Introduction
Amazonian dark earths (ADEs), also known as terra preta, are among the most studied soils in tropical archaeology (Glaser et al., Reference Glaser, Haumaier, Guggenberger and Zech2001, Reference Glaser, Guggenberger, Zech, Lehmann, Kern, Glaser and Woods2003; Silva, Reference Silva2018; Tavares, Reference Tavares2022). They are characterized by properties that contrast sharply with the surrounding nutrient-poor soils, including high fertility, elevated organic matter, and enriched nutrient content. For this reason, ADEs are widely interpreted as the result of long-term pre-Columbian occupation and management, including biomass burning, food preparation, and waste disposal (Kern, Reference Kern1996; McClain et al., Reference McClain, Victoria and Richey2001; Behling and Costa, Reference Behling and Costa2004; Fageria and Stone, Reference Fageria and Stone2006; Madari et al., Reference Madari, Cunha, Novotny, Milori, Martin Neto, Benites, Coelho, Santos, Teixeira, Kern, Madari, Lima and Woods2010; Glaser and Birk, Reference Glaser and Birk2012; Dos Reis, Reference Dos Reis2015; Negreiros et al., Reference Negreiros, Pessoa Júnior, de Souza, Justi, Ramos and da Silva2020; Moura, Reference Moura2022).
Despite this consensus, the extent to which ADEs are exclusively anthropogenic remains debated (Lehmann et al., Reference Lehmann, da Silva, Steiner, Nehls, Zech and Glaser2003a; Glaser and Birk, Reference Glaser and Birk2012). Recent studies suggest that natural processes may play a more significant role than previously recognized. In particular, Silva et al. (Reference Silva, Corrêa, Wright, Bomfim, Hendricks, Gavin and Muniz2021) identified microcharcoal enrichment and external inputs of phosphorus and calcium predating human occupation, with magnitudes exceeding those expected from anthropogenic activity alone. These findings indicate that ADE formation may involve contributions from natural depositional and pedogenic processes, challenging traditional interpretations and calling for a more integrated, hybrid model of soil formation.
Mineralogical approaches provide a critical but still underexplored line of evidence for evaluating such a model. While most studies have focused on chemical and organic proxies, mineral assemblages offer direct information on sediment provenance, geological inheritance, and postdepositional transformation (German, Reference German2003; ICOMANTH, 2005; Heckman and Rasmussen, Reference Heckman and Rasmussen2011; Delarmelinda et al., Reference Delarmelinda2017). More importantly, because minerals can be used to distinguish between natural and occupation-related inputs, a systematic mineralogical provenance analysis offers a direct test of whether ADEs are products of human engineering alone or instead result from the interaction of geological inheritance, alluvial deposition, pedogenesis, and cultural practices (Tomazelli, Reference Tomazelli1978; Caspari et al., Reference Caspari, Bäumler, Norbu, Tshering and Baillie2006; Teixeira et al., Reference Teixeira, Kern, Madari, Lima and Woods2010; Barros et al., Reference Barros, Santos, Souza Júnior, Delamerlinda, Araújo Filho and Câmara2018).
To address this question, this study develops a mechanistic, four-process hybrid model for ADE formation, supported by preliminary mineralogical data from four archaeological sites in Pará, Brazil (Terra Preta 2, Ilha de Terra, Jacarequara, Jabuti). Using X-ray diffraction (XRD) and scanning electron microscopy with energy-dispersive spectroscopy (SEM-EDS), we (1) analyze the mineralogical provenance of ADEs, distinguishing local bedrock inheritance from alluvial inputs via accessory mineral assemblages; (2) identify detrital apatite as a potential natural phosphorus source, acknowledging that this evidence is preliminary and does not definitively refute anthropogenic phosphorus enrichment given the limited number of sites and analytical constraints; and (3) propose a testable model in which geological and sedimentary inputs supply mineral-bound nutrients, pedogenesis regulates their release, and human activity enhances their stabilization. By linking mineral provenance to soil-forming mechanisms, this study provides a framework to evaluate the relative contributions of natural and human factors in ADE formation across Amazonia.
