Habitat Quality Dynamics in Urumqi over the Last Two Decades: Evidence of Land Use and Land Cover Changes

Abstract

The integrity of habitat quality is a pivotal cornerstone for the sustainable advancement of local ecological systems. Rapid urbanization has led to habitat degradation and loss of biodiversity, posing severe threats to regional sustainability, particularly in extremely vulnerable arid zones. However, systematic research on the assessment indicators, limiting factors, and driving mechanisms of habitat quality in arid regions is notably lacking. This study takes Urumqi, an oasis city in China’s arid region, as a case study and employs the InVEST and PLUS models to conduct a dynamic evaluation of habitat quality in Urumqi from 2000 to 2022 against the backdrop of land use changes. It also simulates habitat quality under different scenarios for the year 2035, exploring the temporal and spatial dynamics of habitat quality and its driving mechanisms. The results indicate a decline in habitat quality. The habitat quality in the southern mountainous areas is significantly superior to that surrounding the northern Gurbantunggut Desert, and it exhibits greater stability. The simulation and prediction results suggest that from 2020 to 2035, habitat degradation will be mitigated under Ecological Protection scenarios, while the decline in habitat quality will be most pronounced under Business-As-Usual scenarios. The spatial distribution of habitat quality changes in Urumqi exhibits significant autocorrelation and clustering, with these patterns intensifying over time. The observed decline in habitat quality in Urumqi is primarily driven by anthropogenic activities, urban expansion, and climate change. These factors have collectively contributed to significant alterations in the landscape, leading to the degradation of ecological conditions. To mitigate further habitat quality loss and support sustainable development, it is essential to implement rigorous ecological protection policies, adopt effective ecological risk management strategies, and promote the expansion of ecological land use. These actions are crucial for stabilizing and improving regional habitat quality in the long term.

1. Introduction

Land use change is a key factor influencing regional habitat quality [1]. Habitat quality refers to the capacity of the natural environment to provide the essential conditions necessary for species’ survival [2]. Land use change can reflect the status of regional biodiversity and ecosystem health to a certain extent, serving as a crucial guarantee and fundamental prerequisite for the survival and reproduction of organisms [3]. Therefore, the level of habitat quality is directly related to the maintenance of biodiversity and the function of ecosystem services. With the intensification of human activities, especially the rapid advancement of urbanization, the dramatic expansion of impermeable surfaces has had a significant impact on the quality of regional habitats, which in turn poses a great challenge to biodiversity conservation [4], and coupled with global climate change, the quality of habitats and biodiversity poses a serious threat to ecosystems [5]. In particular, urban expansion over the past two decades has encroached upon extensive areas of arable and natural land with high ecological value, resulting in the loss of over 80% of biologically important habitats [6]. This problem is exacerbated in extreme arid regions, where limited water resources and land desertification intensify habitat degradation under human-induced disturbances, making habitats more vulnerable [7]. Without timely intervention, habitat quality will be subjected to substantial threats in the future [8].

To enhance regional habitat quality, numerous scholars have conducted extensive research. Amphibian biologists were among the first to focus on this topic, primarily assessing specific wildlife species or communities through field surveys [9,10]. Swiss scholar Glenz et al. proposed an application of a predictive wolf habitat model to address the issue of large-scale migration and the establishment of small populations in Canis lupus populations [9]. With the intensification of urbanization’s impact on habitat quality since the Anthropocene, current research primarily focuses on two aspects: First is quantifying habitat quality to identify priority conservation areas [11,12]. Ikaunece et al. conducted a forest key habitat inventory and established a forest network of endangered tree species in order to establish priority protected areas for Nordic endangered tree species [11]. To effectively protect the biodiversity of coastal areas, Zhu et al. utilized geomorphological models, species habitat models, and other methodologies to determine the priority sequence for conservation in response to sea-level rise and land use changes within the Matanzas River Basin [12]. On the other hand, current research explores the impacts of land use changes on habitat quality from both micro and macro perspectives during the urbanization process. At the microlevel, biodiversity is regarded as the best indicator of habitat quality. Researchers have primarily analyzed the effects of land use changes on biodiversity, focusing on plant and animal communities [13,14]. Urban growth’s conversion of ecological land to construction land significantly fragments habitats, disrupts ecological flows, and diminishes biodiversity [14]. At the macrolevel, most studies combine InVEST (Integrated Valuation of Ecosystem Services and Trade-offs)and predictive models to assess the impacts of historical and future land use changes on habitat quality [15,16]. Habitat quality is closely related to changes in land use types, with some researchers evaluating the relationship between land use and habitat quality through changes in habitat quality caused by land use transitions [17,18]. Chinese researchers, utilizing Geographically and Temporally Weighted Regression (GTWR) and Multiscale Geographically Weighted Regression (MGWR) models, have identified urban expansion as a primary driver of shifts in habitat quality [19]. Habitat quality is closely related to changes in land use types, and some researchers have evaluated the relationship between land use and habitat quality through changes in habitat quality caused by land use conversion. Wu et al. [20] and Zhang et al. [21] developed frameworks to assess future land use impacts on habitat quality. However, it is important to note that the contribution of land use conversion to changes in habitat quality has rarely been measured across different historical and future scenarios. This is particularly true in varying developmental contexts, where the changes in habitat quality and their driving mechanisms remain poorly understood. Consequently, there is a limited understanding of the long-term dynamic changes in regional habitat quality, as well as inadequate support for regional ecological quality enhancement and sustainable economic development.

