Alternative vegetation states on the Loess Plateau and implications for large-scale afforestation success

Large-scale afforestation on the Loess Plateau, costing hundreds of billions of Chinese yuan, has increased vegetation cover but also depleted soil water, raising concerns about the long-term ecosystem sustainability. While debates continue over suitable afforestation areas and precipitation thresholds, the potential role of alternative stable states, a captivating nonlinear dynamical phenomenon, in afforestation success has been largely overlooked. Here, we combined a systematic field survey (4,875 sites, survey mileage of 80,000 km) with a minimal model to explore potential alternative vegetation states, using tree cover as a state variable along a mean annual precipitation gradient. The results showed a clear signature of alternative states of tree cover across the Loess Plateau: Within the 350 to 500 mm mean annual precipitation range, three vegetation states coexist, which are identified as treeless (cover < 5%), open woodland (5 to 50%), and forest (cover > 50%). For areas with mean annual precipitation between 500 and 700 mm, the ecosystem displays bistability consisting of an open woodland and a forest state. Our minimal model revealed that vegetation-precipitation positive feedbacks expanded the range over which alternative vegetation states are permitted and shifted the associated thresholds. Regime shifts between the alternative vegetation states have a strong impact on carbon storage potential, suggesting that afforestation strategies should prioritize bistable and tristable zones where restoration is feasible. These findings provide a framework for afforestation planning and advance the theory of alternative stable states in dryland forests.