This work was supported by the Aix-Marseille University - Labex ECCOREV - LOST project. This work also received support from the French government under the France 2030 investment plan, as part of the Initiative d’Excellence d’Aix-Marseille Université - A*MIDEX - Institute for Mediterranean Archaeology ARKAIA (AMX-19-IET-003, project MediteRANat). This work was supported by the Tellus Program of CNRS-INSU and received financial support from the Labex Dynamite.

1 - Introduction

1In the Mediterranean area, a great disparity in the distribution of archaeological sites is observed both temporally and geographically, including for periods where the environmental contexts are not necessarily hostile for human occupations.

2Several causes may be responsible for these disparities such as cultural (Guilaine & Manen, 2007; Lachenal et al., 2018; Guilaine, 2018; Perrin et al., 2018), environmental (Roberts et al., 2019), or taphonomic factors (Bintliff, 2002; Brädtmöller  et al., 2012; Schmidt et al., 2012; Banks et al., 2013; Bicho et al., 2017; Guilaine, 2018; Weninger et al., 2006; 2014). To discriminate these factors, this study proposes an integrated approach targeting two sedimentary markers synonymous of climatically temperate periods and attractive environments for potential past occupations or even particularly favorable for the remains preservation (depending on the nature of the remains - oxidization of organic remain, dissolved bones): red paleosols horizons and calcareous tufa sequences. Among all the studied sites of this ongoing research (e.g., Luberon, Lure, Sainte-Victoire) (fig. 1), two sites already investigated (radiochronology, anthracology, malacology, geomorphology, sedimentology) are presented (fig. 1): 1/ the paleosol horizon of the Upper Pleistocene MIR-11 section (Luberon) (Ollivier, 2006; Ollivier et al., 2014; Sirdeys et al., 2022) and 2/ the Holocene calcareous tufa sequence of Pont de Joux in the Huveaune Valley (Ste-Baume) (Bonifay, 1986; D’Anna et al.,1988; Ollivier et al., 2006).

3The red paleosols are located at the base of cryoclastic formation dated to MIS-3 and are contemporary to Greenland Interstadials (Rasmussen et al., 2014) which temperate climate conditions likely allowed soil development (Ollivier, 2006; Ollivier et al., 2014). Similar to “stasis” in sedimentary dynamics, the pedological processes at the origin of these horizons strongly contrast with the intense and powerful sedimentary productions (mainly cryoclasts) affecting the foothills and attributed to the colder periods that followed, and sometimes preceded, their emplacement. The “red-soil markers” are considered indicators of interest for geomorphology, paleoclimatology, or archaeology studies due to their environmental context (development during temperate climates and topographic surfaces of circulation) and their capacity to record archaeological remains and evidence of human frequentation (for instance lithic industries, bones remain, trampling marks on the soil, traces of fire hearths) (e.g., Mandel & Bettis, 2001; Walkington, 2010).

4The second markers are the calcareous tufa sequences developed during the Late glacial-Holocene periods (Vaudour, 1994; Ollivier, 2006; Ollivier et al., 2006). These carbonated deposits are formed at the emergence of karstic resurgences, in the river valleys, cascade and depressions through physico-chemical and biochemical processes (Pedley, 2009; Capezzuolli et al., 2014). Tufa sequences are particularly sensitive to variations directly correlated to local to wider scale environmental conditions of deposition, resulting in a succession of facies organized following climate, hydrological rhythms and/or impact of human activities (Dabkowski, 2020; Ollivier, 2006; Ollivier et al., 2004a,b, 2006; Goudie et al., 1993; Pentecost, 2005). Rich in palaeoecological records (malacofauna, plant macro-remains, charcoal, and sometimes pollens), the Postglacial (Borel et al., 1984) tufa formations reflect attractive environments 

5(quasi permanent water flow, fertile soils, cynegetic areas), and frequently contain archaeological remains up to the Mesolithic period (Vaudour, 1994; Dabkowski, 2014). Calcareous tufa seal macro archaeological remains such as lithic artefacts, bones, or ceramics. They also preserve evidences of local anthropic activities such as burnt sediments, charcoals, fire hearths (e.g., Degaugue & Guendon, 1992; Dabkowski, 2014) as well as wider-scale anthropic pressures/practices such as deforestation and land-use (Ollivier, 2006). However, the anthropogenic changes in the tufa formations must be read in parallel with natural changes that also affect tufa accumulation (e.g., availability in dissolved CaCO₃, efÏcient water hydrodynamism leading to carbon dioxide degassing from water, suitable substrate for CaCO₃ precipitation, catchment morphology) (Goudie et al., 1993; Ollivier et al., 2006; Dabkowski, 2020).

6These variations in natural and anthropogenic sedimentary processes are not always apparent from a visual stratigraphic description, which requires a finer scale reading of sedimentary processes and paleoenvironmental conditions.

7In the light of the paleoenvironmental and sedimentological importance of the selected markers, this work emphasizes the identification of tracers in the characterization of sedimentary dynamics within two sedimentary markers of climatic temperate phases in south-eastern France (paleosols and calcareous tufa) and questions their capacity to record anthropogenic signals at different levels (local to watershed) for key periods of low occupation data. The specific and adaptable analyses (spectrocolorimetry, semi-quantitative geochemical measurements, calcimetry) confronted with classical sedimentological methods (grain size, loss on ignition, and determination of insoluble fractions) test the efÏciency of these methods in the fine reading of sedimentary dynamics. The aim is to recognize and accurately read of the sedimentary processes within sedimentary markers. In application to paleoenvironmental and archaeological contexts, relationships with taphonomic conditions will be discussed to relate taphonomies and archaeological site distribution in the French Mediterranean context.

2 - General context

8The studied sites are both located in southeastern France: in the Grand Luberon and the Huveaune valley (fig. 1). Located at the frontier between the southern Alps and the Mediterranean domains, the Luberon is 70 km-long E-trending range. It is a ramp anticline southly bounded by a south-verging reverse fault. Composed of an envelope of Cretaceous calcareous rocks, its heart is made of marly-calcareous rocks (Clauzon et al., 2011). The range is divided in the Grand and the Petit Luberon, respectively reaching 1125 m and 727 m asl. Strongly eroded by alluvial organisms, the southern piedmont of the Grand Luberon offers an efÏcient landing area joining the Durance Valley in which morphosedimentary dynamics are well-recorded (Ollivier, 2006). The Mirail Valley represents one of the hydrosystems among which sedimentary dynamics are recorded since the Upper Pleistocene (fig. 1).

