The Amazon Dieback Threshold

node · earth · Amazon Basin, with the southern and southeastern basin most exposed

The hypothesis that the Amazon has a point beyond which its own rainfall generation fails and large areas convert irreversibly to a degraded, fire-maintained savanna-like state - and the long, unresolved argument about where that point is. The mechanism is not in dispute and follows directly from moisture recycling: less forest means less transpiration, which means less rainfall downwind, which kills more forest. The number is entirely in dispute. Coupled climate models produced Amazon dieback as an output as early as 2000, though later model generations mostly did not reproduce it. Nobre and colleagues proposed in 2016 a boundary around forty per cent deforestation combined with substantial warming; Lovejoy and Nobre revised that sharply downward in 2018 to twenty to twenty-five per cent, arguing that fire and climate change interact with clearance rather than adding to it. Flores and colleagues reframed the question in 2024 in terms of compounding disturbances and proposed safe boundaries well below the earlier figures. Observational work has reported declining resilience across most of the forest since the early 2000s - slower recovery from perturbation, which is the generic statistical signature that precedes a critical transition in many systems. What no one has is an observed threshold, because a threshold can only be observed by crossing it.