Regional setting
Study area
The study area, shown in Figure 1, covers northeastern and northwestern Pará, encompassing four archaeological sites: Jacarequara (Barcarena, adjacent to the Tocantins River), Jabuti (Bragança, northern Pará), Terra Preta 2 (Juruti, right bank of the Amazon River, Lower Amazon), and Ilha de Terra (Portel, within the Caxiuanã National Forest). All four occupy lowland settings near major watercourses, consistent with pre-Columbian settlement strategies favoring well-drained, riverine locations. The ADEs at these sites are Late Holocene to Early Quaternary anthropogenic A horizons enriched in charcoal, ceramics, and organic residues, overlying Neogene–Cretaceous substrates (Lehmann et al., Reference Lehmann, Kern, German, McCann, Martins, Moreira, Lehmann, Kern, Glaser and Woods2003b; Silva, Reference Silva2022). The region is climatically humid equatorial, with mean annual rainfall between 2000 and 2500 mm and consistently high relative humidity (IBGE, 1993; Costa, Reference Costa2011; Silveira et al., Reference Silveira, Oliveira and Piccinin2010; Costa et al., Reference Costa, Costa and Kern2013; Barcarena, 2016). Relief ranges from gently undulating plateaus in Barcarena and Bragança to fluvial plains and lateritic surfaces in Juruti, and dense forested lowlands in Caxiuanã (Embrapa et al., Reference Embrapa, Rodrigues, Oliveira Júnior, Silva, Valente and Cardoso Júnior1999; Bragança, 2023). These physiographic contrasts in soil fertility, drainage, and erosional dynamics influenced both the preservation and transformation of ADEs (Brasil, 1993; CNEC, 2002; Simonian, Reference Simonian2006).
Sample location map and adjacent regional geological units of the studied Amazonian dark earth (ADE) sites. The map situates the four sampled areas within their regional geological frameworks (Alter do Chão Formation, Barreiras and Post-Barreiras deposits, and crystalline sources) based on Vásquez et al. (Reference Vasquez, Souza and Carvalho2008).
Figure 1 Long description
A composite image with four maps detailing Amazonian dark earth sites and geological units in Pará, Brazil. The top left map shows the Terra Preta 2 site near Juruti, marked with a black circle, adjacent to the Amazon River. The top right map highlights the Jacarequara site near Barcarena, with city limits and geological units like Alter do Chão Formation and Barreiras sediments. The bottom left map displays the Ilha de Terra site near Portel, with mature lateritic cover and alluvial deposits. The bottom right map features the Jabuti site near Bragança, showing Barreiras Group and coastal deposits. The legend identifies city limits, Amazon River, black earth sites and geological units with specific colors and symbols. The inset map of Pará locates the four sites within the region. Datum SIRGAS 2000 is referenced, with Vásquez et al., 2008 as the source.
Preliminary geological context and detrital mineral sources
The four sites studied here represent only a fraction of Amazonia’s lithological diversity, and their limited number does not allow regional generalization regarding lithological sources. Rather, the geological descriptions presented for each site serve as preliminary local context for interpreting the mineral assemblages observed in the corresponding ADEs and provide a baseline against which the proposed hybrid model can be evaluated. Each site offers a distinct detrital mineral background against which anthropogenic inputs were incorporated during ADE formation.
