A new hypothesis for the origin of Amazonian Dark Earths

The study challenges the prevailing view that Amazonian Dark Earths (ADEs) are principally anthropogenic soils created by pre-Columbian peoples through biomass burning and soil amendment. The research question asks how persistent, highly fertile soil patches could arise within nutrient-impoverished Amazonian landscapes and whether human activity alone can account for their genesis, timing, and magnitude. The context includes the dominance of acidic, nutrient-poor Oxisols and Ultisols in Amazonia, the long-term presence of productive native ecosystems, and archaeological evidence indicating domestication of plants >10,000 years ago with complex agricultural societies emerging <4000 years ago. ADEs are concentrated in central and eastern Amazonia, far from early Andean settlements, and biochar experiments have failed to reproduce ADEs’ enduring mineral fertility. The study aims to test whether exogenous, natural geomorphic processes (e.g., alluvial deposition) predate and underpin ADE fertility, with humans later identifying and using these fertile patches.

Prior work characterizes ADEs as Anthrosols due to frequent artifacts and high pyrogenic carbon, with a dominant hypothesis that pyrolysis of nutrient-rich biomass created extensive fertile patches in otherwise dystrophic soils. However, the timing, scale, and mechanisms of such human activity remain poorly constrained. Archaeological and paleoecological evidence indicates domestication of native plants >10 ka BP and regionally variable emergence of intensive cultivation mostly within the last 4000 years, with ADEs rare near some early Andean settlements and more common in central/eastern Amazonia where intensive management evidence is younger (2500–500 years BP). Experimental biochar additions have not replicated ADEs’ long-lasting mineral fertility, suggesting missing processes in prevailing models. Geomorphological models have predicted ADE distribution using river proximity and elevation alone, hinting at landscape controls. Syntheses of regional pollen, charcoal, and paleoflood archives document Holocene shifts in monsoon intensity, fire regimes, and floodplain dynamics that could influence soil carbon and nutrient deposition, motivating tests of exogenous, fluvial contributions to ADE formation.

Study site: EMBRAPA-CPAA near the Solimões River (central Amazon), on a terrace ~40 m a.s.l., up to ~27 m above river level and ~10 m above adjacent floodplain. A bluff interpreted as a natural levee borders a ~12 ha ADE patch. Sampling design: From EMBRAPA reference profiles, ~4.5 ha were intensively sampled using six transects (~100 m apart), each with five soil profiles (~10 m apart), totaling 30 profiles under secondary evergreen rainforest. Each profile was sampled every 10 cm from 0–100 cm, yielding 300 samples, with uncompressed cores for bulk density. Soil chemistry and stocks: Extractable P (Mehlich-1) and Ca (1 N KCl) were measured; total elements were determined after multi-acid digestion (HNO3, HClO4, HF protocols) and analyzed by ICP-MS (Ag, Al, As, Ba, Be, Ca, Cd, Co, Cu, Cr, Cs, Fe, K, Li, Mg, Mn, Mo, Ni, P, Pb, Rb, Se, Sr, Ti, Tl, U, V, Zn). Stocks to 1 m were calculated as concentration times bulk density summed by 10 cm increments. Carbon analyses: Total organic C via combustion GC; pyrogenic carbon (microcharcoal) estimated after acid-peroxide digestion of fine soil fraction with mass-balance of residual carbon. Particle size observations and clay mineralogy by XRD were used to assess texture and mineral assemblages. Isotopes: Bulk soil organic matter δ13C by GC-IRMS; contributions of C3 vs C4 sources estimated via two-endmember mixing using regional δ13C values, accounting for diagenetic fractionation in surface layers. Radiocarbon (14C) age-depth models for bulk SOM and pyrogenic carbon were developed using AMS on 15 samples (8 bulk, 7 PyC), with pretreatments and standards; calibrated ages reported as years BP. Sr and Nd isotopes: Soil mineral phase digested; 87Sr/86Sr measured after chromatographic separation (SR-B50-A resin), and Sm-Nd isotopes measured by multi-collector mass spectrometry; εNd(0) calculated. Mixing considerations used to apportion riverine vs local sources (including evaluation of Sr/Ca ratios and potential fishbone vs sediment contributions). Spatial analysis: First-order kriging mapped extractable P and Ca to identify depositional gradients relative to a paleochannel levee and to guide maximum-contrast ADE vs neighboring Ultisol sampling. Topographic context derived from SRTM and nearby ICESAT-2 data. Human input estimation: Nutrient excess (P+Ca) in ADE relative to Ultisol (~32 Mg ha−1) was used to back-calculate required inputs from potential human-derived sources (fish waste, feces) under assumed consumption rates and population densities (20–60 people ha−1), incorporating a protein conversion factor and considering potential leaching and combustion losses.

