关键词: 生态安全格局, 生态安全预警, 多情景模拟, 深度学习, 鄱阳湖流域

Abstract:

In the context of rapid global industrialization and urbanization, the high concentration of industries and populations has profoundly altered regional land use and land cover patterns. The rapid expansion of cities and the intensification of agricultural activities continue to encroach upon ecological spaces, triggering a series of ecological and environmental issues such as land degradation, habitat fragmentation, and the decline of ecosystem services. These changes exacerbate the conflict between socio-economic development and ecological conservation, posing a serious threat to regional ecological security. Against this backdrop, research on ecological security early warning has become a critical pathway for defining the baseline of sustainable development, optimizing watershed management, and supporting scientific decision-making. As the spatial foundation for regional sustainable development, the scientific construction of an ecological security pattern is an important approach to reconciling the contradiction between ecological protection and socio-economic development. Therefore, improving the accuracy of ecological security warning identification, constructing region-specific ecological security patterns, and implementing tailored ecological protection and risk management have become urgent needs for ensuring regional sustainable development. Using the Poyang Lake Basin as a case study, this research, from the perspective of dynamic ecological security warning and pattern optimization, constructed a research framework of “warning assessment-source extraction-element identification-scenario simulation-pattern construction”. First, an ecological security early warning indicator system was built across four dimensions: stress risk, spatial pattern, ecological function and dynamic process response. The coefficient of determination method was used to screen negative samples to establish spatial constraints, and the CNN-LSTM model was combined to achieve a dynamic ecological security early warning assessment. Secondly, the warning results were integrated with Morphological Spatial Pattern Analysis (MSPA) to identify ecological sources, extracting core areas with structural integrity, functional importance, and significant ecological pressure, thereby balancing conservation priority and restoration urgency. Subsequently, based on a resistance surface and circuit theory, ecological corridors, pinch points, and barrier points were identified to construct an ecological network for the Poyang Lake Basin. Finally, the Patch-generating Land Use Simulation (PLUS) model was employed to simulate land use changes in 2030 under four scenarios: inertial development, economic development priority, ecological protection priority, and cropland protection priority. By overlaying these with the spatial distribution of the ecological network elements, priority conservation and restoration areas within the basin were identified, forming a spatial governance and decision-support plan. The results indicate that: 1) the area under the Receiver Operating Characteristic (ROC) curve (AUC) for the CNN-LSTM model was 0.92, indicating that the model possesses reliable predictive capability. The ecological security warning index shows a spatial differentiation pattern of “higher in the south and lower in the north”. High-susceptibility areas are mainly distributed in hilly and mountainous regions and their interfaces with rivers and lakes. Specifically, higher and highest susceptibility areas are concentrated around the Jiuling Mountains in the northwest, the Jitai Basin in the central part, and the Poyang Lake area in the north. 2) A total of 16 ecological sources were identified, covering a total area of 26405 km². These sources are mainly located in the Xiushui River basin, the Jitai Basin, the Jiuling Mountains in southern Jiangxi, and mountain ranges such as the Luoxiao Mountains in western Jiangxi. They possess structurally intact natural ecosystems and are key patches for maintaining regional ecological security and carrying important ecosystem services. Eleven key ecological corridors were identified, with a total length of 1132.21 km, and 18 important ecological corridors, with a total length of 860.73 km. The key ecological corridors are spatially widespread and extend over longer distances, generally running northwest-southeast across the basin. They primarily connect the mountain ranges in the northwest, northeast, and east, passing through the Xiushui, Fuhe, and Xinjiang River basins. These corridors effectively link distant ecological sources but may also increase risks and uncertainties during species migration and energy flow. The important ecological corridors are relatively concentrated and shorter in length, running north-south through the southern part of the basin and east-west through the northern part. They mainly connect the western and southern mountain ranges, passing through the Ganjiang and Raohe River basins. This layout facilitates rapid migration and exchange of species over relatively short distances but also subjects adjacent ecological sources to higher pressure from material and energy flow. A total of 149 ecological pinch points were extracted, located in areas with lower resistance that are easier for species to traverse, making them key areas for conservation to maintain ecological connectivity. In addition, 968 ecological barrier points were identified, mostly distributed at road intersections and around construction land, requiring targeted ecological restoration and landscape pattern optimization. 3) Based on the PLUS model, land use in 2030 was simulated for four different development scenarios in the Poyang Lake Basin: inertial development, economic development priority, ecological protection priority, and cropland protection priority. Combining this with the distribution of ecological network elements, 97 priority conservation areas and 457 priority restoration areas were identified. Ultimately, a security pattern of “Five Zones, Three Belts, Multiple Points” was constructed. Here, “Multiple Points” refers to the ecologically critical nodes within the basin requiring priority management, including the 97 priority conservation areas and 457 priority restoration areas. Priority conservation areas are mainly distributed at the intersections of ecological sources and corridors in counties such as Suichuan, Yongxin, and Nancheng. Priority restoration areas are concentrated in the mid-sections of corridors within cities like Dexing, Yudu, and Fenyi. The “Five Zones” includes the Northwest Jiangxi Large Ecological Corridor Connectivity Zone, the Northeast Jiangxi Abandoned Mine Ecological Restoration Zone, the West Jiangxi Industrial Green Transformation and Upgrading Zone, the Central-East Jiangxi Agricultural Land Ecological Quality Improvement Zone, and the South Jiangxi Ecological Tourism Economic Development Zone. The “Three Belts” includes the east-west Poyang Lake Ecological Corridor Construction Belt, the northwest-southeast Xiushui-Fuhe River Basin Ecological Corridor Construction Belt, and the north-south Large Contiguous Ecological Corridor Construction Belt. The findings can provide ecological risk warnings and management scenario predictions for territorial spatial planning in rapidly changing environments, supporting the achievement of ecological security and sustainable development goals in the basin.

Key words: ecological security pattern, ecological security early warning, multi-scenario simulation, deep learning, Poyang Lake Basin