Redacción HC
17/01/2024
The Amazon rainforest, often dubbed “the lungs of the planet,” plays a vital role in stabilizing Earth’s climate. But new research warns that this global climate regulator may be approaching a catastrophic turning point—one where the forest could irreversibly transition into a fire-prone savanna. The study, published in the Proceedings of the National Academy of Sciences (PNAS), uses an advanced Earth System Model to project Amazonian biomass trends under different climate mitigation scenarios. The findings are stark: under low mitigation efforts, up to 40% of the forest could be lost by 2100.
For decades, scientists have debated the resilience of the Amazon. Field-based ecological models have predicted a "tipping point," beyond which feedback loops—like increased fires and grass dominance—would lock the system into a degraded state. In contrast, many climate models have painted a more optimistic picture, suggesting that rising CO₂ levels might enhance forest growth through increased photosynthesis.
This new study bridges the gap between these diverging projections by simulating the interactive effects of plant competition, fire feedbacks, and CO₂ fertilization under both high and low emissions scenarios. The central question: will the Amazon remain a carbon sink, or will it turn into a carbon source?
The research uses GFDL-ESM4.1, one of the most sophisticated Earth System Models available, to simulate forest dynamics through the 21st century. Unlike simpler models, this one accounts for:
Simulations span both high and low emissions pathways, incorporating real-world phenomena such as El Niño (ENSO) and the Atlantic Multidecadal Oscillation (AMO), both of which are known to intensify droughts and fires in the Amazon.
Under stringent mitigation (low emissions), the model projects a continued increase in tropical biomass through 2100. CO₂ fertilization boosts productivity, and fire disturbances remain relatively contained. In this scenario, the Amazon continues to act as a robust carbon sink, buying us precious time in the fight against climate change.
The narrative shifts dramatically under high emissions. After 2060, carbon gains are reversed as fire incidence accelerates and forest recovery slows. By 2100:
This feedback loop—fire promotes grass, which promotes more fire—is described as a "nonlinear ecological trap" from which forests may not recover.
This transformation is not just an ecological issue. The implications for biodiversity, Indigenous communities, and global climate stability are profound.
As the authors emphasize, “The cost of inaction is not abstract—it is a planetary risk.”
Unlike previous projections, this model combines real-world ecological thresholds with global climate feedbacks. It reveals that forest resilience is not guaranteed, especially when recovery mechanisms (like canopy regrowth) are overwhelmed by fire and drought. It also contrasts with more optimistic Earth System Models, such as EC-Earth3-Veg, which project faster post-fire recovery but less abrupt biomass loss.
These contrasting outcomes highlight a crucial point: how we model ecological recovery matters. Improving post-fire recovery representation in models is a pressing research frontier.
The call to action is clear: drastically reduce carbon emissions now. Forest conservation policies must be paired with:
The forest’s future will be decided in this generation. If high-emission trends continue, the Amazon could tip within decades, not centuries.
The Amazon is closer than ever to a dangerous tipping point. But the future is not fixed. Ambitious mitigation can preserve this irreplaceable ecosystem and its climate-regulating functions. The science is clear, the risks are real, and the window for meaningful action is closing.
We can still choose forest over fire—if we act now.
Referencia: Martínez Cano I, Shevliakova E, Malyshev S, John JG, Yu Y, Smith B, Pacala SW. Abrupt loss and uncertain recovery from fires of Amazon forests under low climate mitigation scenarios. Proc Natl Acad Sci U S A. 2022;119(52):e2203200119. Available on: https://doi.org/10.1073/pnas.2203200119
![]()