At Barcarena, soils overlie the Neogene Barreiras Formation sandstones and Quaternary fluvial deposits, both rich in quartz and kaolinite, contributing sandy detritus and aluminosilicate precursors to ADE mineralogy (Brasil, 1974; Embrapa, 1997, 2011; Manzatto et al., Reference Manzatto, Freitas Junior and Peres2002; Silveira et al., Reference Silveira and Marques2004). At Bragança, situated between the Pará–Maranhão and Bragança–Viseu basins, Miocene limestones (Pirabas Formation), Barreiras siliciclastics, and Quaternary tidal deposits are supplemented by inputs of resistant heavy minerals (zircon, ilmenite) derived from nearby crystalline sources such as the Tracuateua Intrusive Suite (∼2.09 Ga) (Góes, Reference Góes1981; Faria, Reference Faria1982; Behling and Costa, Reference Behling and Costa2004; Rossetti, Reference Rossetti2006; Vasquez et al., Reference Vasquez, Souza and Carvalho2008; Embrapa, 2011; Jorge, Reference Jorge2017; Sousa, Reference Sousa2021; Bragança, 2023;). At Juruti, dominated by the Cretaceous Alter do Chão sandstones, intense laterization has produced bauxite deposits and hematite–kaolinite-rich soils, whose strong weathering signatures are reflected in the local ADE (Klinge, Reference Klinge1965; M.L. Costa, Reference Costa1991; Lucas, Reference Lucas, Carvalho, Boulangé, Melfi and Lucas1997; J.A. Costa, Reference Costa2011; Vasquez et al., Reference Vasquez, Souza and Carvalho2008; Costa et al., Reference Costa, Carmo, Oliveira, Lima, Kern, Goeske, Teixeira, Kern, Madari, Lima and Woods2010). At Portel (Caxiuanã), Post-Barreiras sediments and deeply weathered Latosols are dominated by kaolinite, quartz, and iron oxides (Rodrigues and Oliveira, Reference Rodrigues and Oliveira1996; Silveira et al., Reference Silveira, Kern, Quaresma and Lisboa2002; Carmo, Reference Carmo2008; Vásquez et al., Reference Vasquez, Souza and Carvalho2008; Gama et al., Reference Gama, Valente, Oliveira Júnior, Cravo, Carvalho, Rodrigues, Cravo, Viégas and Brasil2020).
Materials and methods
Sample preparation
Approximately 150 g of soil sample from the A horizon of ADE soils was collected from each of four archaeological sites in Pará, Brazil. Four representative samples were selected from the ADE horizons of each site for mineralogical analysis. Representative samples were selected and labeled as follows: TPA-1 (Terra Preta 2 site, Juruti), TPA-2 (Ilha de Terra site, Portel), TPA-3 (Jabuti site, Bragança), and TPA-4 (Jacarequara site, Barcarena) (Fig. 2). Sampling focused on the A horizon, at a depth of 0–20 cm within the soil profile, associated with archaeological dark earth formation, where anthropogenic enrichment is most pronounced. Samples were then sieved using an electromagnetic sieve shaker (Bertel model 220v) into fractions >0.250 mm, 0.250–0.125 mm, 0.125–0.062 mm, and <0.062 mm, and shaken for 7 minutes per sample. The analytical focus was on identifying provenance indicators (e.g., quartz, chromite, kaolinite) and occupation-related inputs (e.g., apatite, pyrogenic hematite, ceramic fragments).
Representative field context and soil profile of ADEs in the eastern Amazon. The upper image shows a typical anthropogenic horizon rich in organic matter and archaeological inclusions, overlying lighter mineral horizons. The lower image illustrates the surrounding lowland landscape with forested areas and seasonally flooded zones characteristic of the study region.
Mineralogical identification by XRD
Samples were analyzed at the Mineral Technology Center (CETEM) in Rio de Janeiro. X-ray data of powdered samples were obtained using a Bruker D8 Endeavor diffractometer under the following conditions: CuKα radiation (40 kV/40 mA); goniometer step of 0.02° 2θ with counting time of 0.5 s per step, scanned from 4° to 80° 2θ, using a LynxEye position-sensitive detector. Phase identification was carried out by comparison with reference patterns from the Crystallography Open Database (COD, 2023) using HighScore Plus software v. 5.3a (Malvern Panalytical). Quantitative phase analysis was performed by the Mineral Characterization Laboratory (LCM) of the Federal University of Pará using the Rietveld refinement method, a full-pattern fitting approach in which a calculated diffraction profile, derived from crystal structure models, peak profile functions, and instrumental parameters, is iteratively fit to the measured pattern by least-squares minimization of the difference between observed and calculated intensities (Moore and Reynolds, Reference Moore and Reynolds1989; Young, Reference Young1993; McCusker et al., Reference McCusker, Von Dreele, Cox, Louer and Scardi1999; Albinati and Willis, Reference Albinati, Willis and Prince2006). Phase weight fractions were extracted from the refined scale factors following Hill and Howard (Reference Hill and Howard1987). Structural models for all identified phases were retrieved from the COD database.