Despite current research addressing habitat quality changes across various temporal and spatial scales—such as provincial, metropolitan, protected areas, and river basins—there is relatively little focus on arid regions, and there is a lack of systematic assessment indicators. Arid regions are among the most vulnerable natural landscapes in the world, covering approximately 40% of the Earth’s total area. These regions are characterized by scarce precipitation, high evaporation rates, low runoff, and poor soil quality. Their ecological environments are exceptionally fragile and highly sensitive to both climate change and human activities [22]. Since the 21st century, with the advancement of the national Western Development Strategy and the implementation of the Belt and Road Initiative, Urumqi has experienced rapid economic development, population growth, and intensified the development and utilization of water and soil resources. As a result, the city’s ecological environment issues have become increasingly prominent [23]. The dynamics and future trends of habitat quality in the region are critical for the country’s ecological security and biodiversity conservation. In response, this study focused on Urumqi, an oasis city in Northwest China’s arid region. First, the land use transfer matrix from 2000 to 2022 was used to identify the direction and extent of land use changes, as well as analyze their temporal and spatial characteristics. Next, the InVEST-PLUS model simulated land use changes under four scenarios for 2035, measuring historical and future habitat quality to analyze its temporal and spatial evolution: Business-As-Usual (BAU), cropland protection (CP), ecological protection (EP), and economic development (ED). Finally, this study quantitatively assessed the impact of land use changes on habitat quality. The findings provide a theoretical basis for stakeholders to optimize land use structures and improve habitat quality.

2. Materials and Methods

2.1. Study Area

Urumqi, located in the arid northwest region of China, holds a pivotal position within the core area of the Silk Road Economic Belt. The municipality governs seven districts and one county, covering an area of approximately 1.42 × 10⁴ km², with 545.1 km² developed. The permanent population is 4.085 million. The climate is classified as a mid-temperate continental arid climate, with an average annual temperature of 25.7 °C and annual precipitation of 286.3 mm. Urumqi is surrounded by mountains on three sides, with the Gurbantunggut Desert to the north (Figure 1), which is a typical composite system of “Mountains-Rivers-Forests-Fields-Lakes-Grass-Sand, and Ice” [24]. As a result, it is rich in flora and fauna. Urumqi is home to approximately 1100 species of wild plants and more than 300 species of vertebrates. As the capital city of Xinjiang, Urumqi, driven by the Urumqi metropolitan area, has cultivated a favorable environment for economic growth. However, nature faces numerous challenges. Research on the spatiotemporal evolution and driving mechanisms of habitat quality in oasis cities under the impact of urbanization can provide valuable theoretical guidance for regional sustainable development and ecological balance.

2.2. Data Description

The types of data used in this study mainly included spatial data and statistical data. The spatial data were mainly land use data and land use simulations of factors affecting land use. The data on land use and land cover were sourced from the Geospatial Data Cloud. Land use data were obtained through the interpretation of remote sensing imagery. To ensure optimal data accuracy, a series of preprocessing tasks were conducted. Each image must have had less than 10% cloud cover, and the imaging period was selected from July to September. Missing images were replaced with those from adjacent years, with a spatial resolution of 30 m. The main preprocessing steps included geometric correction, radiometric calibration, atmospheric correction, image mosaicking, and cropping. The ENVI software was primarily used for radiometric calibration, with the purpose of eliminating errors inherent in the sensor during imaging and converting image brightness grayscale values into radiance values and other physical quantities. Atmospheric correction used the model tool “Atmospheric Correction Module”-“FLAASH Atmospheric Correction”. Subsequently, image mosaicking was performed using the “Mosaicking”-“Seamless Mosaic” tool. Finally, cropping was performed using the “Regions of Interest”-“Subset Data from ROIs” tool with the administrative boundary vector data of the study area, thereby obtaining the regional image data for Urumqi in the years 2000, 2010, 2020, and 2022. All image data were unified using the World Geodetic System 1984 geodetic coordinate system. The PLUS (Patch-level Land Use Simulation Model, PLUS) model’s influencing factor data included a total of 13 categories, such as temperature and precipitation. Details of the data are provided in

Table 1. Data sources of the land use/land cover changes and its impact factors.

Sub-Data	Year(s)	Resolution	Database Sources	Access Date
Land use/Land cover	2000\2010\2020\2022	30 m	https://www.gscloud.cn/	20 December 2023
DEM/Slope/Elevation	2020	30 m	https://www.gscloud.cn/	20 December 2023
NDVI/Soil	2020	1000 m	https://www.resdc.cn/	25 April 2024
Precipitation/Temperature	2020	30 m	http://data.cma.cn/	22 April 2024
GDP/Population	2020	1000 m	https://www.resdc.cn/	22 April 2024
Railway/Highway/Road/ Settlement/Water	2020	1:1,000,000	https://www.webmap.cn/	20 December 2023

. Before using the PLUS model for simulation and prediction, it is necessary to rasterize the driving factors used in this study. First, in ArcGIS 10.8 software, the projection tool was used to ensure that the projection coordinates of the road network, railway network, residential points, and river system vector data were consistent with the land use data of the study area Next, the Euclidean distance tool was used to rasterize the four accessibility factors. Finally, all raster data were unified in terms of spatial extent and resolution using the resampling and cropping raster tools, with the projection coordinates set to Albers Conic Equal Area and the spatial resolution standardized to 30 m.

The methodology of this article consists of four steps, as illustrated in Figure 2. The first step involves preprocessing land use data and land use impact factor data. The second step defines four scenarios based on regional planning policies, including ecological red lines, the Urumqi 14th Five-Year Plan, and permanent basic farmland: BAU, CP, EP, and ED. The Markov chain model is used to predict land use demand for 2035, and the PLUS model simulates future land use changes under these four scenarios. The third step involves using the InVEST model to evaluate and compare the spatiotemporal distribution of habitat quality for the years 2000, 2010, 2022, and 2035. The fourth step is to analyze the research results and propose optimization recommendations.

