Nonlinear dynamics and threshold responses of Xing’an permafrost to climate and human disturbance in Northeast China: Insights from a multi-risk framework

The ecosystem-dominated Xing’an permafrost of Northeast China has degraded under rapid climate warming and intensifying human activities, threatening ecological security and infrastructure. Yet most assessments emphasize thermal states or geoenvironment−permafrost interactions, offering limited system-level appraisal. Here, we integrate multi-source datasets (19902020) to quantify landscape ecological risk (LER; disturbance and vulnerability components) and an entropy-weighted permafrost vulnerability index (PV), and combine them with equal weights into a comprehensive permafrost degradation risk metric (CPDR). CPDR effectively captures both temporal trajectory and spatial mosaic of degradation risk: Elevated CPDR consistently preceded major permafrost losses, demonstrating an early-warning capability, with high-risk zones concentrated in ecotones and human-disturbed corridors, while permafrost in mountainous forests remained comparatively stable. Thermal drivers, especially mean annual air temperature (MAAT) and mean annual minimum air temperature (MAMinAT), were the dominant accelerators, whereas vegetation cover (normalized difference vegetation index, NDVI) and snow cover (snow cover duration, SCD) generally mitigated risk. Threshold analysis revealed phase-dependent, nonlinear responses of the Xing’an permafrost to environmental forcing, with some critical thresholds (e.g., MAAT ≈ 0°C, NDVI ≈ 0.4 – 0.52, SCD ≈118 days), beyond which degradation risk rises sharply. PV provides the structural susceptibility, while LER modulates and amplifies local disturbances, and their coupling shapes CPDR. These findings highlight the need for integrated, threshold-aware risk management that links early warning to adaptive interventions and aligns permafrost protection with ecosystem-service co-benefits.

中国东北以生态系统为主导的兴安多年冻土，在气候快速变暖与人类活动加剧的双重驱动下发生退化，对区域生态安全与基础设施安全构成威胁。然而现有多数评估多聚焦于热状态或地质环境-多年冻土相互作用，对系统层面的综合评估仍较为不足。本研究整合1990年至2020年的多源数据集，对景观生态风险（landscape ecological risk, LER，包含干扰与脆弱性两个组分）以及熵权法构建的多年冻土脆弱性指数（permafrost vulnerability index, PV）进行量化，并以等权重将二者耦合为综合多年冻土退化风险指数（comprehensive permafrost degradation risk metric, CPDR）。CPDR可有效捕捉退化风险的时间演变轨迹与空间镶嵌格局：高风险值始终超前于大规模多年冻土流失事件，展现出良好的预警能力；高风险区集中分布于生态交错带与人类活动干扰廊道，而山地森林内的多年冻土则相对稳定。气候热因子，尤其是年平均气温（mean annual air temperature, MAAT）与年极端最低气温（mean annual minimum air temperature, MAMinAT），是驱动退化风险加剧的主导因子；而植被覆盖（归一化植被指数，normalized difference vegetation index, NDVI）与积雪覆盖（积雪持续天数，snow cover duration, SCD）则通常会缓解退化风险。阈值分析表明，兴安多年冻土对环境胁迫的响应具有阶段依赖性与非线性特征，存在若干关键阈值（如年平均气温≈0℃、归一化植被指数≈0.4~0.52、积雪持续天数≈118天），超过这些阈值后退化风险会急剧上升。PV体现了多年冻土的内在结构脆弱性，LER则调节并放大局地干扰效应，二者的耦合作用共同塑造了CPDR的空间分布与时间变化。本研究结果表明，亟需构建兼顾阈值效应的综合风险管理框架，将早期预警与适应性干预措施相结合，并使多年冻土保护与生态系统服务的协同效益相统一。