How resilience signals abrupt declines in vegetation productivity: globally uneven yet predictable

Abstract Background Abrupt declines in vegetation productivity risk irreversible transitions. However, the magnitude of such declines (abruptness)—a key determinant ranging from nonlinear reversible transitions to nonlinear hysteresis transitions—remains poorly quantified. Methods Using two long-term satellite datasets (VODCA and GIMMS NDVI3g), we quantified the abruptness of global vegetation productivity declines and evaluated the early warning performance of resilience indicators, including autocorrelation at the first lag (ACF1) and variance. We further investigated how climate, soil, vegetation, and human activities regulate abruptness across gradients of early warning capacity. Results Hotspots of high abruptness were identified in South America, Africa, and central Eurasia. ACF1 exhibited strong but regionally variable early warning signals, with peak performance in regions of intermediate vegetation productivity (0.88 ≤ VOD < 1.18). In these regions, intensified interactions among vegetation, climate, soil, and human activities acted to amplify abruptness. In contrast, the early warning capacity of ACF1 declined in both low- and high-productivity regions. Conclusions These findings establish ACF1 as a context-dependent early warning of abrupt declines in vegetation productivity, advancing targeted resilience-based management of vulnerable ecosystems under ongoing global change.