2.3. Habitat Quality Evaluation Indicator System

The InVEST model is a comprehensive valuation model of ecosystem services and trade-offs that provides users with a variety of options for evaluating ecosystem functioning. It consists of three main modules, each containing lots of valuation items: terrestrial, marine, and freshwater ecosystem valuation. The main purpose of this study was to evaluate the habitat quality in Urumqi using the habitat quality module of the InVEST model. The habitat quality module assumes that the higher the quality of the habitat, the greater the biodiversity in the area, and vice versa [25]. This model correlates the land use type with the threat factors of the ecosystem and derives the degree of the habitat degradation of the ecosystem based on the sensitivity of the ecosystem to the threat sources, the distance of influence of the threat sources, the degree of interaction between the threat sources, etc. (taking the range of values from 0 to 1), and the value is directly proportional to the level of habitat quality, as specified in the following formulas:

$$Q_{xj}=H_j\left[1-\left({\displaystyle \frac{D_{xj}^Z}{D_{xj}^Z+k^Z}}\right)\right]$$

(1)

where Q_xj: habitat quality of individual raster x at habitat or land use type j; H_j: habitat suitability at habitat or land use type j; D_xj: weighted mean of threat levels of raster cell x at habitat or land use type j; k: half-saturation parameter, taken as 1/2 of the maximum value of D_xj; and Z: normalization constant, taken as 2.5.

$$D_{xj}={\sum }_{r=1}^R{\sum }_{y=1}^{Y_r}\left(w_r÷{\sum }_{r=1}^R{w}_r\right)\times r_y\times i_{rxy}\times {\beta }_x\times S_{jr}$$

(2)

$$i_{rxy}=1-\left({\displaystyle \frac{dxy}{d{r}_{max}}}\right)$$

(3)

where D_xj: weighted average of the total threat level to which raster x is exposed, a certain threat factor specifically; R: all grids on the raster layer of threat factor r; Y_r: phase element set on the raster layer of threat factor r; W_r: normalized threat weight, ranging from 0 to 1; r_y: used to determine whether a raster y is the source of the threat factor r; i_rxy: distance function between the habitat class and threat factor; β_x: under the relevant environmental protection state; β_x: the accessibility level of the threat source to grid x under the relevant environmental protection status; and S_jr: the sensitivity to threat factor r when the habitat or land use type is j.

Based on the land use data, the model estimates habitat quality by considering the distance and intensity of threat factors, as well as the sensitivity of different habitat types to these factors. In this study, we refer to the InVEST model manual, consider the actual conditions in the study area, and draw on relevant research to define the threat source factors (

Table 2. Threat factor parameter.

Threat Factors	Weight	Influence Distance (km)	Spatial Decay Type
Cropland	0.7	8	linear
Built-up land	1	10	exponential
Bareland	0.2	3	exponential

) and sensitivity factors (

Table 3. Sensitivity of different land use types to habitat threats.

Land Use Type	Habitat Suitability	Threats
		Cropland	Built-Up Land	Bareland
Cropland	0.5	0	0.5	0.4
Forest land	1	0.8	0.9	0.5
Grassland	0.7	0.5	0.6	0.5
Water body	0.9	0.7	0.8	0.2
Built-up land	0	0	0	0
Bareland	0.1	0.1	0.2	0

).

In the geographic information map model, cropland, forest land, grassland, water body, built-up land, and bareland are assigned values from 1 to 6. In ArcGIS 10.8, the land use conversion map cells were algebraically overlaid to integrate spatial information from the map-coded values [15]. Based on the geographic information map model, the map algebra overlay operation was performed on the Urumqi 2022 and future land use atlas grid cells, and the habitat quality changes were analyzed based on the results of the operation. The formula is

$$W=10B+Q$$

(4)

where W is the map cell grid map of land use class changes; B is the 2022 land use map cell grid value; and Q is the attribute value of the future land use mapping unitary grid.

The relationship between land use change and habitat quality was determined through the Habitat Quality Dynamic index. The formula is as follows [15]:

$${HQDI}_{ij}={\displaystyle \frac{∆H_{ij}}{∆L_{ij}}}$$

(5)

where HQDI_ij is the dynamic index of habitat quality; positive values indicate positive impacts of land use change on habitats; negative values indicate negative impacts; and the larger the absolute value, the larger the impacts. ΔH_ij indicates the change in habitat quality caused by land use type conversion, and ΔL_ij indicates the area of land transformation in the region during the same period.

We used the contribution index (CI) to characterize the extent to which land use type conversion contributes to changes in habitat quality, calculated as follows [15]:

$$CI=Q_{ij}\times {HQDI}_{ij}$$

(6)

where CI is the contribution index, and Q_ij is the proportion of the area converted from land use type i to land use type j to the total converted area. Positive and negative values of the CI represent positive and negative contributions, respectively. The higher the absolute value of the CI, the greater the impact of land use conversion on habitat quality.