9Downstream from the folded Cretaceous limestone massif of the Ste-Baume (1148 m asl), the Huveaune valley follows a synclinal gutter and cuts through Oligocene and Cretaceous formations of the Marseille basin (Laville et al., 2018) over 50 km to reach the Mediterranean Sea (Vaudour, 1961). Well-developed karst aquifers, lead to the development of calcareous tufas in the valley widely described in the Huveaune watershed and to which the Pont de Joux formation belongs (Bonifay, 1986; d’Anna et al. 1988; Vaudour, 1994; Ollivier et al. 2006) (fig. 1).

3 - Material and methods:

3.1 - Sampling strategies

10The two sequences were selected according to several criteria:

111.The sedimentary significance of these markers from the perspective of sedimentary dynamics. The parameters responsible for the formation of soils (e.g., Wagner et al., 2015) or calcareous tufa accumulations (Goudie et al., 1993; Dabkowski, 2020) require quite stable sedimentary environment and morphogenic trends (e.g., low detritism, temperate climate). They also form in transfer zones with high sensitivity and reactivity to variations in geomorphological conditions such as slopes (southern piedmont of Luberon) and fluvial plains (Huveaune valley) which makes them first rank archives to finely record shorter or lower amplitude changes in sedimentation modalities (e.g., micro-erosive phases, facies changes related to variations in hydrology).

122.The climatic significance of these two sedimentary markers, which is mainly induced during temperate climatic periods associated with low sedimentary dynamics (Pentecost, 2005; Muhs, 2007). Warmer temperate periods also favour the development of well-forested environments, which are part of the bioclimatic conditions of their establishment. It favours the development of pedogenesis (Rohdenburg, 1989) and the dissolution of limestone in the massifs and the reprecipitation of carbonates in calcareous tufa (Vaudour, 1994).

133.The widespread presence of these formations in the Quaternary heritage of south-eastern France, from mountainous or subalpine environments (e.g., Queyras; Ali et al., 2003; Col du Lautaret, Mlakar et al.,1999) to valleys (e.g., St-Antonin, Guendon et al., 2003) underlines the important representativeness of these markers in various contexts of the region

144.The available corpus of paleoenvironmental data on these two sequences.

15The MIR-11 section is in the Mirail valley (43°47.580’N-5°32.220’E), on the southern slope of the Grand Luberon (fig. 1). It encompasses red soil horizons reaching a meter thick and located at the base of thirty meters thick cryoclastic accumulation. Previously studied by V. Ollivier (in Ollivier, 2006; Ollivier et al., 2014), this sequence is one of the best preserved and developed MIS-3 pedosequences able to record climatic or local events (from slopes to regional watershed scales) in southeastern France (fig. 1). Thus, this section should be representative of sedimentary mechanisms allowing to discuss potential conservation of Upper Palaeolithic sites at a local scale.

16The Pont de Joux tufa sequence (43°22.043’N- 5°36.952’E) in the Huveaune valley has already been studied; to highlight an infilling phase in the Huveaune during the Holocene Valley (Bonifay, 1986; Vaudour 1994; Nicod, 2009) as well as the importance of integrating the geomorphologic context in the conservation of stratigraphic data (Ollivier et al., 2006). The Pont de Joux formation is located at the edge of a narrow and sinuous fluvial corridor and is one of the preserved sections reaching a thickness of almost 12 m. In addition, on the other side of the river, archaeological artifacts attributed to the Neolithic have been found (D’Anna et al., 1988).

17The sampling strategy consists of continuous column sampling of sediments for each sequence. For the MIR-11 sequence, 23 samples of 500 g sediment were collected with a sampling step corresponding to a 5 cm layer thickness. Five additional samples were taken from the surrounding lithologies of the catchment. For the Pont de Joux Formation, 54 samples were collected and the sampling interval for sedimentological and malacological analyses varied from 5 to 25 cm depending on the stratigraphy and sedimentary facies.

3.2- Radiocarbon datings

18This study compiles new and previous radiocarbon dates to correlate the evolution of the studied sedimentary markers with their chronological contexts. New ages were measured on charcoal at the AMS Poznan ¹⁴C radiocarbon laboratory. The old and new dates are calibrated using the Chronomodel software v. 2.0.18 (Lanos et al., 2017, 2018) at two σ with the IntCal20 calibration curve (Reimer et al., 2020) (tab. 1). The chronology of the MIR-11 sequence is supported by three dates, including two in the studied pedosequence (fig. 2). The Pont de Joux section has six radiocarbon dates including five ages on charcoal and one on shell fragments (fig. 3).

3.3 - Sedimentological analyses

19The sediment analyses respond to an integrated approach applied on the PlaSedO sedimentology platform (UMR 7269 LAMPEA laboratory, Aix-en-Provence).

20This integrated approach aims to: 1/ characterize pedologic and calcareous tufas features through fastacquisition methods using spectrocolorimetry, X-ray fluorescence, grain size imagery and digital calcimetry; 2/ identify changes in sediment transport and depositional processes, through specific compositional properties within paleosols and tufas; 3/ to discuss their scale effects with paleoenvironmental and archaeological data.

21Spectrocolorimetric properties were extracted to obtain preliminary and non-destructive detection of the sediment composition in samples retrieved on the two sequences (e.g., estimation of carbonate contents, carbon, or even changes in mineralogy). Measurements have been acquired using a portable spectrocolorimeter Konica Minolta CM-700d on previously 24h-dried sediments at 50°C. Due to the small measurement area and to obtain a homogeneous signal throughout the sample, three measurements were taken on each sample and then averaged. The raw data are composed of the diffuse spectral reflectance of a material surface in the visible spectra (from 400 to 700 nm) and their chromatic properties in the chromatic space defined by the International Commission on Illumination (CIE) in 1976 and called the CIEL*a*b*. The L* parameter reflects the measurement of the total reflectance, from darker or lighter shades, symbolized by values from 0 % to 100 % respectively. The a* parameter reflects green to red colors (negative to positive values respectively) and b* represents blue to yellow shades (negative to positive values respectively). Prior calibration was made by measuring a zero and using a standard white calibration cap Cm‑A177.

22To identify the different facies (soil, calcareous tufas), read their evolution, variations and determine their sources, the total semi-quantitative elemental compositions were obtained using a portable X-ray fluorescence unit (pXRF) Vanta C on raw sediment and surrounding lithologies previously dried at 50°C during 24h. We report major element compositions (Si, Ca, Al, Fe, K, Ti, Mg in %) with the trace element Sr (in ppm) in the limit of detection (e.g., Mg undetected below 3000 ppm) (tabs. 2 & 3). Light elements (LE) with an atomic number Z<12 are not considered (because they are not differentiated). The major element contents were recalculated without the LE.

