Further constraints on the Chauvet caveartwork elaborationBenjamin Sadiera,1, Jean-Jacques Delannoya, Lucilla Benedettib, Didier L. Bourlèsb, Stéphane Jailleta, Jean-MichelGenestec, Anne-Elisabeth Lebatardb, and Maurice Arnoldb

aLaboratoire Environnements et Dynamiques des Territoires de Montagnes, Université de Savoie/Centre National pour la Recherche Scientifique, Campusscientifique, 73376 Le Bourget-du-Lac cedex, France; bAix Marseille Univ., Centre Européen de Recherche et d'Enseignement des Géosciences del'Environnement, Unité Mixte de Recherche Centre National pour la Recherche Scientifique, 7330, BP 80, 13545 Aix-en-Provence, France; and cCentreNational de Préhistoire—Direction générale des patrimoines-Unité Mixte de Recherche 5199 de la Préhistoire à l'Actuel: Culture, Environnement etAnthropologie, 38, rue du 26e R.I., 24 000 Périgueux, France.

Edited by Kenneth Farley, California Institute of Technology, and accepted by the Editorial Board March 20, 2012 (received for review November 12, 2011)

Since its discovery, the Chauvet cave elaborate artwork called intoquestion our understanding of Palaeolithic art evolution and chal-lenged traditional chronological benchmarks [Valladas H et al.(2001) Nature 413:419–479]. Chronological approaches revealinghuman presences in the cavity during the Aurignacian and theGravettian are indeed still debated on the basis of stylistic criteria[Pettitt P (2008) J Hum Evol 55:908–917]. The presented 36Cl CosmicRay Exposure ages demonstrate that the cliff overhanging theChauvet cave has collapsed several times since 29 ka until the seal-ing of the cavity entrance prohibited access to the cave at least21 ka ago. Remarkably agreeing with the radiocarbon dates ofthe human and animal occupancy, this study confirms that theChauvet cave paintings are the oldest and the most elaborate everdiscovered, challenging our current knowledge of human cognitiveevolution.

absolute dating ∣ cosmonuclide ∣ remote sensing ∣ rockart ∣ rockfall

Chauvet cave, in Vallon Pont d’Arc, Ardèche, France, is a siteof exceptional scientific interest for a number of reasons: (i)

the variety of its majestic parietal; (ii) very good conservation ofthe floor and wall ornamentations, exhibiting human and animalimprints; (iii) revelations of unknown techniques in Palaeolithicrock art (such as stump drawing); (iv) predominance of rarethemes such as felines and rhinoceroses; and (v) unequalled aes-thetic delivery (1). On the basis of stylistic comparison withknown elements, the Chauvet cave rock ornamentations were in-itially estimated as being Solutrean (22–17 ka BP) and Magdale-nian (17–10 ka BP) (2). The first radiocarbon dates ranging fromapproximately 30 to approximately 32 14C ka BP (3, 4) thus dis-agreed with stylistic analyses such as formalized by Leroi-Gouran(5). Although currently confirmed by 82 radiocarbon dating (6,7), by crossing dating methods (8, 9), and supported by the recentdiscovery of manifestations of sophisticated Aurignacian prehis-toric art (40–28 ka BP) in the Swabian Jura (10), it still remainsunclear whether the current stylistic framework should be aban-doned in favor of radiocarbon chronologies. Additional robustchronological constraints are therefore critical in establishingChauvet cave as a reliable benchmark in the absence of compar-able equivalent (1, 11, 12). If corroborated by an independentmethod, the absolute chronological framework of the Chauvetdrawings will indeed establish them as the oldest and most ela-borate Aurignacian paintings ever discovered. This will further-more confirm the existence of an already extremely mature art atthat time period during which only few elaborate engraving areknown (10), but no other paintings (13, 14). To fill this knowledgegap, a geomorphological study combined with 36Cl dating ofrockslide surfaces overhanging the cave entrance was conducted.

The Chauvet Cave ClosureGeomorphological studies carried out at Chauvet cave during thelast 10 y have unambiguously demonstrated that only one en-trance to the cave existed at the time it was visited by humans

(1, 15). Delannoy et al. (15, 16) definitively rule out the possibilityof another entrance, demonstrating that any other hypothesizedaccess would imply either going through an 80 m vertical access orentering on the other side of the plateau with a 2-km walk under-ground through karst systems disconnected from the Chauvetcave one (Fig. 1).

