Brancier et al 2014 Zf G

B Zeitschrift für Geomorphologie, Supplementary Issue Published online February 2014 Fast Track Article A Micromorphological Assessment of Anthropogenic Features in Pre-Columbian French Guiana Dark Soils (FGDS): First Results Jeanne Brancier, Cécilia Cammas, Dominique Todisco and Eric Fouache with 17 figures and 5 tables Abstract. In order to document site formation processes at microscale and to characterise pre-Columbian French Guiana dark soils (FGDS), micromorphology was performed on three sites. For the first time it was possible (i) to microscopically characterise pre-Columbian Anthrosols in different physical contexts and (ii), to identify anthropogenic features associated with past human occupation. Microfeatures of the Holocene alluvial terrace of the lower Maroni River witnessed (i) several episodes of clay enrichment and/or redistribution, (ii) seasonal waterlogging, and (iii), post-depositional biological activity. Clay enrichment and organic matter inputs together with biological activity processes might have alternated, probably in relation to vegetal cover and/or anthropogenic activities. On top of the alluvial terrace, bioturbated dark layers are enriched in fine brown organic matter and charcoals. Cumulic soil development was favoured when successive sediment inputs due to episodic flooding and/or overland flow was possible (Chemin Saint Louis site). On a lateritic hill, under rainforest, at the MC87 ring-ditched mountain (Montagnes Couronnées or Crowned Mountain), microscale identification of yellowish unburnt oxic B horizon aggregates together with anthropogenic features related to fire such as charcoals and burnt soil fragments (rubefied and dark brown aggregates) stress that lateritic soil acted as support for activities in the enclosure, and as reworked material in the ditch. These components could result from clearance for settlement, agricultural management and cultivation, or domestic activities. The obtained results allow first comparisons to be drawn between pre-Columbian FGDS and Brazilian dark earths (BDE). Except a similarity in colour, the former is revealed to be less rich in anthropogenic components with an absence of phosphatic elements such as bones. Key words: Dark soils, Micromorphology, Anthrosols, Formation processes, Pre-Columbian, French Guiana 1 Introduction Archaeological research in tropical humid rainforest of Amazonia revealed black soils whose characteristics and formation have been strongly influenced by the effects of pre-Columbian human activities (Arroyo-Kalin 2008a). These Anthrosols (WRB soil classification 2006) are termed Terra Preta do Indio, or Amazonian Dark Earth (ADE). They are found in three kinds of pedological contexts: Oxisols (US Soil Taxonomy) (or Ferralsols, WRB soil classification 2006), Ultisols (Acrisols) and Inceptisols (Nitisols). During the 1960s, these Anthrosols were divided in two different types: (i) Terra Preta (TP), which are dark, rich in artefacts, with high concentrations of nutrients (such as phosphorus, nitrogen and calcium), large amounts of stable soil organic matter (Glaser et al. 2001), near neutral pH, due to remains of kitchen middens in housing areas and (ii), Terra Mulata (TM), that are remains of previous agricultural management (Sombroek 1966). Following Ruivo et al. (2003: 252) “inasmuch as Amazonian Dark Earth (ADE) are Anthrosols formed in the past, they are part of the archaeological record and hence © 2014 Gebrüder Borntraeger Verlagsbuchhandlung, Stuttgart, Germany DOI: 10.1127/0372-8854/2014/S-00159 eschweizerbart_xxx www.borntraeger-cramer.de 0372-8854/14/S-00159 $ 7.75 2 Jeanne Brancier et al. merit the attention of Geoarchaeology”. Although many chemical analyses were conducted on ADE, micromorphological studies are few (Lima et al. 2002, Ruivo et al. 2003, Schaefer et al. 2004, Arroyo-Kalin 2008a, 2008b). North of Brazil, in equatorial French Guiana, pre-Columbian occupational remains (e.g., pottery, charcoals, funerary urns) and structures (e.g., pits, ditches) are often found in association with dark-coloured and thick organic soils. Until now, such Anthrosols which contrast with surrounding non-archaeological soils were not precisely studied and hence not clearly defined as being Terra Preta or Terra Mulata. The main objective of the present paper is to use a geoarchaeological approach, and especially soil micromorphology, to assess human influence on the formation of French Guiana archaeological dark soils (FGDS). Such investigation, in addition with other geoarchaeological methods, appears essential to establish a framework for discussion and interpretation of past human impact on tropical landscapes (Ruivo et al. 2003). More specifically, this study aims to (i) microscopically characterise pre-Columbian FGDS, (ii) Fig. 1. Up right: location map of French Guiana in South America. Bottom left: zoom on French Guiana with location of studied archaeological sites: alluvial terrace of the Maroni River, Balaté (BAL) and Chemin Saint Louis (CSL), hilltop forest site, MC87. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 3 identify anthropogenic features associated with past human occupation and (iii), outline differences in Anthrosol features in different geomorphological contexts. 2 Pedo-geomorphological and archaeological contexts Pre-Columbian Anthrosols were studied on three selected open-air sites in different pedo-geomorphological and archaeological contexts (Fig. 1). The climate of the regions investigated is humid tropical with two dry seasons and two rainy seasons due to latitude fluctuations of the Intertropical Convergence Zone (Groussin 2001). Both sites were surveyed and then excavated by the french Institut National de Recherches Archéologiques Préventives (Inrap). 2.1. The Holocene Maroni River terrace: Chemin Saint Louis (CSL) and Balaté (BAL) Chemin Saint Louis and Balaté sites are located in western French Guiana, along the Maroni River, south of Saint-Laurent-du-Maroni, facing the Suriname border. The mean annual rainfall is about 2500 mm over the Maroni watershed and at Saint-Laurent-du-Maroni, where the mean annual air temperature is about 26.5 °C. The studied sites are situated on the top of the Fig. 2. Alluvial sites: CSL and BAL. A. DEM of the lower Maroni region with location of BAL and CSL at the south of the town of Saint-Laurent-du-Maroni. B. Aerial image with location of the archaeological sites: diagnostic areas (surrounded in white) and excavated areas (coloured in white). C. Schematic cross section of terraces along the lower Maroni River. eschweizerbart_xxx 4 Jeanne Brancier et al. Holocene terrace of the lower Maroni River next to the Balaté Creek (Fig. 2A and B). This lower alluvial terrace, termed T4, is observed in eastern Suriname and western French Guiana, and formed between 10,000 and 6,000 BP (De Boer 1972, Palvadeau 1999). It was identified and dated, as the other Quaternary terraces of the Maroni, according to its topographic position, its sedimentological features (facies, granulometry), the degree of cohesion and weathering of deposits, with very few 14C dates (De Boer 1972, Palvadeau 1999). The relative altitude of this lower terrace is generally about 5 – 6 m above the Maroni River and nowadays it is not flooded (by non-extreme floods) as indicated at the regional scale by the presence of several villages on its top (e.g., Pilima Pata, Nasson, Bossou) (Palvadeau 1999). The lateral extension of this terrace, perpendicular to the Maroni’s banks, is limited (≤ 1 km). It has a typical sequence of non-cohesive clayey sand (1.5 – 2.5 m, up to 6 m thick) overlaying the bedrock or, partly, the deposits from the older Pleistocene T3 terrace (Fig. 2C). On top of the T4 terrace, pre-Columbian Anthrosols of CSL and BAL are acidic and have a low fertility (Briand et al. in press, van den Bel et al. 2011, 2012). 