23In carbonate-rich lithologic contexts, the quantification of calcium carbonates (CaCO₃) contents is a key indicator of sedimentary facies evolution. Our method incorporates this analysis to discriminate the pedologic sediments from the substratum according to the principle that the main source of CaCO₃ is primarily related to limestones dissolution under the action of pedogenesis (Zamanian et al., 2016). The determination of the carbonate contents in calcareous tufa is used to characterize the production of biogenic carbonates along the sequence. CaCO₃ content is mainly controlled by warmer temperatures, water rich in dissolved CaCO₃, upstream degree of karstification, detrital inputs and variations in hydrodynamism (Pentecost, 2005). To determine the carbonate content of our paleosol and calcareous tufa sequence samples, a FOGII Digital Soil Calcimeter™ was used on 1 g of 24-hour dried and crushed sediments mixed with HCl solution (6 M). The calcimeter was calibrated using a 0.8 g of a commercial standard of a 99.1 % pure calcium carbonate.

24To quantify the organic matter (OM) content of MIR-11 samples, the loss on ignition method provides a useful characterization of the sediment composition and sedimentary facies (Santisteban et al., 2004; Boyle, 2004). Here, bulk samples were prepared following the protocol of Dean (Dean, 1974). First, the samples are weighed before and after (A) a drying at 105 °C and a short cooling in a desiccator for 30 minutes to obtain the moisture content. The samples are then heated to 550 °C for 2 hours, cooled in a desiccator for 30 minutes and weighed again (B). The loss of organic matter is calculated using the following equation: OM (%) = (B / A) * 100.

25The grain size properties on the 5 µm to 2 cm range were acquired to detect high-energy sedimentary dynamics in the matrices of the MIR-11 section samples. The grain size measurements were obtained using the particle sizer Analysette 28 Fritsch Imagesizer which combines wet process grain size measurements (for sediments ranging between 5 µm to 3 mm) and dry process measurements (for 20 µm - 20 mm grain size sediments). The preparation of the samples required a wet sieving at 250 µm (fine to medium sand limit, according the Blott and Pye granulometric classes - Blott & Pye, 2001- ) to optimize the quality of the data (e.g., applying the dry process without thinner particles allows for more accurate results on coarser fractions). The fraction larger than 250 µm was 24h-dried at 50°C, weighed and measured using the dry process. Approximately 5 g of sediment from the fraction smaller than 250 µm directly by the wet process without any prior preparation. The particle size proportions of the two data sets (wet and dry) were deduced by weighting the percentage classes of each data set by the masses of each fraction (fraction > 250 µm and < 250 µm). This was done to obtain a combined particle size data set representative of the proportions of fine and coarse sediments.

26In addition to complementary analyses such as calcimetry, X-rays and spectrocolorimetry, we conducted semi-quantitative analyses to determine the quartz content in different layers. The aim of these analyses is to isolate the true detrital signal resulting from watershed erosion from the carbonate sedimentation (Irion & Müller, 1968 ; Ollivier, 2006). The insoluble fraction was then isolated using 6 M hydrochloric acid on a 20 g sample previously sieved between 250 and 800 µm. The samples were manually sorted to separate the insoluble fraction from organic material and counted using a DinoLite microscope. The results indicating the number of quartz grains per 100 g of sediments are presented in figure 3.

27To determine correlations between measured sedimentological properties and samples, we present two correlation biplots, one for each section, based on Principal Component Analysis (PCA). They were obtained using the package FactomineR (Lê et al., 2008) in the RStudio software. The first biplot (fig. 4) combines the samples from the MIR-11 section and is based on the comparison of the following variables: the spectrocolorimetric parameters (L*, a*, b*), the grain size (silt, sand, gravel), the X-ray elemental composition (Si, Ca, Al, Fe, K, Ti, Sr), the carbonate content (CaCO₃) and the OM content. The second biplot combines all the samples from the Pont de Joux formation in terms of the spectrocolorimetric parameters (L*, a*, b*), the elemental X-ray composition (Si, Ca, Al, Fe, K, Ti, Sr), the carbonate contents (CaCO₃) and the insoluble fraction (quartz). Correlations between variables in each section are then tested using Pearson’s correlation. It provides correlation coefÏcients and the significance of the correlation (p-value) between each variable and highlighting the positive or the negative linear dependencies between two quantitative variables (fig. 4; R package Hmisc; Harrell & Dupont, 2019).

3.4 - Paleoenvironnemental analyses

3.4.1 - Malacology

28In the Pont de Joux Formation, the malacological study was carried out on 5 liters of sediment per sample, wet-sieved (500 µm mesh) and sorted under the microscope according to palaeomalacological methodology (Evans 1972; Limondin 2002; Limondin‑Lozouet & Moine 2014) (fig. 5).

3.4.2 - Available data from previous publications

3.4.2.1 - For the two sections (MIR-11 and Pont de Joux)

29The data set available from the MIR-11 sequence (Ollivier, 2006; Ollivier et al., 2014) includes a description of the geomorphological setting and a stratigraphic survey that constrained the sampling of the section to the pedological facies. Radiocarbon dates from the sequence, together with malacological, anthracological and magnetic susceptibility data, were used to discuss chronologies and the paleoenvironmental context in which the sequence were developed. For the Pont de Joux formation, available radiocarbon dates (two on the sampled section, one on the other bank of the river), anthracological and archaeological data are based on D’Anna et al. (1988). Additional radiocarbon dating and geomorphological interpretation is based on Ollivier et al. (2006). These data have been included in this study in order to start the study in a well-defined chronological and paleoenvironmental context.

3.4.2.2 - For the regional archaeological data

30The PATRIARCHE Database (French National Archaeological Map, Ministry of Culture) was used to produce archaeological maps for the Eze and the Huveaune watersheds. It represents various archaeological entities (such as occupation sites, burials, ruins, walls, etc.) per archaeological period (fig. 6).

4 - Results

4.1 - The MIR-11 pedologic horizons (mirail valley, luberon): sedimentological results

4.1.1 - Stratigraphy

31On figure 2, the 4b1 level consists of an alluvial deposit with small centimetric pebbles caught in a sandy-clay matrix, pedological alteration (C?) then comprises facies becoming predominantly pinkish sandy-clay as well in which carbonate nodules gradually appear (4b2, 4b3). A more organic level (close to the pedologic B horizon) composed of brown clays with polyhedral debitage indicates a continuity in the pedological processes (4b4). Variations are noted in the appearance of coarser sandy grains towards the top of the pedological sequence (4b5) until a clear delimitation is reached in erosive contact by a unit composed of coarse limestone blocks and pebbles in a brown sandy matrix dated at 41600 ± 1400 BP (4702742388 cal. BP) (Poz-134104) (above 4b5) (tab. 1).

