The Chauvet-Pont d’Arc cave is situated in France in the Ardèche valley near Vallon-Pont-d’Arc (44° 23′ 14′′ N, 4′ 25′ 07′′ E, Figure 1A, B, D). Discovered in 1994, it is universally known for the richness and the age of its Paleolithic paintings [Clottes 2001; Clottes and Geneste 2007; Delannoy and Geneste 2020]. The 14C datings of charcoal and paintings indicate two periods of occupation attributed to the Aurignacian and to the Gravettian (from 37.5 to 33.5 ka cal BP and from 31 to 28 ka cal BP; Clottes et al. 1995; Cuzange et al. 2007; Quiles et al. 2016; Valladas et al. 2005). These dates coincide with one of the activity phases of the Bas-Vivarais volcanism. Several recent 40Ar/39Ar datings have actually dated these eruptions between 29 ± 10 ka and 35 ± 8 ka (95% probability uncertainty Nomade et al. 2016; Sasco 2015; Sasco et al. 2017). These dates fully cover the 14C AMS datings [Quiles et al. 2016] and the thermoluminescence (TL) datings [Guibert et al. 2015], which correspond to the human settlements responsible for numerous parietal representations in the cave. Among them, spray-shape signs (Figure 1C) were hypothetically ascribed to depiction of strombolian volcanic eruptions [Nomade et al. 2016].
In the context of the pluridisciplinary study of the cave financed by the French Culture Ministry the analysis of heavy minerals of the sediments was undertaken. This study led to the discovery of volcanic minerals attributed to a cryptotephra subcontemporaneous with the sedimentary infilling of the entrance area of the cave. These results, presented here, corroborate the contemporaneity of the recent volcanism of the Bas-Vivarais with the history of the cave during a period close to its occupation by Aurignacians.
2. Stratigraphical and sedimentological context
The volcanic minerals were identified in the sediments of a limited archeological pit located at about twenty meters from the old entrances which are now filled by deposits related to several processes (Figure 2). Four phases were distinguished in the clogging of the entrance [Debard et al. 2016]:
- The first one constitutes the origin of a gravity scree ( = lower scree) mainly due to the gelifraction of the porch and entry area, to the mechanical expansion of the vaults as well as to the reworking of old superficial deposits. The spreading of the deposits on about forty meters up to the entrance of the Chamber of the Bear Hollows (“Salle des Bauges”) is attributed to mobilization by creep and solifluction.
- The second one also corresponds to a gravity scree (intermediate scree) less expanded than the lower one. Their very similar components suggest a same origin.
- The third phase corresponds to blocks generated by the collapse of the cliff. This collapse mainly occurred in two main phases determined by 36Cl cosmogenic dating (23,500 ± 1200 and 21,500 ± 1000 years—Sadier et al. 2012) and gave rise to the definitive cave closure. This set was later covered by fine brown sediments originating from reworking of superficial deposits.
- After the closing of the porch, the ultimate registered events correspond to the desquamation of the equilibrium vaults resulting in small sedimentary clasts, to water runoff, to recent concretions growth, to the settling down of the scree.
The archeological pit (named GE1) carried out between 2006 and 2010 [Gély and Maksud 2009; Gély et al. 2012] is located in a small gallery connecting the old porch to the Chamber of the Bear Hollows (“Salle des Bauges”) and corresponding to one of the natural paths used by large vertebrates and prehistoric humans (Figure 2). Of small dimensions (50 × 50 cm), 80 cm deep, the pit intersected the lower scree and, near the surface, a gravel of clasts issued from the desquamation of the vault.
Six beds have been defined (Figure 3a, b):
- Bed 1: near the surface, this layer is mainly composed of small plates of desquamation (c.1a). Underneath is found dry gravel of elements righted in counterslope (c.1b). Among the clasts of the first level was a fistulous concretion which was dated by U/Th to 13,315 ± 79 years [Genty 2008]. This dating as well as the one obtained on the stalagmite Chau-stm 8, which seals the scree near the archeological pit and whose basis is dated at 11,500 ± 170 years [Genty et al. 2005] allow to date this bed between the Bølling–Allerød warming and the beginning of the Holocene (Lateglacial). Level c.1b supplied a flint attributed to the Upper Paleolithic, a trace of the last human incursion before the porch was finally clogged up and attributed to the Gravettian (31,000–28,000 cal BP; Quiles et al. 2016).