2.1.1 Chemin Saint Louis (CSL) Chemin Saint Louis (Fig. 3) is a large multi-component black earth site on the top of a sandy levee associated with the T4 terrace as well as the lower backslope. An excavated area of about 5000 m2 was opened comprising both depositional contexts. This higher riverbank represented an attractive site for human settlement and it is likely that the first Amerindians settled by the Fig. 3. CSL excavation. A. Excavated area with soil texture variations cartography and location of the Logs sampled for micromorphological analyses. B. Schematic cross section of the southern part of the site with Log 1 location. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 5 beginning of the fifth millennium BC, just after the formation of the T4 terrace. Three occupations have been clearly defined during excavation and emphasized by thirty 14C dates: (i) a first phase, dated between c. 2500 and 2000 cal BC, corresponding to a very Early Ceramic Age settlement; (ii) a second phase, dated between c. 300 cal BC and 400 cal AD, with hitherto unknown ceramic series; and (iii), a third phase, dated between c. 900 and 1300 cal AD, featuring a Late Ceramic Age regional complex, likely contemporary with the Balaté site (van den Bel et al. 2011, 2012). 2.1.2 Balaté (BAL) The 20,000 m2 archaeological site of Balaté is situated on the lower T4 terrace at the confluence of the Balaté Creek (south) with the Maroni River (west and north) (Fig. 4). The site is located on a clayey sand spit (Pointe Balaté) on the right side (outer curve) of the Balaté Creek. This natural elevation was recently perturbed by industrial activities (especially a gravel factory) causing a loss of the site’s integrity with a conspicuous anthropoturbation of the uppermost part of the archaeological layer. Oral tradition of the inhabitants from the Arawak village of Balaté indicates an ancient Lokono occupation (van den Bel 2008). During excavation conducted by Inrap, a funerary zone with urns and diffuse spreading of fragmented ceramics due to ancient Amerindian occupation was found. The site is 14C dated at c. 1100 –1300 cal AD. Dates and ceramic chronology indicate a Late Ceramic Age (cal. AD 900-1500) occupation featuring Barbakoeba ceramics. In addition, the site may have known a younger occupation, possibly c. 1500 cal AD (Briand et al. in press). Fig. 4. Archaeological map of the BAL excavation and location of the Logs sampled for micromorphological analyses. eschweizerbart_xxx 6 Jeanne Brancier et al. 2.2 Ring ditched Mountain site in lateritic hill context: MC87 Investigations were conducted in the northeast of French Guiana, in the Approuague watershed and Regina region, where mean annual rainfall is about 3800 mm and mean annual air temperature is about 27 °C. Two ring ditched mountains (Montagnes Couronnées) were studied in the interdisciplinary COUAC Project involving the Inrap, Institut National de la Recherche Agronomique (Inra), Centre International en Recherche Agronomique pour le Développement (Cirad), UMR Ecologie des Forêts de Guyane (EcoFoG) and Centre National de la Recherche Scientifique (Cnrs) (Fig. 5). These sites are located in lateritic hills covered by equatorial rainforest and underlain by a Palaeoproterozoic bedrock (migmatites, gneiss, paragranite) belong- Fig. 5. Ring ditched mountain site: MC87. A. Digital Elevation Model (DEM) of MC87 area. B. Schematic map of MC87 and location of logs sampled for micromorphological analyses.. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 7 ing to the Guiana shield. Ring ditched mountains were detected by pedestrian survey and appear as a large U-shaped ditch (~2 m depth by 2 m across) surrounding the top of a hill. In French Guiana about forty sites were identified. Some sites were also found in Brazil and Suriname suggesting a wide geographical distribution. In most cases, 14C dating indicates that ditches were built around the first century AD (Mazière 1997, Versteeg 1981), but pottery remains indicate that various ethnic groups or cultures may have occupied the sites (Mestre et al. 2013). The latter may have been places for villages or necropoleis such as the Pointe-Morne site, near the Brazilian border (Inrap excavation). The presence of the ditch may suggest a defensive role in case of settlements, or a symbolic role for necropoleis such as a separation between the living and the dead. Until now, no clear archaeological evidence supports the hypothesis concerning the presence of a fence together with the ditch. Such sites may have been a form of land acquisition by numerous and structured pre-Columbian societies (Mestre et al. 2013). The large amount of energy required to excavate the ditch, technical constraints associated with building such a feature, and managing the soil and topography, suggest a long duration settlement. Located at 98 m above sea level on top of a lateritic hill, the MC87 site is considered to be representative of this type of settlement. Radiocarbon dates indicate occupations between c. 500 and 1100 cal AD (Jérémie & Dambrine 2011). The physical context is favourable to the development of Oxisols, which are acidic, nutrient-poor, with low fertility. 3 Method 3.1 Micromorphology Micromorphology concerns the microscopic identification, description and interpretation of fabrics, features and components in soils (Bullock et al. 1985, Fitzpatrick 1993). In archaeological contexts, soil micromorphology is an essential tool for studying site-formation processes (Courty et al. 1989, Courty 2001, Cammas & Wattez 2009). However, its application to pre-Columbian sites of French Guiana has not been documented while it has been used for diverse cultural periods to resolve various on-site geoarchaeological problems (Courty et al. 1989, Cammas & Wattez 2009). In order to perform a detailed microstratigraphic and microfacies analysis, undisturbed and oriented soil samples were taken from the three selected pre-Columbian sites. Collected samples were manufactured for 30 μm thick large-format thin sections (13 × 6 cm) using standard methods (Guillore 1983). The samples were oven-dried, impregnated with polyester resin, and prepared in the Inrap Unit of Micromorphology of AgroParisTech (Thiverval-Grignon, France). Thin sections were studied using a Nikon E200 Pol petrographic microscope (20 to 400×) in plane-polarized light (PPL), cross-polarized light (XPL), and oblique incident light (OIL). Micromorphological descriptions were mainly based on Fitzpatrick (1993), Bullock et al. (1985), and Stoops et al. (2010). For each thin section, micromorphological characteristics were investigated in a qualitative way with visual estimates for mineral components, pedofeatures and anthropogenic components. Each thin section was divided in 1 cm thick horizontal lines with the help of a grid. For each line, abundance of anthropogenic components were visueschweizerbart_xxx 8 Jeanne Brancier et al. ally estimated in surface percentage: charcoals, burnt brown to black aggregates, burnt rubefied aggregates, reworked yellow non-burnt aggregates, and pottery fragments. Charcoal percentage size classes from silt to gravel are based on Stoops (2003). Microfeatures were interpreted following Courty et al. (1989), Goldberg & Macphail (2006), Stoops et al. (2010), Cammas & Wattez (2009), and by comparison with various archaeological, pedological and experimental thin sections of the laboratory. Comparison with Terra Preta was mainly made using works of Lima et al. (2002) and Arroyo-Kalin (2008a, 2008b). 