The Amazon’s Carbon Collapse: How the World’s Biggest Rainforest Is Turning From Sink to Source

For a long time, the Amazon rainforest was described as one of Earth’s great climate stabilizers. That was not romantic language. It was a physical truth. The forest stores vast amounts of carbon in trunks, roots, soil, and living biomass, and for decades it helped slow warming by absorbing more carbon dioxide than it released. The Amazon also helps regulate rainfall across South America and holds an amount of carbon equivalent to roughly 15 to 20 years of global CO2 emissions. In other words, this was never just a forest. It was one of the planet’s core life-support systems.

Now that system is starting to fail in places, and sometimes dramatically. The shift did not happen all at once. It came through years of deforestation, fragmentation, deliberate burning, hotter temperatures, and repeated droughts. Scientists have been warning for years that the eastern and southeastern Amazon were under especially severe stress, and that warning has become harder to dismiss. Some regions are no longer reliably acting as carbon sinks. In bad years, and in some damaged areas even beyond those years, they are becoming carbon sources instead.

That change matters far beyond Brazil. A rainforest that absorbs less carbon leaves more of humanity’s emissions in the atmosphere. A rainforest that starts emitting more than it absorbs does something worse: it adds fuel to the crisis it once helped contain. What makes the Amazon story so alarming is that it is not only about trees being cut down. It is about a feedback loop. Deforestation and warming dry the forest. A drier forest burns more easily and grows less efficiently. Burning and weaker growth release more carbon. More carbon helps warm the planet further. And that warmer planet, in turn, makes the forest less resilient.

The atmospheric backdrop to this crisis is historically extreme. Before industrialization, atmospheric carbon dioxide was about 280 parts per million. Ice-core records show that over at least the last 800,000 years, CO2 never rose above roughly 300 ppm. Today, NASA lists the latest Mauna Loa measurement at 429 ppm for February 2026 and notes that atmospheric CO2 has risen by more than 50% since preindustrial times, primarily because of fossil fuel burning. A major 66-million-year synthesis cited by Columbia Climate School concluded that the last time atmospheric CO2 consistently reached today’s levels was about 14 million years ago. That means the Amazon is being asked to function under atmospheric conditions unlike anything modern humans or their societies evolved under.

The most important thing to understand is that the Amazon’s carbon crisis is not a single event. It is a transition. In 2021, a landmark study led by Luciana Gatti found that southeastern Amazonia had become a net carbon source, with total emissions in eastern Amazonia driven largely by fire and the southeast in particular acting as a source even when fire emissions were stripped out. Reuters’ later investigation summarized the finding bluntly: the southeastern Amazon was not only a net producer of carbon, but even the forest alone, apart from fires, had become a carbon source in that region. That was a turning point in the scientific and public understanding of what was happening. The issue was no longer just future risk. Parts of the change were already underway.

The geography of that warning matters. The western Amazon remains healthier in many areas, largely because it is more remote and less heavily fragmented. But the east and southeast have been hit much harder by ranching, soy expansion, logging, road-building, land grabbing, and associated fires. That difference has created what might be called two Amazons: one still trying to function as a carbon sink, and another increasingly damaged, dried, and weakened by human pressure. Scientists cited by Reuters warned that the healthier west is no longer enough to offset the losses and emissions of the more heavily disturbed east.

The role of fire is central, and often misunderstood. The Amazon is a humid tropical forest. It is not supposed to burn the way fire-adapted ecosystems do. When it does burn, it is usually because people have cleared or degraded land, left forest edges exposed, and introduced ignition into a system that evolved without frequent large fires. Reuters’ explainer on Amazon fires notes that the destruction of old-growth trees releases large amounts of carbon dioxide, much of it immediately as the remaining vegetation is burned. Fire is not merely a symptom of Amazon decline. It is one of the engines driving it.

And fire has become more dangerous because drought has become more severe. In early 2024, Reuters reported that climate change was the main driver behind the Amazon’s record drought, which affected all nine Amazon countries, lowered river levels to historic lows in some places, killed endangered river dolphins, and created conditions likely to worsen fire risk. Scientists told Reuters that the drought’s scope was unprecedented across the entire basin. They also warned that when drought combines with deforestation and heat, it can push the Amazon closer to a point where large areas stop functioning as lush rainforest altogether.

That warning turned into a carbon story in 2023. In February 2026, the Max Planck Institute for Biogeochemistry summarized new work showing that the Amazon turned into a weak carbon source during the extreme 2023 drought. Using tower measurements, satellite data, and vegetation models, the researchers concluded that the region released between 10 and 170 million tons of carbon in 2023 instead of absorbing it. Crucially, they found that the anomaly was driven mainly not by unusual fire emissions but by weakened vegetation uptake during extreme heat and atmospheric dryness. The forest simply stopped pulling in as much carbon as it normally would. That is a subtler and, in some ways, more frightening signal, because it suggests the climate machinery of the forest itself is faltering under stress.

That distinction is easy to miss in public debate. When people hear that the Amazon became a carbon source, they often imagine smoke alone. But in 2023, according to the Max Planck team, the bigger story was that trees under extraordinary dryness and heat were less able to photosynthesize and grow. The sink weakened so sharply that the basin crossed over into source territory. Reuters reported something similar in mid-2024 when it covered a broader land-carbon study showing that forests and other land ecosystems failed to curb climate change in 2023. Scientists involved in that work described the land sink as a pump that had started choking. The implication was stark: in hot, dry years, nature may simply stop bailing us out the way we have long assumed it would.

