Did 1.79% of Brazilian voters just save the Amazon?

11 minutes ago

9 min read

Last Sunday, 25 October, Luiz Inácio Lula da Silva, Brazil’s incumbent president, pulled off something extraordinary with the help of millions of determined Brazilians. And the outcome might have just temporarily saved the Amazon.

Rewinding to 4 October, Flávio Bolsonaro was winning the presidential elections in Brazil with 47% votes, placing him ahead of Lula by 2 percentage points. For background if you haven’t been following Brazilian politics, Flávio Bolsonaro is Jair Bolsonaro’s son, running for presidency in his place.

Jair Bolsonaro’s 2019–2022 term added 45,600km2 to the Amazon’s total forest loss (INPE PRODES). That’s an average of about 11,400km2 a year, around 60% higher than under Lula’s current presidency — deforestation rose sharply in Bolsonaro’s first three years and stayed high, whereas it has fallen every year under Lula (see Fig. 1).

Fig. 1: Taxas de desmatamento – Amazônia Legal – Estados (annual deforestation rate by state, km²). Source: INPE PRODES, TerraBrasilis.

In 2022, an analysis published by Carbon Brief estimated that Bolsonaro-era policies would mean around 76,000km2 more forest loss by 2030 than full enforcement of Brazil’s Forest Code.

This would have been huge, because the Amazon is a climate tipping point: a system that, once lost, cannot be recovered on any timescale that matters to us.

Is the Amazon close to a tipping point?

The word tipping point is a straightforward and quite explicit one: if you are reading this on a chair, push yourself on the two back legs until you hit the narrow balance where a gentle push would be enough for you to fall backwards. You’ve successfully reached a tipping point, one where the slightest movement can either mean a return to balance or a violent fall. If the Amazon “tipped” over the edge, the largest rainforest of the planet would essentially turn into a savanna.

How would it work? The Amazon makes much of its own rain: trees release moisture that falls again further west. Cut down enough forest and this cycle breaks down. The forest dries, burns more easily, and can turn into a degraded, savanna-like landscape that would take centuries to recover, if it ever does.

The Amazon is close to its tipping point. Scientists have now put numbers on its threshold: with 1.5-2C of global warming — which the world is currently on course to exceed by a massive margin (UNEP Emissions Gap Report 2025) — it would take only 22-28% of the Amazon to be deforested, according to the latest research, published in Nature in 2026. Fire and drought make that threshold easier to reach. On the same measure, we are at roughly 19–20% today — and a second Bolsonaro-type term may have pushed the Amazon to 20–23% by 2030: into a danger zone that starts at 22%, where even touching the edge is too much risk to accept.*

In other words, we are a few percentage points from the edge. And in the eastern Amazon, where clearing is heaviest, the warning signs are already visible: less rain, hotter dry seasons, and, in the southeast, forest that now releases more carbon than it absorbs.

Bolsonaro’s first term cleared an area roughly the size of Denmark. Had he won a second term in 2022, deforestation at the same pace would have added another area about the size of Wales by 2026. Continued to 2030, those policies would have brought the Amazon to the edge of the danger zone within this decade, just as what could become the strongest El Niño on record drives drought and fire across the forest.

How the Amazon may have just been (temporarily) saved from collapse

On October 4th, Flávio Bolsonaro was winning the elections in the first round. And against all odds, Lula won in the second round thanks to the population’s mobilisation. He won with 51.79% of the votes. Had fewer than 2 in every 100 voters switched sides, the result would have flipped: a slim win that might have just given the Amazon a reprieve.

Had Bolsonaro won, deforestation would have increased. Flávio Bolsonaro has said he doesn’t believe climate change is real, and has promised to ‘modernise’ environmental laws and give Amazon landowners more decision-making power, changes environmentalists warn would weaken forest protections (AP). Combined with the super El Niño and changes in climate, loss of Amazon forest could have been enough to trigger tipping.

