Are We About to Lose the Amazon?

Deforestation at Novo Progresso, Pará State, in Brazilian Amazon. Image: Coordenação-Geral de Observação da Terra/INPE via Flickr

Recent drying over the Amazon could be the “first warning signal” that the rainforest is approaching a tipping point, new research says.

Ayesha Tandon | Carbon Brief

The Amazon is the largest rainforest in the world and receives 2-3 meters of rain every year. However, intensifying droughts and human-driven deforestation mean parts of the forest are beginning to dry out.

The study, published in Science Advances, finds that deforestation is delaying the start of the South American monsoon, leading to reduced rainfall over the Amazon.

The authors warn that continued deforestation could push the region past a tipping point in which a further, rapid reduction in rainfall would kill vast swathes of trees.

Over the past 40 years, the Amazon’s dry season has already become longer, the paper finds. This might be the early warning signal that the combined rainforest and South American monsoon systems are approaching a critical threshold, the authors say.

The authors also stress the importance of ongoing experimental work to quantify the impacts of increasing temperature and CO2 on the Amazon rainforest, so that scientists can produce more accurate models of the links between deforestation and rainfall.

Amazon water cycle

The Amazon is the largest rainforest in the world.

The region contains around 400bn trees and is home to at least 10% of the world’s known species. It is also a key carbon store, holding more than 120bn tonnes of carbon in its vegetation and soil.

Tropical rainforests are warm and humid all year round. The Amazon basin receives around 2-3 metres of rainfall every year on average. It “recycles” much of this rainfall back into the atmosphere through evapotranspiration – the movement of water from the land to the atmosphere through a combination of evaporation and transpiration.

Much of this rainfall comes from the South American monsoon, which is driven in part by the temperature difference between the warm Amazon rainforest and cooler Atlantic ocean.

Sun and rain over the Amazon rainforest, at Uatumã Sustainable Development Reserve, located in São Sebastião do Uatumã (AM). Image: André Luís Matos

However, as droughts become more intense and frequent, the humid Amazon climate is beginning to dry, killing trees or making them less resilient to future changes. The ongoing drying trend is exacerbated by deforestation and wildfires.

Around 20% of the Amazon has already been deforested and a further 6% is “highly degraded”.

Dr David Lapola is a research scientist at the State University of Campinas in Brazil and a Carbon Brief contributing editor. Lapola, who was not involved in the study, tells Carbon Brief that the Southern and south-eastern Amazon are “currently experiencing a crisis in terms of changing climate and land use”.

Scientists have warned for decades that human-caused climate change could push key components of the Earth system – such as ice sheets, rainforests and monsoons – past critical thresholds and into new states.

Identifying these “tipping points” is an active area of research. 

Previous studies suggest that the Amazon could be pushed beyond its tipping point if forest loss exceeds 40%. At this level of deforestation, evapotranspiration in the Amazon would reduce significantly, leading more trees to die from lack of water.

This self-perpetuating cycle could see large areas of tropical forest turn into dry grasslands in just decades, in a process called “dieback”.


To investigate the link between Amazon deforestation and rainfall, the study authors produce a model of moisture transport across South America that simulates how air moves through the Amazon. The model includes key feedbacks between vegetation, soil moisture and the atmosphere.

The authors find that deforestation reduces the amount of water released into the atmosphere through evapotranspiration. The drop in atmospheric moisture drives a reduction in rainfall.

To form raindrops, water vapour in the atmosphere condenses into liquid water, releasing energy in the form of heat. The reduction in rainfall means that less energy is released in this way, limiting warming in the atmosphere above the region.

As a result, the temperature difference between the warm Amazon rainforest and cooler Atlantic ocean becomes less pronounced. This can cause delays in the onset of the Amazon’s wet season and a lengthening of the dry season, resulting in drier soils and higher tree mortality.

Overall, this feedback means that deforestation in the Amazon weakens the South American monsoon, further reducing rainfall over the Amazon.

The link between deforestation and rainfall in models with (dashed line) and without (solid line) coupled atmosphere-vegetation dynamics. Source: Spracklen and Coelho (2023).

Prof Dominick Spracklen is a professor of biosphere-atmosphere interactions at the University of Leeds and co-wrote a “focus” article on the new study. 

He tells Carbon Brief that including this complex feedback between the forest and atmosphere makes the rainforest-monsoon system “more sensitive to deforestation, compared to many previous studies that did not include this feedback”.

