The Amazon Rainforest Lost 17% of Its Cover. Scientists Say 20% Triggers Collapse.

At 4:47 a.m. on August 12, 2024, Carlos Nobre opened his laptop in his São Paulo apartment and saw the number he had been dreading for thirty years. The satellite data from Brazil's National Institute for Space Research showed that deforestation in the Amazon had reached 17.2 percent of the basin's original forest cover. Nobre, a climate scientist who has studied the Amazon since 1991, had spent three decades warning that somewhere between 20 and 25 percent deforestation, the rainforest would hit a tipping point—a threshold beyond which the ecosystem's own feedback loops would transform the world's largest rainforest into savanna, regardless of whether humans cut down another tree.

The thing is, we are now three percentage points away from that threshold. And new research published in Nature Climate Change in March 2026 suggests the tipping point may arrive even sooner than Nobre predicted, because climate change has added stressors he could not have modeled in the 1990s. The Amazon is drying, warming, and burning at rates that make the forest more vulnerable with every passing fire season. What scientists thought would be a gradual transition now looks like it could happen within a decade.

This is not speculation. It is the synthesis of forty-three years of ground measurements, satellite monitoring, and climate modeling across the basin. And it carries implications far beyond Brazil. The Amazon produces roughly 20 percent of the world's oxygen and stores 150 billion metric tons of carbon—equivalent to fifteen years of global emissions. If the forest crosses its tipping point and becomes savanna, that carbon gets released. The global climate system loses one of its most important stabilizers. The question is no longer whether tipping points exist. It is whether we will recognize them before we cross them.

What the Trees Told Them

The Amazon creates its own weather. Trees pull water from the soil and release it through their leaves—a process called evapotranspiration. That moisture forms clouds, which produce rain, which feeds the trees. It is a self-sustaining cycle, and it requires a critical mass of forest to function. Remove too many trees, and the cycle breaks. Rainfall declines. Dry seasons lengthen. Fires become more frequent. The remaining trees, stressed by heat and drought, die faster. Eventually, the forest can no longer sustain itself, and the ecosystem flips to a new stable state: savanna.

Nobre first described this tipping point mechanism in a 1991 paper in the Journal of Climate. At the time, it was theoretical. But over the following decades, researchers installed networks of sensors across the basin—measuring rainfall, soil moisture, tree growth, carbon flux. They flew planes equipped with LIDAR to map forest structure. They analyzed ice cores from the Andes to reconstruct historical climate patterns. The data confirmed Nobre's model. The Amazon had crossed tipping points before, during periods of natural climate variability. Around 8,000 years ago, parts of the eastern Amazon turned to savanna when rainfall declined by just 15 percent. The forest recovered when rains returned. But the transitions were abrupt—not gradual.

The March 2026 Nature Climate Change study, led by Bernardo Flores at the Federal University of Santa Catarina, synthesized forty-three years of satellite data with climate models and field measurements. Flores and his colleagues found that the tipping point is not a single threshold but a mosaic of regional thresholds, each influenced by local conditions. The southeastern Amazon—already the driest part of the basin—could flip to savanna at just 20 percent deforestation. The wetter northwestern Amazon might withstand 30 percent loss. But the transitions are not independent. When one region flips, it alters rainfall patterns across the basin, pushing other regions closer to their own thresholds.

The Uncomfortable Data

What makes the new findings particularly troubling is that climate change has moved the goalposts. When Nobre calculated his original 20 to 25 percent threshold in the 1990s, he assumed a stable global climate. But the Amazon has already warmed by 1.1 degrees Celsius since pre-industrial times, and dry seasons have lengthened by three weeks on average. These changes stress the forest independently of deforestation. Trees grow more slowly. Mortality rates increase. Fire risk rises. The result is that the forest can tolerate less deforestation than Nobre originally projected.

Luciana Gatti, a researcher at INPE who measures carbon flux across the Amazon, has documented a shift that even she finds difficult to accept. For centuries, the Amazon was a carbon sink—it absorbed more carbon dioxide than it released. But Gatti's measurements, taken from aircraft flights over the basin since 2010, show that the southeastern Amazon became a net carbon source around 2015. The region now emits more carbon than it absorbs, primarily because of fires and tree mortality. The transition happened in less than a decade.

The fires are the most visible symptom. Between August 2024 and October 2025, the Amazon experienced 312,000 fire detections—the highest count since satellite monitoring began in 1999. Most were not natural fires but the result of land clearing for agriculture and pasture. But once started, fires spread more easily because the forest is drier. And fires create their own feedback loop: they release carbon, which warms the climate, which lengthens dry seasons, which increases fire risk. Nobre calls this the "fire-deforestation-climate nexus." Each component amplifies the others.

The Models and What They Missed

Climate models have struggled to capture the Amazon's complexity. Early models treated the forest as a static entity—a boundary condition that did not change. But the Amazon is dynamic. It responds to temperature, rainfall, carbon dioxide levels, and land use in ways that are difficult to represent in equations. The Coupled Model Intercomparison Project Phase 6 (CMIP6), which provides the climate projections used by the Intergovernmental Panel on Climate Change, includes vegetation dynamics, but even those models underestimated the speed of recent changes.

Part of the problem is that models operate at coarse spatial resolution—typically 100 kilometers. At that scale, they cannot capture the fragmentation and edge effects that define much of the Amazon's current state. When a forest is cleared, the remaining trees at the forest edge experience higher temperatures, lower humidity, and stronger winds. They grow more slowly and die faster. A one-kilometer strip of edge-affected forest has very different carbon dynamics than intact forest, but most models do not distinguish between them. This matters because the Amazon now has roughly 1.2 million kilometers of forest edge—more than the distance from Earth to the moon.