- Bed 2: this bed is a small gravel secondarily packed in a sandy-silty reddish matrix. Some clasts show a soft surface identified as bears’ polish. Bed 2b corresponds to the reworking of the uppermost part of bed 3.
- Bed 3: this bed is about 10 cm thick, it is less gravely, with a sandy and micaceous matrix very rich in charcoals responsible of its dark color. This layer contrasts with the reddish color of the remaining part of the sequence. It corresponds to the reworking of hearths probably at the origin of the significant thermal impacts observed close to the archeological pit on the ceiling and the walls [Ferrier et al. 2014]. Rubefied clasts or with blackish marks have been found in this bed. Dating of several charcoals (i.e. ranging from 36,000–37,000 cal BP; Quiles et al. 2016) attributes these hearths to the Aurignacian occupancy.
- Bed 4: this bed is divided in two, the upper part corresponding to a small dense gravel with angular granules due to runoff, while its lower part is coarser. The matrix is sandy-silty reddish brown becoming more yellow toward the bottom.
- Bed 5: this bed is made of an angular gravel packed in a red sandy sediment, more silty toward the base. At its base, a charcoal dated from between 36,361 and 34,797 cal BP [Quiles et al. 2016] is contemporaneous to the first Aurignacian occupation of the cave.
- Bed 6: also, coarse, contains wall elements (one block and clasts) polished by bears. The matrix is more sandy, brownish.
- In beds 2 to 5, the presence of sorted lamellar microstructures and occasionally layered caps at the surface of the clasts shows the role of cold weather (mainly frost periods) in this part of the scree and its spreading by solifluction toward the Chamber of the Bear Hollows.
The volcanic minerals characterizing the Chauvet cryptotephra were identified in the beds 6 to 1b, indicating a reworking of the initial tephra fall from the outside.
3. Material and methods
The mineralogical study was carried out on the 63 μm–2 mm sand fraction separated by wet sieving in an aqueous solution of sodium hexametaphosphate (dispersant) and removing of carbonates (postdeposition calcite coatings) by HCl 1N/2 hours treatment. Results are based on the study of ten samples distributed in the various beds of the archeological pit. The percentages between the silty-clayey fraction and the clayey fraction characterize clayey-silty sands and clayey-sandy silts. The examination of the light sandy fraction reveals quartz-rich sands (angular or slightly blunt quartz) poor in feldspars, issued from the regional granitic and metamorphic basement. Heavy minerals were separated with bromoform (samples CHAU 1–4 = beds 1b, 2, 3, 4) or with sodium polytungstate (samples CHAU 8–9 = beds 5a and 6, samples ARD 25–28 = beds 3, 4, 5, 6). Thus, they represent the heavy fraction (d > 2.8), including some minerals of the basement and all the volcanic minerals potentially present, except possible plagioclases of basaltic origin, not analyzed. These volcanic minerals either correspond to minerals typical of basaltic magma or to mantellic minerals issued from ultramafic xenoliths (peridotites/lherzolites). The percentages of heavy minerals vary between 0.85 and 3% of the sand fraction. The minerals have been identified with a polarization microscope and a binocular lens. In order to establish their frequency, which is slightly variable due to the grain size of the samples, several samples were counted on a squared grid. To specify their respective characterizations and origins, the various minerals identified by optical study were then analyzed with an electron microprobe. The handpicked minerals were included in epoxy resin and polished with abrasive papers and diamond powders. They were then analyzed with a WDS Cameca SX 100 microprobe at the Camparis center (Sorbonne University, Paris VI/CNRS) with column conditions 15 keV, 10 nA. These results were then compared to those of the analysis performed on minerals issued from tephras or peridotitic nodules from the Upper Pleistocene volcanism of Bas-Vivarais (Ardèche).
