Response characteristics and threshold identification of desert steppe degradation to multiscalar drought in Yinshanbeilu

Under ongoing climate change, drought-induced grassland degradation has become a major threat to dryland ecosystem stability. However, the nonlinear response of grassland degradation to drought across different temporal scales and the associated ecological thresholds remain insufficiently quantified. Taking the Yinshanbeilu desert steppe of Inner Mongolia as a case study, we integrated long-term remote-sensing and meteorological data from 1985 to 2024 to construct a Grassland Degradation Index (GDI) and multi-scale Standardized Precipitation Indices (SPI). Generalized Additive Models and piecewise regression were used to identify scale-dependent drought responses and critical thresholds. The results showed that six-month cumulative drought was the dominant climatic driver of degradation, explaining 67.4% of the variation in GDI, which was higher than the explanatory power of one-, three-, and twelve-month drought indices. Threshold analysis identified an early-warning threshold of approximately −0.26 to −0.28 and a degradation-acceleration threshold of −0.42 for SPI-6. A reverse GDI threshold of 0.51 indicated a transition from a drought-resistant to a drought-promoting state. Recovery showed strong hysteresis, as highly improved grassland required much wetter conditions, with SPI-6 values ranging from 1.0 to 2.2. These findings demonstrate that the coupled degradation–drought threshold framework can serve as an ecological indicator for detecting early-warning signals, identifying tipping points, and guiding adaptive restoration in arid and semi-arid grasslands.