4.1.2 - Sedimentological results

32All measurements made on the MIR-11 sequence are classed in tab. 2 and represented on figure 2. The following description presents the main variations observed along the stratigraphy for each analysis (fig. 2).

33At the base of the sequence (S01 to S05), the L* (brightness) values indicate a relatively stable pattern, with percentages ranging from 59.05 % to 57.94 %. Until 50-55 cm (S12), L* presents a gradual decrease with a minimal value at 34.27 % to slowly increase up to 43.20 % between 55 and 95 cm, suggesting darker sediment colors. A slight decrease of 2 % in the L* values between 95 and 105 cm, indicates slight darkening in the sediments tones before abruptly become lighter at the top of the sequence (S22). From 0 to 25 cm (S01 to S05), the a* (green-red component) and b* (blue-yellow component) values indicate a relatively stable pattern, with a* values ranging from 6.07% to 7.83% and b* values ranging from 17.65 to 19.25. Until 50-55 cm (S11), both a* and b* depict a gradual decrease, followed from 55 to 70 cm (S15), by a gradual increase up to 9.63 and 17.67 suggesting a first shift towards less intense red and yellow tones followed by a slight opposite trend. From 70 to 95 cm, there is a gradual decrease in both a* and b* values reaching 6.61 and 12.98 respectively. Finally, the last trend describes rises until the maximum of a* and b* values (11.23 and 24.20, respectively). This indicates a return to more intense red and yellow sediments colors.

34At the base of the sequence (S01 to S04) the OM content is similar with percentages ranging from 1.61 % to 1.97 % suggesting a relative stability. At 20-25 cm, the OM shows a slight increase at 2.82 % (S05) which turn into a gradual increase reaching 5.33 % at 55-60 cm (S12). A slight decrease occures between 55 and 70 cm (3.82 % for S14) followed by a rise in OMC up to 6.01 % at 90 cm (S18). Then, OM slightly decreases until 5.17 % at 105 cm and fall to 2.23 % at 110 cm (S22).

35The sediment grain size parameter describes the proportions (in %) of silts (5 - 63 µm), sands (63 µm - 2 mm) and gravels (2 mm - 2 cm) and depict five distinct trends within the sequence of MIR-11. At the bottom of the sequence (S01 to S04), the sand fraction is dominant with values ranging from 58.9 % to 72.9 %. It also has the highest gravel content of the sequence studied, reaching up to 1.5 %. The silt content ranges from 25.6 % to 40.1 %. At 20-25 cm (S05), there is a significant shift in the sediment composition, with a sharp increase in the silt content to 92.0 % and a decrease in sand content to 7.9 %. The gravel fraction becomes negligible (close to 0) at this height. This shift suggests a transition to finer-textured sediment dominated by silt. From S06 to S10, the silt content remains high, ranging from 87.7 % to 94.4 %, while the sand content decreases further, ranging from 5.6 % to 12.2 %. Gravel content remains negligible. This indicates a continued dominance of silt and a decrease in sand content, meaning that the shift observed at S05 towards finer sediment particles remains. Between 45 and 55 cm (S10 and S11), the grain size properties describe an increase in the proportions of coarser grains with sands and gravels contents up to 24 % and 0.5 % respectively. Silts contents decreases to 75.2 %. Starting from 55 cm to 90 cm the sediment composition shows a remarkable consistency, with silt content exceeding 96 % and sand content decreasing to less than 4 %. Gravel content remains negligible or absent. This suggests a nearly pure silt sediment, indicating very fine-textured deposits. At the top of the sequence (S21 and S22), there is a slight increase in sand content with 14.0 % and 11.2 %, respectively, while the silt content decreases to 85.7 % and 88.8 %. Gravel content remains negligible. These changes indicate a shift towards coarser sediments compared to the previous depths, although silt still dominates the composition. In summary, the sediment composition within the sequence shows a progressive increase in silt content, a decrease in sand content, and negligible gravel content. This suggests a transition from fine to very fine-textured sediments, with a slight variation towards coarser sediments at the top of the sequence.

36The table 2 and the figure 2 also provide the semiquantitative concentrations of the major elements (Si, Al, Ca, Fe, K, Ti and Mg) in the different samples labeled from S01 to S22. Si contents have values ranging from 32.03 % to 72.92 %. It depicts a global increase along the profile and vary from 32.03 % (S01) to 72.92 % (S19) to finally decrease in the top of the sequence (with values of 63.69 % and 56.43, for S21 and S22 respectively). On an opposite trend, the highest values of the Ca are at the base of the sequence with values close to 50 % (from S01 to S05) and decrease to reach 4.1 % at 95 cm (S19) to finally increase up to 23.30 % at 110 cm (S22). Al contents range from 8.36 % (S01) to values close to 16.01 % (at S17) to finish at 11.29 % at the top of the sequence (S22). Fe content vary gradually along the profile from 2.79 % (S01) to 9.46 % (S17) and fluctuate in the top with a slight decrease at 95 cm (around 5.21 %) to finally reach 6.03 %. K content range from 0.72 % (S01) to 2.15 % (S17) and slightly decrease to 1.60 % at the top of the sequence. Ti concentrations range from 0.46 % (S01) to 1.20 % (S19) to reach 1 % at 110 cm (S22). Mg is only detected between 50-55 cm with a high value of 3.81 %. Traces of Sr depict the same trend as Ca content with high values at the base of the profile ranging from 2051 to 2790 ppm (S01, S02, S03) and abruptly decrease to 1790 ppm at 15-20 cm (S04). Values then gradually decrease to 558 ppm at 90-95 cm (S19) to finally increase up to 1196 ppm at the top of the profile (S22). The major elements composition of the surrounding lithologies located upstream of MIR-11 (Cretaceous limestones and marls and Miocene peridotites) are also reported on table 3 (tab. 3) and figure 2 (fig. 2). Cretaceous limestones (Berriasian, Valanginian, Hauterivian) are mainly composed of Ca contents (between 57.61 % and 64.84 %). Si values are ranged between 25.64 % and 30.29 %, Al content vary from 5.80 % to 7.55 %, Fe have values from 1.75 % to 3.93 %, K content from 0.59 % to 1.06 %, Ti from 0.27 to 0.43 %. No Mg have been detected, and traces of Sr content vary from 771 ppm (for the Hauterivian limestones) to 3566 ppm (for the Valanginian limestones). The elemental composition of Valanginian marls (Cretaceous) reveals high level of Si with 50.57 %, a Ca content of 38.23 %, 6.86 % of Al, 2.07 % of Fe, 1.11 % of K, 0.38 % of Ti and no-detection of Mg. 3566 ppm of Sr were also measured. The Miocene peridotite is the most distant lithology from MIR-11 but also the most different. It is composed of 37.48 % of Si, 20.33 % of Ca, 15.38 % of Fe, 11.65 % of Al, 9.65 % of Mg, 2.78 % of Ti, 0.49 % of K and 2530 ppm of Sr.