It should be noted that the Rietveld refinement quantified only the crystalline phases, with the reported weight fractions normalized to 100% of the crystalline assemblage. No internal standard was added to determine the absolute amorphous content; consequently, the amorphous component, attributed to pyrogenic carbon (biochar) and visible as a broad, low-angle hump in the diffractograms, was assessed only qualitatively through identification of its characteristic diffuse scattering and was not included in the quantitative results.
Microanalysis by SEM
SEM-EDS techniques assisted in identifying soil minerals and supported chemical analyses of minerals in the samples studied. Images were obtained at the Institutional Laboratory of Scanning Electron Microscopy of the Emilio Goeldi Museum, using a TESCAN Mira3 microscope with a field emission gun, interpreted according to Parfenoff et al. (Reference Parfenoff, Pomerol and Toureno1970). Samples were coated with Au for 1′30″, depositing a 10- to 15-nm-thick layer. Images were generated using secondary electron detection, with 5 kV acceleration voltage and a working distance of ∼15 mm. EDS analyses were performed at 15 kV acceleration voltage and 15 mm working distance, using an X-Act detector and AZtecOne software (Oxford Instruments, UK). Qualitative mineralogical interpretations, supported by comparative geological context, were primarily employed over statistical analyses on the mineralogical distributions.
Results and discussion
X-ray mineralogical characterization
XRD mineralogical analysis of the four ADE samples (Fig. 3) revealed a mineral assemblage uniformly dominated by quartz and kaolinite, with site-specific accessory iron and Fe-Ti oxides resolved by Rietveld refinement. Quartz is the dominant phase in every sample (83.7–98.6 wt%, lowest at Terra Preta 2 and highest at Ilha de Terra), reflecting its resistance to chemical weathering and its abundance in the sandstone-dominated substrates (Barreiras and Alter do Chão Formations) and Quaternary fluvial deposits that supply detritus to these soils. Kaolinite is the principal secondary phase (1.4–13.4 wt%, same range), corresponding to the kaolinite-1A polytype associated with the triclinic structural form commonly reported in tropical weathering environments. Its dominance over less stable aluminosilicates is diagnostic of advanced chemical weathering under intense leaching, where only thermodynamically stable minerals persist (Kronberg et al., Reference Kronberg, Fyfe, Leonardos and Santos1979; Martinelli, Reference Martinelli1986; McClain et al., Reference McClain, Victoria and Richey2001; Lehmann et al., Reference Lehmann, Kern, German, McCann, Martins, Moreira, Lehmann, Kern, Glaser and Woods2003b), in agreement with prior analyses of ADEs in the Marajó communities (Negreiros et al., Reference Negreiros, Pessoa Júnior, de Souza, Justi, Ramos and da Silva2020), southeast Pará (M.M.C. Silva, Reference Silva2012; Silva et al., Reference Silva, Pereira, Oliveira, Melo, Kern, Costa and Kämpf2012), and the Santarém region (U.S. Silva, Reference Silva2012), and reflecting the role of tropical weathering in generating the stable clay minerals associated with ceramic raw materials (Novotny et al., Reference Novotny, Hayes and Madari2009; Silveira et al., Reference Silveira, Oliveira, Kern, Costa and Rodrigues2011; Gonçalves, Reference Gonçalves2018).
X-ray diffraction (XRD) patterns of the studied ADE sites: TPA-1 (Terra Preta 2), TPA-2 (Ilha de Terra), TPA-3 (Jabuti), and TPA-4 (Jacarequara). Quartz is the dominant mineral phase in all samples, with subordinate kaolinite and minor hematite and anatase in TPA-1. Broad features indicate the presence of amorphous material and biochar.
Figure 3 Long description
A line graph with three overlaid traces and one additional trace shown in the legend. The y-axis label is, Counts. The y-axis scale shows 100, 1000 and 10000. The x-axis label is, Position theta degrees Copper Cu. The x-axis shows labeled ticks at 10, 20 and 30. Four sample labels appear in the legend: TPA 1, TPA 2, TPA 3 and TPA 4. Each sample entry includes a list of phases and percentages. The plotted traces show repeated narrow peaks at multiple positions across the full x-axis range. Several peaks have vertical text labels above them reading, Quartz. One peak near the left side has a vertical label reading, Anatase. A region near the left side is labeled, Amorphous. A lower left region is labeled, Biochar. Near the right side, one peak label reads, Kaolinite. The traces show a higher baseline at lower position values that decreases toward the middle of the plot, with narrow peaks superimposed throughout.
