What is it about?
Tropical forests in the Amazon face growing threats from fires, droughts, and deforestation. Using a 20-year fire experiment in southeastern Amazonia, we show that even heavily degraded, grass-invaded forests can recover once fires stop — and without transitioning to savanna. However, recovering forests are dominated by generalist species with lower diversity, and remain vulnerable to new fires, droughts, and windstorms. These findings highlight the urgent need to prevent fires and protect remaining forests.
Featured Image
Photo by David Clode on Unsplash
Why is it important?
We show that Amazon forests subjected to repeated fires, severe droughts, windstorms, and edge effects over 20 years can recover without transitioning to savanna, directly challenging theoretical predictions of irreversible forest collapse. Two key findings are: a) forest recovery is driven by generalist species with acquisitive strategies rather than savanna-affiliated species, even near the Amazon-Cerrado ecotone; and b) recovered communities exhibit more susceptibility to future drought, fire, and windstorm events.
Perspectives
This work would not have been possible without two decades of dedication from an extraordinary field team at IPAM. When I started this research, the dominant narrative was that the Amazon was approaching an irreversible tipping point. Our data tell a more nuanced story: forests are more resilient than we feared, but more vulnerable than we hoped. I find that both reassuring and deeply concerning. Reassuring because there is still time to act. Concerning because the window is closing. If this paper convinces even one policymaker to prioritize fire exclusion in Amazonian forests, it will have been worth every hour spent in the field.
Leandro Maracahipes
Yale University
Read the Original
This page is a summary of: Forest recovery pathways after fire, drought, and windstorms in southeastern Amazonia, Proceedings of the National Academy of Sciences, April 2026, Proceedings of the National Academy of Sciences,
DOI: 10.1073/pnas.2532833123.
You can read the full text:
Contributors
The following have contributed to this page