Protecting the Amazon is vital:

It stores 150–200 billion tonnes of carbon (Science Panel for the Amazon), equivalent to roughly 15–20 years of global fossil-fuel CO₂ emissions (Global Carbon Project, via Carbon Brief)
It holds about 10% of the world’s biodiversity (Flores et al. 2024, via Carbon Brief)
About 47 million people live there (Science Panel for the Amazon)
And it acts as a rain machine for the region, contributing 24-29% of wet-season precipitation in the La Plata basin (parts of Argentina, Bolivia, Brazil, Paraguay and Uruguay).

The forest may still tip, through wildfires and other effects of climate change, or through clearing elsewhere in the Amazon. But it has gained some breathing space, and the world some room for manoeuvre: a luxury it likely wouldn’t have had if the election had gone the other way.

What does this mean for global climate action and for membership organisations

In a nutshell, we have entered a new era compared with 1980–2015 and then 2015–2025. The former was an era where action was urgent but where the world could get away with a year or two of insufficient action, as long as we made up for it later. The 2015–2025 decade was one of ambition, but also heightened urgency as the world started running out of time and clearly attributable climate related damage and deaths began. From 2026 onwards we are in a time of climate reckoning. We are now committed to overshooting 1.5C, we have seen unprecedented catastrophes across all continents (floods, heatwaves, wildfire, storms…) and every misstep can be deadly for millions.

Brazil was close to such a misstep — but the mobilisation of its people, against all odds, saved the day, at least for this time. Brazilians were so determined to vote that abstention fell noticeably between the two rounds. Their reasons went far beyond the Amazon, but in rejecting Bolsonaro’s vision for the country, a majority also rejected his record on the forest.

Under this new overshoot era, every step in climate mitigation must be concrete, complete, and brave. Brazil has shown that it was possible, with enough determination, to keep moving forward. We call on associations to uphold the same level of mobilisation, to safeguard the climate that all industries, professions, and at the end of the day, the people depend on for their lives, livelihoods, and prosperity.

Join a growing collective of committed membership organisations doing incredible work to leverage the power, expertise and techniques within each of their sectors to advance environmental progress and safe living and working conditions for all. Register here.

*Methodological note: how close is the Amazon to its tipping threshold?

Purpose

This article discusses how close the Amazon is to the tipping threshold identified by Wunderling et al. (2026): 22–28% of the Amazon cleared, combined with 1.5–1.9°C of global warming. Satellite estimates put current forest loss at about 15% of the original forest (MAAP 2026). That figure can't be compared directly with the threshold, for three reasons:

the studies use different boundaries for the Amazon
they count different kinds of forest loss
the threshold comes from a model, not from observations

To keep the comparison consistent, we estimated today's level of clearing on the model's own scale. These are our own approximate calculations, not figures reported in the study.

All percentages are calculated against the Amazon's original forest area of approximately 647.6 million hectares (≈ 6.48 million km²), as estimated by MAAP (2022; 2026). This is the forest cover before European colonisation, across the nine Amazon countries, within the RAISG Amazon boundary.

Step 1: the model's starting point

Wunderling et al. (2026) use deforestation scenarios covering the wider Amazon region (Soares-Filho et al., LBA-ECO LC-14). In these scenarios, 18% of the Amazon basin had been cleared by 2020. In the model, a basin-wide transition happens when clearing reaches 22–28% and global warming reaches 1.5–1.9°C.

Step 2: forest lost since 2020

We added observed forest loss since 2020 to the model's starting point. We did this in two ways, because sources disagree on whether forest lost to fire should count as "deforestation".