Beyond this shared matrix, Terra Preta 2 (TPA-1) was the most mineralogically diverse, containing hematite (2.4 wt%) and anatase (0.6 wt%) alongside its elevated kaolinite content (13.4 wt%), reflecting the intense laterization of the Cretaceous Alter do Chão sandstones in Juruti that produced the bauxitic, iron oxide–rich soils characteristic of the region (Costa, Reference Costa1991; Lucas, Reference Lucas, Carvalho, Boulangé, Melfi and Lucas1997). Jacarequara (TPA-4) resolved a minor hematite content (0.1 wt%); at this concentration, hematite could derive from a pedogenic/detrital source or a pyrogenic, anthropogenic origin, and bulk XRD alone cannot discriminate between them.
The most robust crystallographic signature of human activity, however, is non-crystalline: a broad, low-angle amorphous hump present in all four diffractograms, attributed to pyrogenic carbon (biochar). Its presence across every site provides the clearest XRD-based evidence of anthropogenic charcoal enrichment, in contrast to the crystalline phases, which are governed primarily by natural weathering and provenance.
The predominance of quartz and kaolinite in the bulk patterns reflects their high abundance and crystallinity, while accessory detrital minerals and minor iron oxides are only partially resolved or fall below the refinement’s detection limit due to their low concentrations and, in some cases, poor crystallinity. As noted by U.S. Silva (Reference Silva2012) for the Da Mata and Porto de Santarém ADE sites, such phases are frequently camouflaged within the diffractogram background, underscoring the need for complementary high-sensitivity techniques to characterize the provenance-diagnostic accessory assemblage.
Minerals identified by SEM
SEM-EDS observations reveal a range of mineral morphologies and textures consistent with the mineral phases identified by XRD, while also capturing accessory minerals not resolved in bulk analyses. Overall, quartz grains commonly display irregular surfaces and conchoidal fractures, with some grains showing subrounded edges. Kaolinite occurs as lamellar platelets, frequently forming aggregates with pseudo-hexagonal outlines. Iron and aluminum oxides and hydroxides appear as granular to botryoidal coatings on mineral surfaces.
At Terra Preta 2 (Figs. 4 and 5), quartz, kaolinite, hematite, zircon, and ilmenite were identified. Quartz grains range from angular to subrounded morphologies and are commonly coated by fine-grained material. Kaolinite occurs as lamellar aggregates, while hematite appears as coatings and granular masses. Zircon and ilmenite occur as discrete accessory grains, the latter showing Fe-Ti compositions consistent with partial alteration to anatase. This is in line with the XRD detection of anatase, rather than ilmenite, at this site, because progressive structural breakdown during weathering broadens ilmenite’s diffraction peaks below detection, while the secondary anatase remains well crystallized.
Microphotographs of ADEs from the TPA-1 site (Terra Preta 2) showing (A) a general representative image of the soil; (B) the chemical composition of A; (C) zircon (1) embedded in quartz grains coated by clay; and (D) zircon (1) and ilmenite (2).
Microphotographs of ADEs from the TPA-1 site (Terra Preta 2), showing (A) ilmenite (1), kaolinite (2), quartz (3), and kaolinite (4); (B) ilmenite (1); and (C and D) clay coating quartz crystals.
At Ilha de Terra (Fig. 6), quartz, kaolinite, ilmenite, and chromite were observed. Chromite occurs sporadically, typically as isolated grains embedded within quartz, with smooth to slightly altered surfaces. Kaolinite maintains a lamellar habit, and ilmenite occurs as accessory grains associated with the sandy matrix.
Microphotographs of ADEs from the TPA-2 site (Ilha de Terra) showing (A) quartz (1, 2, and 3), quartz with a clay coating (4), ilmenite (5), and kaolinite (6); (B) energy-dispersive spectroscopy (EDS) image of A; (C) chromite embedded in quartz (1); and (D) a representative image of clay coatings on quartz.
