In 2024, the Brazilian Amazon faced one of the worst fire seasons in its history: 15.6 million hectares (38.5 million acres) were burned, an area the size of Portugal, of which 43% consisted of forest vegetation. According to data from MapBiomas, a collaborative network mapping land use in Brazil, the area destroyed by flames was 117% higher than the historical average.
Now, a group of scientists has found a surprising capacity for forest regeneration after fires, although biodiversity loss persisted. The results were presented in a two-decade study in the southeastern Amazon that attempts to predict the risks the world’s greatest rainforest faces as deforestation, reduced precipitation and long-term droughts become more frequent.
“Our study brings a message of hope,” said lead author Leandro Maracahipes, a researcher at the Amazon Environmental Research Institute (IPAM) in Brazil and at Yale University in the U.S. “It demonstrates that a highly degraded forest can recover even after many disturbances.”
Maracahipes emphasized, however, that this regeneration gives rise to a new type of forest. “It is a new ecosystem. Although it can recover, this forest is much poorer, with a greater number of generalist species, much more vulnerable to new disturbances,” he told Mongabay.
The experiment began in 2004, during a peak of deforestation in the Amazon, and was published in April 2026 in the journal Proceedings of the National Academy of Sciences. The study focused on the southern part of the biome, an area in the Brazilian state of Mato Grosso, where climate models indicated greater drought impacts. The region is home to the Tanguro Research Station, one of the main field laboratories used by Maracahipes and team, located near cattle ranches.
For their study, the researchers divided a 150-hectare (371-acre) forest area into three equal parts: one plot that had burned annually, another every three years, and the third an unburned control site.
According to Maracahipes, although the experiment demonstrated rapid ecological recovery, analysis also detected a change in the diversity of species making up this new forest. This alteration was most intense at the edges of the forest; here, the study mapped a loss of 31.3% to 50.8% in plant diversity.
In these edge zones, more generalist, fast-growing and less fire-resistant trees, such as Mabea fistulifera and Tachigali vulgaris, became more abundant. There was also a reduction in forest-specialist vegetation, including Ocotea guianensis, Ocotea matogrossensis and Micropholis egensis.
“Forest specialists [trees] are generally species with high wood density, which are more resistant to fire and strong winds, and long-lived, that is, they store carbon for long periods,” Maracahipes said, adding that these species are also key to feeding local fauna.
In recent decades, many scientists have warned about the Amazon’s risk of savannization — the process in which the typical dense rainforest landscape transforms into a sparse, lower-vegetation ecosystem, similar to the neighboring Cerrado savanna biome. Among the experts supporting this thesis is Brazilian climatologist Carlos Nobre, a senior researcher at the Institute of Advanced Studies at the University of São Paulo (USP) and co-chair of the Science Panel for the Amazon (SPA).
“The risk of savannization is the synergistic interaction of deforestation and global warming, especially the large increase in the frequency of severe droughts,” Nobre, who wasn’t involved in the study, told Mongabay. “Before, we observed one severe drought every two decades. Now we have had four very strong droughts in two decades: 2005, 2010, 2015-16, 2023-24.”
Nobre added that the Amazon climate could become characteristic of that of a tropical savanna, where the dry season lasts for six months. “Throughout the southern Amazon, the dry season has already increased by four to five weeks in 45 years,” he said.
One of the objectives of the study was to find out whether vegetation savannization was occurring. In the end, however, researchers didn’t find evidence of it; the anticipated greater presence of grasses, characteristic of savannas, just wasn’t there, Maracahipes said.
“From 2019 to 2020 onward, we observed an increase in tree cover and high competition with grasses, which began to be eliminated. Grasses do not withstand a closed canopy. When the forest canopy closes, they are excluded from the system,” he said.
The authors found “little evidence that fire-degraded forests transition toward persistent savanna, although recurrent fires or future climatic changes could drive long-lasting degradation or human-derived savannas.”
Study co-author Paulo Brando, a forest engineer and associate professor at Yale who has studied the impacts of drought and fragmentation in the Amazon for more than 20 years, said the experiment’s results pointed to a glass half-full outlook.
“We were able to show that ecological resilience is extremely high in these systems, if we give the forest a chance,” he told Mongabay. “Even in this degraded area, even in this system that burned, under the effects of climate change and exposed to agricultural areas, we still have a version of forest returning — far short of a primary forest — but we have it.”
Another recently published similar study also pointed to the quick regrowth and regeneration of lowland rainforest in Ecuador, while showing that full ecological recovery — especially the return of the original species composition — might take much longer. In this case, the analysis looked at the diversity of not just plant species, but also animals and microbes.
According to Brando, the path to restoring degraded areas of the Amazon is to shield it from pressure. “If we create a strategy for fire prevention and management, even degraded areas can return to being a new version of a forest,” he said. “So it is very important that we understand the resilience of the forest to create public policies.”
Brando said recovering a degraded forest produces a much faster effect than starting a restoration project from scratch. “If the right elements are there, such as the presence of animals or seed dispersal, for example, it will be more resilient than a planted forest. And currently we see a whole incentive precisely for planting forests instead of creating a more comprehensive policy that allows degraded ones to return.”