2.4. PLUS Model

The Patch-generating Land Use Simulation (PLUS) model is an advanced version of the Future Land Use Simulation (FLUS) model. It integrates the Land Expansion Analysis Strategy (LEAS) and Cellular Automata model based on multi-type random patch seeds (CARS). This model not only enables a better exploration of the triggers behind various types of land use changes but also improves the simulation of changes among multiple types of land use patches [4]. We used the PLUS model to simulate future land use, which integrates a Markov chain. The advantage of this model is its ability to better explain the mechanisms driving various land use changes and more accurately simulate future land use changes under different policy scenarios [26]. Based on the land use data of Urumqi in 2015 and 2020, the development probability of the six land use types was determined, and then based on the actual situation, 13 factors were selected from natural, socio-economic, and accessibility aspects, including elevation, slope, slope direction, temperature, precipitation, soil type, Normalized Vegetation Index (NDVI), population density, and Gross Domestic Product (GDP), as well as distance to highway, distance to railway, distance to river, and distance to settlements; however, future development plans for the region were not considered when setting land use simulation scenarios. In the LEAS, the probability of development for each land use type was determined using a random forest algorithm (Figure 3). The current land use map of Urumqi in 2020 was simulated based on the 2015 land use data, combined with the development probability of each type of land use incorporating the development probability for each land use type, the projected future demand for each land use, the transfer cost matrix, and the weights of domain factors. The simulated land use data for 2020 were then compared with the actual 2020 data through spatial overlay analysis. The Kappa statistic and Figure of Merit (FoM) were calculated, and the parameters were iteratively adjusted to improve the simulation accuracy. Among them, the Kappa coefficient is an indicator for measuring classification accuracy. Usually, Kappa ranges from 0 to 1 and can be divided into five groups to indicate different degrees of consistency: 0.0 to 0.20 indicates very low consistency (mild), 0.21 to 0.40 indicates general consistency, 0.41 to 0.60 indicates moderate consistency (moderate), 0.61 to 0.80 indicates basic consistency, and 0.81 to 1 indicates almost complete consistency [26]. The FoM coefficient is concentrated on the grid where land use change occurs and is usually used to measure the goodness of fit of changes in land use composition. The FoM value is usually between 1% and 59%, and larger FoM values often match higher model accuracy [26]. Finally, based on the 2020 land use data, the model parameters were repeatedly debugged in the PLUS model to simulate the land use data of Urumqi in 2035 under the B-A-U, CP, EP, and ED scenarios.

2.5. Multi-Scenario Simulation

The Business-As-Usual (B-A-U) scenario refers to a situation where historical land use changes continue without any changes or interventions. In this scenario, land use transition probabilities, transition cost matrices, and constraints on conversion areas remain unchanged. Croplands in farming areas are crucial for food production, agricultural supply, and rural development. Therefore, this study introduces the CP scenario, which focuses on ensuring food security and preserving permanent basic agricultural land. In the CP scenario, the likelihood of cropland converting to urban areas is reduced by 60%, following the Business-As-Usual (BAU) probabilities, while the transition probabilities for other land uses remain unchanged [27]. The EP scenario focuses on the conservation of land types important for ecological restoration, such as forests, grasslands, and water body, with the aim of ensuring ecological security and maintaining ecosystem service functions [28]. The ED scenario requires the rational development of urban boundaries within the constraints of cropland protection and ecological protection lines, with the aim of creating a compact and intensive urban spatial pattern [29]. The proportions of land use transition probabilities are detailed in

Table 4. Transition probability matrix of each land use type in the EP and ED scenarios.

Scenarios	Land Use Type	Cropland	Forest	Grassland	Built-Up Land
Business-As-Usual	not adjusted
Cropland Protection	Cropland				−60%
Ecological Protection	Cropland		+30%	+60%	−50%
	Forest land	−80%	−80%		−90%
	Grassland	+20%			−80%
	Built-up land	+20%	+20%	+50%
Economic Development	Cropland				+60%
	Forest land				+40%
	Grassland				+40%
	Water body				+20%
	Bareland				+10%

Notes: Based on BAU scenario: “+” and “−” indicates “increasing” and “decrease”.

.

In this study, based on the land use data of Urumqi in 2015 and 2020, the development probability (Figure 3) of each land use type was generated using the LEAS module in the PLUS model. This was then further simulated to generate the 2020 land use map by combining the 2015 land use data with the development probabilities in the CARS module. After comparing the actual 2020 land use with the simulated 2020 land use, the following results were obtained: Kappa = 0.8917, FoM = 11.74% (Figure 4).

Consequently, the outcomes of the PLUS Model are closely aligned with the actual land use patterns, exhibiting a high degree of precision, and thus are suitable for forecasting land use in Urumqi by the year 2035.

3. Results

3.1. Characteristics of Historical and Future Land Use Spatiotemporal Changes

Based on the land use transfer matrix of Urumqi from 2000 to 2020 (Figure 5), the land types in Urumqi are ranked by area proportion from largest to smallest as follows: grassland > bareland > cropland > forest land > built-up land > water body. Notably, the area of arable land first increased and then decreased, exhibiting land use dynamic indices of 0.42% and −24.74%, respectively. The main reason is that large-scale reclamation after 2000 led to an increase in arable land area, and after 2010, the policy of converting farmland back to forest and grassland caused most of the arable land to be transformed into forest and grassland; the area of grassland showed a decreasing trend over 22 years, possibly due to overgrazing; the area of water body first decreased and then increased, with land use dynamics of -1.16% and 14.28%. The area of construction land showed a rapid growth trend, with new construction primarily emerging from arable land and grassland. Overall, the study area is characterized by a high coverage of grassland and desert but exhibits poor habitat quality. This trend is compounded by the rapid expansion of construction land, leading to the loss of ecological areas such as forest land, grassland, and arable land.

The model results of land use area changes in Urumqi for the year 2035 under different scenarios are as follows (Figure 6). Under the BAU scenario, land use types that saw an increase in area include built-up land, bare land, and forest land, with built-up land experiencing the most significant growth at 27.55%. In contrast, cropland, water bodies, and grassland areas decreased, with cropland showing the largest reduction at 25.30%. Under the CP scenario, compared to the 2020 land use areas, cropland decreased by 159.63 km², grassland by 563.48 km², and water body by 31.4 km². Forest land, built-up land, and bareland areas increased by 37.61 km², 112.48 km², and 604.41 km², respectively. The CP scenario’s land use evolution pattern is consistent with the BAU scenario. The decline in cropland is attributed to the lack of control over the conversion of permanent cropland areas. Under the EP scenario, forest land, grassland, and water body increased, while cropland decreased by 27.04%, and built-up land and bareland decreased by 1.49% and 3.36%, respectively. Under the ED scenario, built-up land, bareland, and forest land increased, with built-up land exhibiting the highest growth rate of 39.15%. This increase is closely linked to factors such as population density and GDP. In contrast, cropland, water bodies, and grassland decreased, with cropland experiencing the most substantial reduction of 20.10%.