37The highest carbonate contents are located at the base of the sequence with values around 60 % until 15 cm (for S01, S02, S03). At 20 cm, a net CaCO₃ decrease of 5 % (55.1 % for S04) occurred. This decrease gradually remains to reach 16 % at 60 cm (S12). A short increase with 18.2 % and 20.2 % is observed in the followed samples S13 and S14, to abruptly decrease down to the less than 1 % of CaCO₃ (for S17 and S19). From 100 cm (S20), the CaCO₃ content fastly increase to reach the value of 28.3 %.

4.1.3 - Correlations between sedimentologic parameters

38The biplot (Pincipale Composante Analysis - PCA) derived from the multivariate analysis of MIR-11 samples (fig. 4A) explains 81.7 % of the total variance. The main poles detected are: (i) Ca, Sr and CaCO₃ rich samples also bearing lighter-yellowish sediments colors and coarser sediments (b*); (ii) the pole of samples with high Si, Al, Ti, K, Fe and OM content, with finer grain sizes and the most reddish sediments. The dimension 1 (x-axis; fig. 4A) reflects 68.2 % of the total variance and all the variables contribute to this dimension in very close percentages (between 5 and 7.5 %) especially, the carbonates-K-OM and Ca contents parameters. The negative part of the axis gathers samples rich in Ca, CaCO₃, coarse (sands, gravels) with the lightest and yellowest tones (high surface reflectance L* values, and high b* value). They correspond to samples located at the base of the profile (from S01 to S08; level 4b1 and 4b2), and the last sample (S22) is also very close. The positive values of the dimension 1 correspond to high contents of Si, Al, Ti, K, Fe, OM, silts and reddish sediments. Most of the rest of the sample of the sequence is associated to these variables.

39The correlation plot (fig. 4C) highlights the most correlated variables (significance of the positive or negative correlation). It offers an easy way to determine what group of parameters we can focus on to characterize sedimentological facies. It opposes the same parameters (poles) as observed along the first dimension of the PCA (fig. 4A). There are positive and significant correlations (p-value < 0.05) between the following variables: Silts, Ti, Si, OM, K, Al, Fe and a*. In opposition, L*, Sand,

40Gravel, Sr, CaCO₃, Ca contents and b* value are wellcorrelated between them but depict strong negative correlations with the previous parameters. We note that the a* and b* values are weakly correlated with the rest of the parameters in the dataset and seems to vary independently from the rest of the parameters.

4.2 - The tufas sequence of Pont de Joux (Huveaune Valley, Ste-Baume)

4.2.1 - Stratigraphy

41The Pont de Joux sequence is located on the right bank of the Huveaune River and overlies a colluvio-alluvial formation composed of pebbles and silt dated to 7120 ± 40 BP (8017-7918 cal. BP) (Poz-134103) (levels 1a and b, fig. 3; tab. 1). The overlying tufa sequence reaches 7 m thick. Its base (unit 2, fig. 3) is dated at 7180 ± 150 BP (8220-7698 cal. BP) (Poz-130207) (D’Anna et al., 1988; tab. 1). It is composed of alternating white to beige tufa beds more or less indurated and silty levels (up to level 6) deposited in erosive contact with unit 1b. Unit 7 is composed of a white calcareous lime mud frequently observed in the regional tufa sequences (Vaudour, 1994; Ollivier, 2006). We distinguish “lime mud” which refers to calcareous muds that are particularly rich in carbonates (CaCO₃ ≥ 80%) (Vaudour, 1994; Pedley, 2009), from silty tufa facies which consist of more detrital and less carbonated sediments than lime mud. It underlies a sequence of beige indurated and stromatolithic tufas (units 8 and 9), followed by darker silty phases (units 10 and 11). Level 12, composed of lime mud and reworked stromatolithic fragments, precedes a series of indurated tufas rich in phytoclasts (units 13 and 14). Levels 15, 16 and 17 reflect sandy-silt facies with small fragments of burnt tufa. The top of the sequence (units 18, 19 and 20) shows levels of granular tufa interbedded with reworked phytoclasts globally enriched in silt. On the left bank of the Huveaune, a second formation containing at its base fragments of Neolithic ceramics (Late Neolithic) mixed with tufas and sands dated to 3710 ± 160 BP (4447-3638 cal. BP) (Ly- 4515) at 192 m NGF (D’Anna et al., 1988) presents a succession of tufas facies several meters thick composed of reworked plant macro-remains (reeds) and oncolithes.

4.2.2 - Sedimentological results

42All the measurements made on the Pont de Joux sequence are classed in tab. 4 and represented on figure 3. The main variations observed along the stratigraphy for each analysis are described below (fig. 3).

43The stronger variations concern the reflectance parameter (L*, from darker to lighter shades) with values ranging from 40 % to 65 %. At the bottom of the sequence (levels 1a, 1b) the sediments are the darkest, from level 2 the samples become lighter reaching their maximum value in level 8. Then the reflectance decreases sharply reaching values similar to the base of the sequence and finally stabilizing around 60 % with small variations in the upper part of the sequence. The variations observed with the b* parameters present similar but smaller variations around 20 while the a* parameter describes constant values close to 5.

44The quantification of the insoluble fraction of the Pont de Joux sequence is reported on figure 3. Most of the samples have no insoluble fraction. The most important content is located at the base of the sequence, in the colluvio-alluvial part (between 100 and 200 grains/100 gr; from 45 to 115 cm). This amount decreases at the base of the calcareous tufa sequence to reach less than 5 to 0 grains/100 gr between 235 and 485 cm. A slight increase up to 75 grains/100 gr occurs at 505 cm to finally return to 3 grains/100 gr at 515 cm. Between 600 and 780 cm, the insoluble fraction signal is weak but visible, with a maximum of only 4 grains/ 100 gr from 630 to 660 cm. The top of the sequence is depleted of insoluble fraction.