3.2 Sampling 3.2.1 Alluvial terrace sites: CSL and BAL At CSL two sequences were sampled: (i) Log 1, 90 cm thick (six thin sections), at the southeast excavation limit, and (ii) Log 2, 90 cm thick (six thin sections), at the northeast limit of the site Fig. 6. Schematic profiles, field characteristics of Logs 1 (90 cm thick) and 2 (90 cm thick), and location of the micromorphological samples at CSL. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 9 (Figs. 3, 6). At BAL, three sequences were sampled: (i) Log 1, 60 cm thick (three thin sections), at about 30 m southwest of the Maroni River and about 65 m north of the Balaté Creek, (ii) Log 2, 40 cm thick (seven thin sections), at 60 m south/southwest of the Maroni and 20 m north of the Balaté Creek, (iii) Log 3, 60 cm thick (seven thin sections), at 30 m west of the Maroni and 20 m north of the Balaté Creek (Figs. 4, 7). The latter is situated in the vicinity of a zone where funerary urns were excavated. Except for two profiles, all logs were sampled from the sub-surface to the non-organic lower horizon found at 70 cm depth in CSL, 50 cm depth in BAL Log 1, and at 40 cm, under 40 cm of modern backfill, in Logs 2 and 3 of BAL. For the two last profiles, sampling was performed just under the modern backfill of gravels to the non-organic lower horizon. The used zero level was under the 40 cm of modern backfill (Fig. 7). 3.2.2 Ring ditched mountain: MC87 Eight survey pits, three of which were sampled. Logs 3 and 4, 30 cm and 40 cm thick respectively (three thin sections for each profile), were sampled in the enclosure surrounded by the ditch where archaeological remains were found indicating past occupation (Figs. 5B, 8). Log 1, 90 cm thick (eight thin sections), was sampled in the ditch from the sub-surface to the lower limit of digging, around 1 m depth. This log was located near the crest line of the hill (Figs. 5B, 8). Fig. 7. Schematic profiles, field characteristics of Logs 1 (60 cm thick) and 2 (90 cm thick), and location of the micromorphological samples at BAL. eschweizerbart_xxx 10 Jeanne Brancier et al. Fig. 8. Schematic profiles, field characteristics of Logs 1 (90 cm thick), 3 (30 cm thick) and 4 (40 cm thick), and location of the micromorphological samples at MC87. 4 Results and Discussion 4.1 The microfacies of the alluvial terraces sites 4.1.1 The alluvial terrace The sediments of the terrace are mainly composed of unsorted quartz grains (from very fine to very coarse sand) in a yellowish silty-clay matrix (Tables 1 & 2). The grade of pedality is variable (low to moderately developed) and porosity mostly results from both floral and faunal turbation. Soil mixing by animals is revealed by abundant excremental micro-pellets and by the presence of some excremental nodules with oriented and well-sorted quartz grains. Such biological features could be faeces of geophageous fauna (Stoops et al. 2010). Abundant argillic coatings indicate active eluviation especially in the form of microlaminated coatings (Fig. 9A–C). Such eschweizerbart_xxx Table 1. Main micromorphological features of CSL logs. Relative abundance a Qz Qz Qz Qz Qz Qz       Silt Silt Silt Silt Silt Silt       Inf Ench/Coll Ench/Coll Ench/Coll Ench/Coll Ench/Coll Ench/Coll       Igm Igm Igm Igm Igm Pg Ch Ch Ch Ch Ch Ch       S S S S S S U U U U U U Qz Qz Qz Qz Qz Qz       Silt Silt Silt Silt Silt Silt       Ench/Coll Ench/Coll Ench/Coll Ench/Coll Ench/Coll Ench/Coll       CC Relative abundance a Excrement type U U U U U U Type Relative abundance a S S S S S W Nodule Relative abundance a Fine Fraction type      () () () () () () Ml/Dc  RN RN RN RN RN RN       FN/RN FN/RN FN/RN FN/RN FN/RN FN/RN       Descriptive Criteria Relative abundance a Ch En En Ch Ge Ge Type Coarse Fraction Type Igm Igm Igm Igm Igm Sp/Ma Porosity a Sorting Textural Grade of Pedality Biological C/f related distribution Depth (cm) Horizon/Layer Main pedofeatures Microstructure Type eschweizerbart_xxx Log 1 SLO 1 Modern deposit 20-32 SLO 2 Transition/Anthrosol 32-41 SLO 3 Anthrosol 42-53 SLO 4 Anthrosol 54-64 SLO 5H Gradual Limit 65-74 Alluvial Terrace 75-82 SLO 5B Log 2 SLO 6 Anthrosol 25-36 SLO 7 Anthrosol 36-45 SLO 8H Anthrosol 45-54 SLO 8B Anthrosol 54-64 SLO 9 Gradual Limit 64-74 Alluvial Terrace 75-88 SLO 10 Mineral Components () () () () () () 11 a = Relative abundance in thin section: () = absent or practically absent;  = low / few / poorly-developed;  = medium / frequent / moderately-developed;  = high / abundant / well-developed. Microfeature description (After Bullock et al. 1985) Microstructure Type = Co : Complex; Igm : Intergrain microaggregate; Ma : Massive; Pg : Pellicular grain; Sp : Spongy. Related Distribution = C/f: Coarse/fine; Ch: Chitonic; En: Enaulic; Ge: Gefuric. Grade of Pedality (aggregation) = W: Weakly developed; M: Moderately developed; S: Strongly developed. Sorting = U: Unsorted. Coarse Fraction Type = Qz: Quartz. Biological/Excrement Type = Coll: Collembolan; Ench: Enchytraeids; Inf: Infillings. Textural/Type = CC: Clay Coating; Dc: Dusty clay; Ml: Microlaminated. Type Nodule = FN: Ferruginous Nodule; RN: Rock Nodule. Anthropogenic Features in Pre-Columbian French Guiana Dark Soils Thin section Microstructure Table 2. Main micromorphological features of BAL logs. U Qz  Silt/Cl  Anthrosol 30-43 Ma Ge () W BAL 1-3 Alluvial terrace 44-58 Ma Ge U Qz BAL 2-1 Backfill 0-12 (+40) a Ma/Ag Ch/Ge  M W U Qz  Silt/Cl  BAL 2-2 Anthrosol 7-19 (+40) Ma/Ag Cl/En () W U Qz  Silt/Cl  BAL 2-4H Anthrosol 15-27 (+40) Ma Cl/En () W U Qz  Silt/Cl  Ma Cl/En   BAL 2-4B Anthrosol 20-33 (+40) M U Qz BAL 2-5 Alluvial terrace 35-46 (+40) Chm/Ma Cl/En () M U Qz BAL 3-1H BAL 3-1B BAL 3-2 Gradual limit Anthrosol Anthrosol 0-12 (+40) 6-17 (+40) 13-24 (+40) S M M BAL 3-3 BAL 3-4 BAL 3-5 Anthrosol Gradual Limit Alluvial terrace 25-37 (+40) 37-48 (+40) 49-62 (+40) a Igm Ch  Ma/Igm Ch/Po () Ma En () Ma Cl/En () Ma Cl/En  Ma Cl/En () W W M  Silt/Cl   Silt/Cl   Silt/Cl  U Qz  U Qz  U Qz  U Qz  U Qz  U Qz  Silt Silt Silt Silt Silt Silt       Ench/Coll Ench/Coll Ench/Coll Ench/Coll Geo/Inf   CC Lc  CLi/CC Ml/Lc  CC/Pa    HC/CC () CC Ml/Lc/Dc CC Ml/Lc/Dc () () ()  RN RN CC Ml/Lc/Dc RN  RN () () () RN RN/NF RN   () ()     ()  = + 40 cm = thickness of overlying backfills relative abundance in thin section: () = absent or practically absent;  = low / few / poorly-developed;  = medium / frequent / moderately-developed;  = high / abundant / well-developed. Microfeature description (After Bullock et al. 1985) Microstructure Type = Ag: Aggregated; Chm: Chamber; Co : Complex; Igm: Intergrain microaggregate; Ma: Massive. Related Distribution = C/f: Coarse/fine; Ch: Chitonic; Cl: Close; En: Enaulic; Ge: Gefuric; Po: Porphyric. Grade of Pedality (aggregation) = W: Weakly developed; M: Moderately developed; S: Strongly developed. Sorting = U : Unsorted. Coarse Fraction Type = Qz : Quartz. Fine Fraction Type = Cl: Clay Biological/Excrement Type = Coll: Collembolan; Ench: Enchytraeids; Geo: Geophageous; Inf: Infillings. Textural/Type = CLi: Clay Infills; CC: Clay Coating; Cr: Crescent; Dc: Dusty clay; HC: Hypocoating; Lc: Limpid clay; Ml: Microlaminated; Pa: Papules. Nodule/Type = FN: Ferruginous Nodule; RN: Rock Nodule. b= RN () () Ench/Coll  Ench/Coll  Ench/Coll   FN/RN Relative abundance b Cr/Lc Type CC  RN RN RN RN/NF          Jeanne Brancier et al. eschweizerbart_xxx BAL 1-2 Ench/Coll Relative abundance b U Qz  Silt/Cl  Nodule Descriptive Criteria Sorting M Textural Type Grade of Pedality  Biological Relative abundance b Porosity b Ch Main pedofeatures Excrement type C/f Related distribution Co Relative abundance b Microstructure Type 15-30 Fine Fraction type Depth (cm) Anthrosol Relative abnudance b Horizon/Layer BAL 1-1 Coarse Fraction Types Thin Section Mineral Components 12 Microstructure Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 13 alternating limpid and dusty clay microlayers may be related to changes in vegetal covers as observed in other climatic contexts. Limpid clay