4. Mineralogy and characterization of the cryptotephra
The results of the heavy minerals counting of six samples are reported in Table 1 and Figure 4. They make it possible to appraise the frequencies of the various species and possibly to link them to the modal composition of the original host rocks. All the samples are largely dominated by olivine (73.8 to 89.7%). This one is mainly composed of xenomorph grains which can result from the mechanical fragmentation of peridotites nodules or from basaltic magma. Some automorphic specimens clearly originate from the last. The orthopyroxenes (xenomorph enstatite 3.9–7.6%) and the green clinopyroxenes (xenomorph chromiferous diopside, 2.5–6.6%) then represent minerals characteristic of peridotites xenoliths [e.g. Berger 1981]. The brown clinopyroxene (diopside) only plays a minor role (0.7–2%). It probably originates from a basaltic magma together with other pyroxenes such as those occurring in the alkali basalts from the Massif Central. The brown amphiboles, also little abundant (0.2–2%) are probably peridotites amphiboles as those described by Berger .
Heavy minerals spectra of samples containing Chauvet cryptotephra
|Sample||CHAU 1||CHAU 2||CHAU 3||CHAU 4||CHAU 8||CHAU 9|
These minerals are attributable to a basaltic volcanism rich in peridotites nodules (lherzolites) as well as lower crustal material which can only correspond to the Upper Pleistocene volcanism of the Bas-Vivarais located about 35 km away [e.g. Berger 1973, 1981; Sasco et al. 2017]. They come from a tephra brought by the wind at the probable origin of most of the other drifts (basement silts and sands) and scattered in the cave environment. The homogeneity of the minerals assemblage and the freshness of the minerals (e.g. automorphic olivine) lead one to think that this slightly reworked tephra is subcontemporaneous with the sediments including the latter. Thus, it can be defined as a cryptotephra. This fall probably results from a phreatomagmatic eruption associated with the formation of a maar. The explosive products (base surges and falls) are actually usually rich in peridotites minerals (see discussion). The origin of the fall from a strombolian eruption is much less probable. In fact, this type of eruption effectively produces basaltic ashes and scoriae which are not present in the Chauvet samples and are also generally poor in minerals.
The other heavy silicates are basement minerals, which play only a minor role (Table 1). These include garnets (0.7 to 5.9%), andalusite, sillimanite-fibrolite, tourmaline, and rutile. They could be alluvial minerals from the Ardèche river, coming from the Massif Central and reworked from old fluvial terraces by runoff or gravity process. But they were more probably brought by the wind. Some could also come from the cryptotephra, the explosive products of the maars being mainly rich in basement minerals. The opaque minerals, little represented (2.5 to 3.6%), correspond either to Fe–Ti oxides issued from the regional basaltic volcanism (magnetite, titanomagnetite), or to silicates and oxides of unknown origin (work in progress), or to rare chromium spinels (about 23% of Cr2O3) probably issued from peridotites.
5. Microprobe geochemical study of minerals
In order to precisely characterize the mineralogy and chemical composition of the minerals, a mineralogical study using an electron microprobe was undertaken and compared with the available analyses [Berger 1981] and new analyses of proximal volcanic falls (base surges) as well as peridotites from the Ardèche volcanism. Various characteristic analyses are reproduced on Table 2.
Selected microprobe analyses of minerals issued from Chauvet cryptotephra (sample CHAU 4, bed 4)
OL1–2: basaltic olivine—OL3–4: peridotite olivine—EN1–2: enstatite—CD1–2: chromiferous diopside—BD1–2: brown diopside. Olivines: numbers of ions on the basis of 4 O, pyroxenes: numbers of ions on the basis of 6 O.
