The Baix LPS (44∘42′36.30′′ N, 4∘43′20.80′′ E) is located at the western edge of the Rhône Rift Valley, SW of Valence and ca. 4 km west of the Rhône River at an elevation of 155 m a.s.l. (Figs. 1, 2). It is situated on the foot slope of a convex ridge that forms the northern edge of a small basin, consisting of Lower Cretaceous limestone and marl, covered by Cretaceous debris and Pleistocene sediments (Chenevoy et al., 1977; Saint Martin, 2009).

To the south, the basin is bordered by the basaltic plateau of the Coiron volcanic province with the Montagne d'Andance (530 m a.s.l.), a volcano that was active until the Late Miocene to Pliocene times (Chenevoy et al., 1977; Pastre et al., 2004). Several episodic creeks drain the basin via the rivers Ouvèze and Payre in an eastern direction towards the Rhône River (Fig. 2). The present climate is characterised by a mean annual temperature of 13.3 ∘C and mean annual precipitation of 874 mm, with a precipitation maximum in October or November (Meteo-France, 2021). The predominant wind regime is the mistral, a strong northerly wind that is channelled within the Rhône Rift Valley and accelerated southward (Guenard et al., 2005). At the opening of the western shoulder of the Rhône Rift Valley near Baix, a part of the wind flow gets diverted westward into the small basin on whose northern edge the Baix LPS is located (Fig. 2). The Baix LPS is situated in the wind shadow with respect to the wind that blows from the Rhône Rift Valley into the basin immediately south of the low crest on whose foot slope the Baix LPS is located. This situation explains the occurrence of the thick loess patch at this south-facing foot slope.

The Baix LPS was first mentioned by Suen (1934) and named “Baix” by Mazenot (1956). Solely based on field observations, Mazenot (1956) subdivided the Baix LPS into four parts, comprising (1) a palaeosol at the base, containing debris of basalt and Cretaceous limestone, overlain by (2) decarbonated red loamy loess, superimposed by (3) typical yellow loess, and (4) typical loess on top. Mazenot (1956) interpreted the entire Baix LPS as having developed in one single loess deposit, arguing that the intensive alteration of the basal part occurred due to a high groundwater table in former periods.

In this study, we performed luminescence screening to support the stratigraphy based on detailed field observation, hypothesising that the lower ∼ 7 m of the Baix LPS reflect the Eemian to Middle Pleniglacial palaeoenvironmental changes. Therefore, we collected data on the luminescence sensitivity, the (approximate) palaeodoses and the dose rates. In addition, we applied a sensitivity approach following Fitzsimmons et al. (2022) but using the coarse fraction and performing an additional step for inter-aliquot normalisation to obtain a proxy indicating the likely provenance of the material. Here, we give a brief outline of the procedure, while a detailed description is given in the first part of the Supplement (“Methods – Luminescence screening”).

A semi-continuous sequence of material was collected from the lower ∼ 7 m of the Baix LPS in 16 vertically aligned boxes (laboratory codes of the OSL box samples are HDS-1802 (top) to HDS-1817 (bottom) (cf. Sect. 2.2.2). In the dark luminescence laboratory, up to eight ca. 1 cm3 small subsamples (A–H) were extracted at intervals of 5 cm from each box sample (altogether 126 subsamples). Additional material was taken for water content and radionuclide determination, both needed for calculating an effective dose rate. Minimal sample preparation included wet sieving (125 µm mesh size) and preparation of small aliquots (a few hundred grains) from the retained polymineral coarse-grain fraction (>125 µm).

To provide a time frame, we applied the SAR protocol establishing dose–response curves with several test-dose-normalised regeneration-dose points (Lx/Tx), including dose points for testing the recuperation and recycling ratio, once with a BLSL protocol (Murray and Wintle, 2000) after IR stimulation (pIR-BLSL) and signal detection in the UV (Hoya U340) and once with a post-infrared (at 60 ∘C) infrared (at 225 ∘C) protocol (pIR60IR225) (Thomsen et al., 2008) with detection of the blue emission around 410 nm (through interference filter CH-30D410-44.3). All measurements were performed on two automated luminescence Risø reader models, TL/OSL DA15 and TL/OSL DA20 (an update of TL/OSL DA15), respectively (Lapp et al., 2012, 2015). The protocols are illustrated in Figs. 4 and 5.

