turpan region was distributed mainly in the northern part of Gaochang district (Figure 6). During the study period, the living space in the region was narrowly distributed in an east-west direction and was concentrated in the northern township areas of the districts and counties. In Hami, on the other hand, living space was concentrated in the northern part of Yizhou District (Figure 6); the standard deviation ellipse of the region during the 20-year period roughly shows a circle, with the lengths of the long and short axes being equal, indicating that living space was more evenly distributed in all directions. Owing to the rapid economic development of the region between 2000 and 2020 and the implementation of the West-to-East Gas Pipeline and the Xinjiang Power Transmission Project, the population of Xinjiang increased, resulting in the conversion of a large area of ecological space in Turpan and Hami into living space. This process caused its center of gravity to shift in the southwestern and southeastern directions by 9.56 and 6.77 km, respectively, which is a large change.

The ecological space center of gravity of the Turpan region during the period 2000–2020 was located mainly in the central part of Gaochang District (Figure 7), and its ecological space dispersion was greater in the northwest-southeast direction. The ecological space center of gravity of the Hami region, on the other hand, was located mainly in the northeastern part of Yizhou District, near the border of Yiwu County. During the study period, the ecological space of the region was uniformly distributed across all districts and counties, and the standard deviation ellipse had highly similar long and short axes. The center of gravity of the ecological space in Turpan and Hami did not change significantly between 2000 and 2010, and the center of gravity shifted to the southwest and southeast by 0.03 km and 0.11 km, respectively, between 2010 and 2020. Although the area of the ecological space continued to decrease during the 20-year period, the center of gravity did not shift drastically due to the large ecological space base of the area, which accounted for 97% of the ecological space in this region.

3.4. Spatial and Temporal Evolution of Ecological Risk in the Turpan and Hami Areas

Ecological risk can clearly reflect the degree of superiority or inferiority of the environmental conditions of ecosystems. In this study, on the basis of the proposed ecological risk evaluation method, different types of ecological risk scores were calculated and analyzed three times: in 2000, 2010 and 2020. As shown in Figure 8, the potential risk, connectivity risk, resilience risk and comprehensive risk of the Turpan and Hami regions changed significantly during the period of 2000–2020, and the area of high-risk areas gradually increased, which makes it necessary to reduce the ecological risk and improve the ecological and environmental conditions of the region.

The overall trend of potential risk is a stepwise increase, with a gradual decrease in low-risk areas, a gradual increase in medium- and high-risk areas, and a gradual decrease in the ecological security situation (Figure 8a). The low-risk areas are located mainly in Yizhou District; the border zone between Yiwu County and Barkun Hasak Autonomous County, including Baishi Township, Tianshan Township, Corps Hongshan Farm, and townships in the northern part of the counties; and cities in Turpan District. Compared with other areas, these areas have lower levels of social development and sparser population distributions; furthermore, the land cover type is mainly grassland, which has a higher water yield, so the potential risk value is lower. The medium-risk areas are located in areas with more forest and grassland cover within Yizhou and Gaochang districts, which are sparsely populated and subject to fewer anthropogenic impacts and therefore have relatively low potential risk values. High-risk areas, on the other hand, are concentrated in the main townships and townships of the districts and counties in the Turpan and Hami regions, where the terrain is flat, the population density is high, some of the areas are accompanied by a certain degree of construction and population agglomeration during the development process, and the type of land cover is mostly constructed and unutilized so that the potential risk of these areas is relatively high.

The risk of connectivity shows a parabolic trend of increasing and then decreasing, but the overall risk value still increases (Figure 8b). The low-risk areas are located mainly in Nanhu Township and Wubao Township in Yizhou District with Nanshan Mining Township in Shanshan County and the northwestern townships of Barkun Hassak Autonomous County. These places are less disturbed by human interference and are far from construction land and arable land; thus, they have a lower degree of landscape fragmentation and a higher degree of connectivity and agglomeration between various land types. These factors lead them to have low connectivity risk values. The medium-risk areas are close to built-up land and farmland and are more widely distributed but have poorer landscape connectivity. High-risk areas are concentrated in the main urban areas of Gaochang District, Shanshan County and Torkun County; these areas are subject to severe human interference, dense road networks, single land-use types and weak landscape connectivity, and are therefore at high risk of connectivity.