microlaminations generally refer to dense vegetal cover and dusty ones to clearance (Goldberg & Macphail 2006, Stoops et al. 2010). The location of such coatings just under the Anthrosol could thus suggest events of clearance due to human occupation. Redoxic features such as ferruginous nodules or fine red speckles in the matrix indicate that soils were seasonally waterlogged. Charcoals are very few and plant remains fragmentation is typical of a redoxic environment (Briand et al. in press). In thin sections, little difference is noticed between the alluvial terrace in BAL and CSL: only a little more biological activity as well as slightly less clay in CSL. 4.1.2 The archaeological layers of CSL In all the dark archaeological layers of CSL (Table 1), the coarse fraction is mostly composed of unsorted quartz grains (size similar to the alluvial terrace), in addition to rare minerals such as hornblende or andalousite (Fig. 9D). In the two profiles, ~ 5 % of quartz grains are identified as runiquartz, i.e. weathered quartz with cracks filled with iron and aluminium (Eswaran & Sys 1975). Runiquartz is very common in laterites and in most Oxisols in which the material filling cracks is usually different from the enclosing matrix. At CSL runiquartz are likely inherited from lateritic material coming from the Maroni River watershed. Distribution of the matrix around sand grains and in microaggregated form results from pedological activity. The microaggregation is related to mesofaunal activities especially enchytraeids and/or collembola (Fig. 9F). Some planar voids suggest compaction following the aggregation, mostly by faunal and floral reworking. The presence of frequent large chambers of irregular form and channels is likely related to megafaunal activity such as anecic and/or endogeic earthworms (Stoops et al. 2010). In Log 1 (82 cm thick; Fig. 6), micromorphology clearly shows that the archaeological strata (between 33 –73 cm depth) are associated with cumulic soil development. The latter results from fine successive alluvial and/or overland flow sediment inputs (mm to cm thick) alternating with episodes of human-induced organic (plant) inputs and bioturbation (van den Bel et al. 2011). Charcoal counts emphasise a clearly increasing trend in the archaeological layers (Fig. 10). Two phases corresponding to two successive archaeological Anthrosols are identified. The first phase shows a maximum charcoal percentage of ~ 18 % between 54 and 74 cm (SLO 4, SLO 5 H). In the second one the maximum charcoal percentage is ~ 30 % to 35 % between 32 and 54 cm (SLO 2, SLO 3). Each occupation phase seems homogeneous. These observations point out that charcoal input may have been regular, suggesting that human activities did not change during each period. In the area around Log 1, ceramics were scarce. In Log 2 (87 cm thick; Fig. 6), in the archaeological strata (between 32 –77 cm depth), charcoal percentage decreases from top (~ 30 %) to bottom (~ 8 %) with some variations (Fig. 10). At microscopic scale, vertical distribution of charcoals and pottery sherds show two different parts, (i) an older one with ~ 10 –12 % charcoals and pottery sherds (50 – 80 cm, SLO 8 B, SLO 9, SLO 10) and (ii), a more recent one with ~ 18 % charcoals (25 – 50 cm, SLO 6, SLO 7, SLO 8 H). This succession may be linked to changing activities with time. However, Log 2 appears less stratified than Log 1, and with more biological features. These facts as well as the presence of eschweizerbart_xxx 14 Jeanne Brancier et al. Fig. 9. Microphotographs of the alluvial sites thin sections. A. Limpid illuvial clay coating with extinction bands typical for kaolinitic clay (XPL), alluvial terrace, Log 3, BAL (BAL 3-5). B. Microlaminated coating in a channel with mostly dusty lamination and few limpid clay laminations (PPL), alluvial terrace, Log 2, BAL (BAL 2-5). C. Microlaminated coating with dusty and limpid clay (PPL), alluvial terrace, Log 1 (SLO5 B, depth: 80 cm), CSL. D. Loose fine material between quartz grains. Micro-transition in the Anthrosol (PPL), Log 1, CSL (SLO 3). E. Typical faunal excrement: quartz grains and the matrix are curved (PPL), Log 1, BAL (BAL 1-2). F. Typical aggregation of fine material by biological mesofauna activity in the archaeological layer (PPL), Log 1, CSL (SLO 4). the pottery sherds in the alluvial terrace deposits (mostly SLO 10) suggest more perturbations than in Log 1. The difference between the two logs could also be the result of two different activity areas. Three kinds of pottery were identified at microscale with most often quartz grains as mineral temper but also hornblende, andalousite, mica, or micro-charcoals (Table 5). Two sherds from Log 2 (SLO 9, 66 cm depth and SLO 10, 80 cm depth) show a microlaminated matrix, which suggests malaxing processes, while the other sherds (from SLO 6, 27 cm depth, SLO 8 B, 55 cm and 60 cm depth, and SLO 9, 72 cm depth) show a homogeneous matrix suggesting more mixed material. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 15 Fig. 10. Microscopic charcoal percentages per size classes, Logs 1 and 2, CSL. 4.1.3 The archaeological layer of BAL In all the dark archaeological layers of BAL (Log 1, between 19 – 48 cm; Log 2, between 6 – 27 cm; Log 3, between 6 – 49 cm), the coarse fraction is similar to CSL (Table 2) with dominant quartz, rare hornblende as well as possibly inherited runiquartz. Microstructures varied from massive eschweizerbart_xxx 16 Jeanne Brancier et al. Fig. 11. Microscopic charcoal percentages per size classes Logs 1, 2 and 3, BAL. to more developed aggregation towards the top (BAL 2-2, BAL 3-1 H, BAL 3-1 B) of the archaeological layers. There is a weak to medium grade of pedality. Pedality is mostly related to biological aggregation by mesofauna (i.e., enchytraeids and/or collembola) or locally to geophageous fauna (BAL 2-4 H, Fig. 9E) with few chamber voids. Few clay coatings were observed in all Logs 1 and 2 (Fig. 7). In Log 3, they were only identified in the lower part of the Anthrosol (BAL 3-3). Clay coatings were limpid in Log 1, while microlaminated (dusty/limpid) in Logs 2 and 3. The origin of these variations could be linked to ancient occupation or modern human reworking. In Log 2, the limit between the alluvial terrace and the dark archaeological layer presented some V-shaped microincisions. In the lower part of the Anthrosol, close to the limit with the terrace, eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 17 removed sub-angular aggregates derived from the alluvial terrace are observed (BAL 2-4 B). These characteristics indicate ancient mechanical erosion phases which might result from anthropogenic action such as intentional scratching of the soil (Macphail et al. 1990). Charcoal counting emphasises an enrichment in charcoals in the archaeological layer from ~ 10 to 18 % (Fig. 11). In Log 1 (60 cm) charcoals are ~ 8 and 10 %, with a peak at 37/38 cm depth (~ 15 %) (BAL 1-2). Red iron impregnations around some charcoals, mostly in Logs 2 and 3, stress redoxic conditions. In addition, two kinds of pottery were identified at microscale with plant (BAL 2-4 H) or mineral (BAL 3-1 B) temper (Table 5). In Logs 2 (41 cm) and 3 (60 cm), nearer from the Balaté Creek than Log 1, the charcoal percentage tends to decrease from top to bottom of the Anthrosol. In Log 2, two peaks are observed at ~ 18 %, one between 7 and 8 cm (just under the modern backfill) and a second between 11 and 15 cm depth. In Log 3, three peaks appear, the first one at 14 cm depth (~ 18 %), just under the pottery, the second one at 28 cm depth (~ 13 %), and the last one at 40 cm (~ 10 %) (Fig. 11). 