This study aims to date the blocking of the paleo-entrance byrockfall deposits, which definitively closed access to Chauvet caveuntil its rediscovery in 1994. In addition to geomorphological andgeological evidences of rockfall deposits, an original 3D surveymethod was used to map the extension of the rockfall in thishighly secured and protected site. The deposit (between 185 and198 m asl), whose maximum thickness is 10 m, has a total run-outdistance of approximately 80 m from the top of the cliff [meanelevation of 245 m above sea level (asl)], the height of the cliffbeing 47 m (Fig. 2). The location and geometry of the paleo-caveentrance were reconstructed using 3D mapping tools (15). Thetotal rock mass volume obstructing the cave entrance was esti-mated to be approximately 4;500 m3 using a high-density digitalmodel (precision 8� 2 cm) produced by terrestrial Light Detec-tion and Ranging (LiDAR) from the missing cliff volume corre-sponding to the well-preserved subcircular concave rockfall scarsoverhanging the paleo-entrance. In addition, located on lowerCretaceous limestone (Urgonian), the area in front of the paleo-entrance has the geometrical characteristics of an accumulationzone (15). A detailed survey of the 47-m high cliff overhangingthe cave entrance revealed three concave rockslide scars that maybe linked to the dislocation of several limestone monoliths. Themorphology of the overhanging cliff corroborates that the sourcesof the rockfalls lie between the Pilier d’Abraham, a prominentmonolith east of the cave entrance, and another prominent arêteroughly 25 m from the Pilier d’Abraham to the west. In betweenthese two prominent features, two embossed edges as well as sur-face differences allow three surfaces to be defined, from west toeast: E1, E2 and E3. These probably correspond to three differ-ent gravitational events involving debris volumes of 500 m3,1;300 m3 and 2;100 m3 (�10%), respectively (15, Fig. 2).

The blocks resulting from the collapse of the limestone mono-liths have formed an apron of fallen rocks on the hillside as wellas inside the cave, which can be identified thanks to several ob-jects characteristic of the endokarst system in their genetic place.

Author contributions: B.S., J.-J.D., and J.-M.G. designed research; B.S., J.-J.D., L.B., D.L.B.,S.J., and J.-M.G. performed research; B.S., L.B., and D.L.B. contributed new reagents/analytic tools; B.S., L.B., D.L.B., S.J., A.-E.L., and M.A. analyzed data; B.S., and J.-J.D.,L.B., D.L.B., S.J., J.-M.G., A.-E.L., and M.A. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. K.F. is a guest editor invited by the Editorial Board.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. E-mail: Benjamin.Sadier@univ-savoie.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1118593109/-/DCSupplemental.

8002–8006 ∣ PNAS ∣ May 22, 2012 ∣ vol. 109 ∣ no. 21 www.pnas.org/cgi/doi/10.1073/pnas.1118593109

Dow

nloa

ded

by g

uest

on

June

16,

202

0

In addition to the heterogeneous arrangement of the collapsedblocks, the oldest stalagmitic flowstones (>300 ka, 9) on the cavewalls show some block impacts, suggesting the occurrence ofhighly energetic processes (12). These impacts rule out a cryoclas-tic agent on the deposit’s emplacement. Furthermore, the fallenrocks are currently in a progressive compaction phase rather thanaccretion by the addition of cryoclastic elements (15). The closureof Chauvet cave has most likely resulted from several successiverockfalls, the last of which definitively sealed the entrance. In anattempt to date the collapses and the closure of the entrance, pre-vious studies used U/Th dating of speleothems sealing the topof the rockfall deposit, yielding a maximum age of 11.5� 0.2 ka(9, 15). However, interruption of calcite precipitation during theLast Glacial Maximum hampers the use of this dating methodbetween 25.5 to 15 ka (9), and thus only allows a minimum ageestimate for the closure of the cave. Finally, 14C dating of archae-ological layers underlying the rockfall deposits into the cave haveshown Aurignacian occupancy phases at the entrance (15). As inHillaire Room and Megaloceros Gallery (7, 8), these archaeolo-gical layers are indeed older than 27 cal 14C ka BP.