Olivines are mainly characterized by the variation of their MgO contents. The analysis of automorphic grains reveals compositions comprised between 43.67 and 46.30% of MgO (Fo = 83.1–86.6%). Such compositions, poor in Mg, are typical of basaltic phenocrystals as confirmed by their automorphic habitus. After all the cross-checking we have done, it seems that the limit of composition between basaltic and peridotite olivines is at about 47% MgO. The lowering of magnesium in these basaltic olivines (e.g. Table 2, OL1–2) results in their iron enrichment (FeOt = 12.65–15.39% for the most ones) as shown by the diagram of Figure 5. Except for visibly broken specimens from basaltic origin with MgO contents <47%, the other xenomorphic olivines are more magnesian (MgO = 47.40–48.96%) and characterize olivines from peridotites (e.g. Table 2, OL3–4, Figure 3). However, these olivines are slightly less magnesian than the olivine from peridotites of the Vestide du Pal and Chambon maars analyzed by Berger , which mostly exceed 49% of MgO (48.71–49.82%); but based on six analyses performed on the phreatomagmatic tuff of Vestide. We have noticed that they were less magnesian and compatible with the compositions of the Chauvet cryptotephra (44.57–44.98–45.37–47.08–47.44–48.02% MgO). For comparison, the olivines from the Ray-Pic maar peridotites have compositions varying between 47.89 and 49.50% MgO. The olivines of microperidotite enclaves found within the strombolian tuff of Gravenne de Montpezat,were also analyzed. They are slightly more magnesian and mainly exceed 49% MgO. Although small, these variations seem to indicate significant slight differences of composition for the olivines of peridotites xenoliths from the Ardèche volcanism.
The orthopyroxenes necessarily come from peridotites of which they are one of the characteristic minerals with MgO contents near 32–33%. These are enstatites En87.8–89.8, Wo0.93–1.27 (Figure 6). Their analytical results (e.g. Table 2, EN1–2) are close to those of Berger  concerning the orthopyroxenes of peridotites from the Vestide du Pal and Chambon maars. Like them, they are slightly chromiferous (Cr2O3 = 0.22–0.34%). The other orthopyroxenes we analyzed (Ray-Pic maar, Borée maar, Gravenne de Montpezat scoria cone, Moula rockshelter tephra) showed relatively similar compositions.
The green clinopyroxene corresponds to the common clinopyroxene of peridotites [Berger 1981]. This is chromiferous diopside (calcic clinopyroxene Wo45–48, Figure 5). The results of the ten analyses performed (e.g. Table 2, CD1–2) show slightly less chromium-rich pyroxenes than those analyzed by Berger  for peridotites of the Vestide du Pal and Chambon maars: 0.54–0.81% compared to 0.75–1.36%. This difference is not explained. Only one sample differs with 1.34/1.48% Cr2O3.
The brown clinopyroxenes correspond to diopsides Wo44.97–48.30–En40.01–42.34–Fs10.68–13.87 (e.g. Table 2 BD1–2, Figure 6). These are basaltic pyroxenes common in the alkali basalts from the Massif Central and attributed to the basaltic magma pulverized by the phreatomagmatic eruption. They show typical amounts of Al2O3 (8.51–9.46%), TiO2 (2.15–2.53%), and FeOt (6.14–7.86%).
Analysis of some opaque minerals revealed the existence of several minerals of unspecified origin. The first one presents a chemical composition with 27–34.5% SiO2, 5–7% K2O, 15.4–20.3% FeOt, 13.9–17.7% Al2O3, 5.8–8.8% MgO, 3.8–4.5% TiO2. The second one is mainly composed of FeO (67–75% FeOt), but contains low percentages of SiO2, Al2O3, CaO, MnO, and P2O5 and cannot be ascribed to a true oxide.
6. Discussion, origin, and possible age of the cryptotephra
As previously seen, the mantellic minerals (peridotites minerals = magnesian olivine + orthopyroxene + chromiferous diopside) and the basaltic minerals (olivine + brown diopside) characterizing the cryptotephra of the Chauvet-Pont d’Arc cave can be ascribed without any doubt to a phreatomagmatic eruption related to a maar of the Upper Pleistocene volcanism of the Bas-Vivarais. They were pulverized and mixed by the phreatomagmatic explosions. The absence of basaltic ash (microscoriae) attributable to a strombolian eruption, the mixing of mantellic and basaltic minerals and the lower explosivity of this type of eruption led us to exclude a strombolian eruption at the origin of the cryptotephra.