4.2 The microfacies of the ring ditched mountain: MC87 4.2.1 Components and microstratigraphy 4.2.1.1 Enclosure In all thin sections, few quartz grains (very fine to medium sand) and frequent lateritic rock fragments or granules are embedded in a clayey matrix showing a low birefringence (Table 3). Inside the enclosure, Log 3 (30 cm thick, Fig. 8) is largely bioturbated in all the profile with a complex, massive to aggregated microstructure. Some aggregated areas are cut by planar voids likely resulting from compaction due to roots (MC87 3-1 B). Channels with infills of mesofaunal faecal pellets are abundant in the entire profile in addition to some regular chambers related to anecic and/or endogeic activity (Stoops et al. 2010). The coarse fraction is composed of quartz grains (~ 10 %), frequent lithorelics and ferruginous fragments of different sizes, from mm to cm. Organic remains include few fresh plant remains (e.g., rootlets and tissues). Frequent papules and few clay coatings are observed in all the profile of Log 3 (Table 3). Clay coatings and infillings are mostly composed of non-laminated limpid clay. Few exhibit a microlaminated matrix where limpid clay alternates with darker dusty clay laminations. These coatings indicate rhythmic illuviation and suggest possible variations in soil pH in relation to natural and/or intentional fires (i.e., charcoals and ashes inputs) or anthropogenic activities, which may have induced bone inputs. Microlaminated coatings with speckled clay could also suggest bare soil whereas limpid clay coatings could be related to forest soil (Goldberg & Macphail 2006, Stoops et al. 2010). In this log, most of the clay coatings were broken by faunal activity and root compaction indicating alternating phases of illuviation and bioturbation in the entire profile. In Log 4 (40 cm thick, Fig. 8) the coarse fraction is composed of the same components as in Log 3. Microscopic analysis reveals a composite, granular, crumb to massive faunal-turbated microstructure composed of irregularly shaped peds in the entire profile. The latter result from the coalescence of subangular to rounded aggregates of different sizes which may have been produced by earthworms (Stoops et al. 2010). Microaggregates have been affected by root comeschweizerbart_xxx 18 Table 3. Main micromorphological features of MC87 logs. Infilling 29-40 Co Cxp Infilling 46-53 Co Cxp/Chm MC 87 1-4B Infilling 54-60 Ma/Igm Chm eschweizerbart_xxx S U Qz S U Qz S U Qz S U Qz S U Qz S U Qz       MC 87 1-5a Infilling 66-71 Ma Chm MC 87 1-5b Infilling 72-82 Ma Chm  MC 87 1-6 Infilling 82-95 Ma Chm ()  n/m Enclosure MC 87 3-1H Anthrosol 5-13 Ma/Chm Chm U Qz Gradual limit 14-20 Ma/Chm Chm  S MC 87 3-1B M/S U Qz MC 87 3-2 B-Horizon 22-30 Ma Chm () M U Qz MC 87 4-1 Anthrosol 5-15 Ma/Fi Chm S U Qz MC 87 4-2 Anthrosol 15-24 Ma/Chm Chm S U Qz MC 87 4-3 Gradual Limit 29-40 Ma/Chm Chm S U Qz              Clay Clay Clay Clay Clay       Clay  Clay  Clay Clay Clay Clay Clay      Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf Ench/Coll/Inf  RN/FN  RN/FN  RN/FN      CC Ml/Lc CC/Pa Ml/Lc/Dc CC/Pa Ml/Lc/Dc CC/Pa Ml/Lc/Dc   CC Ml/Lc  () RN/FN  () RN/FN     RN/FN  RN/FN  RN/FN  RN/FN () RN/FN () RN/FN  = relative abundance in thin section: () = absent or practically absent;  = low / few / poorly-developed;  = medium / frequent / moderately-developed;  = high / abundant / well-developed. Microfeature description (After Bullock et al. 1985) Microstructure Type = Chm: Chamber; Co: Complex; Fi: Fissural; Igm: Intergrain microaggregate; Ma: Massive. Void Type = Chm: Chambers, Cxp: Complex packing. Grade of Pedality (aggregation) = W: Weakly developed; M: Moderately developed; S: Strongly developed. Sorting = U: Unsorted. Coarse Fraction Type = Qz: Quartz. Biological/Excrement Type = Coll: Collembolan; Ench: Enchytraeids; Inf: Infillings. Textural/Type = CC: Clay Coating; Cr: Crescent; Dc: Dusty clay; Lc: Limpid clay (orange); Ml: Microlaminated; Pa: Papules. Nodule/Type = FN: Ferruginous Nodule; RN : Rock Nodule. a  RN/FN ()     RN/FN    Relative abundance a  Clay Type Clay  Nodule Relative abundance a  Qz Descriptive Criteria Qz U Type U S Textural Relative abundance a S  Excrement type  Main pedofeatures Biological RN/FN      Jeanne Brancier et al. MC 87 1-3 MC 87 1-4H Relative abundance a Cxp Fine Fraction type Cxp Co Relative abundance a Co 12-24 Coarse Fraction type 1-12 Infilling Sorting Infilling MC 87 1-2 Ditch Grade of Pedality Void Type MC 87 1-1 Porosity a Mineral Components C/f Related distribution Microstructure Type Depth (cm) Horizon/Layer Thin Section Microstructure Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 19 paction, which caused the development of broken and angular peds separated by planar voids following the root system (MC87 4-3). At the bottom, the microstructure is mostly massive with scarce clay coatings, some chambers and channels in which faecal pellets of enchytreids and/or collembola are found (MC87 4-3). The upper part of the basal unit is bioturbated (MC87 4-2bis). Such characteristics and the intensity of the bioturbation suggest ancient soil surface/ sub-surface in situ reworked. This horizon is overlain by a non-homogeneous Anthrosol including gravels and pebbles (MC87 4-2bis). In this horizon, the massive microstructure could indicate human-induced reworking in addition to possible sediment inputs by overland flow/ colluviation. The upper part of Log 4 corresponds to the actual bioturbated, sub-surficial and surficial horizons enriched in organic matter. The geomorphological situation of Logs 3 and 4 at the top of the lateritic hill and under rainforest suggests relative soil stability with limited natural erosion. Similar kinds of clay coatings were found in Logs 3 and 4 but their location differs in the profiles. In Log 3, clay coatings and papules were observed in the entire profile. Towards the top of the log, clay coating abundance decreases together with an increase of papules. These characteristics indicate that in Log 3 the actual surficial horizon developed on a clay-enriched horizon. In Log 4, clay coatings are scarce and appear at the bottom of the profile (MC87 4-3), indicating that the clay-enriched horizon is likely located lower in this profile. Differences between the two logs suggest that Log 3 may have been reworked/eroded (i.e., rejuvenated) and that Log 4 may have been better preserved in relation to sediment inputs. Microtopography and spatial variation of human activities might have also played a role in profile differentiation. 4.2.1.2 Ditch In the field, the top of the ditch infilling was at 1 m depth, with ~ 1.5 m of sediment thickness, indicating that the ditch was 2.5 m deep when it was excavated. Thin sections from Log 1 (Fig. 8) reveal that microstructure type is mainly related to differential intensity of bioturbation (mostly mesofauna), and to water content of the sediment, allowing identification of rhythmic sedimentary inputs. At the bottom end of the infilling (MC87 1-5 a, 1-5 b, 6), a massive microstructure with interconnected voids suggests sediment compaction and episodes of water saturation during and/or after deposition. The sediment may be related to fine muddy colluvium without excluding some human inputs. Charcoals (~ 30 %) are of gravel-size, meaning that biological activity/comminution did not have time to develop and that burial was likely rapid. From ~ 50 cm depth up to the surface (MC87 1-4 H, MC87 1-3, MC87 1-2, MC87 1-1), frequent packing voids stress rapid piling up of peds, and suggest a backfill deposit and/or collapse of earth mounds located close to the ditch. The higher degree of pedality, with less aggregate coalescence, could be related to dry sediment inputs and/or limited waterlogging. At 15 cm depth (MC87 1-2), the microstructure is more massive, with sub-horizontal planar voids, suggesting compaction of the clayey matrix and alternating phases of wetting/drying. Because the clayey matrix shows