1. Introduction
Currently, the global environment is facing a number of threats, such as the atmosphere, oceans, freshwater, and land being subject to enormous pressures exerted by humans. Climate change is by far the greatest environmental threat, as it affects almost all other environmental threats to some extent. It is reported that the current global average temperature has risen by about 1 °C compared to the pre-industrial period [
1], and if greenhouse gas (GHG) emissions are not substantially reduced in the coming decades, the global average temperature increase is expected to exceed 1.5 °C or 2 °C during the 21st century [
2,
3]. Climate warming can lead to frequent disasters such as floods [
4], droughts [
5], and heat waves [
6], and it is a key factor in exacerbating the associated ecological risk. The aim of this paper is to explore the impacts of climate change on urban land use/cover change (LUCC) and landscape ecological risk change in arid zones.
Ecological risk is usually characterized as the negative ecological impacts that ecosystems and their components may suffer or are suffering due to exposure to one or more stressors [
7,
8]. Landscape ecological risk assessment is a branch of ecological risk assessment at the regional scale that assesses potential negative outcomes on ecosystems from multiple sources of risk, such as natural disturbances and human activities, by coupling landscape patterns and ecological processes [
9,
10]. Commonly used evaluation methods include the risk “source-sink” method and the landscape ecological risk index (LERI) method [
11]. The former follows the inherent mode of traditional ecological risk assessment, which is “stressor identification–receptor analysis–exposure and ecological effect characterization”, and with it, it is hard to portray risk dynamics under multiple sources of stress [
12]. The latter takes land use as a risk complex and its changes as risk triggers, which puts more emphasis on the spatio-temporal heterogeneity of risk and enables an integrated characterization of multiple sources of risk [
9,
13,
14]. Therefore, the LERI method is the most widely used method at present. Extensive studies have also confirmed a tight link between LUCC and landscape ecological risk [
15,
16].
Previous studies on LUCC simulation have focused on scenarios of natural development, cultivated land protection, economic development, and eco-priority [
17,
18], mostly considering the orientation of different social policies, and the effects of climate change are not yet explicit. However, LUCC is not only influenced by socio-economic policies; the additional driving role of climate factors cannot be ignored [
19]. The Coupled Model Intercomparison Project Phase 6 (CMIP6), which combines the shared socio-economic pathways (SSPs) and the representative concentration pathways (RCPs), provides researchers with a variety of more plausible scenarios for the future development of society in the context of global climate change [
20,
21,
22]. SSP-RCP scenarios are increasingly being used in studies of future LUCC projections [
23,
24].
The following representative models are mainly used to simulate LUCC. These include the Markov model [
25], system dynamics (SD) model [
26], the artificial neural network (ANN) model [
27], etc., for structural prediction, and the cellular automata (CA) model [
28], the conversion of land use and its effects at small regional extent (CLUE-S) model [
29], the future land use simulation (FLUS) model [
30], the patch-generation land use simulation (PLUS) model [
31], etc., for spatial distribution prediction. Among them, the SD model is a top–down improved model [
20], which puts special focus on the explicit expression and simulation of nonlinear feedback mechanisms when solving complicated issues [
32]. Many scholars consider that LUCC is caused by multiple feedback mechanisms of socio-economic and biophysical factors [
32,
33,
34], and thus, the SD model is an excellent tool to simulate future land use demand. The PLUS model integrates a land expansion analysis strategy (LEAS) and a CA model based on multi-type random patch seeds (CARS) [
31]. It effectively improves the information rule mining and landscape pattern simulation strategies [
35], better reveals the intrinsic relationship of LUCC, more accurately simulates the spatial pattern of land use, and achieves higher simulation accuracy [
31,
36].
A review of previous studies showed that many scholars have assessed the landscape ecological risk based on LUCC using various modeling methods, but little attention has been paid to the distribution of urban land use patterns and the evolutionary characteristics of landscape ecological risk under the influence of climate change. This study can effectively fill the gap in research on the impacts of LUCC on landscape ecological risk under future climate change and is of great significance in helping the relevant administrative departments to adjust socio-economic development policies, optimize land use patterns, and seek an optimal sustainable development model for the region.