We aimed to extract a quartz-dominated signal with the pIR-BLSL protocol and a potassium-feldspar-dominated signal with the pIR60IR225 protocol. For adapting the pIR-BLSL and the pIR60IR225 protocols to the Baix samples, we performed a series of tests, including De-range tests, dose recovery tests, normalisation dose tests (pIR60IR225 protocol only) and tests of anomalous fading. Thus, our OSL-screening approach contained all relevant preparatory measurement steps needed for elaborated SAR OSL dating. In contrast to proper OSL dating, however, the De measurements were performed on only three aliquots per subsample for the pIR-BLSL approach and one aliquot per subsample for the pIR60IR225 protocol in order to save measurement time. For dose rate assessment, radionuclide determination was performed with the novel μDose system (Tudyka et al., 2018, 2020; Kolb et al., 2022) on six samples regarded as representative of the Baix sediments and palaeosols. Assuming radioactive equilibrium, the radionuclide concentrations were transformed to dose rates (Guérin et al., 2011), while the cosmic dose rate was calculated with the R package “Luminescence” (Kreutzer et al., 2012). Following standard procedures, we assumed an internal potassium content of 12.5±1.25 wt % for the pIR60IR225 protocol, a values of 0.1±0.02 for the pIR60IR225 approach (Kreutzer et al., 2014) and 0.035±0.02 (Lai et al., 2008) for the pIR-BLSL approach and water content as suggested by Sauer et al. (2016) and applied to the Collias LPS by Bosq et al. (2020b). While our OSL “age estimate” calculation relies on 126 De measurements for the pIR60IR225 protocol and the triple number of De measurements for the pIR-BLSL protocol, a mean effective dose rate was used for each of the two SAR protocols. Due to the reduced number of aliquots used for De determination and the use of a representative mean effective dose rate for each of the two SAR approaches, we called them “age estimates” and put them into quotation marks. We consider our luminescence screening approach for deriving OSL “age estimates” adequate for testing the field hypothesis of whether the lower ∼ 7 m of the Baix LPS contains an Eemian to Middle Pleniglacial record. We will additionally perform OSL dating on selected samples in future studies.

Varying luminescence sensitivities in a loess profile may serve as a proxy for different source areas or materials (Fitzsimmons et al., 2022). We compared the IR signal of the aliquots measured with the pIR-BLSL protocol to the subsequently readout BLSL signal, using all measured test dose signals (Tx) for this procedure. For determining the luminescence sensitivities, first, the Tx signals of a complete SAR measurement (eight measurement points) were averaged for each aliquot (Tx_mean of an aliquot) before a mean with the standard error of the three aliquots of each subsample was calculated (Tx_mean of a subsample). Finally, the ratio of the signals of the two stimulations was derived (Tx_IR/Tx_BLSL). While samples with a dominant IR signal may point to the predominance of feldspar (emitting in the UV), the dominance of a BLSL signal could indicate the predominance of quartz in the polymineral coarse grains. Thus, larger values (ratios) would point to more feldspar emitting around 340 nm and smaller ratios to a quartz dominance emitting in that range.

The overall 13.7 m thick Baix LPS was subdivided into 19 horizons, developed in (mostly reworked) loess and soil sediments that overlie slope deposits with a large share of rock fragments from local bedrock (Table 1). Four levels of soil formation were identified, in the following termed “Holocene soil”, “upper interstadial palaeosol”, “lower interstadial palaeosol” and “interglacial palaeosol complex”. Several erosional discontinuities were observed, e.g. above the 7 Btg and 5 Bw2 horizons. Various forms and sizes of secondary carbonate accumulation, i.e. pseudomycelia, hard nodules and soft carbonate concentrations, were observed (Table 1, Fig. 3b). All horizons showed abundant bioturbation features such as root channels and earthworm burrows, which make it a challenge for luminescence dating.

Examples of pIR-BLSL and pIR60IR225 shine-down, dose–response curves and graphical results of the fading tests are given in the second part of the Supplement (“Results – Luminescence screening”) (all g values normalised to 2 d). While four aliquots measured with the pIR-BLSL protocol showed g values, which within error margins agreed with zero, eight aliquots exhibited g values pointing to fading. For the pIR60IR225 protocol, four aliquots exhibited g values which conform with zero, while the other four aliquots showed g values above zero (mean of all eight aliquots 6.45±1.16). As this is not expected for pIR-IR measurements (Thomsen et al., 2008), we eliminated the prompt readouts for g-value determination. This produced a mean g value over all eight aliquots of 1.24±2.48, conforming with zero (no fading). Therefore, we assumed that the larger g values for the pIR60IR225 protocol were measurement artefacts (cf. Thiel et al., 2011). Such measurement artefacts may be produced if prompt readouts are included in fading measurements and no minimum time after irradiation and preheating is considered (see final public response to discussion of Kadereit et al., 2020). As no correlation between the size of a De of an aliquot and its g value was observed, we expect the pIR60IR225 protocol to deliver reliable data.