limited shrinking and that the sub-horizontal orientation of voids seems to exclude natural processes, sediment compaction may be related, at least partly, to human activities (e.g., trampling). eschweizerbart_xxx 20 Jeanne Brancier et al. 4.2.2 Vertical dynamic of human related components 4.2.2.1 Pottery and charcoals Only one pottery sherd is found in the three profiles (Log 4, MC87 4-1, ~ 10 cm depth, at the limit between surficial horizon and the Anthrosol horizon, Fig. 7). Residual aggregates in the sherd, which have not been homogenized, indicate mixing of the sediment (Table 5). Grey to whitish colour and mineral characteristics, such as the fact that the clay is still birefringent, suggest heating at low temperature (Cammas & Wattez 2009). A very thin red border around the fragment could also indicate a slip. Charcoals are ubiquitous in the archaeological layers (Fig. 12). In Log 3 charcoal percentages reach 20 % at 14 cm, and remains ~15 % to the bottom of the sequence at 28 cm depth. In Log 4, from the top to 25 cm depth, they vary between 30 % (9 cm) and 50 % (15 cm), and between 10 and 20 % in the lower 10 cm of the sequence. A peak of 55 % is observed at ~ 18 cm depth. In Log 1, a peak of charcoals at 35 % is visible at 60 cm depth. Charcoals vary between 20 and 25 % above 60 cm, and between 25 and 35 % below this depth. 4.2.2.2 Aggregates In all profiles, three kinds of massive aggregates (0.2 – 2 mm) with rounded shape and few phytoliths (~ 5 %) were identified at microscale (Fig. 13). Yellowish aggregates have a lighter colour Fig. 12. Microscopic charcoal percentages and aggregate types per size classes, Logs 1, 3 and 4, MC87. eschweizerbart_xxx Log 1 SLO 1 SLO 2 SLO 3 SLO 4 SLO 5H SLO 5B Log 2 SLO 6 SLO 7 SLO 8H SLO 8B SLO 9 SLO 10 Ditch MC87 1-1 MC87 1-2 MC87 1-3 MC87 1-4H MC87 1-4B MC87 1-5a MC87 1-5b MC87 1-6 Enclosure MC87 3-1H MC87 3-1B MC87 3-2 MC 87 4-1 MC87 4-2 MC87 4-3 Phytolithb Gravel-size Charcoalb Balaté (BAL) 15-30 () 30-43 () 44-58 () 0-12 (+40) a () 7-19 (+40) () 15-27 (+40)  20-33 (+40) () 35-46 (+40) () 0-12 (+40) () 6-17 (+40)  13-24 (+40)  25-37 (+40) () 37-48 (+40) () 49-62 (+40) () Chemin-saint-louis (CSL) Rubefied Aggregateb Modern deposit Transition/Anthrosol Alluvial terrace Backfill Anthrosol Anthrosol Anthrosol Alluvial terrace Gradual limit Anthrosol Anthrosol Anthrosol Gradual limit Alluvial terrace Potteryb Horizon Layer BAL 1-1 BAL 1-2 BAL 1-3 BAL 2-1 BAL 2-2 BAL 2-4H BAL 2-4B BAL 2-5 BAL 3-1H BAL 3-1B BAL 3-2 BAL 3-3 BAL 3-4 BAL 3-5 Depth (cm) Thin Section Table 4. Main micromorphological features related or possibly associated to to human activities in the Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 21 three studied sites. () () () () () () () () () ()  () () ()               () ()  () () () () () () ()   () () Sub-surface Anthrosol Anthrosol Anthrosol Gradual limit Alluvial terrace 20-32 32-41 42-53 54-64 65-74 75-82 () ()  () () () ()  () ()  ()       ()  () () () () Sub-surface Anthrosol Anthrosol Anthrosol Gradual limit Alluvial terrace 25-36 36-45 45-54 54-64 64-74 75-88 MC87 () ()      () () () ()         () () () () () Infilling Infilling Infilling Infilling Infilling Infilling Infilling Infilling 1-12 12-24 29-40 46-53 54-60 66-71 72-82 82-95 () () () () () () () () Anthrosol Gradual limit B-Horizon Anthrosol Anthrosol Gradual limit 5-13 14-20 21-30 5-15 15-25 30-40 () () ()  () () Modified after Arroyo-Kalin 2008 a BAL 2-1 to BAL 3-5 : + 40 cm thickness of overlying backfills b = relative abundance in thin section: () = absent or practically absent  = low / few / poorly-developed  = medium / frequent / moderately-developed  = high / abundant / well-developed n/m = not measured eschweizerbart_xxx n/m ()  () ()     () () ()  () () 22 Table 5. Main micromorphological features of pottery sherds in the three studied sites. Thin Section Dept (cm) Size (cm) Temper Colour Matrix Mineralogy Organic fragments Interpretations Quartz grains (sand) Mixing Mixing Chemin Saint Louis (CSL) Log 1 42 1 x 0.6 Mineral: Sand Orange to brown with darkish impregnations Massive clayey 27 1 x 0.4 Mineral Black, grey, brown Massive to aggregateclayey Quartz grains SLO 8 B 55 0.5 (+ fragts) Mineral Reddish-orange Massive clayey Green minerals (hornbende) white (andalousite), and quartz grains Micro-charcoals Mixing SLO 8 B 60 1 x 0.8 Mineral Reddish-orange Massive clayey Green minerals (hornbende), white (andalousite), and quartz grains Micro-charcoals Mixing SLO 3 Log 2 SLO 6 66 0.2 Mineral: Sand Dark-brown Massive clayey SLO 9 72 0.5 x 0.5 Mineral: Sand Light yellowish Massive clayey Sequined mica, quartz grains SLO 10 80 2x1 Mineral: Sand Grey with reddish-orange microlaminations Clayey massive Quartz grains (sand) BAL 2-4H 20 1.8 x 0.5 Vegetal Colour gradient (top to bottom): orange/grey/brown dark Clayey massive Few quartz grains Mostly plant residues with phytoliths Malaxing BAL 3-1B 10 2.1 x 0.6 Mineral Colour gradient (top to bottom): grey/Brown with black impregnations Clayey massive Green minerals (hornblende), sequined mica and quartz grains Pollens and phytoliths Malaxing Mineral Colour gradient: white/light grey/grey Clayey massive Quartz grains with microparticles of iron oxides Micro-charcoals and phytoliths Mixing Malaxing Mixing Malaxing with others micro-pottery fragments like ‚chamotte‘ Balaté (BAL) MC 87 MC87 4-1 9 1.8 x 0.5 Jeanne Brancier et al. eschweizerbart_xxx SLO 9 Quartz grains and black opaques microparticles Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 23 Fig. 13. Microphotographs of aggregates in MC87 thin sections. A. Human induced juxtaposition of heterogeneous aggregates: yellow (Y), burnt rubefied (R) and burnt black (B) in a yellowish to brownish matrix, with some roots (Rt) fragments (OIL), infilling of the ditch, Log 1 (MC87 1-3, depth: 30 cm). B. Closer view of a yellow aggregate, possibly from an oxic horizon (OIL), Log 4 (MC87 4-1, depth: 8 cm). C. Closer view of a black burnt aggregate: dark colour results from heating in reducing conditions. Abundant burnt rootlets indicate the aggregate comes from a surficial horizon (PPL), Log 1 (MC87 1-4 B, depth: 60 cm). D. Closer view of a burnt rubefied aggregate. The reddish to brownish colour results from heating in oxidising conditions (PPL), Log 4 (MC87 4-1, depth: 10 cm). E. Packing voids with aggregates (Agg) linked as result of mesofauna activity and human fill of the ditch (PPL), Log 1 (MC87 1-3, depth: 35 cm). F. Microlaminated clay coatings with alternating dusty and limpid clay in an aggregated matrix resulting of rhythmic illuviation suggesting possible variations in soil (PPL), Log 3 (MC87 3-2, depth: 28 cm). eschweizerbart_xxx 24 Jeanne Brancier et al. Fig. 14. Microphotographs of components related or possibly associated to human activities in the three studied sites. A. Wood charcoal at the bottom end of the ditch infilling, (PPL), Log 1 (MC87 1-6, depth: 80 cm), MC87. B. Wood charcoal in the Anthrosol, Log 1, BAL (PPL), (BAL 1-2, depth: 35 cm). C. Charcoal in the Anthrosol, Log 1, CSL (PPL), (SLO 3, depth: 50 cm). D. Phytoliths (possibly linked to human activities) indicated by arrows (PPL), Log 1, CSL, (SLO 5 H, depth: 70 cm). E. Pottery sherd with vegetal fragment as temper. Matrix shows a gradient colour from orange to brown dark (PPL), Log 2, BAL (BAL 2-4 H, depth: 20 cm). F. Pottery sherd with black iron speckles (PPL), Log 4, in the enclosure, (MC87 4-1, depth: 7 cm). G. Pottery sherd with quartz grains (sand) as temper in the Anthrosol (PPL), Log 1 (SLO 3, depth: 42 cm), CSL. H. Pottery sherd at the bottom end of the Anthrosol. Successive composing layers with different components include quartz grains (in white), hornblende and mixed clay (mC, bottom right). Such components might have been used as a temper (PPL), Log 2 (SLO8 B, depth: 60 cm), CSL. