Claims

Carvajal's chronicle of the 1541–42 Orellana descent describes populous riverbank settlements, large food stores, travelled roads, fine ceramics, and multi-settlement political territories along extended reaches of the Amazon — interspersed with stretches he reports as empty.

The account's most extreme figures — 50,000, 10,000, and 8,000 warriors, and villages four or five leagues long — appear in the Medina/Muñoz manuscript branch but not in Oviedo's version, which preserves the populous-settlement picture and one figure of about 5,000 warriors.

Oviedo states that he supplemented the friar's written account with particulars learned orally in Santo Domingo from Captain Orellana and the gentlemen who came with him.

Independent sixteenth-century testimony — Diogo Nunes's 1538 entry into the Machiparo province (recorded in a letter of 1553 or 1554 to João III) and the ten accounts of the 1560 Ursúa expedition — describes dense settlement, vast food stores, long straight travelled roads, and renowned ceramics on the same middle-Amazon reaches Carvajal described.

Medina (1894) records that López de Gómara cited Orellana's relación of the voyage and called it 'mentirosa'.

Achmatowicz (2018) assesses Carvajal's manuscript as a trustworthy historical source on the basis of a study of its geographic reference points.

Carvajal is substantially credible about the qualitative social geography he described — populous, organized, road-connected riverine societies in specific reaches — while his extreme numerical claims are textually unstable and should not be treated as counts.

Canopy-penetrating LiDAR flown over 4,430 km in 2024 documented 432 earthworks (396 previously unknown) in southwestern Amazonia; the 2026 study delimits an Aquiry cultural area of roughly 183,000 km² and predicts approximately 24,000–30,000 earthworks within it.

An 1,800 km stretch of the southern Amazon rim was occupied by earth-building societies in fortified villages around AD 1250–1500; the study's population estimate of roughly 500,000–1,000,000 for this arc explicitly depends on site contemporaneity, ethnographic conversion formulae, and model-projected extrapolation.

The Casarabe culture of the Llanos de Mojos (Bolivia) practiced intensive year-round maize monoculture supported by engineered drainage canals and savannah farm ponds, documented by remote sensing, field survey, and phytolith analysis.

Anthropogenic dark earths, raised-field complexes, and human-made forest islands document long-term pre-Columbian landscape engineering in multiple Amazonian regions, and domesticated tree species are concentrated on and around archaeological sites.

Raised fields in the Llanos de Mojos show intermittent construction and use from as early as AD 570–770, consistent with flood-risk mitigation built "only in those areas where there were no alternatives" — challenging readings of them as continuously cultivated, permanently intensive farmland.

Pre-Columbian societies of the Llanos de Mojos engineered fisheries at landscape scale: zigzag weirs and straight causeways cross the San Joaquín floodplains, and most of the 382 detected ponds sit within 35 meters of such a feature — a system whose V-shaped structures functioned to concentrate fish as floodwater receded.

Zooarchaeology at Loma Salvatierra — a monumental mound 50 km from the Mamoré River, occupied AD 500–1400 — analyzed more than 17,000 fish remains and documents year-round exploitation of interfluvial aquatic resources, including species that persist in small water bodies through the dry season.

The 2026 study's proposal of 1.25–3 million people at the Aquiry peak (AD 100–300) is highly model-sensitive: it rests on an assumed simultaneous-use fraction and roughly 300 people per average earthwork, and the paper's own one-family-per-house alternative yields about 250,000–600,000 (315,000–760,000 with canopy correction).

Published basin-wide population estimates are model families sharing assumptions rather than independent measurements: Denevan himself later rejected the habitat-density method behind his influential 5.1–6.8 million figure — while still holding "at least 5 to 6 million" reasonable on other grounds.

Summed radiocarbon-date distributions for the Amazonian biome fit logistic growth approaching carrying capacity in the centuries before European colonization; the proxy's terminal decline is attributed by the authors to sampling bias compounded by post-European collapse, not to a pre-European endogenous crash.

Immediately before European contact, Amazonia was a strongly heterogeneous human landscape: substantial sedentary populations, engineered subsistence, and settlement systems along favored river corridors, floodplains, savannas, and bluffs, embedded in a much more lightly occupied forest matrix.

Several major settlement systems were occupied into the contact era or beyond: the Belterra Plateau sites near Santarém from ca. AD 1300 to seventeenth-century European colonization (with dark earths, ponds, and trodden roads); Marajó's Cacoal site AD 1300–1650; and Upper Xingu settlements inhabited to ca. 1770 with layouts matching living Kuikuro villages.

Region-specific colonial reconstructions document catastrophic but uneven post-contact decline: the Napo–Aguarico population fell from roughly 49,000–51,000 to about 6,000 by circa 1700 (≈88%), and Mainas declined 62–64%, through differing mixtures of epidemics, slave raids, missionization, forced service, revolt, and flight.

Colonial-era epidemics in sparsely settled lowlands commonly "faded out" rather than sweeping uniformly, with slave raids and missionization repeatedly reintroducing infection and adult mortality heaviest where disease arrived late — collapse was an interactive process, not a single wave.

Across 7,363 recorded archaeological sites, distance from a river is the most important predictor of pre-Columbian occupation, and interfluvial areas were more sparsely populated than areas close to rivers.

Cumulative phytolith and charcoal studies across 14 lowland regions — concentrated in western and interfluvial Amazonia — find little to no evidence of annual crop agriculture, significant forest clearing, anthropic soils, enrichment of favored trees, or recurrent extensive fires in most sampled terra-firme settings during the pre-Columbian era.

Machine-learning prediction over known sites finds earthwork and dark-earth extents "limited, albeit impressive," with other archaeological site types predicted to dominate more than two-thirds of Amazonia.

The sampling behind "sparse western Amazonia" conclusions has been formally criticized as thin: roughly 300 soil cores clustered in five small areas — about 1.17 cubic meters of sediment — standing in for two million square kilometers.

In the Upano Valley, "a long period of forest recovery largely eradicated the influence of Pre-Columbian occupations" on the ecological record, while the later Colonial-period signal persisted in the same profiles.

In the same basin-wide model, almost 80% of Amazonia has a predicted probability of 0–1% for earthwork presence per 1 km² cell, with these low-probability areas concentrated in northwestern, northern, and central Amazonia.