A comparative analysis of land use data for Urumqi in 2022 and 2035, based on a geographic information system model (Figure 7), shows that unchanged land (coded as 11, 22, 33, 44, 55, 66) accounted for 91.02%, 90.35%, 94.39%, and 90.64% across the four time periods. From 2022 to 2035, the BAU scenario’s total land use change area is 1367.38 km², with the largest changes occurring from grassland to bareland (36) and from bareland to grassland (63). In the CP scenario, the total land use change area is 1326.38 km², with the greatest change from grassland to bareland, followed by the conversion of cropland to built-up land. The ED scenario shows a total land use change area of 1273.48 km², with the most significant change from grassland to unused land (667.05 km²), followed by cropland to built-up land (148.87 km²). In the EP scenario, the total land use change area is 795.04 km², with the largest change from bareland to grassland (253.17 km²), followed by the conversion of cropland to grassland (181.62 km²). This reduction in the conversion of water body, forest land, and grassland to arable and built-up land under the EP scenario controls excessive growth in construction and cropland, resulting in significant ecological improvements. In summary, the simulation of Urumqi’s land use situation in 2035 indicates that future changes in land use will primarily focus on the oasis transition zones and ecological intersection areas.

3.2. Spatiotemporal Changes in Habitat Quality in the Past and Future

From 2000 to 2022, the habitat quality scores in Urumqi showed a declining trend, with scores of 0.4180, 0.3937, and 0.3838, reflecting change rates of −5.8% and −2.5% (

Table 5. Average habitat quality (HQ) for 2000–2022 and different scenarios (BAU, CP, EP, and ED).

HQ	2000	2010	2022	BAU	CP	EP	ED
Mean	0.4180	0.3937	0.3838	0.0898	0.0893	0.0906	0.0892

). During this period, the implementation of the Western Development Strategy and rapid urbanization significantly contributed to the continuous decline in habitat quality. After 2010, the continuous advancement of the “Grain-to-Green” policy slowed the rate of decrease in arable land and grassland areas, thus mitigating the downward trend in habitat quality. During the study period, areas with excellent habitat quality consistently constituted 1% of the total area. The proportion of good habitat quality areas decreased from 37% in 2000 to 32% in 2010, remaining stable thereafter. The proportion of areas with moderate habitat quality initially increased and then decreased, rising from 26% in 2000 to 28% in 2010, before returning to 26% in 2022. The proportion of areas with poor habitat quality gradually increased from 36% in 2000 to 41% in 2022. Overall, habitat quality in Urumqi displayed a trend of deterioration (Figure 8).

The spatial distribution of habitat quality in Urumqi exhibits significant regional heterogeneity. The distribution pattern is characterized by a high-density zone along the northern side of the Tianshan Mountains, a low-density zone in the Gurbantunggut Desert, and transitional distribution features in other areas (Figure 9). High-quality habitats are concentrated on the northern slopes of the Tianshan Mountains and in the northwestern part of Tianchi, primarily due to the dominance of forest land use types in these areas, which have high forest coverage and thus ensure a higher level of habitat quality. Other regions exhibit sporadic distributions of high-quality habitat. Relatively good habitat quality areas are primarily found within the Tianshan Mountains, where grassland vegetation dominates, and urban development and population density remain relatively low. These areas maintain better ecological conditions, contributing to higher habitat quality compared to urbanized zones. The effective implementation of ecological protection policies has further stabilized grassland landscapes, promoting the improvement of habitat quality. Moderate habitat quality regions are primarily found within the central oasis area. Poor habitat quality regions are predominantly distributed in the northern desert zone and adjacent areas to Turpan, where land use types are primarily bareland and low-vegetation grasslands, with limited potential for ecological improvement. Overall, the spatial distribution pattern of habitat quality in Urumqi is predominantly influenced by land cover types, showing significant spatial variability. Areas with natural vegetation, such as grasslands and forests, tend to have higher habitat quality, while urbanized areas and regions with high human activity, like built-up lands and agricultural zones, typically exhibit lower habitat quality. This spatial variability underscores the impact of land use and land cover on the ecological balance and habitat quality within the region.

Based on the changes in habitat quality over four categories, the habitat changes from 2000 to 2022 are categorized into five types (Figure 10a). The majority, 86.01%, are stable zones with minimal anthropogenic disruption. Significant degradation and significant improvement areas account for 2.62% and 0.81%, respectively, while minor degradation and minor improvement areas account for 8.52% and 2.03% (Figure 10b). These changes are largely concentrated in regions of heightened human activity, reflecting the dual influence of human actions on habitat quality. The hot-spot analysis of habitat quality was employed to study the spatial variation in habitat quality. With little spatial fluctuation in habitat quality from 2000 to 2022, the year 2022 serves as an illustrative case (Figure 10c).

The results indicate that hot-spot areas are primarily found in the Tianshan Ecological Protection Zone and the Salt Lakes, with a focus on forest–grassland mosaic zones. These areas have higher habitat quality due to their natural vegetation and relatively low human disturbance. In contrast, cold-spot areas are concentrated in the oasis aggregation development zones, the periphery of the Gurbantunggut Desert, and the outskirts of the Salt Lakes. These cold spots are dominated by agricultural lands, with grasslands occupying secondary positions. The largest proportion of cold spots is located at the edges of oases, characterized by bareland, which indicates a significant loss of ecological quality in these regions due to urban expansion and agricultural development.

Different scenarios (BAU, CP, EP, and ED) for 2035 show a different spatial distribution of habitat quality (Figure 11). Among them, the EP scenario demonstrates the highest habitat quality, attributed to significant increases in grasslands, forest lands, and water bodies. Spatially, all scenarios show that the Tianshan Ecological Protection Zone has high habitat quality, while other areas have lower quality.