At Jacarequara (Fig. 7), quartz, kaolinite (in lower abundance), ilmenite, and apatite were identified, with the last notably occurring as well-defined prismatic grains. Kaolinite appears as small lamellar aggregates, and quartz grains show similar textural characteristics to those observed at the other sites.
Microphotographs of ADEs from the TPA-4 site (Jacarequara) showing (A) ilmenite (1), quartz with a clay coating (2), quartz (3), kaolinite (4), and apatite (5); (B) energy-dispersive spectroscopy (EDS) image of A; (C) ilmenite (1), quartz (2 and 5), and kaolinite (3, 4, and 6); and (D) quartz with a clay coating (1), ilmenite (2), and kaolinite (3, 4, and 5).
At Jabuti (Fig. 8), quartz, kaolinite, zircon, and ilmenite were observed. Quartz grains are abundant and exhibit angular to subrounded shapes. Kaolinite occurs as lamellar aggregates, while zircon and ilmenite are present as discrete accessory grains.
Microphotographs of black earth soils from the TPA-3 site (Jabuti) showing (A) quartz (1), kaolinite (2 and 3), and zircon (4 and 5); (B) energy-dispersive spectroscopy (EDS) image of A; (C) zircon (1) and kaolinite (2); and (D) EDS image of C.
Geological inheritance
The presence of accessory minerals such as zircon, ilmenite, apatite, and chromite provides crucial information about sediment provenance and can be compared to the mineral assemblage of adjacent rocks, shown in Table 1. However, they derive from two distinct natural sources: local bedrock weathering and exogenous alluvial deposition.
Classification of minerals in studied ADE sites compared to the main geological units of the Amazon region.
Table 1 Long description
Mineral assemblages are grouped by location or geologic unit into ultra-stable, stable, and unstable minerals or pedogenic products, plus a brief note on which group predominates. All ADE sites list quartz as the main ultra-stable mineral; zircon is also present at Terra Preta 2 and Jabuti, while Ilha de Terra uniquely includes chromite. Stable minerals in ADE sites are mostly ilmenite, with hematite and anatase added at Terra Preta 2 and apatite added at Jacarequara. Unstable minerals and pedogenic products in ADE sites are dominated by kaolinite, with gibbsite also reported at Terra Preta 2. Predominance notes indicate extreme lateritization at Terra Preta 2 and Jabuti, a mix of stable and ultra-stable minerals with anthropic influence at Jacarequara due to apatite, and preserved mafic input at Ilha de Terra alongside ultra-stable minerals. Regional units show similar patterns: Alter do Chão and Post-Barreiras combine quartz and zircon or quartz alone with kaolinite and Fe/Al oxides, consistent with intense alteration, while Barreiras includes tourmaline and apatite among resistant detrital minerals. Tracuateua Intrusive Suite stands out for including magnetite and apatite among stable minerals and lacking listed pedogenic products. Interpretive labels such as lateritization, mafic input, and anthropic influence summarize likely processes and should be read as qualitative rather than quantitative dominance.
a This study.
b Mendes (Reference Mendes2010).
c James et al. (Reference James, Santos Júnior and Costa2018).
d Palheta et al. (Reference Palheta, Abreu and Moura2009).
Local bedrock inheritance is indicated by minerals that can be linked to the immediate geological substrate of each site. At Jabuti and Juruti, zircon and ilmenite are typical of granitic and metamorphic rocks; these match the Paleoproterozoic Tracuateua Intrusive Suite and the Alter do Chão Formation, respectively. At Jacarequara, prismatic apatite occurs in the Barreiras Formation sandstones and is interpreted as detrital, inherited directly from the local parent material. Chromite at Ilha de Terra, however, cannot be explained by local bedrock alone, because no mafic/ultramafic sources are known in the immediate vicinity of Portel.