The graphic above shows the relationship between deforestation and rainfall. The dashed line shows the model used in the study with all feedbacks included, while the solid line shows a model which does not include the atmosphere-vegetation feedbacks.

The study authors find that if deforestation crosses a “critical threshold”, rainfall could drop by 30-50% over just a few years, pushing the system past a tipping point and damaging or killing large areas of the forest.

The model shows that Amazon rainfall is more sensitive to deforestation when key feedbacks between the atmosphere and vegetation are taken into account. This indicates that the tipping point could be crossed sooner than previously thought, the authors warn.

Dr Nils Bochow – a researcher at the Arctic University of Norway and co-author of the study – tells Carbon Brief that “changes in the South American monsoon have a strong influence on the rainforest and vice-versa”. He adds:

“If we do not include these interactions and feedbacks, then we might strongly underestimate the response of the rainforest. This might give a false sense of security or undermine the urgency to act.”

Early warning signs

If an Amazon tipping point is crossed, large sections of lush rainforest could transform into a dry savannah. This process of “savanisation” would take decades to take full effect, but once underway the process is difficult to reverse. The knock-on impacts for the rest of the planet could be profound.

“Tipping points are notoriously hard to understand or predict,” Spracklen tells Carbon Brief. 

However, he says there are often early warning signs when a tipping point is approaching. He likens these to the wobble of a spinning top before it falls over.

After using the models to determine what this “wobble” would look like in the Amazon, the authors analyse decades of ERA5 reanalysis data to search for it.

Change in soil moisture, in kilogram per metre squared per year, over 1979-2019 from ERA5 reanalysis data. Source: Bochow and Boers (2023).

The map below shows the change in soil moisture between 1979 and 2019. Red indicates a drying trend over the four-decade period, while blue indicates wetting.

The authors find that over 1979-2019, soil in the Amazon has become drier. They also find that the dry season now lasts between five and 15 days longer than it used to – meaning that the region is receiving less rainfall, on average, than it was four decades ago.

This indicates that the monsoon-rainforest system has been losing stability in the last decades, the authors say. This might be the first warning signal that a tipping point is approaching, they add.

“The results of this study underline the need to double down on efforts to stop deforestation and help the Amazon region develop in a way that does not lead to additional deforestation,” Spracklen tells Carbon Brief.

Forest loss in the Amazon is beginning to slow. Deforestation in the Brazilian Amazon fell by at least 60% in July 2023 compared to the same month last year, after a new administration led by Luiz Inácio Lula da Silva took power in Brazil.

Leaders of the eight Amazon basin countries met in August 2023 to agree on the need to sustainably develop while preventing further deforestation in the region, and formally recognised that the Amazon is approaching a tipping point.

Model uncertainty

This study is “one of the first” to simulate the feedbacks between the monsoon and Amazon, according to the study authors.

Bochow tells Carbon Brief that including these “non-linear” components is key, because when small changes “reinforce each other” they can lead to significant impacts.

However, the model is unable to account for everything. Most notably, it does not include the impact of rising CO2 levels or temperatures on the forest. This omission is a notable “gap” in the study, according to Lapola.

He tells Carbon Brief that elevated CO2 levels can have a significant impact on the forest through changing evapotranspiration levels, reducing rainfall and inducing plants to use water more efficiently.

Lapola adds:

“[We should] have more experimental studies, in which we manipulate the ecosystem to test its limits in terms of resilience – for example, testing [the effects of] higher temperature and CO2.”

Spracklen tells Carbon Brief that researchers need more observational data, but warns that observations also have their “limitations”. A combination of observations and models are needed to make better predictions, he says.

Bochow points out that climate models show “a big spread in the response of the Amazon rainforest to climate change and deforestation” and agrees that there is an “urgent need to constrain the models better by doing more field experiments and observations”.

He also emphasises that “the exact numbers of our model are not to be taken for granted”. He tells Carbon Brief:

“The model simulations are used as guidance where to look for characteristic changes of stability loss in observations. Our study really focuses on the observed changes in the historical data and these do not depend on our employed model.”


Like What You’re Seeing? Become a patron for as little as $1 per month. Explore ways to support our mission. Sign up for our newsletter (for nothing!). Subscribe to our podcast at iTunes. Share this story with others.