Resilience risk did not significantly change from 2000 to 2020 (Figure 8c). The low-risk areas are located mainly in the border zones of the Hami districts and counties and the northern areas of the Turpan districts and counties, which have complex and diverse vegetation types, mainly grasslands and woodlands, with a low probability of drought risk and high-quality habitats. The medium-risk areas are widely distributed, and the unused land along the Turpan and Hami regions is distributed on the periphery of the high-risk areas, with medium levels of the habitat quality index, the ecosystem resilience index and the probability of drought, and a certain degree of resilience. The high-risk areas are mainly arable land and construction land, which have higher risk values because of the higher intensities of development and construction in these areas, higher probabilities of drought risk, lower habitat quality and ecosystem resilience indices, and weaker abilities to recover if the ecosystems are damaged.

Comprehensive risk showed an inverted U-shaped trend of decreasing and then increasing over the past 20 years, with the area of high-risk areas increasing and the ecological security situation gradually deteriorating (Figure 8d). The low-risk areas are located mainly in Barkun Kazakh Autonomous County, the northern part of Gaochang District and the southern and northern parts of Shanshan County, where the ecosystems are relatively stable; thus, the risk values are low. The medium-risk areas are concentrated around the medium- and higher-risk areas. The high-risk areas are located mainly at the edge of the development centers of Yizhou District, Shanshan County, Gaochang District and Torkun County, where the land use cover types are relatively homogeneous, mostly consisting of built-up land and unused land, and are subject to greater interference from human activities, which increases their risk values.

3.5. Analysis of the Spatial and Temporal Evolution of Ecological Risk from a PLES Perspective

During the period 2000–2020, although the average values of various ecological risks in Turpan and Hami increased but then decreased, in general, the average value in 2020 increased compared with that in 2000 (Figure 9), indicating that the ecological risk situation in the Turpan and Hami regions gradually deteriorated. During this period, the changes in ecological risk in Turpan were as follows: connectivity risk > comprehensive risk > potential risk > resilience risk. The changes in ecological risk in Hami were as follows: potential risk > connectivity risk > comprehensive risk > resilience risk.

Between 2000 and 2020, the average value of PLES potential risk in Turpan gradually decreased, while the average values of resilience risk, connectivity risk and composite risk showed an overall increasing trend (

Table 5. The average ecological risk value of PLESs in the Turpan–Hami region from 2000 to 2020.

Type of Risk	Year	Turpan PLES			Hami PLES
		Production Space	Living Space	Ecological Space	Production Space	Living Space	Ecological Space
Potential risk	2000	0.45262	0.57231	0.54444	0.40821	0.57181	0.53462
	2010	0.48536	0.58288	0.55051	0.43897	0.56619	0.53352
	2020	0.42111	0.49725	0.54638	0.41463	0.50578	0.54803
Connectivity risk	2000	0.68612	0.67752	0.59846	0.70109	0.72343	0.60770
	2010	0.69797	0.70928	0.63211	0.72410	0.74493	0.63664
	2020	0.69490	0.70692	0.61889	0.71606	0.74541	0.62740
Resilience risk	2000	0.77431	0.80165	0.46869	0.73972	0.83046	0.46958
	2010	0.81977	0.85800	0.46328	0.82072	0.85797	0.45998
	2020	0.83642	0.85777	0.46318	0.83445	0.86154	0.46019
Composite risk	2000	0.63758	0.68307	0.53739	0.61607	0.70803	0.53765
	2010	0.66777	0.71602	0.54881	0.66115	0.72245	0.54371
	2020	0.65134	0.68753	0.54303	0.65565	0.70444	0.54542