45The major element analysis presents a dominant proportion of Ca contents with values ranging from 43 % to 93 %. While the base of the sequence (levels 1a and 1b) shows lower contents, the Ca content suddenly increases reaching 70 % from level 2 and rising to 93 % at levels 6, 7 and 8. The Ca content then falls to 50 % and stabilizes between 70 and 80 %. A strict opposite trend is described by the Si, Al and Fe elements. For instance, Si shows higher values (over 10 %) in units 1a and 1b, then a net decrease between units 2 to 9 reaching values lower than 2 %. From level 10, Si content increases over 5 % and globally stabilizes around this value with small noisy variations of ± 2 %. Al, Fe and Ti which are present in smaller proportions, also show similar variations throughout the sequence.

46The Sr trace element content varies from 624 to 1023 ppm and seems to follow approximatively the same trend as the Ca content, except in levels 1a, 1b and in the upper part of the sequence (from levels 14 to16).

47The lowest calcimetric values are found in units 1a and 1b (CaCO₃ < 50 %). From units 2 to 9, there is a net increase from 50 % to 75 %. The carbonate content then strongly decreases to reach 50 % at level 11 and finally stabilize around 75 % in the latest part of the sequence (between units 15 to 20).

4.2.3 - Correlations between sedimentologic parameters

48The biplot of the Pont de Joux samples (fig. 4B) explains 76.9 % of the total variance. It displays the samples in two categories: (i) Ca-CaCO₃ rich samples also bearing lighter reddish and yellowish sediments colors; and (ii) the Si-Al-Fe-Ti-K and Quartz rich samples. The dimension 1 (x-axis; fig. 4B) reflects 59.5 % of the total variance and opposes samples with rich Ca-CaCO₃ contents and lighter/reddish and yellowish sediments on its negative part, from samples with high Si-K-Al-Fe-Ti and Quartz contents. The colluvio-alluvial levels 1a and 1b are represented in this latest group with levels 2, 9 to 11. The other stratigraphic levels of the Pont de Joux belong to the Ca-CaCO₃ rich category.

49The correlation plot (fig. 4D) expresses positive correlations between Si, Ti, K, Al, Fe and Quartz (with p-values < 0.05). In opposition, these parameters exhibit negative correlations with Ca, CaCO₃ contents, and L values. Positives correlation coefÏcients relates Ca, CaCO₃ contents and L* values (with p-values < 0.05). Sr content appears to be positively correlated with the Ca-CaCO₃-L* group and the Si-Ti-K-Al-Fe group, but p-values close to 0.05 mean that these correlations are not significant, and that Sr should not be considered as a determining element in the discrimination of the different stratigraphic levels.

4.2.4 - Malacological results

50A total of 55 taxas are identified and 3 malacozones are defined based on ecological and biostratigraphic analyses of the molluscan successions (fig. 5).

51In levels 1a and 1b, the first malacozone is defined by the diversity and abundance of xerothermic molluscs (e.g., Granopupa granum, Candidula gigaxii, Candidula rugosiuscula) and a noteworthy lack of palustrine species. It describes an environment of open and dry shrubby vegetation with small water inputs. Then, from level 2 to the beginning of the 5^th level, the diversity and abundance of xerothermic species decrease in favor of palustrine and mesophile species (e.g., Oxyloma elegans, Vertigo antivertigo) expressing a palustrine shallow water environment with open marshy vegetation. The uppermost part of level 5 until level 10 do not yields enough shells to determinate a malacozone. The last malacological assemblages between levels 11 to 20 present a significant proportion of palustrine (e.g.,  O. elegans, C. minimum, V. antivertigo and Z. nitidus) and freshwater species (e.g., Ancylus fluviatilis). The associated environment is consistent with well-developed riparian palustrine vegetation in an open wetland.

5 - Interpretations and discussion

5.1 - The MIR-11 sequence: soilforming processes, age, taphonomies implications and human frequentations.

52The pedogenic nature of the sequence is supported by a progressive change in the nature of the samples, contrasting coarse-grained, light-matrix, carbonate-rich levels corresponding to the alteration front with finer, darker, redder levels rich in Si, Al, Fe, K, Ti and OM since the level S05. As widely discussed in literature (Verheye & De La Rosa, 2005), soil-formation processes on rich-carbonate parent material alter and transform the prior minerals into secondary minerals rich in Si-, Al-, and Fe- that will predominantly compose the soil horizons. Here, the parent material is interpreted as the base of the sequence (S01 to S04) because of their facies (lighter colors and coarser cryoclasts) and their elemental composition similar to the closest lithologies with high Ca and low Si contents (see fig. 2, Berriasian limestones and Valanginian marls). The increase in OM content is attributed to the development of a forest cover required for pedogenesis and confirmed by the malacological and anthracological study already carried out by Ollivier et al. (Ollivier et al. 2014). This, combined with the decrease in matrix grain size (over 90% silts versus less than 35% in the cryoclastic parts), argues for a radical shift in sedimentary processes towards calmer environmental dynamics leading to pedogenesis. Colorimetric parameters are also indicative. Dark red brown lithochromic horizons can be assimilated to fire-generated rubified sediments (Ferro-Vázquez  et al., 2022) an active fire will also produce alterations of the underlying substrate to varying degrees. To date, however, few studies have addressed how the characteristics of burned substrates relate to pyrotechnology. Here, we systematize the use of colour to identify burnt sediments by performing quantitative colour measurements in the CIELab system of experimentally heated soils and sediments. The experimental design included different temperatures, different heating durations and substrates with varied chemical and mineralogical compositions, including naturally red soils and sediments with different degrees of pedogenesis. The measured colours were analysed by multivariate statistics for diagnosing whether sediments have been heated or not, and to which temperature. We achieved an accurate identification of heated versus unheated samples independently of their composition. The determination of the temperature of heating required prior knowledge of basic mineralogy and chemical properties in the targeted sediments (silicate or carbonate material, total and secondary Fe, Ca contents and the amount and kind of organic matter, or be the product of reworking of surrounding lithologies (Legout et al., 2013) or even originate from karst purges (e.g., decalcified sediments). Here, the pedological origin of the color is evidenced by the positive correlation between the chromatic parameter L* and CaCO₃ contents (figs. 4A & 4C). This means that lighter tones are linked to higher carbonate content, as has already been commonly used in marine core studies (Mix et al., 1992). Thus, the decrease in L* observed along the profile clearly indicates a loss of CaCO₃ and, conversely, more Fe and Al, which is related to the transformation of primilary minerals into Fe and Al bearing minerals responsible for the rubefaction (Fedoroff & Courty, 2013).