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 25 than the matrix of the soils and ditch infill, some of them showing yellow and red dots. The absence of clay coatings and their colour suggest that they might derive from a B oxic horizon. Reddish burnt aggregates, with a composition similar to the surrounding matrix but darker in colour (red in OIL), include frequent microcharcoals with yellow and red speckles. Dark brown burnt aggregates are composed of abundant microcharcoals and abundant burnt rootlets, with black speckled matrix in OIL. The presence of rootlets in dark brown aggregates indicates that they probably derive from a burnt soil surface, possibly in relation to clearance and/or domestic fires. In some aggregates planar sub-horizontal voids suggest compaction by floral turbation without excluding the possibility of past trampling (Log 4, MC87 4-2). In the enclosure, the difference of aggregates concentration between Logs 3 and 4 is apparent. In Log 3, reddish and dark brown burnt aggregates are less than 5 %. Yellowish aggregates reach ~ 10 % around 13 cm and are only present from top to ~ 14 cm depth. In Log 4, a peak of 20 % of dark brown burnt aggregates is observed at ~ 21 cm depth, which is the maximum percentage of the three profiles. This peak, with fine gravel-size burnt aggregates, is located just under the charcoal peak at ~ 18 cm. In this thin section, reddish aggregates are less than ~ 5 % and yellowish aggregates less than 10 –15 %. Between 30 and 40 cm, reddish and dark brown burnt aggregates are few (< 5 %) while yellowish ones are absent. Such a distribution from top to bottom with charcoals, non-burnt and burnt aggregates, may suggest an ancient soil surface that might have burnt before reworking by biological activity, anthropogenic processes and/or short distance colluviation. In the ditch (Log 1), the location of aggregates results mostly from secondary redistribution. Less than 5 % of dark brown burnt aggregates are identified from the top to 35 cm depth, underneath which they are very few. Such distribution may suggest intentional anthropogenic redistribution of sediments (i.e., refuse). Reddish aggregates are identified in the entire profile with peaks of ~ 5 % between 11 and 60 cm. Yellowish aggregates are ~ 5 % up to 39 cm, with a peak at 10 % at 35 cm, before becoming very few at ~ 2 % up to 80 cm depth. 5 Synthesis and conclusion 5.1 Microfeatures in pre-Columbian French Guiana dark soils (FGDS) Micromorphology allows for the identification of anthropogenic features in pre-Columbian French Guiana dark soils. It provides new data even if results are primarily microlocal without encompassing all stratigraphic and spatial variations for each studied site. Microfeatures of the Holocene alluvial terrace of the Maroni River reflect (i) several episodes of clay enrichment and/ or redistribution (i.e., elu-illuviation, coatings), (ii) seasonal waterlogging and (iii), post-depositional biological activity. Clay enrichment and organic matter inputs together with biological activity processes might have alternated, probably in relation to vegetal cover and/or human past activities. At BAL, post-depositional processes in the Anthrosol are complex, with differential conservation of archaeological layers due to pre-Columbian and modern human activities together with natural processes (especially faunal and floral-turbation) (Fig. 16). In the studied logs, anthropogenic stratigraphy is less developed as compared to CSL, likely in relation to more limited sedimentary inputs by flooding and/or overland flow. At CSL, succession of microlayers eschweizerbart_xxx 26 Jeanne Brancier et al. Fig. 15. Schematic formation processes of CSL dark archaeological layers. in the levee backslope is identified in the black sandy cumulic Anthrosol. Good preservation of the archaeological sequence is likely due to successive sedimentary inputs by flooding and/or overland flow (Fig. 15). eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 27 Fig. 16. Schematic formation processes of BAL dark archaeological layers. At ring ditched mountain MC87 in the three studied logs, the presence of yellowish unburnt aggregates is likely related to an oxic B horizon. This micro-scale identification stresses that the lateritic soil acted as support (i) for activities in the enclosure and (ii), as reworked material in the ditch (Fig. 17). Anthropogenic features related to fire are apparent with charcoals and burnt soil fragments (rubefied and dark brown aggregates). These components could result from clearance for settlement, agricultural management and cultivation, or domestic activities. Preliminary observations of wood cell characteristics of charcoals strongly support the human fire hypothesis as wood cells are not coalescent by vitrification (Tardy 1998). In Oxisols, Acrisols or Ultisols, clay coatings and infillings are usually composed of limpid yellowish clay, most eschweizerbart_xxx 28 Jeanne Brancier et al. Fig. 17. Schematic formation processes of MC87 dark archaeological layers (enclosure) and ditch infilling. often not being laminated (Marcelino et al. 2010). The presence of crescent clay coatings and infillings, as in Log 3, may indicate rhythmic illuviation. This suggests possible variations in soil pH in relation to past human activities such as intentional fires (charcoals and ashes inputs) or bone inputs (Arroyo-Kalin 2008b). Microlaminated coatings with speckled clay also suggest bare soil in contrast to limpid clay coatings, which relate more to forest soil (Goldberg & Macphail 2006). 5.2 Diferences between pre-Columbian French Guiana dark soils (FGDS) and Brazilian dark earths (BDE) The preliminary micromorphological results on Anthrosols of French Guiana give original data about the nature of the archaeological sites. They also highlight similarities and differences eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 29 with Brazilian dark earths (BDE). Microscopic observations show that charcoal fragments constitute by far the most significant anthropogenic inputs observed in studied pre-Columbian FGDS. This character is also observed in all Terra Preta/Terra Mulata of the Amazonian basin (Arroyo-Kalin 2008a, 2008b, Glaser & Birk 2012). The presence and ubiquity of microscopic charcoals is of special relevance because (i) the dark colour of these soils primarily results from pre-Columbian charcoals inputs (Glaser & Birk 2012) and (ii), many studies suggest that black organic carbon contained in charcoals enhances soil fertility (Glaser et al. 2001, Glaser & Birk 2012). Comparison between FGDS and BDE also reveals noticeable differences. In FGDS anthropogenic components and artefacts are clearly less abundant as compared to typical Terra Preta, suggesting that (i) studied FGDS might be more similar to Terra Mulata, or (ii) that the studied areas/logs were located at the fringe of the habitation site (e.g., CSL). The absence of bone remains and phosphatic microfeatures in FGDS as compared to Brazilian dark earths such as fish bones (Arroyo-kalin 2008a, 2008b) must also to be underlined. Such observation might be related (i) to more acidic condition in FGDS which prevented bone preservation (if bones were associated with kitchen midden or soil amendment) and/or (ii), to the absence or limited anthropogenic bone inputs in FGDS. 5.3 Perspective Because FGDS are part of the pre-Columbian soils of the Amazonian basin they might be considered as a specific category of Amazonian dark earths (ADE). Nonetheless, the first micromorphological results obtained in French Guiana have to be supported by further studies in similar and different cultural as well as physical contexts including sites on the coastal plain. In order to provide a better interpretation of pre-Columbian