Continental palaeoenvironments during MIS 2 and 3 in southwestern France: the La Ferrassie rockshelter record P. Bertran a, b, *, L. Caner c , R. Langohr d , L. Leme ´e e , F. d’Errico b, f a INRAP, Centre d’activite´ les Echoppes,156 Avenue Jean Jaure `s, 33600 Pessac, France b PACEA, UMR 5199 CNRS, ba ˆtiment de ge´ologie, Avenue des faculte´s, 33405 Talence, France c Hydrasa, UMR 6532 CNRS, Universite´ de Poitiers, ba ˆtiment des Sciences Naturelles, 40 Avenue du Recteur Pineau, 86022 Poitiers, France d Laboratory of Soil Science, Earth Science Institute, University of Gent, Krijgslaan 281/S8, 9000 Gent, Belgium e SRSN, UMR 6514 CNRS, Universite´ de Poitiers, ba ˆtiment de Chimie, 40 Avenue du Recteur Pineau, 86022 Poitiers Cedex, France f Department of Anthropology, The George Washington University, 2110 G Street, NW, Washington, DC 20052, USA article info Article history: Received 12 March 2008 Received in revised form 2 July 2008 Accepted 29 August 2008 abstract Reappraisal of La Ferrassie rockshelter sequence provides novel information on the continental palae- oenvironments in southwestern France during the end of Marine Isotope Stage 3 and the beginning of MIS 2. Cycles of sedimentation and soil development have been recognised, and are tentatively inter- preted as reflecting the Dansgaard/Oeschger climatic cycles. From bottom to top, the sequence shows a trend to progressive reduction in length (and/or intensity) of the soil-forming periods and conversely, an increase in frost-related processes and sedimentation. Cambisols, associated with significant clay weathering and decalcification, developed at the base of the sequence (35 to w30 14 C kyr BP). They are then replaced by simple carbonate-rich humus horizons during the interval w30 to w28 14 C kyr BP. At the beginning of MIS 2, the talus dynamics shifts from rockfall accumulation with intense snow-induced washing to stone-banked solifluction lobes in connection to a semi-desert-like periglacial environment. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Debris talus in cave entrances and rockshelters has long been a focus of interest for archaeologists and geomorphologists. Since the beginning of the last century, these sequences and associated archaeological assemblages have progressively allowed establish- ing the cultural stratigraphy of the Palaeolithic and documentation of the physical and biological environments during the last climatic cycle of the Pleistocene. However, this interest slightly declined in the last decades due to better resolution and continuity for palae- oclimatic reconstruction provided by marine, lacustrine, and ice records. Although these non-cave archives have proved to be very successful in reconstructing the palaeoclimates, the continental sedimentary sequences remain of great interest and still arouse abundant works (see Woodward and Goldberg, 2001). This is due to the fact that (i) marine and ice archives do not yield direct infor- mation on the physical environments on the continents; lake- derived palaeoenvironmental sequences are still scarce, and for example, none is available for the Upper Pleistocene in south- western France, (ii) in many regions sediment accumulations in rockshelters and caves are the thickest available continental records and have therefore a potential to characterise past land- scapes and their evolution, particularly during cold stages; with the exception of loess deposits, such accumulations have few equiva- lents on the continents, (iii) Palaeolithic open air and cave sites are not necessarily similar. Analysis of both types of sites is needed to fully understand land use by Palaeolithic societies. One of the main difficulties that arises from the analysis of these deposits is the complexity of the sedimentary processes and factors involved in their formation. However, studies during the last 20 years on talus dynamics in active sedimentary environments and subsequent development of lithofacies models (solifluction: Bene- dict, 1970; Matthews et al., 1986; Van Vliet-Lanoe ¨, 1988; Bertran et al., 1995; debris flows: Bertran and Texier,1994; Van Steijn et al., 1995; Blair, 1999b; dry grain flows: He ´ tu et al., 1995; He ´ tu and Gray, 2000; rock avalanches: Yarnold, 1993; Blair and McPherson, 1994; Bertran, 2003; snow avalanches: Blikra and Nemec, 1998; Blikra and Selvik, 1998; Jomelli and Bertran, 2001; slush flows: Gilbertson and Sims, 1974; Washburn, 1979; Nyberg, 1989; steep alluvial deposits: Wells and Harvey, 1987; Blair, 1999a; overland flow: Mu ¨ cher and De Ploey, 1977; Lenoble, 2005) have significantly improved our understanding of such processes. A general overview of the lith- ofacies of slope deposits is available in Bertran et al. (2004). These studies particularly showed that grain-size variations within the deposits mostly reflect particle segregation during redistribution * Corresponding author. INRAP/PACEA, Centre d’activite ´ les Echoppes, 156 Avenue Jean Jaure `s, 33600 Pessac, France. Tel.: þ33 15 57 01 00 35. E-mail address: pascal.bertran@inrap.fr (P. Bertran). Contents lists available at ScienceDirect Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev 0277-3791/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.quascirev.2008.08.022 Quaternary Science Reviews 27 (2008) 2048–2063