In comparing habitat quality ratings across scenarios, the BAU scenario has good and excellent habitat quality areas totaling 458.15 km², or 3.23% of the total area. In contrast, the EP scenario shows an increase in good and excellent habitat quality areas by 20.72 km², representing 3.38% of the total area. The ED scenario integrates elements of both the CP and BAU scenarios, aiming to balance ecological and economic development. Overall, the average habitat quality in Urumqi is significantly declining (Table 5). Reasons include various factors such as land use transformation, urbanization and economic development, water resource scarcity and pollution, climate change, overgrazing and irrational development, and insufficient ecological sensitivity and protection [5,23,24,30]. These factors, including urbanization, agricultural expansion, and land degradation, are intertwined and contribute to the continuous deterioration of Urumqi’s ecological environment quality. The rapid urban sprawl, especially in the oasis transition zones, leads to the loss of critical habitats, including grasslands and forest areas. This, combined with the pressures of intensive farming and the expansion of infrastructure, has significantly impacted the overall ecological health of the region. Moreover, the development of areas near deserts and salt lakes exacerbates soil degradation and reduces the capacity of ecosystems to support biodiversity. The ongoing decline in habitat quality underscores the need for effective land use planning and sustainable development practices to protect the region’s natural resources and biodiversity. Although the ED scenario may better reflect the actual conditions of the city, the EP scenario offers the best habitat quality in Urumqi. To improve this situation, it is necessary to adopt comprehensive ecological governance measures, optimize the sustainable use of land resources, and strengthen ecological and environmental protection.

3.3. The Influence of Land Use Alterations on Habitat Quality Dynamics

The desert–oasis transition zone is the area where environmental quality changes are most pronounced. The interconversion between farmland, grassland, desert, and construction land is the main factor affecting ecological environment quality. Land use changes have both positive and negative impacts on habitat quality. Land use changes contribute both positively and negatively to habitat quality changes. Under the BAU scenario from 2022 to 2035, the positive contribution index to habitat quality change is 0.43, while the negative contribution index is −41.46, yielding a total contribution index of −41.03 (

Table 6. Contribution index for the impact of land use conversions on HQ-CA from 2022 to 2035.

Habitat Quality Change Type	Land Use Transfer	BAU		CP		EP		ED
		CA (km2)	CI	CA (km2)	CI	CA (km2)	CI	CA (km2)	CI
Reduction	12	0.22	−0.01	0.22	−0.04	0.22	−0.01	0.22	−0.01
	13	98.94	−3.16	34.04	−1.12	33.71	−1.16	33.71	−1.16
	14	2.27	−0.08	0.48	−0.02	0.49	−0.02	0.49	−0.02
	15	109.07	−3.3	173.46	−5.48	148.87	−4.87	148.87	−4.87
	16	6.49	−0.2	2.41	−0.07	1.50	−0.05	1.50	−0.05
	21	0.21	−0.01	0.22	−0.01	0.23	−0.01	0.23	−0.01
	23	27.35	−1.43	27.42	−1.47	27.39	−1.53	27.39	−1.53
	26	0.17	−0.01	0.16	−0.01	0.16	−0.01	0.16	−0.01
	31	39.28	−1.09	42.35	−1.22	43.24	−1.30	43.24	−1.30
	32	63.58	−2.62	63.67	−2.70	63.48	−2.81	63.48	−2.81
	34	2.68	−0.02	2.60	−0.03	2.59	−0.03	2.59	−0.03
	35	36.12	−0.89	33.13	−0.85	38.52	−1.02	38.52	−1.02
	36	712.82	−25.70	635.72	−23.49	667.05	−25.78	667.05	−25.78
	41	1.73	−0.04	0.67	−0.02	0.65	−0.02	0.65	−0.02
	43	5.62	−0.23	5.54	−0.23	5.54	−0.24	5.54	−0.24
	45	20.32	−0.48	0.95	−0.02	1.01	−0.02	1.01	−0.02
	46	29.11	−1.63	46.20	−2.66	46.71	−2.80	46.71	−2.80
	53	27.14	−0.02	26.23	−0.02	18.92	−0.01	18.92	−0.01
	61	1.23	−0.01	2.07	−0.01	5.25	−0.02	5.25	−0.02
	63	140.40	−0.49	134.26	−0.47	138.26	−0.51	138.26	−0.51
	65	8.70	−0.06	6.50	0.05	13.37	−0.10	13.37	−0.10
Improvement	56		73.14	0.02		46.71	−2.80
	64	10.45	0.43	9.16	0.38	9.22	0.40	9.22	0.40
Total		1333.88	−41.03	1238.27	−39.54	766.46	-28.08	1257.16	−42.92

Notes: HQ-CA means Habitat Quality Change Area; numbers 1 to 6 in the table represent cropland, forest land, grassland, water body, built-up land, and bareland, respectively.

).

This indicates that land use changes will severely degrade habitat quality during this period. The main cause of habitat quality degradation is the transfer of grassland and cropland to other land types, with a cumulative change area of 1071.91 km², contributing 89.4% of the total impact. In the CP scenario, the positive contribution index to habitat quality change is 0.4, and the negative contribution index is −39.94, resulting in a total contribution index of −39.54. This suggests that land use changes will lead to habitat quality degradation, with the main cause being the transfer of grassland to other land types, with a cumulative change area of 777.46 km², contributing 70.81% of the total impact. Under the EP scenario, the positive contribution index to habitat quality change is 1.26, while the negative contribution index is −29.34, resulting in a total contribution index of −28.08. This indicates that habitat quality will still decline, though less severely compared to other scenarios. The reduction in the transfer area of grassland to other land types is significant, with a cumulative change area of only 206.17 km². In the ED scenario from 2022 to 2035, the positive contribution index is 0.4, the negative contribution index is −42.31, and the total contribution index is −41.92. This indicates that land use changes will lead to habitat quality degradation, with the main cause being the transfer of grassland to bareland and cropland to built-up land, with a cumulative change area of 815.92 km², contributing 72.44% to the total impact.