The results of the effective dose rate determination for the six selected samples are presented in Fig. 8. They show a relatively narrow spread, justifying the use of a mean effective dose rate for the OSL screening. For pIR-BLSL measurements, supposed to sample quartz, the effective dose rates ranged between 2.43±0.18 and 2.93±0.23 Gy ka-1 (dark-blue symbols), giving a mean value of 2.80 Gy ka-1 (blue line). For pIR60IR225 measurements, likely sampling potassium feldspar, the effective dose rates range between 3.52±0.19 Gy ka-1 and 4.15±0.24 Gy ka-1 (red symbols), providing a mean value of 3.98 Gy ka-1 (red line). These values were used for OSL “age estimate” calculation. For the lowermost silt loam sample, a higher water content was also assumed (light-blue and yellow symbols) for dose rate assessment, accounting for its hydromorphic features. While for “age estimate” calculation, De errors and beta source calibration errors were taken into account, only the mean values of the dose rate without assuming an error were considered. Thus, the OSL screening gives an idea of the course of the burial doses along the lower 7 m of the Baix LPS, divided by an effective dose rate regarded as representative, to produce a profile of approximate “age estimates” (Fig. 9). A graph summarising OSL “age estimates” block-wise for each of the 16 boxes HDS-1802 to HDS-1817 following the central age model approach of Galbraith et al. (1999) is provided in Fig. S5 in the second part of the Supplement. Throughout the investigated part of the profile, pIR-BLSL “age estimates” are smaller than pIR60IR225 “age estimates”.

While the course of “age estimates” gained with both the pIR-BLSL and the pIR60IR225 protocol shows a trend of increasing data values from top to ca. 12.3 m depth, surprisingly, the values do not increase further below that depth for the pIR-BLSL protocol. Further, several aliquots measured with the pIR-BLSL SAR protocol show “age estimates” younger than the Last Glacial Maximum (LGM) limit (cf. yellow line in Fig. 9), which is not logical for loess deposits, especially not at depths several metres below the ground level. Although the incorporation of quartz grains from the surface and younger deposits by bioturbation cannot be excluded, with quartz being bleachable faster than feldspar (Godfrey-Smith et al., 1988), this assumption does not seem appropriate as an explanation for the comparably young(er) “age estimates” produced by the pIR-BLSL SAR protocol. In general, we suppose that the pIR-BLSL “age estimates” are less reliable than the pIR60IR225 “age estimates” produced by our OSL-screening approach. We assume that the IR depletion of the polymineral coarse grains was too incomplete to get a more or less “pure” BLSL signal from quartz. It is likely that the polyminerals emitted an unstable UV emission from feldspar, especially as preheating and pIR-BLSL readout occurred at significantly lower temperatures and for considerably shorter times than for the pIR60IR225 protocol. This assumption agrees with the partly considerable g values observed by the fading tests. For these reasons, we disregard the age assessments associated with the pIR-BLSL measurements and only consider the pIR60IR225 “age estimates” for the discussion. The derived “age estimates” of the OSL screening are tentatively reflected against the MIS as defined by Lisiecki and Raymo (2005) (Fig. 9). As the MIS stratigraphy rather reflects global climate changes, we correlated our OSL “age estimate” results with stadial and interstadial phases derived from European terrestrial (LPSs, pollen and lacustrine sequences) archives (Woillard, 1978; Beaulieu and Reille, 1992; Guiter et al., 2003; Haesaerts et al., 2016).

The OSL sensitivity values (IRSL / pIR-BLSL ratios) originate from coarse-grain separates and therefore do not represent the far-transported aeolian main component of the LPSs but rather accessories. They ranged between ca. -0.5, likely pointing to subdued contribution from feldspar, and ca. +1.5 (Fig. 10), likely pointing to an increased contribution from feldspar (or feldspar having a stronger signal in the UV range if the depletion of the OSL signal was insufficient by the IR stimulation).

