New research shows that fire in the Amazon rainforest over the past 10,000 years has been almost exclusively human- driven, not a natural process, and has substantially shaped modern forest composition.

The study, “The pyrogeography of Amazonia: a Holocene perspective” by Crystal N. H. McMichael and colleagues (including Mark B. Bush), was published in June 2026 in the open-access journal Frontiers of Biogeography.

It is based on a compilation and analysis of 1,361 radiocarbon (¹⁴C) dates from 303 sites, covering soil charcoal fragments and burned archaeological materials.

Fire is essentially human- ignited in these wet forests. Pre- human (glacial- period) paleoecological records show charcoal absent or present only in minute amounts.

Examples include sites such as the Hill of Six Lakes (Brazil), Serra Sul dos Carajás, and especially the Campo Libre core on Ecuador’s Andean slopes, which recorded ~30,000 years without fire until humans arrived around 4,500 years ago. Researchers describe human arrival as introducing fire to a “naïve landscape” of evolutionarily fire-sensitive plants.

Spatiotemporal spread of fire (Holocene phases):

• ~10,000- 6,000 years ago: Highly restricted, mainly in peripheral areas and along the main Amazon River channel.
• ~6,000- 4,000 years ago: Rapid geographic expansion across most regions, coinciding with early maize cultivation in northwestern and southwestern Amazonia.
• ~3,000- 2,000 years ago: Fire reached the deep interior (central Amazonia had been a relative fire-free refuge for millennia).

Decline in the last ~700 years (two phases): An initial drop ~600- 700 years ago supports an “early abandonment” or major pre- Columbian land- use shift (linked to a pollen surge indicating reforestation in some lake records). A larger decline followed the post- 1541 “Great Dying” (estimated 90- 95% Indigenous population loss from disease, warfare, and enslavement after European contact), reducing ignitions and allowing further forest recovery.

Ecological legacy: Intensified human- driven fire over the last ~2,000 years caused vegetation changes on a scale comparable to major climatic events (e.g., deglaciation or Pleistocene megafauna extinctions). It selectively favored more fire- tolerant species (such as certain palms) over fire-sensitive ones, leaving a lasting imprint on today’s plant communities.

This work challenges the long- standing idea of a largely “pristine” pre- Columbian Amazon.

Fire histories were not fully synchronous across regions, but the overall pattern of human introduction, expansion from periphery to interior, and later declines is clear from the summed probability distributions and related analyses of the dated materials.

The paper is open access via the journal (Pensoft and Frontiers of Biogeography).

The findings align with broader evidence that natural lightning- ignited fires are rare in intact humid Amazonian forest, while human activity has long been the dominant ignition source.

_____________________________________________________________________________________

Glacial- period (Last Glacial and earlier Pleistocene) paleoecological records from Amazonia and adjacent areas show that fire was rare or absent before human arrival, supporting the interpretation that Amazonian wet forests evolved largely without fire.

These records come primarily from lake and wetland sediment cores analyzed for charcoal (macroscopic and microscopic), pollen, phytoliths, and other proxies. Key examples cited in the 2026 pyrogeography paper and supporting studies include:

Campo Libre (eastern Andean slopes, Ecuador, ~2,700 m elevation)

• A ~30,000- year sediment core from an alluvial terrace in a mid- elevation valley of the Quijos River.
• No macroscopic charcoal was found in 122 samples spanning the full record.
• Microscopic charcoal was also absent in samples dating between ~12,000 and 30,000 cal yr BP.
• Fire (regional microscopic charcoal) first appears only around ~3,300- 3,400 cal yr BP, with clear human signals (maize phytoliths) beginning ~4,500- 4,700 cal yr BP.
• The site underwent major vegetation shifts driven by climate and river dynamics (e.g., from active floodplain to palm forest around the Pleistocene- Holocene transition, then back toward swampier conditions), but these occurred without fire until humans arrived.

Hill of Six Lakes (northwestern Brazilian Amazon)

• Multiple long lake records (e.g., Lakes Pata, Verde, Dragão, and more recent work on Lake Malachite) spanning up to ~170,000 years or the full deglacial period.
• Charcoal is rare and present only in very low abundances (often tiny microparticles <40- 80 μm, consistent with distant sources) during glacial times.
• The landscape remained continuously forested (with compositional shifts due to cooler temperatures, ~4- 5°C colder at the Last Glacial Maximum). Fire increases notably only in the Holocene, sometimes linked to possible early human activity around ~10,200 cal yr BP in newer records.

Serra Sul dos Carajás (southeastern Amazonia, Brazil)

• Lake and bog records from this drier, plateau, ecotone region covering the last ~25,000- 26,000 years.
• Charcoal was absent or present only in very low amounts in some locations until the deglacial period and onset of the Holocene.
• Where glacial- period fire occurred, abundances were much lower than during the Holocene. Fires became more frequent during climatic transitions (e.g., dry- to- wet shifts around 11,000- 10,200 years ago) and later under seasonal conditions, with possible early human influence in the Holocene.