4. Discussion

4.1. Spatial and Temporal Variations in Habitat Quality and Driving Mechanisms in the Arid Zone

This study provides a comprehensive analysis of land use spatiotemporal changes in Urumqi from 2000 to 2022 and used the PLUS model to simulate land use under four scenarios for the year 2035. The Urumqi oasis ecosystem encompasses a diverse range of environments, including mountains, water bodies, forests, fields, lakes, grasslands, and deserts. Throughout the study period, both natural factors and human activities have jointly influenced changes in land use, thereby impacting habitat quality. With the acceleration of urbanization, the area of built-up land has consistently increased, while grassland areas have steadily declined. Spatially, urban expansion has exhibited a trend of moving from south to north and from west to east [31]. The deterioration in habitat quality is closely related to the reduction in grassland area and the increase in built-up land. The deterioration in habitat quality is closely related to the reduction in grassland area and the increase in built-up land.

From the perspective of natural factors, elevation, precipitation, and the NDVI impose significant constraints on ecological land use over long-term scales, thereby influencing overall habitat quality. Elevation affects factors such as slope and annual precipitation, which in turn impact land use types and habitat quality. Generally, habitat quality is higher in mid- to low-elevation areas and lower in high-elevation areas. Previous studies have indicated that high elevation and steep slopes provide favorable conditions for the growth of natural vegetation such as forests and grasslands, resulting in higher vegetation cover and habitat quality, which aligns closely with the findings of this study [21]. Temperature and precipitation affect species’ habitat suitability, with significant impacts on species composition, ecological functions, processes, and surface vegetation growth. Adequate rainfall promotes habitat suitability [32]; however, excessive rainfall can trigger landslides, mudslides, and other disasters, exacerbating soil erosion and negatively impacting habitat quality. This study finds that ecological land use in the region is more susceptible to precipitation factors. Since 1980, the northwestern arid region has experienced a continuous trend of warming and increased moisture [33], though precipitation remains unevenly distributed spatially. For desert plants in arid areas, water is a critical limiting factor for survival. Compared to studies in the eastern regions, precipitation is crucial for habitat quality in Urumqi. The NDVI serves as a key indicator of vegetation growth, with elevation and precipitation affecting vegetation growth and types in arid regions [30]. Soil type is a determinant of terrestrial ecosystem habitat quality, with soil pH and organic matter showing negative and positive correlations with habitat quality, respectively. A reduction in soil organic matter may lead to decreased soil fertility and quality, thereby impairing habitat quality [5]. In arid regions, where most soils are aeolian, soil fertility is low and soil types are homogeneous, which limits their impact on desert plants.

From a socio-economic perspective, the rapid urbanization in Urumqi over the past two decades has led to the expansion of urban built-up areas and transportation networks, which have encroached upon large areas of grassland, thereby disrupting grassland ecosystems [34]. Additionally, the implementation of the Western Development policy has resulted in population growth in the region. Noteworthy are the negative impacts of population migration on receiving areas, such as increased cropland, urban expansion, rising household waste, and greater human disturbance of natural environments. Although various ecological protection policies have been implemented, restoration efforts remain challenging. In summary, the spatial variation in habitat quality in Urumqi is directly influenced by factors such as elevation, precipitation, NDVI, construction land, and population density, while other factors exert indirect effects. From 2015 to 2020, the contribution of various factors to land use changes highlights that all considered driving factors impact land category conversion, though to varying degrees across different land category conversions (Figure 12). Future research should explore the driving forces behind land cover changes and habitat quality variations in Urumqi in greater depth [17]. This will help elucidate the mechanisms of habitat quality evolution and provide a solid foundation for improving habitat quality and promoting sustainable regional development.

Urumqi plays an important role in the northern sand control zone of China’s “two screens and three belts” ecological security pattern, and the habitat in this region has a positive impact on China’s ecological civilization construction [35]. We analyze the changes in various land use types in Urumqi, with the main purpose of better understanding the spatial and temporal distribution patterns of the land use status and habitat quality of different land use types in the region, and providing a theoretical basis for the urban development and ecological protection of Urumqi. The expansion of construction land in Urumqi is based on the premise of reducing the area of grassland and arable land. This is obviously not in line with the concept of sustainable development. Urumqi should firmly grasp the two development opportunities of the “Western Development Strategy” and “the Belt and Road Initiative”, make full use of the preferential policies of the country and the autonomous region, adhere to the concept of ecological civilization construction, correctly guide all efforts, and achieve a win–win situation in regional economy and ecological protection.

4.2. Multi-Scenario Projections of Habitat Quality in Arid Zones and Policy Recommendations

The PLUS model effectively simulates the spatial changes in land use structure [4]. In this study, the LEAS module of the PLUS model was employed to identify 13 selected driving factors. The default setting for the number of decision trees in the random forest was set to 20, with a sampling rate of 1% and 13 features for training the random forest, thereby generating probability distribution maps for different land use types. Parameters such as the transition cost matrix and neighborhood weights were set in the CARS module of the model. Following relevant studies, domain range parameters were set to 3, and the attenuation coefficient of the decay threshold, probability of random patch seeds, and other adjustments were made, with maximum proportions of random seeds for 2020 simulations set to 0.8, 0.1, and 0.0001, respectively, and parallel thread numbers set to 12 as initial assumptions [25].