53Considering the overlap of the two dates in the profile (fig. 2) and their values close to the limit of the radiocarbon dating method (50 ka BP) (Godwin, 1962), the duration of the pedological processes and the chronology are limited to the lack of precision regarding the exact age of the formation and the moment when the cryoclastic cover eroded the upper part of the soil profile. However, based on the dating within the soil profile (38 900 ± 600 BP), it belongs to the MIS 3 and seems to be associated to the climatic Hengelo Charbon interstadial (36 000 and 38 600 BP, Vandenberghe & Van Der Plicht, 2016) corresponding to the glacial interstadials (GI) n°10 and/or 11 on the Greenland ice core timescale (Rasmussen et al., 2014). In Mediterranean context, soil horizons formed on limestone-rich substrates require long periods of time (several centuries or even millennia), low detrital input, a temperate climate, and the development of vegetation (Lamouroux, 1972 Verheye & De La Rosa, 2005; Zielhofer et al., 2009) estimate the durations of Calcic Luvisols soil formation between 10 000 and 40 000 for the Late Pleistocene in an alluvial Mediterranean floodplain of Tunisia. These interstadials warming of several hundred years seems consistent with the soil evolution time-process at MIR-11. Morevover, the concomittent anthracological (Quercus t. pubescens) and malacological (Pomatias elegans, Rumina decollata) identifications characterized a temperate paleoecological environment in the Luberon (Ollivier et al., 2014).

54These morphosedimentary “pauses” during the MIS 3 have already been observed in the southern piedmont of the Luberon (51 ± 3 ka BP and 46.6 ± 1.6 ka BP at TerreRouge - Ollivier, 2006; Ollivier et al., 2014 -), in Haute Provence ( 45 ± 4 ka BP at Forcalquier - Dubar, 2008 - ; between < 37 ka BP and > 30 ka BP at La Tuilière - Dubar, 1979 - ) or in Basse-Provence (e.g., 31.9 ± 1.7 ka BP at La Vautubière - Ambert et al., 1974 - , fig 1 & 6A) and confirm the regional morphogenic response to the various rise of temperatures driven by warming events during the MIS 3 (Sánchez Goñi, 2022; Ollivier, 2006). They are also located at the base of cryoclastic glacis or sandy cryoclastic formations (loess or aeolian sands), whose upper parts are eroded by strong evidence of erosion/transport processes generated during colder periods, as in the MIR-11 profile (S22). For instance, the estimated slopes and watersheds erosional rates are 0.2 to 1.2 mm/year ranges for MIS 3 and MIS 2 (Ollivier, et al., 2014) whereas the long-term denudation rates for the Grand Luberon massif ranges between 55-117 mm/ka (Thomas et al., 2018). It highlights the strong contrasts in continental Mediterranean sedimentary archives depending on the period.

55On the basis of the correlations analysis, the MIR-11 soil-formation processes present an opposite trend in the middle of the sequence (S10 and S11). This could express two different processes: either a “slowdown” in pedogenesis followed by a pedologic rebound leading to the formation of a second soil, or the recording of a single detrital event draining coarser material rich in Ca-CaCO₃. Our analysis favours the first hypothesis for three reasons. The increased content of coarser sediments reveals a rise in transport energy assimilated to detrital event. The OM-Si-Al-Fe contents before and after this event follow a continuous linear increasing trend, indicating continuity in soil-forming processes. Finally, the punctual Mg signal (> 1 %, S10) can only be attributed to a punctual external input. It is suggested that this may be related to the presence of Mg-rich Miocene volcanic material (~ 5 %) in the upstream catchment (Sylvestre, 1977; Cocheme, 2011) (figs. 1B & 2) which could have been drained during a short erosional event bringing higher Mg contents. As the Hengelo-Charbon complex is characterized by an increase in atmospheric precipitation leading to flooding episodes, as already observed in the Dutch fluvial sequences (Van Huissteden et al., 2001), and although impossible to prove, we question the possible impact on the destabilization of the slopes translated by diffusive runoff in the southern Piedmont of the Luberon.

56The MIR-11 paleosol corresponds chronologically to the transition between the Middle Palaeolithic ending at 41030-39260 cal. BP (Higham et al, 2014) and the beginning of the Upper Palaeolithic (Bazile, 2002) corresponding to a homogenization of the presence of H. sapiens throughout Europe and as far as the Mediterranean Levant (Slimak, 2023), making these paleosols important markers in Mediterranean landscapes. In the Lubéron massif, occupations have been documented since the Middle Palaeolithic (e.g., La Baume des Peyrards - Deydier & Lazard, 1910; de Lumley, 1957 -; La Combette from 78.3 to 39.4 ka BP - Buisson-Catil & Texier et al., 1996; Texier et al., 2003; Kreutzer et al., 2021 -). The Adaouste cave site in the Durance Valley (Onoratini et al., 1997; fig. 7A), is the closest occupation attributed to early Upper Palaeolithic periods.

57These shelter or cave sites are relatively well preserved. Based on the composite reconstruction of morphogenic trends in the southern Luberon piedmont proposed by Ollivier et al. (2014), an ante (45 ka cal. BP) and post (32 ka cal. BP) Hengelo-Charbon interstadial incisions are documented. The power of the 32 ka cal. BP incision is estimated at around 15 m high, and therefore question the preservation capacities of open-air archaeological sites.

5.2 - The Pont de Joux sequence: calcareous tufa processes, age, taphonomies implications and human frequentations

58The transition between coarse colluvio-alluvial signal to calcareous tufa is reflected by the strong rise in carbonate contents and the contrasted decrease of the covariant parameters Quartz, Si, Ti, K, Al and Fe contents since level 2 (figs. 3 & 4). These latter parameters are considered as the expression of the surrounding lithologies of the drainage area constituted of Triassic, Cretaceous, Jurassic limestones (french geological map, 1/1 000 000, INSPIRE, BRGM, 2019) as they are also widely considered as detrital indicators in the literature (e.g., Wedephol, 1995; Cary et al., 2008). The malacological results agree with this environmental change by indicating a transition from a terrestrial environment with more than 70 % of xerothermic species (e.g., Granopupa granum) toward a shallow palustrine environment with marshy vegetation through the expression of mesophilous and palustrine fauna (figs. 3 & 5). The radiocarbon date around 7.9 ka cal. BP for the colluvio-alluvial and the beginning of the development is associated to a temperate climate probably following the 8.2 cold event (Kobashi et al., 2007; Thomas et al., 2007) as well as many other tufa deposits across the region (e.g., Luberon, Ollivier, 2006; Meyrargues, Magnin et al., 1990; Saint-Antonin - Guendon et al., 2003; Vaudour, 1994 -; fig. 6B).