Additional context from related sites

• Rainforest- savanna ecotones in Bolivia show very low and infrequent glacial- period fire.
• Even in more fire- prone high Andean settings, pre-human charcoal levels are described as “miniscule” compared with post- human levels.

Natural (lightning- driven) fire was exceptional in these wet tropical systems.

Glacial climates were often cooler and, in many areas, drier or more seasonal, yet charcoal remained scarce.

This contrasts sharply with the Holocene, when human ignition dramatically increased fire frequency and spatial extent.

Amazonian plants are therefore largely evolutionarily fire- sensitive, making the introduction of regular anthropogenic burning a novel and transformative disturbance.

These glacial records form a critical baseline for the argument that Holocene pyrogeography in Amazonia is overwhelmingly human- driven.

_____________________________________________________________________________________

The pyrogeography of Amazonia: a Holocene perspective

The pyrogeography of Amazonia: a Holocene perspective is a 2026 open-access research article, published in Frontiers of Biogeography.

Fire in Amazonian forests rarely (if ever) ignites or spreads without human activity. Humans introduced fire to a previously “naïve” (fire- sensitive) landscape beginning in the late Pleistocene, early Holocene.

The study reconstructs the spatiotemporal patterns of this human- driven fire (pyrogeography) over the last ~10,000 years using a large compilation of radiocarbon dates.

Data and methods

Compiled 1,361 radiocarbon (¹⁴C) AMS dates from 303 sites across Amazonia (after excluding 32 modern dates post-1950, analyses focused on 1,329 dates spanning the last ~12,000 years).

Sources: individually dated charcoal fragments from soil surveys (paleoecological) and burned archaeological materials (described as charcoal or charred).

New dates added from soil charcoal in northwestern and southwestern Amazonia.

Analyses: summed probability distributions (SPDs), composite kernel density estimations (CKDEs), and regional permutation tests.

Regions examined (following ter Steege et al. 2013): northwestern (WAN), southwestern (WAS), central (CA), eastern (EA), southern (SA) Amazonia, and Guiana Shield (GS).

Main results

Earliest fires were concentrated in peripheral areas and along the main Amazon River channel (~10,000–6,000 years ago).

Geographic expansion accelerated ~6,000- 4,000 years ago (linked in part to early maize cultivation), with fire reaching the deep interior (including central Amazonia) mainly ~3,000- 2,000 years ago.

Peak fire likelihood occurred roughly 1,300- 500 cal BP (AD 700-1450).

Two-phase decline in the last ~700 years:

• First decrease ~700- 500 years ago (supports “early abandonment” or major pre- Columbian land- use change/reforestation, consistent with some pollen records).
• Second, larger decline after European contact (~AD 1541 onward), corresponding to the “Great Dying” (estimated 90- 95% Indigenous population loss).

Fire histories show some regional variation and are not fully synchronous, but the overall basin- wide pattern of peripheral- to- interior spread followed by later declines is robust.

The intensification and spread of human- ignited fire over the last ~2,000 years induced vegetation change at rates and magnitudes comparable to major climatic shifts (e.g., deglaciation or Pleistocene megafauna extinctions).

This selectively favored more fire- tolerant species (notably some palms) and left a lasting imprint on the composition and structure of modern Amazonian forests.

The work synthesizes and expands prior archaeological and paleoecological databases.

It reinforces that natural (lightning- driven) fire was exceptional in these wet forests before human arrival, as shown by glacial- period records with absent or minimal charcoal (e.g., Campo Libre, Ecuador: ~30,000 years without fire until humans appeared ~4,500 years ago).

The findings challenge notions of a largely pristine pre- Columbian Amazon and highlight the long- term role of Indigenous fire management in shaping today’s ecosystems.

The full paper is freely available via the journal site (Pensoft).

A contemporaneous press release from Pensoft Publishers summarized these results under the headline “Human- driven fires have shaped the Amazonian forest for over 10,000 years.”

Published: Frontiers of Biogeography (volume 19, e169863)

DOI: 10.21425/fob.19.169863

Authors: Crystal N. H. McMichael, Britte M. Heijink, Nina H. Witteveen, Annabel Zwarts, Mark B. Bush

Abstract

Amazonian forest fires rarely, if ever, ignite and spread in the absence of humans. The first colonization of people into South America happened in the late Pleistocene, and by 10,000 years ago, humans had introduced fire into a naive landscape. Here we build on previous databases of burned ¹⁴C dated archaeological material and ¹⁴C dated charcoal fragments retrieved from soils and reconstruct spatiotemporal patterns of fire in Amazonia for the last 10,000 years. We ask whether fire histories are regionally synchronous, and whether major demographic events such as the ‘Great Dying’ or ‘Early abandonment’ appear in the fire histories within or between regions. We found that most of the earliest fires occurred in the peripheral areas and spread into central Amazonian forests mostly around 2000–3000 years ago. The frequency and recurrence of burned ¹⁴C dated material also showed a two-phase decline over the last 700 years. Most regions show a decrease in fire occurrence ca. 700–500 years ago and then again during the last 400 years when European influence spread across Amazonia. The restructuring of Amazonian pyrogeography over the last 2000 years has likely played a large role in shaping modern forests.