, Figure 10); in contrast, the average values of all types of PLES risk in Hami showed an increasing trend (Table 5, Figure 11). Over the past 20 years, the average resilience risk in the Turpan–Hami region exhibited significant fluctuations. The average risk values for the three phases of PLES are as follows: living space > production space > ecological space. This pattern primarily arises because ecosystems in living spaces have a lower capacity to resist and recover from risks than do those in ecological spaces. Consequently, the average resilience risk in living spaces is the highest. In the potential risk assessment, the average risk value for living space is the highest, followed by that for ecological space, with production space having the lowest average risk value. Living spaces are typically located in densely populated urban areas, where human disturbance is significant, leading to the highest average risk value. Owing to its unique geographical location, the Turpan–Hami region’s ecological spaces primarily consist of land use types such as sandy deserts, Gobi and bare rocky terrains with sparse populations, resulting in a slightly lower risk value than living spaces. Throughout the study period, connectivity risk showed an increasing trend. The average risk value for living space was the highest at 0.73792, followed by production space at 0.71375, and ecological space at 0.62391, which was the lowest average risk value. This phenomenon is attributed to the richer vegetation types and stronger landscape connectivity in ecological and production spaces, which result in slightly lower average risk values than those in living spaces. The overall risk exhibited the smallest variation during the study period; the average risk values for the three phases of PLESs were in the order of living space > production space > ecological space.

4. Discussion

4.1. Construction of an Ecological Risk Evaluation System Based on the Potential–Connectivity–Resilience of SDGs

The SDGs cover three dimensions, namely, ecological sustainability, economic sustainability and social sustainability, each of which contains land-related indicators [37]. As a good vision for global development, these indicators provide directional guidance for the social development of countries. Depending on the specific conditions of different regions, these targets need to be improved and refined by localizing the indicators [38]. In this study, the indicator system is subdivided into six dimensions on the basis of the exposure and disturbance effects of risk sources: exposure of potential, disturbance of potential, exposure of connectivity, disturbance of connectivity, exposure of resilience and disturbance of resilience. We consider exposure and disturbance to be two critical factors influencing the stability and health of ecosystems. Exposure typically refers to potential threat factors, such as slope, landscape connectivity or land use changes, which may have long-term adverse impacts on ecosystems. Disturbance, on the other hand, refers to actual destructive events or processes, representing the real impact of exposure on ecosystems, such as land use risks, road network density or drought risks. Exposure focuses on identifying potential threats to facilitate proactive prevention, while disturbance emphasizes assessing the actual impact of events that have already occurred to develop appropriate response measures. Dividing each dimension into exposure and disturbance helps in future research on ecological risk-driving mechanisms, enabling clearer identification of the sources and mechanisms of different types of ecological pressures, thereby improving the accuracy and effectiveness of ecological risk assessment and management strategies.

SDG15 explicitly states that its goal is “to protect, restore and promote sustainable use of terrestrial ecosystems, sustainably manage forests, combat desertification, halt and reverse land degradation, and halt biodiversity loss.” However, relying solely on SDG15 does not fully reflect the ecological security status of a study area. Constructing a comprehensive ecological risk assessment system requires the integration of other closely related sustainable development indicators from the SDG framework. Among them, SDG6 emphasizes sustainable water resource management, and SDG11 emphasizes building sustainable cities and human settlements, both of which are closely related to regional ecological security. On this basis, this study integrated relevant policy documents and the unique geomorphological features of arid regions to construct an SDG-based ecological risk assessment model that focuses on potential–connectivity–resilience. An in-depth analysis of the spatiotemporal changes in ecological risk in the Turpan–Hami region from 2000 to 2020 was conducted. The results (Figure 8) revealed that ecological risk in the Turpan–Hami region has increased stepwise over the past 20 years. This phenomenon may be attributed to climate change, resulting in rising temperatures and changes in precipitation patterns, and land degradation and desertification caused by human activities such as agricultural development and urbanization. These factors have exacerbated the vulnerability of ecosystems, disrupted their connectivity and integrity, and significantly increased regional ecological risks. However, an SDG-based ecological risk assessment indicator system can be used to gain a deeper understanding of the specific factors contributing to increased regional risk, comprehensively and meticulously assess ecological risk, and provide more targeted guidance and references for the sustainable development of different regions.