Urumqi is one of the farthest cities from the ocean in the world, a typical arid inland area. Climate warming has a particularly prominent impact on arid zones [
37], with potentially devastating effects on local agriculture, water resources, and ecosystems [
38]. As the bridgehead of China’s opening to the West, Urumqi has achieved leapfrog social and economic development, which has also had a non-negligible influence on the eco-environment. In the context of global climate change and accelerated urbanization, understanding how to promote socio-economic development on the basis of efficient resource use and green low-carbon development has become a major challenge for Urumqi. Therefore, we constructed a synthesized framework based on the SD model, the PLUS model, and the LERI to explore the impacts of LUCC on landscape ecological risk under the SSP126, SSP245, and SSP585 scenarios in Urumqi from 2020 to 2060. The specific objectives of this study were (1) to construct an SD model encompassing climate, economic, population, and land subsystems to predict the trajectory of land use demand changes under three SSP-RCP scenarios; (2) to simulate spatial distribution patterns of land use under different SSP-RCP scenarios by integrating with the PLUS model; and (3) to evaluate the landscape ecological risk of Urumqi under different SSP-RCP scenarios by using the LERI and describe its spatio-temporal evolution characteristics.
4. Discussion
4.1. Comprehensive Impacts on Landscape Ecological Risk
SSP126, as a sustainability scenario with low GHG emissions, is closer to a model that implements the concept of ecological priority and green development. This scenario has appropriate carbon dioxide emissions, which can effectively increase leaf area, reduce transpiration rate, improve water utilization in arid regions, and facilitate vegetation growth [
53]. The continuous outward extension of woodland and grassland is conducive to the maintenance of biodiversity and ecological balance, reduces land vulnerability, and ultimately leads to a reduction in the landscape ecological risk of the region. Moreover, forest trees have a strong carbon-sequestration capacity, which can alleviate the pressure brought about by climate change. The growth of vegetation around the desert is also of vital significance to prevent soil erosion and strengthen desertification control. Additionally, this scenario emphasizes the harmonization of urban expansion with the preservation of arable land, and the situation of construction land wantonly encroaching on cultivated land is obviously less than that of the other two scenarios, which to a certain extent mitigates the double pressure on cultivated land brought about by “eating” and “ building”.
SSP245 is a moderate development scenario with medium GHG emissions, analogous to a development pattern that continues the historical development trend of the region [
43]. SSP585 represents a development scenario with high GHG emissions, equivalent to a development pattern that is dominated by rapid economic growth and neglects ecological and environmental protection. In both scenarios, the high demand for urban buildings due to population growth and socio-economic development continues to drive the spread of construction land, and the same type of landscape patches are gradually clustered and become more regular and simple in shape, leading to strengthened landscape connectivity, increased internal stability, and continuously decreasing landscape ecological risk in the city center. However, the blind outward spread of built-up land will encroach on massive farmland and damage the ecological environment of cropland. High concentrations of carbon dioxide will lead to increased soil acidity and decreased land fertility [
54]. The dual influence of climate change and socio-economic development will put enormous pressure on agricultural production. As the population keeps growing, the demand for food will increase accordingly, which does not match the large-scale reduction in farmland and will inevitably exacerbate the conflict between people and land. At the same time, excessive carbon dioxide concentrations will accelerate vegetation growth and shorten the vegetation life cycle [
55], causing the area of forest and grass growth in the SSP585 and SSP245 scenarios to be noticeably smaller than that in the SSP126 scenario. Furthermore, rising temperatures will trigger the Tien Shan glacier to shrink and the vegetation at its end to change [
56]. Glacier melting may cause natural disasters, including floods and glacial mudslides. Glacier runoff will continuously decrease, leading to a gradual scarcity of freshwater resources downstream, as well as diversified and fragmented development of land use types, and will increase the landscape ecological risk.
4.2. Recommendations for the Future Development of Urumqi
Although rapid urbanization can reduce the landscape ecological risk in the built-up area of the city center, the deepening process of industrialization and frequent human activities will change the regional landscape pattern to a certain extent and damage the ecological environment, which will keep the overall landscape ecological risk at a high level. In terms of this research, the SSP126 scenario is the optimal model for future sustainable development in Urumqi. In order to achieve this goal, the local government can try to achieve the following three aspects.
First, the supervision of urban construction land has been strengthened to guide the orderly expansion of cities and towns. Adhering to the principles of government-guided and market-led expansion, we should vigorously revitalize the urban land stock, cut down the occupation of cultivated land for all kinds of structures at the source, and curb the trend of rapidly diminishing arable land in the process of urbanization. Meanwhile, we should exploit reserve resources of cultivated land in accordance with the principle of land setting by water, transform medium- and low-yield farmland, organize idle and abandoned land, improve the integration of land types, and strengthen the ability of cultivated landscapes to resist disturbance.