The volcanism of the Bas-Vivarais is known since the work of Faujas de Saint-Fond . Its volcanic products are mainly alkaline basalts derived from a unique enriched mantle source [Chauvel and Jahn 1984; Downes 1987]. It benefited in particular from the works of Berger [e.g. Berger 1973, 1981, 2007] and is distinguished by the frequency of its ultramafic xenoliths [Berger 1981; Berger and Forette 1975]. It comprises 17 eruptive centers. Its activity is characterized by phreatomagmatic eruptions (maars) preceded or generally followed by a strombolian activity associated with the emission of basaltic flows filling pre-existing valleys.
The karst of the Bas-Vivarais, situated to the southeast, is characterized by the importance of its Upper Pleistocene infillings [Debard 1988, 1997]. The signature of volcanic eruptions from the Bas-Vivarais was evidenced in three of these karstic infillings: the Moula rockshelter in Soyons [Debard and Pastre 2008; Pastre et al. 1994], the Marzal II aven (or Flahault aven), and the Devès du Reynaud aven in Saint-Remèze [Debard and Pastre 2008]. Among them, the in situ tephra found in Soyons is the more representative example of a phreatomagmatic fall comparable to that detected in Chauvet. Among the recent maars of the Bas-Vivarais, four present an explosive activity and can be at the origin of the cryptotephra we described above: the Ray-Pic, the Vestide du Pal, the Chambon, and the Sapède [Berger 1981]. This last explosion crater can a priori be excluded, due to its essentially basaltic type [Berger 1981] as confirmed by our own observations and analyses. Derived tephras are therefore mainly composed of basalt clasts and their olivines are mainly basaltic in composition with 44.01–46.55% MgO (Fo = 86.3–87%). The Chambon maar emitted a base surge and lahars visible in the Fau valley above the hamlet. The mineralogical and microchemical study shows that these products both contain peridotite and basaltic minerals similar to Chauvet’ ones. The peridotites olivines (48.43–48.97% MgO) are however more abundant than the basaltic ones in the sample we analyzed. This maar cut in the basement could have been large enough to disperse widespreadly its products, but, due to little extended dispersal in the proximal area, it was probably less extended than those of the following maars. Among the following eruptions, the Vestide du Pal maar constitutes one of the largest explosive craters in the Massif Central (diameter = 1700 m, Berger 1973, 1981). It emitted important surges visible in its eastern and southern parts (Suc du Pal). The mineralogical study of these products rich in basement minerals pulverized by explosions, show that they contain both peridotite and basaltic minerals. These ones however comprise peridotite olivines (47.05–50% MgO) and peridotite pyroxenes. The basaltic olivines (44.57–45.37% MgO) are rare. However, this composition makes them compatible with the composition of Chauvet cryptotephra ones. The Ray-Pic maar also emitted large surges visible along the road D 215 to the east of the Ray-Pic cascade. The mineralogical study of these tephras also rich in basement minerals shows that they contain basaltic and peridotite minerals. The olivine shows a composition varying between 44.63 and 49.78% MgO (Fo = 85.2–90.9%), which clearly argues for these two origins and compositions close to those of Chauvet.