Rockfall events that expose deeply buried limestone can bedirectly dated using in situ-produced 36Cl concentrations accu-mulated in scar surface samples, because they have been continu-ously exposed to cosmic rays (17, 18). Twenty-two samples werecollected on the E1, E2, E3, and the Pilier d’Abraham surfaces;two samples were collected on the top surface above the studiedcliff to estimate the local denudation rate, and a sample was takenfrom the Treuil cave, which constitutes part of the same karsticnetwork as the Chauvet cave (Fig. 1). Samples were prepared

according to Stone et al. (19), and 36Cl concentrations were mea-sured at the national AMS facility ASTER (CEREGE, Aix-en-Provence, France). The raw data are presented in Table S1. The36Cl concentrations measured in the Pilier d’Abraham samplesrange from 3.1 to 3.4 × 105 atoms 36Cl∕g rock. The 36Cl concen-trations measured in the scar samples are significantly lower, ran-ging from 1.6 to 3.0 × 105 atoms 36Cl∕g rock. Compared to thescars and the Pilier d’Abraham samples, the 36Cl concentrationsmeasured in the top surface samples are one order of magnitudehigher, at 1.2 and 1.1 × 106 atoms 36Cl∕g rock, while that mea-sured in the Treuil cave sample is at least one order of magnitudelower, at 1.2� 0.2 × 104 atoms 36Cl∕g rock. This latter concen-tration was measured in a sample laterally shielded by 12.4 m oflimestone and thus represents the natural background 36Cl con-centration at that depth from in situ muogenic production.

Both samples collected on the top surface, old enough to havereached steady-state (20) with respect to the accumulation of 36Cl(21), yielded similar denudation rates of 18.5 and 21.5 mm∕ka.This indicates that the mean denudation rate for a horizontal lime-stone surface at our site is approximately 20 mm∕ka. Consideringthat the scars and the Pilier d’Abraham samples lie on an almostvertical cliff on which water does not stagnate, thus experiencing alower denudation rate than on an horizontal surface and accordingto previously published values on similar environments and settings(22), the exposure age of the samples collected on the rockfall scarsand the Pilier d’Abraham were calculated using a denudation rateof 10 mm∕ka.

Moreover, the samples from the scar surfaces have probablyaccumulated nucleogenic production underneath the shielding

Fig. 1. Chauvet cave surroundings. (A) Topographical map showing the Chauvet cave network in red and the other karst systems in yellow. (B) Simplifiedgeological cross-section. (C) Map view of the Chauvet cave network.

Sadier et al. PNAS ∣ May 22, 2012 ∣ vol. 109 ∣ no. 21 ∣ 8003

EART

H,A

TMOSP

HER

IC,

AND

PLANETARY

SCIENCE

SANTH

ROPO

LOGY

Dow

nloa

ded

by g

uest

on

June

16,

202

0

provided by rock slab that has then subsequently fallen to the cliffbase. To calculate accurate rockfall scar sample surface exposureages, it is therefore necessary to subtract this 36Cl contribution.The LiDAR-based 3D reconstruction of the cliff overhangingChauvet cave entrance constrained the minimum shielding depth,thus the rock slab thickness, at each sampling point. They rangefrom 4.5 to 6.5 m (Table S2). Maximizing the nucleogenic produc-tion underneath the rock slab implies considering that the expo-sure time beneath the rock slab is similar to the calculated oldestage (35.4� 2.1 ka, Table 1) of the most stable closest arête, thePilier d’Abraham. This latter age was in addition maximized bysubtracting the minimum muogenic 36Cl background componentof 1 × 104 atoms 36Cl∕g rock, as deduced from the Treuil cavesample. Applying this maximum exposure duration, the nucleo-genic 36Cl contribution from the shielded rock slab were then cal-culated for each rockfall scar surface sample at the reconstructedshielded depths prior to the rockfall and subtracted from themeasured concentration. The obtained values range from 1.4to 2.7 × 104 atoms 36Cl∕g rock (Table S1), about 10% of the total36Cl inventory on those samples. The geographic location, pro-duction rates (23), scaling (24) and shielding factors (25), and

CRE ages of the rockfall scar surface samples accounting for bothdenudation rate and maximized (muogenic 36Cl backgroundcomponent: 1.4 × 104 atoms 36Cl∕g rock) shielded 36Cl contribu-tion, are presented in Table 1 and Tables S1 and S2. In a veryconservative way, because in all calculations all possible contribu-tions of 36Cl production prior to the rockfalls have been maxi-mized, the presented ages are minimized rockfall scar ages.