FGDS, only a multidisciplinary and integrative approach will help to specify the nature of the Anthrosols (e.g., kitchen midden in housing area versus agricultural management). In this perspective, micromorphology must be supplemented by specific geoarchaeological methods including sedimentological and pedochemical analyses. In addition, palaeoecological analyses such as anthracology or phytolith identification are also necessary. Acknowledgements We would like to express our gratitude to colleagues of INRAP and of the COUAC Project for allowing Jeanne Brancier to participate to their fieldwork and investigation, including Martijn van den Bel, Jérôme Briand, Mickaël Mestre, Sylvie Jérémie, Etienne Dambrine and Bruno Hérault. Valuable comments on the manuscript were made by Vincent Freycon. Observations presented here are first results of a PhD conducted by Jeanne Brancier (Université Paris I), under the direction of Jean-Paul Demoule and Eric Fouache, with the support of the French Ministry of Research and the INRAP Unit of Micromorphology, housed by the Sol-DMOS team at AgroParisTech (Thiverval-Grignon, France). Thanks to Florian Chabauty for his help in drawing figures. We also really thank Lydie Clerc, Jean-Marc Gilliot and Dalila Hadjar for their help in GIS. M. Engel and two anonymous reviewers are warmly acknowledged for their helpful comments. eschweizerbart_xxx 30 Jeanne Brancier et al. References Arroyo-kalin, M. (2008a): Steps towards an ecology of landscape: the pedo-stratigraphy of anthropogenic dark earths. – In: Woods, W., Teixeira, W., Lehmann, J., Steiner, C., WinklerPrins, A. & Rebellato, L. (eds.): Amazonian Dark Earths. – Wim Sombroek’s Vision: 33 – 83; Kluwer, Dordrecht. Arroyo-kalin, M. (2008b): Steps towards an ecology of landscape: a geoarchaeological approach to the study of anthropogenic dark earths in the Central Amazon region. – PhD thesis, University of Cambridge, 342 pp. Briand, J., Todisco, D., Brancier, J. & Delpech, S. (in press): Balaté (Saint-Laurent du Maroni, Guyane). – Rapport final d’opération de fouille archéologique, Inrap. Bullock, P., Fedorrof, N., Jongerius, A., Stoops, G. & Tursina, T. (1985): Handbook for soil thin section description. – Waine Research Publications, Wolverhampton, 152 pp. Cammas, C. & Wattez, J. (2009): L’approche micromorphologique: méthode et application aux stratigraphies archéologiques. – In: Ferdière, A. (ed.): La géologie, les sciences de la terre appliquées à l’archéologie: 181– 216; Editions Errance. Courty, M.-A. (2001): Microfacies analysis assisting archaeological stratigraphy. – In: Goldberg, P., Holliday, V.T. & Ferring, C.R. (eds.): Earth sciences and archaeology: 205 – 239; Kluwer Academic/ Plenum, New York. Courty, M.-A., Goldberg, P. & Macphail, R.-I. (1989): Soil micromorphology in archaeology. – Cambridge University Press, 340 pp. De Boer, M.W.H. (1972): Landforms and Soils in Eastern Surinam, South America. – Agricult. Res. Rep. 771: 1–169. Eswaran, H. & Sys, C. (1975): A pedogenetical study of soil developped on ultrabasic rocks from Sabah, Malaysia. – Proc. 4th ASEAN Soil Conf. Kuala Lumpur, Malaysia. FitzPatrick, E.A. (1993): Soil Microscopy and Micromorphology. – Wiley, New York, 304 pp. Glaser, B. & Birk, J. (2012): State of the scientific knowledge on properties and genesis of Anthropogenic Dark Earths in Central Amazonia (terra preta de Indio). – Geochim. Cosmochim. Acta 82: 39 – 51. Glaser, B., Haumaier, L., Guggenberger, G. & Zech, W. (2001): The ‘Terra Preta’ phenomenon: a model for sustainable agriculture in the humid tropics. – Naturwissenschaften 88: 37– 41. Goldberg, P. & Macphail, R.-I. (2006): Practical and theoretical geo-archaeology. – Blackwell Scientific, Oxford, 454 pp. Groussin, J. (2001): Le climat guyanais. – In: Barret, J. (ed.): Atlas illustré de la Guyane: 46 – 49; IRD, Paris. Guilloré, P. (1983): Méthode de fabrication mécanique et en série des lames minces. 2nd edition. – Thiverval-Grignon, 22 pp. Jérémie, S. & Dambrine, E. (2011): Impact des occupations amérindiennes anciennes sur les propriétés des sols et la diversité des forêts de la frange côtière guyanaise. – Rapport Intermédiaire n° 2, Inra, Inrap, Cirad, 33 pp. Lima, H.N., Schaefer, C.E.R., Mello, J.W.V., Gilkes, R.J. & Ker, J.C. (2002): Pedogenesis and preColombian land use of “Terra Preta Anthrosols” (“Indian black earth”) of western Amazonia. – Geoderma 110: 1–17. Macphail, R.-I., Courty, M.-A. & Gebhardt, A. (1990): Soil micromorphological evidence of early agriculture in north-west Europe. – World Archaeol. 22: 53 – 69. Marcelino, V., Stoops, G. & Schaefer, E.G.R. (2010): Oxic and related materials. – In: Stoops, G., Marcelino, V. & Mees, F. (eds.): Interpretation of micromorphological features of soils and regoliths: 305 – 327; Elsevier, Amsterdam. Mazière, G. (1997): Rapport de fouille programmée, Montagne Favard. Service Régional de l’Archéologie, Cayenne. Mestre, M., Delpech, S. & Hildebrand, M. (2013): Yaou. Rapport de diagnostic. Inrap, Service Régional de l’Archéologie, Cayenne, 214 pp. Palvadeau, E. (1999): Géodynamique quaternaire de la Guyane française. PhD thesis, Université de Bretagne Occidentale. – Documents du BRGM 287: 1– 232. eschweizerbart_xxx Anthropogenic Features in Pre-Columbian French Guiana Dark Soils 31 Ruivo, M. L. P., Arroyo-Kalin, M. A., Schaefer, C. E. R., Costi, H. T., Arcanjo, S. H. S., Lima, H. N., Pulleman, M. M. & Creutzberg, D. (2003): The use of micromorphology for the study of the formation and properties of Amazonian Dark Earths. – In: Kern, D.C., Glaser, B. & Woods, W.I. (eds.): Amazonian Dark Earths: Origin, Properties, Management: 243 – 254; Kluwer, Dordrecht. Schaefer, C., Lima, H.N., Gilkes, R.J. & Mello, J.W.V. (2004): Micromorphology and electron microprobe analysis of phosphorus and potassium forms of an Indian Black Earth (IBE) Anthrosol from Western Amazonia. – Aust. J. Soil Res. 42: 401– 409. Sombroek, W.G. (1966): Amazon soils: a reconnaissance of the soils of the Brazilian Amazon Region. – Center for Agricultural Publications and Documentation, Wageningen, 292 pp. Stoops, G. (2003): Guidelines for Analysis and Description of Soil and Regolith Thin Sections. – Soil Science Society of America, Madison, 184 pp. Stoops, G., Marcelino, V. & Mees, F. (2010): Interpretation of micromorphological features of soils and regoliths. – Elsevier, Amsterdam, 720 pp. Tardy, C. (1998): Paléoincendies naturels, feux anthropiques et environnements forestiers de Guyane française du Tardiglaciaire à l’Holocène récent: Approches chronologique et anthracologique. – PhD thesis, Université de Montpellier II, 343 pp. van den Bel, M. (2008): Balaté, Commune de Saint-Laurent-du-Maroni. Rapport de diagnostic. Inrap, Service Régional de l’Archéologie, 71 pp. van den Bel, M., Birk, J., Brancier, J., Bruxelles, L., Cammas, C., Frenette, J., Glaser, B., Knippenberger, S., Jiménez, J. P., Verdin, P. (2011): Chemin Saint Louis, Rapport final d’opération. Inrap, Service Régional de l’Archéologie, 2 tomes, 788 pp. van den Bel, M. (2012): Archaeological investigations at the Saint-Louis site on the lower Maroni River: the Ceramic Series. Archaeology and Anthropology. – J. Walter Roth Mus. Anthropol. 17: 1– 52. Versteeg, A. (1981): A fortified pre-Columbian village in east Suriname? – Stichting Surinaams Museum: 38–48. Addresses of the authors: Jeanne Brancier, Université Paris I-Panthéon-Sorbonne, UMR 8215CNRS, France, email: [email protected]. Cécilia Cammas, Institut National de Recherches Archéologiques Préventives, UMR 5140CNRS, DMOS AgroParisTech, France, email: [email protected]. Dominique Todisco, Département de Géographie, IDEES, UMR 6266CNRS, Université de Rouen, France. Eric Fouache, UFR de Géographie et Aménagement, UMR ENeC 8185CNRS, IUF, Université de Paris-Sorbonne, France. eschweizerbart_xxx