Second, the protection of ecological land in suburban and mountainous areas is emphasized. It should divide prohibited development zones at water wetlands, strengthen water-quality monitoring and water-pollution prevention, make full use of reclaimed water for wetland replenishment, and make every effort to promote the protection and restoration of water wetlands. We should establish a scientific ecological protection mechanism, improve the order of ecotourism in natural scenic areas, systematically manage and protect wildlife and plants in nature reserves, ameliorate biodiversity, continuously enhance ecological quality and ecological functions, and reduce ecosystem vulnerability.
Third, a large number of emission reduction measures need to be taken in the future, and the promotion of new energy development should be given a more prominent position. It is necessary to vigorously develop wind power and photovoltaic power generation and minimize the use of traditional fossil fuels. We need to focus on promoting the green transformation of development patterns and steadily push forward carbon peaking and carbon neutrality, in order to address the damage that climate change poses to ecosystem security.
4.3. Strengths and Limitations
We used data from different SSP-RCP scenarios in the CMIP6 dataset to establish a comprehensive framework based on the SD model, th PLUS model, and the LERI to simulate future LUCC and predict landscape ecological risks, which fills the gap in research on the impacts of LUCC on landscape ecological risk under future climate change. Related studies showed [
57,
58] that multi-model ensemble average values were more representative of the actual observations than the values from a single model. Therefore, we calculated the average values of temperature and precipitation for 11 GCMs in order to achieve more accurate simulation results. In addition, some scholars used temperature, precipitation, population, and GDP data to set the future simulation parameters of the SD model [
20,
43,
59]. For the sake of more accurately probing the quantitative relationship between climate, economy, population, and LUCC in this study area, after several experiments, we added vegetation coverage and runoff data to the above four types of data, which yielded powerful simulation results.
The accuracy of land use simulation results directly affects the evaluation results of landscape ecological risk. However, as a new model developed in recent years, PLUS is still in the primary stage of comprehension [
36]. The optimal simulation results of this model are identified by repeatedly debugging the parameters, which is somewhat subjective. In future studies, further comparisons with other models should be conducted to increase credibility. The precision of the satellite data products themselves that are input into the model also affects LUCC simulations to some extent, and it is essential to explore higher-quality land-use products in the future [
60]. Although the LERI method is favored by most scholars for its ability to measure both the vulnerability within the ecosystem and the degree of interference from external factors [
61], it is only concerned with static landscape pattern characteristics and lacks the ecological connotation of ecological risk [
51], which still needs to be continuously perfected for the actual situation in the future. In addition, this study only considered the impacts of LUCC on landscape ecological risk under three SSP-RCP scenarios, which is not comprehensive enough. In the future, we will further quantify the impacts of different types of strategies, resource limitations, and extensive management in land use demand forecasting to improve the accuracy of landscape ecological risk assessment.
5. Conclusions
In this study, we coupled the SD model with the PLUS model to simulate the land use demand of Urumqi under three SSP-RCP scenarios from 2020 to 2060, and we predicted landscape ecological risk changes in Urumqi based on the LERI. The study results indicate that the overall LUCC characteristics driven by different scenarios are similar, but the magnitude and rate of change differ in each land use type. The SSP126 scenario has low GHG emissions and suitable temperatures; therefore, the vegetation growth rate is faster, and there is a significant tendency for woodland and grassland to spread to unused land. Under the other two scenarios, especially the SSP585 scenario, rapid urban development pushes the construction land to increase continuously, and the neighboring cultivated land decreases speedily. Excessive GHG concentrations affect the normal growth of vegetation, and, thus, woodlands and grasslands increase to a lesser extent. Meanwhile, the area of water increases as rising temperatures trigger the melting of glaciers. In terms of the landscape ecological risk response, the risk level of urban construction areas and nature reserves with a favorable ecological foundation is relatively low, whereas the risk level of unused land with a poor ecological environment and water regions such as glaciers and lakes susceptible to disturbance by human and climatic factors is quite high. The rapid expansion of built-up land and massive vegetation growth are the dominant factors that cause a decrease in the overall risk level in Urumqi, and the disturbance and destabilization of water and unused land are crucial reasons for exacerbating the risk in localized areas. In summary, the SSP126 scenario has the lowest landscape ecological risk index. In the future, we should be aware of the need to rationalize land use planning, protect ecological land use, continuously promote green and low-carbon development, and maintain the stability of ecosystems in arid regions.