Low values of around -0.5 occurred in the lowermost horizon, 7 Bt1. The largest concentration of values around 1.5 occurred in the overlying 7 Btg horizon. “Balanced” values scattering closely around 1 occurred in the overlying horizons, i.e. 6 BCk/Btg to 5 Ck2. Values alternating between these two extremes are observed in the 5 Ck2 to 3 BCk3 horizons. Thus, the OSL screening seemed to serve as a proxy for variations in the sediment sources, as suggested by Fitzsimmons et al. (2022), and supported a subdivision into four parts marked in Fig. 10.

The basal interglacial palaeosol complex was composed of three horizons, whereby the lowermost horizon (8 Bt2) had developed not in loess but in a slope deposit with high content of local rock debris. Its difference in parent material is also reflected in its different OSL sensitivity ratio (Fig. 10). Most likely, its high clay content was partially inherited from the pre-weathered slope deposit. The reddish-brown colour and the presence of in situ clay coatings in the overlying 7 Bt1 horizon suggested warm and at least seasonally humid conditions. Relocated aggregates and broken clay coatings in the matrix indicated intense bioturbation. The uppermost reddish- to orange-brown 7 Btg horizon of the basal interglacial palaeosol complex showed similar characteristics to the 7 Bt1 horizon but in addition exhibited strong stagnic properties (Table 1). The sediment in which these two upper horizons of the interglacial palaeosol complex had developed yielded penultimate-glacial to Eemian “age estimates” and exhibited an above-average OSL sensitivity ratio over the entire thickness of the two horizons. Thus, the sediment was clearly different from the last-glacial slope-wash loess that made up the overlying part of the Baix LPS. We thus interpreted the basal interglacial palaeosol complex as an Eemian palaeosol complex, developed in slope deposits with predominant penultimate-glacial loess component, corresponding to the basal red palaeosol in the Collias LPS ∼ 100 km further south (Bosq et al., 2020b). The declining OSL “age estimates” at the base appear to indicate an age reversal. However, in the simplified approach applied in the present study, a common dose rate was assumed for all samples, neglecting the stagnic conditions that clearly prevailed for some time at the base of the Baix LPS, and thus reducing the effective dose rate and increasing the “age estimates” accordingly. Assuming an increased water content (Fig. 8), the “age estimate” at the base would be in agreement with the expected stratigraphy. This issue needs to be addressed in more detail in future studies with more sophisticated OSL approaches. The apparent Eemian “age estimates” for the lower and central part of the 7 Btg horizon may be owed to reworking by short-distance slope-wash processes. The uppermost part of the 7 Btg horizon may have even developed in a younger deposit, possibly reworked and deposited during the late early glacial period (e.g. Mélisey I, with subsequent pedogenesis during St Germain I), as incorporation of soil peds eroded from the underlying horizon and re-sedimentation of insufficiently bleached material could lead to age overestimation. Possible reworking phases correlate most likely with the stadials Mélisey I and Mélisey II (MIS 5d, 5b; Herning, Rederstall (Guiter et al., 2003); St Nicolas, Saint Haon (Reille et al., 2000)) of the early glacial period (Beaulieu and Reille, 1984b; Guiot et al., 1993). The subsequent periods of landscape stability, reflected in the Baix LPS, again by soil formation, possibly correlate with the interstadials St Germain I and St Germain II (MIS 5c, 5a; Amersfoort–Brørup, Odderade (Guiter et al., 2003); St Geneys 1, St Geneys 2 (Reille et al., 2000)) (Beaulieu and Reille, 1984b).