Results and DiscussionThe CRE age distribution of all rockfall scar surface samples(Table 1 and Fig. 3) attests to three temporally distinct events thatspatially correlate with the three surfaces previously identified bythe LiDAR study and their surface state of evolution. Those agesyielded minimized weighted mean ages of 29.4� 1.8 ka for theE1 surface, 23.5� 1.2 ka for the E2 surface, and 21.5� 1.0 kafor the E3 surface. The particularly well-constrained E3 surfaceage implies that the most recent rockfall occurred at least 21 kaago and that no major gravitational event affecting the overhan-ging paleo-entrance has occurred since then. Considering that thefinal closure of the paleo-cave entrance could only be attributedto rockfall talus, and that a more recent collapse would have left amorphological imprint associated with younger CRE ages on

Fig. 2. 3D LiDAR view of the modern Chauvet cave area showing the sample positions on the rockfall scars and the Pilier d’Abraham. Purple and brownsurfaces correspond to rockfall deposit and bedrock, respectively. Continuous red lines mark the limit of the two prominent arêtes. Dashed lines (yellowand white) delineate the contour of the three scar surfaces, E1, E2, and E3. Dashed green line marks the top surface. Samples are located with yellow point.Corresponding 36Cl ages are given in Table 1.

8004 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1118593109 Sadier et al.

Dow

nloa

ded

by g

uest

on

June

16,

202

0

the rockwall, closure of the Chauvet cave entrance must at leastcoincide with the last rockfall event dated at 21.5� 1.0 ka.

The presented 36Cl CRE ages are not only coherent withthe minimum ages derived from U/Th results obtained on spe-

leothems that sealed the rockfall deposit (9, 15) but, more impor-tantly, they also agree with the reported 14C dating and thussupport the deduced phases of human and animal occupancy ofthe cavity. Indeed, none of the available radiocarbon dates (32determinations on large mammal bones and 82 on charcoal sam-ples) calibrated using the INTCAL09 curve (26) is younger than23 cal 14C ka BP. The distribution of ages obtained on charcoalfrom both the walls and floor (3, 7, 8), some of which being pro-duced in the framework of a cross-comparative project betweenseveral independent laboratories confirming the validity of radio-carbon dating (8), indicate two successive phases of human occu-pancy: the first one from 37.5 to 33.5 cal 14C ka BP, and thesecond one from 32 to 27 cal 14C ka BP (Fig. 3).

None of the charcoal dates is younger than 27 cal 14C ka BP.14C dating of large mammal bones, all species being included(Ursus spelaeus and Capra ibex), range from 42 cal 14C ka BPto 23 cal 14C ka BP (27, 28, 29)—the youngest age being asso-ciated with an ibex bone (19 105� 150 14C yr BP). This date fitstightly with the last two rockfall events. Despite the number ofsamples taken for floor surface dating, the absence of more re-cent ages (Fig. 3) strongly supports a direct link between the endof human and large animal occupancy and the closure of theChauvet cave entrance by the two major rockfalls dated at 23.5�1.2 ka and 21.5� 1.0 ka. In fact, this remarkable concordancebetween the end of the animal occupancy in the cave and the ob-struction of its unique entrance by rockfalls (Fig. 3) inevitablyleads to the conclusion that the Chauvet rock art is most likelyolder than 21 ka and thus at least Gravettian.

Although not specifically dedicated to the dating of the Parie-tal ornamentation of the Chauvet cave, this interdisciplinary ap-proach has yielded important data for the chronology of Chauvetcave by directly dating the different collapses that affected therock wall overhanging the entrance. The results corroborate pre-vious 14C and U/Th chronologies on paintings, charcoals on theground, animal bones, speleothems andmean that aMagdalenianand/or Solutrean age for the cave artwork can be discarded. Allthese internally consistent dates obtained on several materialtypes (bones, charcoals, charcoal pigments, speleothems, rockfallscars) using three different and independent methods demon-strate that the Chauvet cave can now be considered as a bench-

Table 1. 36Cl ages of the surface samples from the rockfall scarsoverlying the Chauvet cave entrance. Yellow, orange and bluecorrespond to samples belonging to the three scars, E1, E2 and E3,respectively, while light yellow outlines the two Pilier d’Abrahamsamples.