Alluvial/exogenous contributions are therefore required to explain chromite at Ilha de Terra and possibly some of the rounded quartz and heavy mineral grains at other sites. The most likely source is fluvial transport from distant mafic terrains, such as the Carajás Province or the Tracuateua Suite, with subsequent incorporation into Post-Barreiras deposits. The presence of subrounded quartz grains and mixed heavy mineral assemblages across multiple sites further supports periodic alluvial reworking. As proposed by Silva et al. (Reference Silva, Corrêa, Wright, Bomfim, Hendricks, Gavin and Muniz2021), such alluvial sediments can carry not only accessory minerals but also plant-essential nutrients (e.g., phosphorus, calcium, and exchangeable bases), contributing directly to the baseline fertility of ADEs.
Furthermore, the mineralogical differences between sites reveal variations in the degree of weathering and local pedogenic evolution. In Juruti, the occurrence of hematite associated with kaolinite indicates extreme laterization of the Cretaceous sandstones of the Alter do Chão Formation, a process that also produced the bauxite deposits exploited in the region (Lucas, Reference Lucas, Carvalho, Boulangé, Melfi and Lucas1997). In contrast, the soils of Bragança and Barcarena preserve resistant minerals such as zircon and ilmenite, suggesting a more recent input of detrital material from relatively unweathered crystalline sources, in agreement with previous studies by Silva (Reference Silva2010).
Anthropogenic modification
The mineralogical and microstructural features identified in the ADEs provide direct, though nonexclusive, evidence of anthropogenic modification superimposed on a preexisting mineral skeleton. Rather than forming entirely new mineral assemblages, human activities altered the chemical environment in ways that enhanced the reactivity, redistribution, and stabilization of both inherited and introduced phases (Costa, Reference Costa1991; Kern and Kämpf, Reference Kern and Kämpf2005; Silva et al., Reference Silva, Corrêa, Wright, Bomfim, Hendricks, Gavin and Muniz2021; Silva, Reference Silva2022).
A key implication concerns phosphorus-bearing phases: although phosphorus enrichment in ADEs cannot be attributed exclusively to human inputs, their stabilization and possible reprecipitation within the soil matrix may have been favored by phosphorus enrichment from bone and organic waste (Schmidt et al., Reference Schmidt, Goldberg, Heckenberger, Fausto, Franchetto, Watling and Lima2023). Human activity likely increased the availability and cycling of phosphorus, while pedogenic processes controlled its redistribution and stabilization.
Combustion-related processes provide a second line of evidence. Burning associated with habitation and waste disposal can promote the formation or transformation of iron oxides, contributing to the occurrence of hematite alongside naturally weathered phases. Elevated temperatures also modify soil structure, increase mineral surface reactivity, and facilitate the formation of organo-mineral associations. These effects enhance nutrient retention and contribute to the persistence of ADE horizons under conditions of intense tropical leaching.
Geoarchaeological significance
The mineral assemblages documented in this study support a refined hybrid model for ADE formation in which the four previously described processes interact as a coupled system rather than acting independently (Table 2). The key contribution of this model lies in explicitly linking mineral provenance to soil-forming mechanisms and nutrient dynamics.
Main minerals identified in the studied ADE sites and their interpreted origins and formation processes.a
a The assemblage integrates the four proposed processes in the hybrid ADE origin.
The system is initialized by geological inheritance and sedimentary input. Detrital minerals derived from local bedrock and transported alluvial sediments establish the baseline mineral composition of the soil. Some of these minerals, particularly apatite, represent primary sources of nutrients such as phosphorus and calcium that are released during weathering. The relative contribution of local versus exogenous material varies by site, depending on lithology and proximity to fluvial systems.
Pedogenic processes then act continuously on this inherited assemblage. Under humid tropical conditions, intense weathering, laterization, and leaching promote the formation of secondary minerals (e.g., kaolinite, iron oxides) and the redistribution of elements within the soil profile. These processes control the transformation of primary minerals, the formation of reactive surfaces, and the baseline capacity of the soil to retain or lose nutrients. In the absence of human activity, this system tends toward nutrient depletion despite ongoing mineral weathering.
Anthropogenic modification and episodic alluvial contributions alters this trajectory by acting on a preconditioned mineral system, enhancing the retention and persistence of nutrients supplied by natural sources, and changing the chemical boundary conditions of the soil. Inputs of organic matter, ash, bone material, and ceramics increase nutrient concentrations, raise pH locally, and introduce additional reactive surfaces. These changes enhance key retention mechanisms, including adsorption onto iron oxides, complexation with organic matter, and stabilization within organo-mineral aggregates.