Additionally, the InVEST model was used to assess habitat quality, with spatial comparisons between assessed habitat quality and landscape patterns revealing consistency, further validating the feasibility of the model. Through ecological process modeling, we observed that under the ED scenario, the area of built-up land experienced the most significant increase, by 40.64%. In the EP scenario, the area of forest and grassland increased substantially by 334.35 km², aligning with the land use conversion probabilities set for this scenario. The simulation results for the four scenarios met the expected outcomes, further confirming the reasonableness of the PLUS model’s parameter settings. Additionally, significant spatial variation in habitat quality was found. The Tianshan Ecological Protection Zone exhibited higher habitat quality, characterized by forest and grassland dominance, low population density, and minimal human disturbance. Furthermore, the influence of ecological protection policies enhanced the stability of forest and grassland ecosystems, allowing these areas to maintain high habitat quality. Low habitat quality was predominantly observed in the northern Gurbantunggut Desert Conservation and Restoration Area, where land use types are primarily bareland, with low vegetation cover and exposed surfaces, making it highly sensitive to environmental changes and limiting habitat quality. The spatial distribution of historical and future habitat quality was consistent with ecosystem patterns, validating the model’s evaluation. Finally, under the B-A-U scenario for 2022–2035, the greatest changes were observed in grassland to bareland (36) and bareland to grassland (63), with the most severe habitat quality degradation. This is attributed to the historical overgrazing and other human activities that have damaged grasslands, putting pressure and risks on the ecosystem, leading to habitat quality degradation in some areas [36]. At the same time, certain unused lands with favorable soil conditions have allowed resilient desert plants to thrive and reproduce, or have even accelerated the spread of these plants through human intervention [37]. In the EP scenario for 2022–2035, the most significant change was observed in cropland to grassland (13), primarily due to policies promoting cropland conversion to grassland.

Habitat quality is a crucial indicator of regional ecology [2]. Therefore, forecasting habitat quality under various future scenarios is crucial for providing scientific evidence to support government policies aimed at ecological environmental protection and preventing regional habitat quality degradation. By integrating the InVEST and PLUS models, future habitat quality was predicted under the B-A-U, CP, EP, and ED scenarios. Under the B-A-U scenario, the average habitat quality is expected to decline by 0.3039 compared to 2022, with a contribution index of −41.03 for land use change on habitat quality, which is consistent with previous research findings [15]. Without adjustments to land use structures, allowing them to evolve freely with natural trends, Urumqi’s ecological habitat quality will continue to deteriorate. Compared to in the B-A-U scenario, it is projected that by 2035, the area of habitat with good and excellent quality in the EP scenario will expand by 20.72 km², yet the overall average habitat quality will still deteriorate. It is a fact that, in this scenario, only the conversion area from cropland to forest and grassland has increased, while the conversion amount from forest and grassland to other types has decreased, with the transition probability between other types remaining unchanged. Moreover, certain areas under the EP scenario maintained higher habitat quality, which is consistent with increased forest and grassland areas, indicating the effective mitigation of habitat quality degradation.

To promote the sustainable development of ecological economy in Urumqi, it is crucial to integrate land use regulation policies and actions. Therefore, this study proposes three measures to enhance habitat quality in arid regions: (1) Control the amount of land converted between different land types. In formulating land use policies, it is important to consider the trend and extent of transfers between different land use types, as well as the impact of land use conversions on habitat quality, with the aim of striking a balance between ED and EP. (2) Strictly control urban expansion. During urbanization, it is essential to tightly regulate the construction land within the urban development boundary to prevent outward expansion and encroachment on cropland. In particular, attention should be given to protecting existing forest, grassland, and water resources from unreasonable human activities, preventing arable land abandonment, and construction land encroaching on arable land to ensure stable arable land areas [38]. (3) Plan urban green space construction comprehensively. Urban green spaces are potential green areas. By establishing ecological corridors to connect various green patches across the region, we can provide refuges for oasis flora and fauna, migratory birds, and more. Especially through vertical greening and green roofs, creating green spaces as much as possible will significantly contribute to biodiversity conservation, ecosystem stability, and regional ecological security.

4.3. Restriction and Uncertainty

Due to data limitations, the spatial resolution for most driving factors used in the PLUS model simulations of future land use is 30 m, which precludes the acquisition of higher-resolution data. Consequently, some data were resampled during the simulation process, which introduced potential uncertainties. The InVEST model, with its advantages over traditional methods, is widely used for visualization and dynamic studies [34]. During the habitat quality assessment using the InVEST model, parameters for the habitat quality evaluation model were set based on the model manual and literature. The resulting spatial distribution of habitat quality was consistent with ecosystem patterns, indicating good simulation results. However, there remain uncertainties in model performance evaluation and parameter settings, and further exploration of model performance evaluation methods and appropriate parameters is needed. Additionally, despite considering four different simulation scenarios for land use in 2035, there is still some uncertainty in the simulation results. First, although 13 driving factors were selected for the PLUS model simulations, actual land use driving factors are more complex. On the other hand, future development plans for the region were not considered when setting land use simulation scenarios. Future research should place greater emphasis on the complexity of land use driving factors, conduct more detailed investigations, and develop land use simulation scenarios that align with regional development plans to support medium- and long-term growth.

5. Conclusions

This study examines landscape dynamics in Urumqi under various scenarios and their potential impacts on habitat quality. Using the PLUS model, we analyzed multiple driving factors, including natural, social, and proximity influences, to simulate land use evolution. From 2022 to 2035, Markov chain analysis was employed to derive land use transition probabilities and simulate land use changes under different future scenarios. The InVEST model was subsequently used to quantify habitat quality across diverse landscape patterns. The results indicate that the ongoing loss of natural landscapes, such as forests, grasslands, and water bodies, is the primary driver of habitat quality decline in Urumqi. The effects of built-up land changes on habitat quality exhibit both quantitative and spatial variations across different scenarios. Under the BAU scenario from 2022 to 2035, habitat quality is expected to significantly decrease due to extensive grassland degradation. Areas around the Gurbantunggut Desert should focus on ecological protection to mitigate this trend. In contrast, the EP scenario shows a marked improvement in habitat quality, with only 95.45 km² of grassland transitioning to bareland. Overall, while land use changes under all scenarios from 2022 to 2035 will result in varying degrees of habitat quality degradation, the EP scenario effectively minimizes the extent of this decline.