Estimate A: clearing only (closest to what the model simulates)

Brazil's official clear-cut figures (INPE PRODES, Legal Amazon), for the monitoring years 2021–2025 (August 2020 to July 2025):

Monitoring year	Area cleared (km²)	Source
2025	5,796	Agência Brasil
2024	6,518	INPE PRODES
2023	9,064	INPE PRODES
2022	11,568	Mongabay 2022
2021	13,038	Mongabay 2022
*Total*	*≈ 46,000*

Other Amazon countries: we scaled the total up on the assumption that Brazil accounts for 79.4% of all Amazon forest loss (MAAP 2026). This gives ≈ 58,000 km² for the whole Amazon.
As a share of the Amazon: ≈ 0.9% of the original forest area.
Result: 18% + 0.9 ≈ 19%

Estimate B: including forest lost to fire and other causes

MAAP (2026) reports 13.04 million hectares (≈ 130,000 km²) of primary forest lost in 2022–2025, including fire. That's ≈ 2.0% of the original forest area.
For 2021, we used the clearing-only figure from Estimate A (≈ 0.25%).
Result: 18% + 2.0 + 0.25 ≈ 20%

Result

On the model's scale, current clearing is roughly 19–20%. That puts it about 2–3 percentage points below the lower end of the threshold range.

These figures run to mid-2025 (Estimate A) and end-2025 (Estimate B). Forest loss in 2026 is not yet published. It would add roughly 0.1 points on the clearing-only measure, and more if fire is included.

Scenario: a Bolsonaro-type government, 2027–2030 (hypothetical)

A Bolsonaro-type government would add roughly 0.5–1.4 percentage points by 2030 compared with current rates. On the fire-inclusive measure, it would bring the Amazon to the lower edge of the 22–28% threshold range by the end of the decade.

Assumption: clearing returns to the 2019–2022 average of about 11,400 km² a year (INPE PRODES; Mongabay 2022). That's ≈ 45,600 km² (≈ 4.56 million hectares) in Brazil over four years.

Path to 2030	Added: clearing only (Estimate A)	Added: incl. fire (Estimate B)	Level by 2030 (model's scale)
Bolsonaro-type government	+0.9 points (≈ 57,000 km² for the whole Amazon)	+1.8–2.7 points	*≈ 20–23%*
Current (2025) clearing rates	+0.45 points	+0.9–1.35 points	*≈ 19.5–21.5%*

Estimate B applies a factor of 2–3 to the clearing figure. In 2022–2025, total forest loss (MAAP 2026) was about three times the clearing figure.

Assumptions and limitations

The model's starting point probably overstates, but the model also misses some loss.
- The model's 2020 map (18%) was produced in 2006 by extrapolating 1997–2002 trends, a period of high deforestation (NASA ORNL DAAC). It probably overstates actual clearing at that date; satellite estimates were lower (13.2% based on data to 2021; MAAP #164).
- On the other hand, the model counts only clearing. It does not capture forest degradation (affecting about 40% of the Amazon; AP, Oct 2026), fire or drought-related tree death, all of which reduce the forest's capacity to recycle rainfall.
- These biases run in opposite directions and cannot be precisely quantified, which is why we present ranges rather than single figures.
Different areas. The model covers a wider "PanAmazon" region (NASA ORNL DAAC), while MAAP covers the Amazon biome. The model's total area is not published as a single figure. Testing totals between 5.5 and 7 million km² changes the estimates by less than half a percentage point.
Brazil's share of loss. The scaling for other countries assumes Brazil's historical share of Amazon forest loss (79.4%) still applies. Brazil's share has probably fallen recently as Bolivia's has grown (MAAP 2026), which would make Estimate A slightly low.
Counting fire loss is a judgement call. Estimate B treats forest lost to fire like cleared forest. That's reasonable if fire-damaged forest recycles less moisture, but the model doesn't simulate it directly. The 2–3× fire factor in the scenario is based on 2022–2025, which included the record 2024 fire season.
Pace of change. Warming is already built into the threshold (1.5–1.9°C), so we have not counted it separately. However, warming makes droughts and fires more frequent. The fire seasons after the 2015 and 2023 El Niño events each burned primary forest equivalent to roughly 0.3–0.4% of the Amazon's original forest in a single year (MAAP, Aug 2026). A severe 2027 fire season, following the strong 2026–27 El Niño, could narrow the margin considerably faster than clearing trends alone would suggest.
These are not results of the study. The numbers are approximate and illustrative. Wunderling et al. state only that the threshold "could be transgressed until mid-century".