The cryptotephra-bearing beds are well dated due to the presence of microcharcoals ascribed to frequenting of the site by Aurignacians. 14C AMS datings gave ages comprised between 36,361–34,797 (bed c5) and 36,100–34,880 (bed c2a) years cal BP [Quiles et al. 2016]. Thus, the cryptotephra must have an age greater than or close to 36 ka. The Bas-Vivarais volcanism benefited from 14C datings [Berger 1973; Berger et al. 1975; Berger 2007], from TL datings [Guérin 1983; Guérin and Gillot 2007; Sanzelle et al. 2000], from paleomagnetism measurements [Rochette et al. 1993; Sasco 2015; Sasco et al. 2017], and from K–Ar and 40Ar/39Ar datings [Sasco 2015; Sasco et al. 2017]. Among them, the 40Ar/39Ar datings (ibid.) seem to be the most accurate. The 40Ar/39Ar age of the Vestide du Pal maar would thus be comprised between 25.8 ± 7.7 and 27.2 ± 8.2 ka (ibid.). It would thus belong to the recent phase of the Ardèche volcanism. This age seems too young to fit with the potential age of the Chauvet cryptotephra. Nevertheless, it was performed on a basalt following the phreatomagmatic eruption and not on the tephras coming from the maar explosions. It is also the same for the age of the Chambon maar dated at 18.8 ± 9.6 ka (ibid.). On the other hand, the age of the Ray-Pic maar comprised between 36.2 ± 11.3 and 32.2 ± 11.1 ka (ibid) is well compatible with the 14C datings of the beds containing the minerals of the Chauvet cave. This maar belonging to the middle phase of the Bas-Vivarais volcanism could be at the origin of the cryptotephra we discovered. Nevertheless, its products do not contain the peculiar minerals which are currently studied to define their origin (terrestrial or extraterrestrial).
The correlation with the other tephras identified in the karstic infillings of the region can be examined. The tephra of the Devès du Reynaud aven [Debard and Pastre 2008] is characterized by the presence of basaltic glass and its richness in biotite to which a peculiar provenance can be attributed. This composition is clearly different from the Chauvet cryptotephra. In the Marzal II (or Flahault), the basaltic and mantellic minerals reworked in set 5 [Debard and Pastre 2008] can also be attributed to a phreatomagmatic emission of the Upper Pleistocene volcanism of Ardèche. The data obtained by biochronology suggest that this fall could belong to the recent or middle phase of Ardèche volcanism. The minerals of this infilling could originate from the same fall as the minerals of the Chauvet cave’s cryptotephra or originate from another eruption like those of Vestide du Pal or Chambon. The tephra identified in the Moula rockshelter corresponds to a beautiful in situ phreatomagmatic fall [Debard and Pastre 2008; Pastre et al. 1994]. Rich in basement minerals, it contains both peridotite and basaltic minerals as illustrated by the composition of its olivines varying between 43.97 and 48.87% MgO. Its initial dating around 40 ka [Pastre et al. 1994] was contested according to a TL dating of 72 ± 12 ka [Sanzelle et al. 2000]. It is indeed worth noticing that this older age is hardly reconcilable with the sedimentological and paleoclimatic data suggesting a periglacial context rather than a temperate one such as the Saint-Germain 2 episode [Debard and Pastre 2008]. In the hypothesis of an age close to 40 ka, its correlation with the phreatomagmatic eruption of the Ray-Pic maar is possible. This hypothesis is in any case compatible with the composition of the products identified in these two entities.
The cryptotephra from the Chauvet-Pont d’Arc cave provides a new example of tephra fall from the Upper Pleistocene volcanism of the Bas-Vivarais, trapped in the karst of this area. Its possible attribution to the phreatomagmatic eruption of the Ray-Pic maar could be supported by the mineralogical and geochemical compositions of its products containing mantellic and basaltic minerals. It is also in agreement with the recent 40Ar/39Ar datings of the Ardèche volcanism [Sasco 2015; Sasco et al. 2017], giving an age subcontemporaneous with the beds containing the charcoals well dated by 14C [Quiles et al. 2016] and contemporaneous with the Aurignacian occupation and art, that is slightly older than or close to 36 ka. In the other hand, the cryptotephra could also originate from the Vestide du Pal maar, as it also contains peculiar minerals whose origin is still unknown and which were also found in sedimentary infilling of the Chauvet cave. The discovery of this slightly reworked tephra in this remarkable Paleolithic cave provides an additional evidence of the contemporaneity of the Ardèche volcanism with the Aurignacian occupation of this region. Our findings are however not the bulletproof that will legitimate the hypothesis according to which the spray-shape signs of the cave can be ascribed to the first depiction of a volcanic eruption [Nomade et al. 2016], but they surely reinforced the hypothesis that Aurignacians have been the witnesses of volcanic activity in the Ardèche region and that volcanic derived material did fall in the vicinity of Chauvet-Pont d’Arc cave about 36 ka ago.