Thus, we assume that the basal Eemian palaeosol complex in the Baix LPS started to develop under conditions of dry, warm summers and wet, mild winters, leading to the 7 Bt1 and 8 Bt2 horizons. After a period of reworking (during Mélisey I and/or Mélisey II), soil formation continued into the middle and late early glacial period in the reworked material of the Eemian soil. Eemian Luvisol formation and subsequent truncation and/or reworking are well known from European LPSs (Antoine et al., 2001, 2009, 2016; Günster et al., 2001; Haesaerts et al., 2016). However, early glacial palaeosol horizons in mid-latitude European LPSs generally reflect drier conditions compared to the Eemian conditions (Mosbacher Humus zones) (Antoine et al., 2001, 2016; Schirmer, 2016), whereas in Mediterranean LPSs, more humid conditions seem to be reflected by polygenetic Eemian to early glacial palaeosol complexes (Günster et al., 2001; Ferraro, 2009; Wacha et al., 2011). Within the Rhône Rift Valley, the Eemian palaeosol complex in the Baix LPS is in accordance with LPSs from the south (Bourdier, 1958; Bonifay, 1965; Bosq et al., 2020b). Especially the reworking of the Eemian soil material as observed in the Baix LPS, followed by another phase of in situ pedogenesis, forming the 7 Btg horizon, agrees well with the deposits at the top of the S1 unit identified at the Collias LPS, ca. 100 km further south (Bwk developed in a reworked soil sediment with an OSL age of ca. 83 ka) and with similar horizons reported from LPSs in the Durance Valley (Fig. 1) (Bonifay, 1965; Bosq et al., 2020b).

Other local archives near the Baix LPS and in the Rhône Rift Valley also reflect our derived Eemian and early glacial climate conditions (Fig. 1). Eemian pollen records display summer-drought-resistant sclerophyllous taxa, with Ilex, Olea, Pistacia and evergreen Quercus (Beaulieu and Reille, 1984a; Tzedakis, 2007) and, at higher altitudes, mixed oak forest with the presence of Buxus and Taxus (Reille et al., 2000). The following change in vegetation towards tundra or steppe (Reille and Beaulieu, 1990; Mologni et al., 2021) was associated with a change to colder, drier, more continental climatic conditions, whereas the temperate deciduous vegetation recorded in between suggests a milder climate with higher precipitation and possibly higher temperatures. The gradual climate deterioration of the late Eemian is also reflected by the erosion of intensely weathered (partly red) sediments and a change within the pollen and human occupation remains derived from archaeological sites along the Ardèche and Payre rivers (Valladas et al., 2008; Rivals et al., 2009; Moncel et al., 2015) (Fig. 1).

The erosional discontinuity above the7 Btg horizon reflects an unstable phase with slope-wash processes, during which material of the 7 Btg horizon was mixed with slope-wash loess, followed by a phase of loess accumulation. In a subsequent more stable phase, the lower interstadial palaeosol (represented by the 6 Bw3 horizon) developed. Loess accumulation again set in, while pedogenesis continued, as indicated by the gradual upper boundary of the lower interstadial palaeosol, which reflects the transition from predominating pedogenesis to predominating loess accumulation.

The OSL screening yielded “age estimates” of around 65–75 ka for the 6 BCk/Btg and 6 Bw3 horizons, suggesting the deposition took place during the transition from the late early glacial period (MIS 5a) to the Lower Pleniglacial (MIS 4). Thus, the erosion phase may correlate with Mélisey II (Guiter et al., 2003), whereas the subsequent more stable phase probably correlates with St Germain II (St Geneys 2 (Reille et al., 2000)) and/or Ognon I (Beaulieu and Reille, 1984b, 1989). The transition from the early glacial period to the Lower Pleniglacial in the Baix LPS is thus characterised by a marked climate deterioration. This agrees with central European LPSs, where unstable conditions are reflected by the Niedereschbach zone or Harmignies colluvial deposit (Antoine et al., 2016; Lehmkuhl et al., 2016). At the Collias LPS in the Rhône Rift Valley, this lower interstadial soil was not observed. However, intense slope-wash processes (Dmm horizon) during this period indicated unstable conditions at Collias as well (Bosq et al., 2020b). The pollen records of Les Échets and Velay (Beaulieu and Reille, 1984a, b) also indicate a shift to cooler and drier conditions, including a warmer phase within the general trend of climatic deterioration.

The following enhanced loess accumulation (highest silt content and U ratio) and decreasing weathering intensities (lowest clay content and clay / silt ratio) reflect drier and colder conditions (Fig. 11). The coarse silt fraction is interpreted as having been transported in short-term, near-surface to low suspension clouds within a relatively limited distance (Tsoar and Pye, 1987; Vandenberghe, 2013; Újvári et al., 2016). As it has been proven for other LPS sites along the Rhône Rift Valley (Bosq et al., 2018, 2020b), we also assume that the major part of the source of the loess deposits at the Baix LPS was the seasonally dried-out riverbed of the Rhône River during glacial times.