Sample identity 36Cl age (ka)* 36Cl weighted mean age (ka)†

1.01 30.2 ± 1.929.4 ± 1.81.02 28.6 ± 1.8

1.03 24.1 ± 1.5

23.5 ± 1.2

2.01 22.6 ± 1.52.02 25.9 ± 1.62.03 25.0 ± 1.62.04 23.3 ± 1.53.01 21.1 ± 1.31.05 19.1 ± 1.3

21.5 ± 1.0

1.06 21.4 ± 1.42.05 19.0 ± 1.32.06 21.0 ± 1.42.07 22.5 ± 1.72.08 14.3 ± 1.12.09 21.9 ± 1.43.02 23.5 ± 1.63.03 24.8 ± 1.63.04 22.2 ± 1.43.06 21.0 ± 1.34.01 22.0 ± 1.46.01 35.4 ± 2.1

34.5 ± 2.06.02 33.7 ± 1.9

*Individual ages are reported with 1σ uncertainty considering analyticalerrors only.

†Weighted means were calculated after disregarding outliers in italic. Theseoutliers are identified using a Chi-square test with 95% confident interval.Mean ages are reported with 1σ uncertainty considering both analytical andthe production rate uncertainties. Note that the identified outlier yields anage younger than the weighted mean which is coherent with minor postgravitational readjustment of the cliff.

1015202530354045

Human occupancy

Animal occupancy

1015202530354045

E3

E1

E2

A

Pro

babi

lity

dens

ity fu

nctio

nP

roba

bilit

y de

nsity

func

tion

Cl age in ka36

C age in ka BP14

2

Fig. 3. Probability density plot of the 36Cl ages in ka of the rockfall scar samples and of the calibrated radiocarbon ages of the human and animal occupationinside Chauvet cave. Note the good agreement between the end of human and animal occupation and the age of the cave’s closure. A stands for Pilierd’Abraham ages, and E1, E2, and E3 are for the ages corresponding to the three scars.

Sadier et al. PNAS ∣ May 22, 2012 ∣ vol. 109 ∣ no. 21 ∣ 8005

EART

H,A

TMOSP

HER

IC,

AND

PLANETARY

SCIENCE

SANTH

ROPO

LOGY

Dow

nloa

ded

by g

uest

on

June

16,

202

0

mark in archaeology and rock art science. These results havesignificant implications for archaeological, human, and rock artsciences and seriously challenge rock art dating based on stylisticcriteria (4, 30).

ACKNOWLEDGMENTS. The 36Cl measurements were performed at the ASTERAMS national facility (CEREGE, Aix-en-Provence) that is supported by theINSU/CNRS, the French Ministry of Research and Higher Education, IRD,

and CEA. The French Ministry of Culture and the Direction Regional del’Environnement, de l’Aménagement et du Logement (DREAL) are acknowl-edged for sampling authorization. We thank G. Perazio for his help duringLiDAR surveys; J. Arnaud, G. Vermorel, B. Jarry, and J. Rambo for their helpwith sampling; V. Guillou, G. Aumaître, and K. Keddadouche for their helpduring chemical extraction and measurements; and N. Marriner for proof-reading the manuscript. We particularly appreciated the constructive com-ments of both reviewers, which enabled us to greatly strengthen ourarguments and thus increase the relevance of our findings.

1. Clottes J (2003) Return to the Chauvet Cave: Excavating the Birthplace of Art (Thamesand Hudson, London).

2. Clottes J (1995) Postface: The Chauvet cave today. La Grotte Chauvet à Vallon-Pont-d’Arc, eds JM Chauvet, E Brunel Deschamps, and C Hillaire pp 81–116 Paris: Seuil Eds,(in French).

3. Clottes J, et al. (1995) The paleolithic paintings of the Chauvet cave in Vallon Pontd’Arc (Ardeche, France); direct and indirect dating with the radiocarbon method.CR Acad Sciences Paris 320:1133–1140 (in French).

4. Valladas H, et al. (2001) Evolution of prehistoric cave art. Nature 413:479.5. Leroi-Gourhan A (1965) (1965) Préhistoire de l’art occidental [The Art of Prehistoric

Man in Western Europe]. (Mazenod, Paris), (in French).6. Valladas H (2003) Direct radiocarbon dating of prehistoric cave paintings by accelera-

tor mass spectrometry. Meas Sci Technol 14:1487–1492.7. Valladas H, et al. (2005) Review of 14C dating performed on charcoals from the Chau-

vet cave. Bull Soc Préhist Fr 102:109–113 (in French).8. Cuzange MT, et al. (2007) Radiocarbon intercomparison program for Chauvet cave.