59Then, the nature of the sedimentation follows a progressive decrease of detritism in favour of a biogenic sedimentation accumulation richer in carbonate contents and stratigraphically characterized by vegetal imprinted facies, wither lime-mud, stromatolithic facies. The low malacological abundance reflects a palustrine environment (figs. 3 & 5). Potential taphonomic processes or a local depletion in malacological abundance may be responsible for the sub-sterile fraction.

60Then, the tufa accumulation signal tends to decrease in favour of a rise in the detrital signal (since level 10) with the appearance of darker silty tufa facies dated at 6646 ± 35 BP (7579 - 7465 cal. BP) (fig. 3). This detritism lasts in the upper part of the Pont de Joux sequence and present variations including several maxima (level 15 and 17) in levels which altitudinally and geometrically correspond to the Neolithic dating of 3710 ± 160 BP made in the same formation by D’Anna et al. (D’Anna et al., 1988) on the left bank of the Huveaune. The associated facies correspond to the succession of lime-mud, reworked tufa debris (stromatolites and crusted plant remains) (fig. 3). Other regional studies on contemporaneous tufa formations (e.g., Vaudour, 1994; St-Antonin, Guendon et al., 2003; Luberon, Ollivier, 2006) also described an increase in detrital inputs of climato-anthropic origins (e.g., rapid climate event such as 4.2 ka BP event, fires, widespread vegetation openings due to Final Neolithic land-use) inducing local impacts traduced by calcareous tufa facies changes on similar dated levels during the Late Neolithic (Ollivier, 2006; Ollivier et al., 2006). However, the global carbonate signal slightly increases during this period while the malacological results indicate an open riparian palustrine vegetation (figs. 3 & 5). According to Vaudour (Vaudour, 1994), this type of evolution (slow, degraded accumulation of tufa) should be attributed to valleys with little or no human presence during the Early and Middle Neolithic (e.g., Meyrargues or Jouques, Magnin 1993).

61The Pont de Joux chronology overlaps two major cultural transitions during the Holocene in southern France including the Mesolithic-Early Neolithic transition (between 8550-7950 BP; Guilaine & Manen, 2005) at its base, and the Late Neolithic-Bronze Age transition (around 4000 BP, Lachenal, 2022) in its upper part.

62The French archaeological database does not record any traces of Mesolithic occupations in the Huveaune watershed, despite some prior proven Palaeolithic occupations (e.g., the Upper Palaeolithic burial of Cuges les Pins - https://www.inrap.fr/​decouverte-d-unesepulture-du-paleolithique-final-cuges-les-pins-5271 - or the Marcouline shelters in Cassis - Escalon de Fonton, 1967 -) and a Mousterian scraper found near the Pont de Joux site and mentioned by D’Anna et al., 1988 (fig. 7B). During the Mesolithic, the south-eastern France is already considered a mysteriously poor settlement area (Guilaine

63et al., 2007, Sénépart, 2014). The few preserved Mesolithic sites are located in rock shelters or caves close to a major river or at low altitudes, leading to important losses of information due to erosion processes (e.g., Espeluche “Lalo”, Berger et al., 2016; Nesque and Coulon valleys - Brochier & Livache, 2004; Binder, 2004 -). However, correlations between potential land-use signal during this period and the recording of these activities in calcareous tufas cannot be excluded (Vaudour, 1994). As observed in other studies (Goudie et al., 1993; Dabkowski, 2020) and at Pont de Joux, the factors involved in the late Holocene tufa decline are difÏcult to discriminate. In terms of its chronology and palaeoenvironmental features, the Pont de Joux formation has undergone major changes, from a dry colluvio-alluvial environment to a marshy-paludal environment, which seems to have been attractive to Mesolithic societies (Brochier & Livache, 2004).

64The Late Neolithic-Early Bronze Age transition is marked by a decrease in the number of remains (from 14 to 7, then 4 in the Middle Bronze Age) in the Huveaune watershed (fig. 7B). Attempts to estimate demographic trends suggest a crisis following the demographic boom at the end of the Neolithic, but it is still poorly understood (D’Anna, 1992), in fact, only a decrease in the number of sites is observed (Lemercier, 2020). Studies carried out in the Rhone valley have highlighted significant morphological changes (sediment overlapping, complete erosion on hilltops or slopes) leading to the destruction of alluvial or palustrine archives built during the Bronze Age in torrential valleys (Berger et al., 2000; Berger et al. 2016). So far, no such evidence has been observed in the Huveaune valley.

6 - Conclusion

65Complementary analyses combining sedimentary analysis and paleoenvironmental results at watershed, stratigraphic and elemental scales were performed on two diachronic sedimentary formations (Pleistocene paleosol glacis and Holocene calcareous tufas) corresponding to widespread morphoclimatic sequences of southeastern France.

66It allows:

671. To precisely characterize various morphosedimentary facies such as pedogenesis vs. cryoclastic discharges and calcareous tufas production vs. alluvial detrital deposits.

682. To strengthen the detection of events (increase in detritism, evolution degree of pedogenesis) that are not always easily detectable at the stratigraphic scale (figs. 2 & 3) by correlating the various measured parameters. For MIR-11 (Luberon), more than 22 samples retrieved along the pedosequence and in the surrounding lithologies were considered to assess the pedological nature of the sequence as well as the detection of a detrital event of the watershed through our methodology. Radiocarbon dating combined with anthracological and malacological data, links the pedogenesis process to the temperate warming phase of the Hengelo-Charbon in MIS 3. The sampling of the Pont de Joux sequence (Huveaune Valley) enabled to detail the Holocene development of the calcareous tufa deposits, to describe the variations in detritism and the environmental evolution recorded by the malacological assemblages. The results follow a similar pattern to other regional sequences, with tufas emplacement around 7 ka, followed by an increase in detritism that continues to the top of the sequence.

69These two markers (paleosol and calcareous tufa) evolved in similar global contexts (calm, with low detrital environment and time periods) and are better able to record small variations on a watershed scale. In periods of low detritism, paleosols record climatic warming, colluvial and forest-cover dynamics, as well as calcareous tufa record the same parameters but in an alluvial environment. Their chronologies cover key transitional periods in the history of human occupation in south-eastern France, making these sedimentary archives a powerful tool for describing potential climatic and morphogenic effects on the preservation and distribution of archaeological sites.

70Further analyses using the same approach will also allow the corpus of data on the natural formations to be strengthened at a regional scale (fig. 1) (N. Sirdeys, ongoing thesis) to better understand the bioclimatic and morphogenic dynamics associated with human occupation and the environmental characteristics of archaeological sites.