An Amazon rainforest was put through a 20-year drought experiment — the forest changed in a way scientists didn’t expect

For more than two decades, scientists have been deliberately starving a small section of the Amazon rainforest of rain to understand how the world's largest tropical forest might respond to a much drier future.

The experiment, hidden inside Brazil's Caxiuanã National Forest, has produced a result that is both alarming and unexpected.

Researchers diverted roughly half of the rainfall away from a one-hectare patch of rainforest. At first, the trees appeared remarkably capable of coping. But after several years, the forest began losing some of its largest trees, shedding a substantial share of its biomass and temporarily changing from a carbon store into a carbon source.


Then something unexpected happened.

Instead of continuing toward complete collapse, the remaining forest eventually settled into a new ecological balance.

The catch? It was a much smaller and less carbon-rich forest than the one that existed before the experiment began.

Scientists created an artificial drought inside the Amazon

The long-running experiment, known as the Caxiuanã throughfall exclusion experiment, began in 2002.

Researchers installed thousands of transparent panels roughly 1 to 2 metres above the forest floor. Together with a system of gutters and drainage trenches, the structure prevented about 50% of rainfall from reaching the soil.

The experimental area covers one hectare of seasonal rainforest in Pará state, while a nearby control plot continued to receive normal rainfall.

The setup was designed to mimic one possible consequence of climate change: a forest facing prolonged reductions in available water.

It became the longest-running large-scale experimental drought study in a tropical forest.

The trees initially appeared to withstand the drought

For several years, the forest showed signs of resilience.

Trees altered how they used water, with some relying on deeper water reserves. Previous research at the site also found evidence that certain trees could absorb moisture through their leaves under suitable conditions.

But that resilience had limits.

About eight years after the drought treatment began, mortality increased sharply, particularly among the forest's largest trees. The damage became increasingly visible as the canopy opened and fallen trunks accumulated on the forest floor.

The biggest trees turned out to be especially vulnerable to hydraulic failure — a breakdown in the plant's ability to transport water through its internal tissues.

Losing the biggest trees changed the entire forest

The disappearance of large trees had consequences far beyond the trees themselves.

By the end of the first 15 years of experimental drought, the plot had lost about 34% of its above-ground biomass compared with its starting condition.

The loss represented a major amount of stored carbon.

During this period, the drought-stressed forest also shifted from functioning as a carbon sink to becoming a carbon source, as dead and decomposing vegetation released carbon that had previously been stored in the ecosystem.

The opening of the canopy also allowed more sunlight to reach the ground, potentially making the forest floor warmer and drier and increasing the risk of fire.

Then the Amazon forest did something scientists did not expect

The researchers had reason to worry that the continuing drought could push the forest into a much more dramatic ecological transition, potentially resembling a savanna.

That did not happen.

Instead, after years of heavy tree mortality, biomass began to stabilise.

Between 2017 and 2023, researchers found that the remaining trees were no longer experiencing the same level of water stress that had characterised the earlier stages of the experiment.

The reason was surprisingly simple: there were fewer trees competing for the limited water available in the soil.

With many of the largest trees gone, more water became available to those that survived.

The forest had effectively adjusted to its new, drier conditions — but only after paying a heavy ecological price.

The forest survived, but it was not the same forest

The stabilisation should not be mistaken for a full recovery.

Researchers describe the later stage as eco-hydrological stability. The surviving vegetation was able to maintain water transport and avoid the severe drought stress seen earlier in the experiment.

But the forest had lost a substantial amount of biomass and carbon.

The canopy was more open, large trees were less abundant and the amount of carbon stored in wood remained lower than in the nearby control forest.

In other words, the forest found a way to survive the drought by becoming less dense.

That distinction could prove important when scientists assess how tropical forests may respond to a changing climate.

What does this mean for the Amazon?

The experiment offers a more complicated picture than either complete resilience or total collapse.

It suggests that a rainforest may be capable of adjusting to prolonged water shortages, but that adaptation can come through the loss of trees and stored carbon.

That matters because the Amazon plays a major role in the global carbon cycle. A forest that contains fewer large trees and stores less carbon is not equivalent to the original ecosystem, even if the surviving vegetation eventually becomes physiologically stable.

The findings also suggest that computer models predicting a simple transformation from rainforest to savanna may not capture every way a forest can respond to long-term drought.

Scientists are now watching what happens after the drought

The artificial drought experiment has now entered another important phase.

Most of the rain-exclusion panels were removed in November 2024, ending more than two decades of experimental rainfall reduction.

Researchers are now monitoring the plot to see whether the forest can rebuild the biomass it lost and whether its original structure can return.

That could be just as revealing as the drought experiment itself.

If the forest gradually regains its former structure, it would suggest that at least some of the damage can be reversed when water becomes available again.

If recovery remains limited, it would indicate that decades of drought can leave a much longer-lasting ecological imprint.

A warning hidden inside an unexpected recovery

The most important lesson from the experiment may be that survival does not necessarily mean recovery.

The Amazon forest demonstrated an ability to reorganise itself after years of severe water stress. But it did so after losing many of its largest trees and a significant portion of its biomass.

The experiment represents only a one-hectare patch of forest and does not reproduce every factor affecting the Amazon under climate change, such as rising temperatures and increasing atmospheric dryness.

Even so, more than 20 years of observations provide an unusually detailed look at what prolonged drought can do to a tropical rainforest.

The forest did not simply die.

It changed — and the question scientists are now trying to answer is whether it can ever become what it was before.