Radiocarbon 49:339–347.9. Genty D, et al. (2004) TIMS U/Th and 14C AMS ages of the Chauvet cave stalagmites:

Interest for the chronology of natural and humanevents of the cave. CR Palevol 3:629–642 (in French).

10. Conard NJ (2009) A female figurine from basal Aurignician of Holhe Fels Cave in south-western Germany. Nature 459:248–252.

11. Moro Abadia O, Gonzales Morales MR (2007) Thinking about “style” in the Poststy-listic Era: Reconstructing the stylistic context of Chauvet Cave. OJA 26:109–127.

12. Chippindale C, Tacon P, eds. (1998) The Archeology of Rock Art (Cambridge Univ. Press,Cambridge), pp 90–111.

13. Pettitt P (2008) Art and the middle-to-upper Paleolithic transition in Europe:Comments on the archeological arguments for an early Upper Paleolithic antiquityof the grotte Chauvet art. J Hum Evol 55:908–917.

14. Balter M (2008) Going deeper in the grotte Chauvet. Science 321:904–905.15. Delannoy JJ, et al. (2010) Chauvet cave’s paleo entrance reconstruction using geomor-

phological and 3D tools. Karstologia 56:21–38 (in French).16. Delannoy JJ, et al. (2004) Contribution to geomorphological study for global under-

standing to prehistorical cave—the Chauvet cave (Ardèche—France). Karstologia44:25–42 (in French).

17. Benedetti L, et al. (2002) Post glacial slip history of the Sparta fault (Greece) deter-mined by 36Cl cosmogenic dating: Evidence for non-periodic earthquake. GeophysRes Lett 29:87.1–87.4.

18. Stock GM, Uhrhammer A (2010) Catastrophic rock avalanche 3,600 years BP from ElCapitan, Yosemite Valley, California. Earth Surf Proc Landforms 35:941–951.

19. Stone JO, Allan GL, Fifield LK, Cresswell RG (1996) Cosmogenic chlorine-36 fromcalcium spallation. Geochim Cosmochim Acta 60:679–692.

20. Clauzon G (1996) Sequence boundaries and geodynamic evolution. Géomorphologie:Relief, Processus, Environnement 1:3–22 (in French).

21. Gosse JC, Phillips FM (2001) Terrestrial in situ cosmogenic nuclides: Theory and appli-cation. Quat Sci Rev 20:1475–1560.

22. Curry AM, Morris CJ (2004) Lateglacial and Holocene talus slope development androckwall retreat on Mynydd Du, UK. Geomorphology 58:85–106.

23. Braucher R, et al. (2011) Production of cosmogenic radionuclides at great depth:A multi element approach. Earth and Planet Sci Lett 309:1–9.

24. Stone JO (2000) Air pressure and cosmogenic isotope production. J Geophy Res105:23753–23759.

25. Dunne J, Elmore D, Muzikar P (1999) Scaling factors for the rates of production ofcosmogenic nuclides for geometric shielding and attenuation at depth on sloped sur-faces. Geomorphology 27:3–11.

26. Reimer PJ, et al. (2009) Intcal09 and Marine09 radiocarbon age calibration curves,0–50000 years cal BP. Radiocarbon 51:1111–1150.

27. Bon C, et al. (2008) Deciphering the complete mitochondrial genome and phylogenyof the extinct cave bear in the paleolithic painted cave of Chauvet. Proc Natl Acad SciUSA 105:17447–17452.

28. Bocherens H, Drucker D, Billiou D, Geneste JM, van der Plicht J (2006) Bears andhumans in Chauvet cave (Vallon pont d’arc, Ardèche, France): Insights from stableisotopes and radiocarbon dating of bone collagen. J Hum Evol 50:370–376.

29. Fosse P, Philippe M (2005) The fauna of the Chauvet cave: Paleobiology and anthro-pozoology. Bull Soc Préhist Fr 102:89–102 (in French).

30. Valladas H, et al. (1992) Direct radiocarbon dates for prehistoric paintings at theAltamira, El Castillo and Niaux caves. Nature 357:68–70.

8006 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1118593109 Sadier et al.

Dow

nloa

ded

by g

uest

on

June

16,

202

0