The interaction between these processes produces a feedback system. Over time, this feedback leads to the formation of stable, nutrient-rich horizons that persist despite climatic conditions favoring leaching. The relative strength of each component varies across sites, explaining the observed mineralogical and geochemical variability among ADEs.
Uncertainties and limitations of the study
This study is subject to several limitations related to sampling, analytical resolution, and interpretative constraints. The small number of studied sites limits capture of spatial variability within ADEs, each of which can span hundreds of hectares with heterogeneous artifact distribution and soil properties; thus, the reported assemblages may not be representative. XRD detection limits (∼2–5 wt%) and lack of raw diffraction data precluded identification of accessory minerals observed by SEM-EDS and prevented quantitative phase analysis. Also, attribution of hematite is non-unique, because it can form through lateritic weathering or be enhanced by burning, while apatite may be detrital or influenced by bone waste. The hybrid model itself carries interpretive uncertainties. The four processes are assumed to interact as a coupled system, but their functional coupling is inferred from static mineralogical co-occurrence rather than demonstrated directly. Furthermore, the uniform application of the model across four mineralogically distinct sites does not account for how the relative weight of each process varies by context, nor does it predict site-specific signatures that would allow falsification. Consequently, the proposed interpretations are supported by consistent mineral associations but cannot be independently tested with the current dataset. Future studies should incorporate systematic spatial sampling, quantitative mineralogy, and isotopic tracers; specifically, apatite fission track analysis could distinguish detrital apatite sources and help quantify natural versus anthropogenic phosphorus contributions.
Conclusions
We propose a hybrid model for Amazonian dark earths (ADEs), in which different processes interact as a coupled system. To test this model, we analyzed whether the mineral assemblage records natural provenance inputs and how these interact with human activity. Using XRD and SEM-EDS, we found that bulk mineralogy (quartz and triclinic kaolinite) reflects deep tropical weathering, while accessory minerals reveal two natural sources: local bedrock inheritance and alluvial transport. Site-specific results include: hematite, anatase, zircon, and ilmenite at Terra Preta 2 from lateritic weathering of Alter do Chão sandstones, with anatase interpreted as a weathering product of detrital ilmenite; chromite at Ilha de Terra indicating alluvial input from distant mafic terrains; prismatic apatite at Jacarequara demonstrating detrital inheritance from Barreiras Formation sandstones, confirming that a portion of phosphorus is naturally derived; and zircon with ilmenite at Jabuti reflecting heavy mineral inheritance from the Tracuateua Intrusive Suite. Ceramic fragments and combustion-related iron oxides across all sites suggest human modification of an already enriched substrate.
Together, these findings support a four-process system in which geological inheritance and alluvial deposition supply mineral-bound nutrients, pedogenesis regulates their redistribution, and human activity enhances their long-term retention. This interpretation treats natural and human processes as functionally interdependent and generates testable predictions: sites with greater detrital apatite or alluvial heavy mineral content should display distinct phosphorus dynamics, while anthropogenic input intensity should control nutrient stabilization rather than total nutrient concentrations. Quantifying these relative contributions across the Amazon basin will require systematic spatial sampling, quantitative mineralogy, and isotopic tracers such as apatite fission track analysis.
Acknowledgments
The authors thank FAPESPA (Amazon Foundation for Research Support) and PIBICTI (Institutional Scholarship Program for Technological Development and Innovation) for granting a Scientific Initiation scholarship to ELG (grant no. 0000090010262024); Museu Paraense Emílio Goeldi (MPEG) for providing the facilities of its Soil Analysis and SEM Laboratory; Deeptech Iasauatec Amazon for obtaining and supplying the samples used in this research; CETEM for performing the XRD analyses; and the LCM from the Federal University of Pará for conducting the XRD interpretations and mineral identification.
Competing Interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: ELG reports equipment, drugs, or supplies were provided by Iasauatec Amazon. MCdM reports a relationship with Iasauatec Amazon that includes board membership and employment. The other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
