La séquence loessique Pléistocène moyen à supérieur d'Etricourt-Manancourt (Picardie, France) : un enregistrement pédo-sédimentaire de référence pour les derniers 350 ka

International audienceIn this study, we describe a new Middle Pleistocene loess-palaeosol sequence uncovered during a preventive excavation at Étricourt-Manancourt (Somme, France; fig. 1). The full 12-m-thick sequence exhibits five stacked glacial-interglacial cycles and integrates five in situ Palaeolithic levels and remarkably one Acheulean level dated by TL on burned flints at about 280 - 300 ka.Detailed field stratigraphic approach has been completed by (1) a high-resolution sedimentological study (grain size, TOC) based on a set of more than 380 5-cm-thick samples collected from continuous sampling columns covering the entire units, (2) more than 110 micromorphological block samples for thin sections analysis (fig. 3) and (3) a set of 37 sub-samples for the study of heavy minerals. Despite specific samples (tubes) having been extracted for future OSL dating, the geochronological control is mainly based on TL dating of heated (archaeological) flints, completed by IRSL dating on K-feldspars.The archaeological excavation, located on the slope of a dry valley, was opened over more than 4500 m² and led to the discovery of two deep sinkholes developed in the chalk bedrock (fig. 6). The bottom of the deepest one is located at more than 11 m from the surface of the topsoil and 5 m below the average chalk surface. The study of four reference profiles distributed throughout the excavation (figs. 6-9), and their correlation using well defined pedological level marks led to the definition of a ca. 12-m-thick cumulative loess-palaeosols succession including eighteen main stratigraphic units. This approach allowed building a global pedosedimentary sequences for the Étricourt site that presently serves as a reference for northern France (fig. 14 & tab. 2).Here we focus mainly on the detailed description and analysis of palaeosols. The Étricourt sequence includes five luvisols attributed to full interglacial conditions and allocated to MIS 11 (unit 17), MIS 9 (unit 14), MIS 7 (units 8b and 8a) and MIS 5 (unit 5). Two luvisols are thus identified for MIS 7. The pedosedimentary records corresponding to MIS 9 and MIS 5 are represented by well-developed pedocomplexes corresponding to the superimposition of a truncated luvisol and a humic soil complex that includes systematically a grey forest soil (greyzem).During MIS 9e, a luvisol developed on loessic deposits. The truncation of this Bt horizon (unit 14), and its degradation by hydromorphy and solifluction, indicate a phase of climatic degradation following the interglacial optimum, which could be correlated with sub-stage 9d. The “Lower Humic Soil Complex” begins then with a thick layer of dark clayey silts including scattered flint gravels (unit 13). It is characterised by a strong increase in TOC concentrations, compared to the underlying Bt horizon and by an important biological activity attested by numerous earthworm chambers, partially filled with clayey pellets. Two types of clay coatings are superimposed in this soil unit: 1) light brown to orange clay coatings and 2) brownish red to blackish micro-laminated clay coatings, typical of grey forest soils (greyzems). This soil type today is associated with forest-steppe environments and continental cool climates. The overlying horizon (unit 12), in which the Acheulean level HUD is preserved, is represented by a compacted grey-brownish loam showing a strong banded fabric of cryogenic origin. However, the lithic artefacts are very well preserved and not gelifracted.This Acheulean level has been dated around 288 ± 21 ka using TL on heated flint. The overlying dark-brown horizon (unit 11) is then characterised by a strong biological activity attested by pluricentimetric burrows and a peak in TOC values (0.20 %) indicating a typical humic steppe soil (Ah) horizon, probably of the chernozem type. According to palaeo-pedological data the Étricourt “Lower Humic Soil Complex” thus exhibits a progressive mutation of the environments from continental forest to forested steppe or steppe associated to a progressive climatic degradation directly following an interglacial (Early Glacial phase). According to both the location of the soil complex within the full sequence and TL dating results, this phase is allocated to the transition between MIS 9 and MIS 8.Within the “Lower Humic Soil Complex” occurs an intense frost episode responsible for the banding of unit 12. Since the lithic artefacts from HUD are not gelifracted, this episode of freezing-thawing must have occurred after the burial of the Palaeolithic artefacts. The following sequence of events may thus be proposed (fig. 17): erosion and degradation of the interglacial Bt horizon (unit 14) at the end of MIS 9e and 9d; formation of unit 13 during MIS 9c (HUZ occupation); erosion of unit 13 during MIS 9d; and formation of steppe soils (units 12 to 10b) during MIS 9a, according to the LR04 stack (Lisiecki & Raymo, 2005), or MIS8 according to Railsback et al. (2015); degradation and intense freeze-thaw episode responsible for the cryoturbation of unit 12 and transition to gleys and loess at the transition between MIS 9a and 8.The following glacial period that can be allocated to MIS 8 is recorded by the deposition of a non-calcareous loess (unit 8c). The overlying luvisol (unit 8b) shows mainly layered grainy clay coatings including alternations of limpid orange clays and layers of quartz grains similar to the surrounding sediment. The subsequent erosion (truncation) of this interglacial Bt horizon is underlined by a flint gravel bed (Cx). The overlying horizon (unit 8a) shows many pedorelics, papules and large soil aggregates resulting from the erosion and the reworking of the underlying Bt (unit 8b). A new illuviation phase then takes place in the channels and inter-aggregate spaces indicating the formation of a luvisol on the colluvium produced by the erosion of the underlying Bt. It is proposed that the first luvisol (unit 8b) correlates with the first peak of the curve LR04 corresponding to a temperate phase of MIS 7, 7e at ca. 240 ka, and the second luvisol (unit 8a) with the two near peaks 7c and 7a (fig. 17).The subsequent erosion of the Saalian luvisol soil complex results in the deposition of a new colluvial unit (unit 7b). The overlying greyish horizon (unit 7a), including a Middle Palaeolithic archaeological level, appears in the same stratigraphic position than the humic soil layers described in regional sequences as Mautort or Gentelles, at the transition between MIS 7 and MIS 6. At Étricourt, the Upper Saalian (MIS 6) is represented by clayey loess (unit 6b), and “limons à doublets” (wavy banded fabric, unit 6a), separated by a strong erosion phase indicated by a bed of gelifracted flints. This gravel bed contains reworked lithic industry, among which a flint likely originating from the underlying Middle Palaeolithic level included in the second MIS 7 luvisol has been dated to 222 ± 20 ka. Finally at the top of the Saalian loess unit (MIS 6) occurs a typical luvisol showing a well-marked Bt argillic horizon (unit 5) allocated to the Last Interglacial optimum (MIS 5e). The overlying Upper Humic Soil Complex exhibits the same pedostratigraphic succession than the Weichselian Early Glacial reference sequence of Saint-Sauflieu (grey forest soils with numerous earthworm chambers and several steppe soils).MIS 8 and MIS 6 loess deposits have been dated using the IRSL dating technique applied to both single and multiple aliquots (IRSLSAR and IRSLMAAD, fig. 15). The MIS 6 loess deposit has been dated to 144-180 ka, in good agreement with its chronostratigraphic attribution. By contrast, the MIS 8 loess deposit yielded an apparent MIS 7 IRSL age (215-251 ka). These underestimated IRSL ages together with the large scatter of the IRSLSAR ages, suggest that the MIS 8 loess has been reworked at the beginning of MIS 7.Using litho-pedostratigraphy and the available geochronological data (TL and IRSL dating), some correlations can be proposed between the Étricourt loess-palaeosols sequence and other sequences from NW Europe (figs. 1 & 20). This approach shows that the “Lower Humic Soil Complex”, corresponding to MIS 9 is definitely better expressed at Étricourt than in other sections of northern France and can be proposed as a new reference for this interglacial. In addition, in the section of Saint-Pierre-lès-Elbeuf in Normandy, the micromorphological study also highlights the development of a grey forest soil during MIS 9/8 transition. MIS 9 would also be represented by the Achenheim III pedocomplex in the Rhine Valley, composed of a non-leached reddish brown soil and one or more humic soil horizons. Further east in Europe, no luvisols are to be observed for MIS 7 in Achenheim and Paudorf sections, likely due to more continental climatic conditions as recorded by the development of chernozem-like soils.Finally, the mineralogical study and especially the green amphibole content allows highlighting similarities between the Étricourt sequence and loess deposits from Belgium (fig. 16). These similarities combined with the available results from other loess sequences from northern France suggest that the five green amphibole groups recently defined in Belgium can be extended to the two regions. In this context, one specific contribution of the mineralogical study in Étricourt is to suggest the existence of a hiatus within MIS 6. Beyond the chronostratigraphic interest of the distribution of green amphiboles, it will be necessary in the future to discuss its implications in terms of sources and paleogeography. Indeed, the Étricourt-Manancourt sequence covers the full Saalian complex, a period that is characterised both by major palaeogeographic changes, including the opening of the Strait of Dover and the transition from the Lower to the Middle Palaeolithic.Cet article se focalise essentiellement sur l’approche paléopédologique d’une nouvelle séquence loessique du Pléistocène moyen découverte à l’occasion d’une fouille archéologique préventive à Étricourt-Manancourt (Somme). Cette séquence, d’une douzaine de mètres cumulés, correspond à cinq cycles glaciaire-interglaciaire et intègre cinq niveaux paléolithiques en place. Elle a fait l’objet d’analyses sédimentologiques et micromorphologiques détaillées ainsi que de datations TL sur silex chauffés et IRSL sur feldspaths potassiques. Elle constitue actuellement une nouvelle séquence de référence pédo-sédimentaire pour la France septentrionale pour la fin du Pléistocène moyen (~ 350-140 ka). L’enregistrement comprend cinq luvisols attribués aux SIM 11, 9, 7 et 5e. Deux luvisols sont identifiés pour le SIM 7. Les enregistrements pédo-sédimentaires des SIM 9 et 5 (complexes de sols) sont dilatés et correspondent à la superposition d’un luvisol et d’un pédocomplexe humifère à sol gris forestier (grisol) de contexte Début Glaciaire. Le complexe de sols humifères inférieur attribuable au SIM 9 (« Complexe de sols d’Étricourt ») représente un enregistrement particulièrement détaillé et unique pour le SIM 9 et la transition avec le SIM 8. Un niveau acheuléen situé au-dessus du sol gris forestier du SIM 9 a été daté autour de 288 ± 21 ka par TL sur silex chauffés. Le loess sus-jacent à ce dernier, corrélé au SIM 8, a donné un âge IRSL apparent SIM 7 (215-251 ka) suggérant un remaniement ou une remobilisation du loess au SIM 7. Un niveau paléolithique moyen conservé dans le second luvisol du SIM 7 a été daté à 222 ± 20 ka par TL. Il est surmonté d’un loess primaire allochtone dont les âges IRSL (144-180 ka) confortent son attribution au SIM 6