4.2. Ecological Risk Assessment from a PLES Perspective

Ecological risk refers to the potential negative impacts of uncertain events or disasters on the structure and function of ecosystems within a specific area. It is an important indicator for assessing the sustainability of the regional ecological environment [45], and conducting ecological risk assessments is highly important for the sustainable development of a region. PLES is a method of land classification that is based on the primary functions of land use; it can be used to achieve efficient resource utilization and sustainable development. Since changes in the utilization and intensity of PLESs are closely related to regional ecological security [46], assessing regional ecological risk from the perspective of PLESs provides valuable insights for the future development of a region [47]. Currently, ecological risk assessment methods can be divided into two main categories: source-sink-based methods and landscape-type-based methods. The source-sink-based method follows the framework of “risk source identification-risk receptor analysis-exposure and hazard analysis,” which is characterized by multiple stress factors, multiple risk receptors and spatial heterogeneity [48]. The landscape-type-based method uses the formula risk = probability × loss, which represents the risk intensity distribution through landscape disturbance and landscape vulnerability indices [49]. While numerous studies have been conducted on ecological risk assessment, most have been based on land use types; research on multi-indicator ecological risk assessment from a PLES perspective remains relatively scarce.

In addressing the aforementioned situation, adopting a PLES perspective can more comprehensively reflect the ecological risk distribution characteristics of different functional spaces within a region. This study integrates multisource data, including annual average temperature, social statistics and land use data, surpassing traditional land use landscape-based ecological risk assessment methods. Using the evaluation model provided in Section 4.1, various risk values of PLESs were calculated. The results (Table 5) show that the risk value of living space is the highest, followed by production space, while ecological space has the lowest risk value. This phenomenon is due mainly to the concentration of living spaces in high-population-density cities and residential areas, where resource consumption is high and environmental pressure is significant, leading to greater ecological risk. In contrast, production and ecological spaces primarily consist of plowland, forests, and wetlands, which have stronger landscape connectivity and higher ecosystem resilience, resulting in lower risk values. This method not only enables a deeper understanding of how PLESs can reasonably avoid ecological risks and expands the study of ecological risks in severely degraded arid regions, but also provides a reference for optimizing national land spatial patterns and reducing ecological risks. Furthermore, it offers a theoretical basis for formulating scientific and reasonable land management and planning policies.

5. Conclusions

In this study, which is based on the existing land use types and primary functions of PLESs, a PLES land classification system for the Turpan and Hami regions was constructed, and a spatiotemporal evolution study of PLESs from 2000 to 2020 was conducted. A potential–connectivity–resilience ecological risk assessment model based on the SDGs was subsequently developed to achieve a multidimensional ecological risk assessment of the Turpan–Hami region and evaluate various risk values from a PLES perspective. The aim of this study was to assess the dynamic changes in PLES and ecological risks over the past 20 years from both spatial and temporal dimensions. The results indicate the following:

(1)

In terms of spatial distribution patterns and temporal changes, PLESs in the Turpan and Hami regions have undergone significant changes. Ecological space has dominated, covering 97% of the study area, followed by production space, with living space occupying the smallest area. Over the past 20 years, production and living spaces have continuously expanded, increasing by 1215 km² and 142 km², respectively, whereas ecological space has gradually decreased, as it has served as the primary source of expansion for living and production spaces.

(2)

From the perspective of the potential–connectivity–resilience assessment system, the maximum average values of various risk indices in the Turpan–Hami region have shown an increasing trend, indicating gradual deterioration in the ecological security status of the area. During the study period, the increases in potential risk and composite risk in the Hami region were greater than those in the Turpan region, with differences of 0.00345 and 0.01291, respectively. The differences in connectivity risk and resilience risk were smaller. This indicates that from 2000 to 2020, the rate of ecological degradation in the Hami region was higher than that in the Turpan region.

(3)

This study revealed that over 20 years, the average values of various PLES risks were as follows: living space > production space > ecological space. This indicates that the risk values are greater in the more densely populated living spaces, primarily because of the homogeneous landscape types, low vegetation cover, high population density, and significant human disturbance in these areas, resulting in higher risk values.

This study overcomes the limitations of previous ecological risk assessments based solely on land use, which have focused only on land use landscapes. By detaching the ecological risk assessment from the landscape, this study provides a scientifically sound theoretical basis for optimizing the ecological pattern in arid regions and enhancing regional ecological security. This approach contributes to achieving regional sustainable development goals.