The only weakly weathered loess (5 BCk5 and 5 Ck2 horizons) reflects the predominance of loess accretion rates over soil formation rates during the Lower Pleniglacial at Baix, indicating generally dry and cold conditions. Also, the observed diverse forms and sizes of secondary carbonate accumulation (i.e. pseudomycelia, hard and soft carbonate nodules) and the upwardly increasing CaCO3 contents suggest a weakening of carbonate leaching under increasingly drier conditions (Fig. 11).

This phase of enhanced loess deposition coincides with increased loess accumulation in most of the European, including Mediterranean, LPSs (Antoine et al., 2001; Ferraro, 2009; Zerboni et al., 2015) and with high sedimentation rates at other sites in the Rhône Rift Valley (Tricart, 1952; Sierro et al., 2009; Bosq et al., 2020b). The associated cold and dry climate is also reflected in the decline of trees to semi-open boreal landscape with partly steppe vegetation in the valley (Beaulieu and Reille, 1989; Reille et al., 2000; Moncel et al., 2015).

The upper interstadial palaeosol in the Baix LPS is interpreted to represent a strongly truncated and partially reworked Calcic Cambisol, including the reworked 5 Bw2 horizon, underlain by the in situ 5 Bk2 and 5 BCk4 horizons. Its basal large carbonate nodules require a corresponding carbonate source, i.e. a thick decarbonated soil horizon above. Therefore, we assume that the most intensively weathered upper part of the Calcic Cambisol was eroded. A phase of increased loess accumulation followed. The deposits in which the horizons 5 Ck2, 5 BCk4, 5 Bk2, 5 Bw2, 4 BCk/Bw and 3 BCk3 developed accumulated during the end of MIS 4 and into MIS 3 (Figs. 9, 11). The uppermost subsamples analysed from the 3 BCk3 horizon may point to slightly younger “age estimates” (∼ 40 ka), as would be expected for sediment deposits above an in situ palaeosol. Fine-tuned alterations of screening “age estimates” may be camouflaged by possible subsample-to-subsample variations in radionuclide concentrations and water contents, which both have an impact on effective dose rates and OSL “age estimates” but were not considered for the simplified approach of the OSL screening. Nevertheless, the OSL screening supports the hypothesis that the strongly truncated upper interstadial Calcic Cambisol represents Middle Pleniglacial palaeosol remains. The soil formation phase might correspond to the interstadial periods of Pile–Goulotte (Moershoofd) dated to 50–43 ka and/or Charbon (Hengelo) (38–37 ka) (Guiot et al., 1993; Guiter et al., 2003). The assumed soil formation intensity of the Calcic Cambisol remains, including the prominent carbonate nodules, suggests temperate conditions with at least one phase of intense carbonate dissolution and reprecipitation. Comparably intensely developed brown truncated soil horizons are found in most of the European and Mediterranean LPSs, e.g. Remagen soils, Gräselberg soils (Antoine et al., 2001; Schirmer, 2016; Fischer et al., 2021; Wacha et al., 2011). In the Collias LPS in SE France, a very similar brown Bwk horizon with similar prominent carbonate nodules was identified. The likely period of this soil formation was after ∼ 55 ka and had ended by ∼ 39 ka, as suggested by OSL dating (Bosq et al., 2020b). During our field surveys, we observed the prominent carbonate nodules not only at Baix and Collias but also in several other LPSs in the Rhône Rift Valley. Therefore, these large cone-shaped carbonate nodules that underlie truncated remains of a brown Calcic Cambisol may serve as a useful stratigraphic marker in LPSs along the Rhône Rift Valley. Another possibly comparable brown soil horizon, although without any age provided, was described for the LPS of the Durance Valley (Bonifay, 1965).

The local assumed temperate conditions are supported by the pollen of the MIS 3 interstadials, pointing to deciduous Quercus (at least from 39.405±2250 BP), associated with Abies and Carpinus (Beaudouin et al., 2005). Thus, around ∼ 40 000 BP, the vegetation in the wider region was mainly composed of broad-leaved temperate taxa, whereas at higher altitudes, the last regional presence of Picea was recorded (Reille and Beaulieu, 1990; Beaudouin et al., 2005). With respect to the local hydrology, malacological analyses of a palustrine level at the site of Mignet College in Aix-en-Provence, correlating also with the Pile–Goulotte (Moershoofd) interstadial, pointed to a marsh habitat, which may support assumed enhanced humidity in the region for this period (Magnin and Bonnet, 2014).