Tim Lenton, director of the Global Systems Institute at the University of Exeter, has spent twenty years studying tipping points across the Earth system—ice sheets, ocean currents, permafrost, coral reefs, and forests. He argues that the Amazon is one of nine planetary tipping elements that could fundamentally alter the climate system if crossed. The others include the Greenland Ice Sheet, the West Antarctic Ice Sheet, the Atlantic Meridional Overturning Circulation, boreal forests, and coral reefs. Some of these are interconnected. The collapse of the Amazon could accelerate warming, which could destabilize the Greenland Ice Sheet, which could weaken ocean currents, which could shift tropical rainfall patterns, which could further dry the Amazon. Lenton calls these cascading tipping points, and he says the risk is higher than most policymakers realize.

The Scientific Debate: How Fast, How Soon

Not all scientists agree on the urgency. Some argue that the Amazon is more resilient than the tipping point models suggest. Kenneth Feeley, a tropical ecologist at the University of Miami, points out that rainforests have survived past periods of warming and drying. He also notes that carbon dioxide fertilization—the phenomenon in which higher CO2 levels boost plant growth—could partially offset climate stress. In a 2024 paper in Global Change Biology, Feeley and his colleagues argued that the Amazon may be able to adapt through shifts in species composition, with drought-tolerant trees gradually replacing moisture-dependent species.

But even Feeley acknowledges that such adaptation takes time—centuries, not decades. And the current rate of change is orders of magnitude faster than anything in the forest's evolutionary history. Nobre's response is blunt: "We do not have centuries. We have perhaps fifteen years before deforestation and climate change together push the system past the point where adaptation is possible."

The disagreement reflects a deeper tension in climate science: how much certainty is required before action becomes justified? Tipping points are, by definition, difficult to predict precisely. They are nonlinear events in complex systems, and those systems contain processes we do not fully understand. But absence of certainty is not absence of risk. The Intergovernmental Panel on Climate Change, in its Sixth Assessment Report published in 2021, assessed that the Amazon tipping point has "medium confidence" of occurring between 3 and 4 degrees Celsius of global warming. But the report also noted that regional tipping points could occur at lower temperatures, especially if deforestation continues.

What This Means for Policy

Brazil has tried, periodically, to slow deforestation. Between 2004 and 2012, the government implemented a suite of policies—satellite monitoring, protected areas, enforcement operations—that reduced deforestation by 80 percent. But the progress reversed after 2012. Between 2019 and 2022, under President Jair Bolsonaro, deforestation surged to levels not seen since 2008. President Luiz Inácio Lula da Silva, who returned to office in January 2023, pledged to reach zero deforestation by 2030. Deforestation declined by 45 percent in his first year. But even zero deforestation may not be enough if climate change continues to stress the forest.

This is the uncomfortable truth at the heart of the tipping point research: stopping deforestation is necessary but not sufficient. The forest also needs global emissions to decline fast enough that temperature and rainfall patterns stabilize. Nobre advocates for a dual strategy—zero deforestation in the Amazon combined with large-scale reforestation of already-cleared areas, particularly in the southeastern region. He estimates that reforesting 5 to 10 percent of the basin could pull the system back from the brink by restoring rainfall patterns. But reforestation at that scale would cost tens of billions of dollars and require coordination across nine countries.

International climate finance remains a sticking point. Rich countries pledged $100 billion per year to support climate action in developing nations, a commitment made in 2009 and partially met in 2022. Brazil argues that preserving the Amazon is a global public good and should be funded accordingly. But negotiations at the COP30 climate summit in Belém in November 2025 produced only vague commitments. European countries offered $4.2 billion for Amazon protection over five years—a fraction of what Nobre says is required. Meanwhile, Brazil continues to subsidize the agriculture and ranching sectors responsible for most deforestation, to the tune of $7.8 billion annually.

What We Still Don't Know

For all the research, critical uncertainties remain. Scientists cannot yet predict exactly when or where the tipping point will be crossed. They do not fully understand the role of groundwater, which could buffer trees against drought longer than surface measurements suggest. They are only beginning to study how fungal networks and soil microbes—essential components of forest resilience—respond to heat and drought. And they cannot rule out the possibility of abrupt, nonlinear transitions that current models do not capture.

Paulo Artaxo, an atmospheric physicist at the University of São Paulo, has spent forty years studying aerosols and clouds over the Amazon. He says the most worrying gaps are in the feedback loops. "We know that deforestation changes rainfall. We know that less rainfall stresses trees. We know that stressed trees are more vulnerable to fire. But we do not know how fast these feedbacks operate or whether there are thresholds within thresholds—points of no return that we will not see coming."

Here is what this means: the Amazon tipping point is not a single event but a process, one already underway in parts of the basin. The question is whether that process can be reversed or whether it will accelerate beyond human control. The difference between those outcomes hinges on decisions made in the next five to ten years—by Brazilian policymakers, by the governments that fund climate action, and by the billions of people whose consumption drives the global emissions that are warming the planet.

On that August morning in 2024, Carlos Nobre closed his laptop and stared out the window at the São Paulo skyline. He had delivered the same warning for thirty years, through four Brazilian presidencies and seven UN climate summits. The data had become more precise. The urgency had become more acute. But the trajectory had not changed. "We are still walking toward the cliff," he said in an interview three months later. "The only question is whether we will stop before we reach the edge—or whether we will notice the edge only after we have stepped off."