Beyond burned area: multidimensional fire monitoring reveals a staged transition in the Xingu region, Brazilian Amazon, 1985–2025
This study reveals that fire dynamics in the Xingu region of the Brazilian Amazon underwent a staged, multidimensional transition between 1985 and 2025, where the total burned area, fire size structure, and seasonal timing shifted at distinct times driven by different factors, demonstrating that relying solely on annual burned area masks critical reorganizations in fire regimes.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the vast, humid forests of the Amazon, fire is a complex force that behaves differently depending on where it starts and what it burns. For decades, scientists and policymakers have tracked the health of these forests by measuring a single number: the total area that burns in a year. This metric, while useful for counting acres, acts like a blunt instrument. It tells us how much land was touched by flames, but it hides the story of how those fires behave. A landscape can burn more because the forest has been replaced by dry grass that burns easily every year, or it can burn more because the remaining trees are drying out and catching fire more often. It can also happen that fires are getting larger, or starting earlier in the year, without the total area changing at all. Understanding these distinct patterns is crucial because each requires a different response. If the problem is simply that there is more flammable grass, the solution involves land management. If the problem is that the forest itself is becoming more prone to burning, the solution requires protecting the trees from degradation. Without separating these factors, we risk applying the wrong tools to the wrong problems.
A team of researchers set out to look beyond the simple total burned area in the Xingu Integration Region of the Brazilian Amazon, an area roughly the size of Portugal. They gathered forty-one years of high-resolution satellite data, from 1985 to 2025, to watch how fire changed in this specific corner of the forest. Instead of just adding up the acres, they broke the data down into several different dimensions: how much burned, how big the individual fire scars were, when during the year the burning happened, and what types of land were affected. By using advanced statistical methods to detect sudden shifts in these patterns, they discovered that the fire regime did not change all at once. Instead, it underwent a staged transition, where different aspects of the fire system reorganized at different times.
The first major shift occurred in the mid-1990s, when the total amount of land burning began to rise sharply. However, the nature of the fires themselves did not change immediately. It was not until roughly a decade later, in the late 2000s, that the size of the fires began to transform. Before this second shift, most fires were small, scattered events. Afterward, the landscape began to see a dominance of massive fire scars, with some exceeding ten thousand hectares. The most extreme events, where single fire scars covered more than fifty thousand hectares, appeared only in the most recent year of the study, 2024. This sequence reveals that the system did not flip into a new state overnight; rather, the volume of fire increased first, followed much later by a fundamental change in the size and intensity of individual fire events.
The researchers also found that the fire season itself was changing. The period when fires typically occur has moved earlier in the year, shifting from a peak in November to a peak in September. Furthermore, the burning has become more concentrated, meaning that a larger total area is now being burned within a narrower window of time. This compression of the fire season creates a tighter deadline for prevention and suppression efforts, shifting the window of risk away from the historical schedules used by fire management teams.
To understand why the total burned area increased, the team separated the causes into two distinct drivers: the expansion of flammable land cover and the intensification of burning within existing land types. They found that the increase was split almost evenly between these two factors. About half of the rise in burned area was due to the sheer expansion of pasture land, which grew seventy-fold over the study period. The other half was caused by existing land types, particularly the forest, burning more frequently. This is a critical distinction because the forest area was actually shrinking during this time, yet the rate at which the remaining forest burned increased dramatically. In fact, the intensification of burning within the shrinking forest accounted for nearly a quarter of the total regional increase, a trend that would have been invisible if the researchers had only looked at the total burned area.
The study also examined how fire behaves across different types of land ownership, such as Indigenous territories, conservation units, and public lands. They discovered that the likelihood of a fire starting is heavily influenced by the type of land cover present, while the size of the fire, once it starts, is driven more by how accessible the area is to roads and towns. Indigenous territories, which are often covered in dense forest and located away from major roads, showed very low rates of fire occurrence. In contrast, public lands and unregistered areas, which often contain more pasture and are closer to highways, experienced fire much more frequently. However, when a fire did occur in these different zones, the size of the burn was surprisingly similar, suggesting that once a fire gets going, the physical landscape and accessibility determine its scale more than the legal status of the land.
The researchers concluded that relying on a single number to track fire is no longer sufficient. The fire regime in the Amazon is not a single, uniform phenomenon but a collection of changing parts that evolve at different speeds. The total burned area can rise because of new flammable land, or because existing forests are burning more often, or because fires are getting bigger and starting earlier. Each of these changes tells a different story and points to a different solution. By tracking these components separately, monitoring programs can move beyond simply counting acres to understanding the true nature of the transformation happening in the forest, allowing for more precise and effective responses to the changing fire landscape.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.