Then came 2024, and the story darkened again. In October 2025, the European Geosciences Union and the European Commission’s Joint Research Centre reported that the Amazon had just endured its most devastating forest fire season in over two decades. The 2024 fires released an estimated 791 million tons of carbon dioxide, roughly equivalent to Germany’s annual emissions, and affected 3.3 million hectares of Amazon forest. For the first time in the 2022–2024 analysis, fire-induced degradation overtook deforestation as the primary driver of carbon emissions in the Amazon. That is a historic shift. It means the forest is no longer being destabilized mainly by outright clearing alone. It is increasingly being hollowed out by degradation, especially fire damage, across areas that may still appear forested from a distance.

Reuters added another layer to that picture in May 2025, when it reported that fires had driven unprecedented global tropical forest loss in 2024. Brazil lost 2.8 million hectares of tropical pristine forest, the most of any country, as the Amazon suffered through what Reuters described as the worst drought ever recorded in the rainforest. For the first time, the annual World Resources Institute and University of Maryland report showed fires as the leading cause of tropical forest loss. That was a grim milestone for a rainforest ecosystem that is naturally too wet to burn extensively under normal conditions.

So when people ask whether the Amazon has “flipped” from sink to source, the honest answer is more complex than a slogan. The entire Amazon basin has not permanently and uniformly switched all at once. But parts of it, especially the southeast, have already crossed that line in scientifically important ways, and basin-wide drought years can temporarily push the whole system into source behavior. What was once an occasional warning sign is becoming a pattern of instability. The sink is weakening. Some regions are emitting. Extreme years are now exposing how vulnerable the system has become.

The tragedy is that much of this is driven by human choices layered on top of climate stress. Deforestation for cattle ranching and cropland has fragmented the forest and created edges that dry faster and burn more easily. National Geographic notes that the Amazon still absorbs vast amounts of carbon, but roughly 30% less than it did in the 1990s, with cattle ranching remaining a major motive for deforestation and cropland such as soy also contributing pressure. In that sense, the Amazon’s carbon crisis is not a natural cycle gone wrong. It is a human-made breakdown in which land use and global warming amplify each other.

The consequences go beyond carbon accounting. The Nature review on critical transitions in the Amazon says the forest is home to more than 40 million people, including about 2.2 million Indigenous people across more than 300 ethnicities. It also stresses that the forest contributes up to 50% of regional rainfall and helps sustain moisture flows that affect large parts of South America. When the Amazon weakens, it is not only biodiversity that suffers. Rivers, farming systems, local economies, hydroelectric stability, health, food security, and Indigenous lifeways all come under pressure.

This is why the phrase “tipping point” keeps returning. Scientists do not all agree on one exact threshold, but the concern is real enough that a 2024 Nature synthesis concluded that by 2050, between 10% and 47% of Amazonian forests could be exposed to compounding disturbances capable of triggering unexpected ecosystem transitions. The same review argued that maintaining the Amazon within a safe operating space requires both local action to end deforestation and degradation and global action to stop greenhouse gas emissions. It is not enough to save the forest from chainsaws if the atmosphere keeps heating it from above. And it is not enough to cut global emissions if local land clearing keeps shredding the forest’s resilience from below.

The most unnerving part of the Amazon story may be how quickly the language has changed. Not long ago, the rainforest was often described in almost passive terms, as if its role as a carbon sink were guaranteed by its sheer size. Now the verbs are harsher: choking, drying, burning, fragmenting, degrading, emitting. Reuters’ 2021 special report captured that shift with unusual force, describing how scientists struggled to accept data showing the southeastern Amazon had become a carbon source even in rainy years. The old assumption was that the forest would always be there, always compensating, always saving some part of the climate ledger. That assumption has been broken.

Source: https://www.nytimes.com/2026/05/06/climate/amazon-rain-forest-deforestation-climate.html

And yet the story is not over. The Amazon is damaged, but it is not gone. Parts of it still function powerfully as carbon sinks. Deforestation can be reduced. Restoration can work. Protected areas and Indigenous territories remain among the strongest defenses against collapse. The 2024 Nature review is explicit that there is still room for action, and it identifies the practical priorities clearly: stop deforestation and degradation, expand restoration, and cut global greenhouse gas emissions fast enough to keep the forest from crossing deeper thresholds of heat and water stress.

That is the real meaning of the Amazon’s carbon collapse. It is not just that a forest is suffering. It is that one of Earth’s great climate buffers is becoming less reliable at the very moment humanity needs it most. When the Amazon burns, the atmosphere changes. When drought weakens its trees, the atmosphere changes. When the forest takes up less carbon than before, the climate math for everyone becomes harder. And when parts of the forest begin emitting more than they absorb, the crisis stops being something the Amazon helps solve and becomes something the Amazon increasingly helps accelerate.

The world still talks about the Amazon as if it were a distant frontier, but the carbon story shows that this is an illusion. The Amazon is inside the global climate system, inside rainfall patterns, inside food security, inside energy systems, and now, increasingly, inside the question of whether natural ecosystems can continue cushioning the damage of fossil-fuel emissions. Its decline is not local news. It is atmospheric news.

So the right way to describe the moment is not hopelessness. It is emergency with evidence. The sink has weakened. The southeast has already flipped. The basin can tip into source behavior in extreme drought years. Fire-induced degradation is rising so fast that in recent analysis it overtook outright deforestation as the main source of Amazon carbon emissions. Those are not abstract warnings anymore. They are measurements.

And that is why the Amazon now stands as one of the clearest climate tests on Earth. If governments cannot protect the world’s largest rainforest from clearing, drying, and burning, then every promise about climate stability starts to look weaker. But if deforestation is cut, degraded areas are restored, and global emissions fall fast enough, the forest may still retain enough resilience to keep doing what it has done for millennia: absorbing carbon, making rain, and holding together one of the richest living systems on the planet.