• ADE fertility: Extractable P and Ca were 1–3 orders of magnitude higher inside the ADE patch than in adjacent Ultisols, forming a clear gradient from a putative levee toward nutrient-poor soils.
• Nutrient stocks: ADE profiles contained, on average, 16.8 Mg P ha−1 and 14.9 Mg Ca ha−1 more than surrounding Ultisols (0–100 cm). Sixteen additional elements (Ba, Ca, Cd, Co, Cs, Cu, K, Li, Mg, Mn, Ni, P, Rb, Sr, Tl, Zn) were significantly enriched in ADEs (p < 0.05). PCA showed strong clustering by soil type with P/Ca-associated variance (~56%).
• Carbon pools and sources: Total organic C was only slightly higher in ADEs near the surface, converging with Ultisols at depth. δ13C profiles indicated 5–19% C4-derived carbon in ADE horizons at 30–100 cm, contrasting with C3-dominated Ultisols. Pyrogenic carbon was 5–10× higher in ADE than in Ultisols.
• Chronology: Radiocarbon dating showed microcharcoal deposition in ADEs began ~7630 ± 80 years BP, predating earliest regional evidence of cultivation (~4000 BP). These dates likely represent minimum ages due to potential later inputs and charcoal diagenesis.
• Exogenous inputs: Depth-coincident enrichments in P, Ca, Sr, Nd, and trace elements (e.g., Ni, Rb, Ti, Zn) imply alluvial deposition into ADE profiles near a natural levee, not in adjacent Ultisols. Isotopic signatures support a fluvial source: ADE 87Sr/86Sr ~0.714 (between Solimões River ~0.709 and Ultisols ~0.7152) and more radiogenic Nd in ADE than Ultisols, consistent with river sediment inputs. Mixing suggests ~24% of Sr mass in ADEs from fishbone or river sediment; Sr/Ca ratios favor river suspended sediments as dominant source given observed Ca and Sr concentrations.
• Sedimentology: ADE profiles are sandy clay loam versus very clayey Ultisols, better explained by alluvial deposition than by charcoal-induced aggregation. XRD found similar dominant clays in both soils, consistent with regionally weathered sediments and no mineral neoformation.
• Human input constraints: To supply the observed P+Ca excess (~32 Mg ha−1) in a 1 ha ADE solely via human inputs would require either ~1100–8000 years of continuous occupation (fish consumption assumptions) or ~3700–11,000 years via feces at 20–60 people ha−1, not accounting for leaching and combustion losses that would greatly increase requirements. Thus, human-derived inputs are unlikely to account for ADE genesis at the observed magnitudes and timelines.
• Regional context: Paleoflood and monsoon records indicate mid- to late-Holocene changes in flood regimes and vegetation (dry period ~8000–4000 BP followed by increased tree cover) consistent with delivery and sequestration of pyrogenic carbon and nutrients in floodplain terraces. ADE distribution patterns align with geomorphic predictors (distance to rivers, elevation).

Findings directly address the origin of ADEs by demonstrating that their exceptional fertility and microcharcoal content largely derive from exogenous, fluvial deposition of mineral nutrients and pyrogenic carbon during mid- to late-Holocene flood events, predating intensive human cultivation. Isotopic (Sr, Nd), elemental, and δ13C-14C evidence converge on a natural depositional mechanism associated with paleo-levee environments. Human occupation likely recognized and capitalized on these fertile patches, adding some waste material and charcoal in more recent times, but the scale and timing required for humans alone to generate the observed nutrient stocks are implausible given regional archaeological chronologies. This reframes ADEs from being created primarily by anthropic pyrogenic amendments to being landscape features generated by Holocene geomorphologic and climatic processes, subsequently intensified or maintained by indigenous land use. The results explain why ADE distribution is predictable from geomorphology and why biochar additions fail to replicate ADE mineral fertility: the key ingredients and their stratigraphic context are products of natural river dynamics, not solely managed inputs.

The study proposes and supports a new hypothesis: at the investigated site, ADEs originated predominantly from natural alluvial deposition of pyrogenic carbon and mineral nutrients thousands of years before intensive cultivation. Indigenous peoples likely identified and preferentially settled these naturally fertile areas and may have contributed additional inputs that maintained or enhanced soil properties. This reinterpretation has implications for understanding human-environment interactions in Amazonia and for sustainable land management models inspired by ADEs. Future research should: (1) test for similar depositional and isotopic signatures at other ADE sites; (2) disentangle riverine versus biogenic sources of elements with refined mixing models; (3) integrate geomorphology, paleoecology, and indigenous knowledge to reconstruct site-specific histories; and (4) reassess biochar-based strategies in light of the dominant role of natural processes in ADE genesis.

• Site specificity: Conclusions are based on a single well-studied site and may not capture regional variability or different ADE formation pathways.
• Chronological uncertainty: Radiocarbon dates on pyrogenic carbon may reflect minimum ages due to later inputs and charcoal diagenesis; precise timing of earliest deposition events remains uncertain.
• Source attribution: Sr isotopic compositions of fishbone and river sediment can overlap; mixing models have uncertainties in endmembers and Sr/Ca ratios, limiting precise apportionment.
• Loss processes: Human input calculations simplify complex leaching and combustion losses; actual nutrient retention efficiencies are uncertain and likely variable through time.
• Geospatial data: Lack of lidar-derived high-resolution topography may limit detailed reconstruction of paleoflood elevations and microtopography.
• Generalizability: Findings do not preclude significant anthropic contributions at other ADE sites; broader surveys are required to assess the prevalence of natural genesis.
• Proxy limitations: δ13C mixing models assume stable endmembers and minimal confounding from atmospheric changes or water-use efficiency; XRD cannot detect subtle mineral neoformation; PCA and correlations infer but do not prove causation.