Symbolic use of marine shells and mineral pigments by Iberian Neandertals

Joo Zilhoa,1, Diego E. Angeluccib, Ernestina Badal-Garcac, Francesco dErricod,e, Floral Danielf, Laure Dayetf, Katerina Doukag, Thomas F. G. Highamg, Mara Jos Martnez-Snchezh, Ricardo Montes-Bernrdezi, Sonia Murcia-Mascarsj, Carmen Prez-Sirventh, Clodoaldo Roldn-Garcaj, Marian Vanhaerenk, Valentn Villaverdec, Rachel Woodg, and Josena Zapatal
a University of Bristol, Department of Archaeology and Anthropology, Bristol BS8 1UU, United Kingdom; bUniversit degli Studi di Trento, Laboratorio di Preistoria B. Bagolini, Dipartimento di Filosoa, Storia e Beni Culturali, 38122 Trento, Italy; cUniversidad de Valencia, Departamento de Prehistoria y Arqueologa, 46010 Valencia, Spain; dCentre National de la Recherche Scientique, Unit Mixte de Recherche 5199, De la Prhistoire lActuel: Culture, Environnement et Anthropologie, 33405 Talence, France; eUniversity of the Witwatersrand, Institute for Human Evolution, Johannesburg, 2050 Wits, South Africa; fUniversit de Bordeaux 3, Centre National de la Recherche Scientique, Unit Mixte de Recherche 5060, Institut de Recherche sur les Archomatriaux, Centre de recherche en physique applique larchologie, 33607 Pessac, France; gUniversity of Oxford, Research Laboratory for Archaeology and the History of Art, Dyson Perrins Building, Oxford OX1 3QY, United Kingdom; hUniversidad de Murcia, Departamento de Qumica Agrcola, Geologa y Edafologa, Facultad de Qumica, Campus de Espinardo, 30100 Murcia, Spain; iFundacin de Estudios Murcianos Marqus de Corvera, 30566 Las Torres de Cotillas (Murcia), Spain; jUniversidad de Valencia, Instituto de Ciencia de los Materiales, 46071 Valencia, Spain; kCentre National de la Recherche Scientique, Unit Mixte de Recherche 7041, Archologies et Sciences de lAntiquit, 92023 Nanterre, France; and lUniversidad de Murcia, rea de Antropologa Fsica, Facultad de Biologa, Campus de Espinardo, 30100 Murcia, Spain

Communicated by Erik Trinkaus, Washington University, St. Louis, MO, December 5, 2009 (received for review October 16, 2009)

Two sites of the Neandertal-associated Middle Paleolithic of Iberia, dated to as early as approximately 50,000 years ago, yielded perforated and pigment-stained marine shells. At Cueva de los Aviones, three umbo-perforated valves of Acanthocardia and Glycymeris were found alongside lumps of yellow and red colorants, and residues preserved inside a Spondylus shell consist of a red lepidocrocite base mixed with ground, dark red-to-black fragments of hematite and pyrite. A perforated Pecten shell, painted on its external, white side with an orange mix of goethite and hematite, was abandoned after breakage at Cueva Antn, 60 km inland. Comparable early modern human-associated material from Africa and the Near East is widely accepted as evidence for body ornamentation, implying behavioral modernity. The Iberian nds show that European Neandertals were no different from coeval Africans in this regard, countering genetic/cognitive explanations for the emergence of symbolism and strengthening demographic/social ones.
behavioral modernity symbolism

| Iberia | Middle Paleolithic | shell ornaments |

ation, imitation without understanding, or independent Neandertal innovation (814). As this evidence comes from near the time of contact with modern humans in Europe, unresolved issues of dating and taphonomy impacting the broader paleoanthropological problem explain the persistence of the debate (1518). Here, we report secure evidence that, approximately 50 ka cal B. P., 10 millennia before modern humans are rst recorded in Europe, the behavior of Neandertals was symbolically organized and continued to be so until the very end of their evolutionary trajectory. This conclusion is based on the same types of nds and criteria used to assess the African and Near Eastern evidence pigment-stained, perforated marine shells and the associated procurement, processing, and application of colorants for body ornamentation. The evidence comes from two sites in the Murcia province of southeast Spain, Cueva de los Aviones (37357.30N, 0598.66W) and Cueva Antn (38351.84N; 12947.20W) (SI Appendix, Sections I and II). Results
The Sites. Aviones is a large cave whose ll has for the most part

tems of body decorationperforated and pigment-stained shells of marine mollusksare a feature of the Middle Stone Age (MSA) of southern Africa and the Middle Paleolithic (MP) of the Maghreb and Near East (1, 2). Dated to 70120 ka calendar (cal) B.P., these nds are widely accepted as evidence of symbolic thinking among the earliest anatomically modern humans of Africa and adjacent regions of Southwest Asia (3). An absence of similar nds from the coeval archeological record of Europe has supported the notion that Neandertals were devoid of symbolism, their cognitive inferiority explaining why modern humans dispersing out of Africa eventually replaced them across the entire continent approximately 40 ka cal B.P. (4, 5). In this view, claimed instances of Neandertal symbolism are disregarded on different groundse.g., as coming from old excavations and too ill-recorded to be certain or of ambiguous interpretation and in any case too scant to be meaningful. A case in point is the Mousterian cemetery of La Ferrassie (6), for which regional culture-stratigraphy indicates an age of approximately 6570 ka cal B.P. (7) and where the burial pit of an adult male contained a bone fragment decorated with four sets of parallel incisions, whereas that of a 3- to 5-year-old child was covered by a cupule-decorated limestone slab. In this context, the presence of body ornaments (namely, pierced and grooved animal teeth) in Neandertal-associated archeological cultures (such as the Chtelperronian of France) has been variously explained by stratigraphic mixing, accultur-

been eroded away by postglacial marine erosion (SI Appendix, Section I, Figs. S1S4). At the time of occupation, the Mediterranean was 5090 m lower (19), implying a distance of 1.57.0 km between site and shore (or more, because the Murcia coast is a subsidence area). A brecciated baulk preserved against the northwest wall of the cave was excavated in 1985 (20, 21), revealing an MP stratigraphy resting on an Eemian rock beach and made up of slope deposits of continental origin interspersed with owstones and carbonate crusts (SI Appendix, Section I, Tables S1 and S2). The archeological levels (IV), separated from the rock beach by a thick, sterile deposit (level VI), date to the approximately 4550 ka cal B.P. interval (SI Appendix, Section III, Table S4, Fig. S13). They yielded abundant lithic assemblages primarily made of quartz but with a signicant component of int sidescrapers and points (SI Appendix, Section I, Fig. S5)in

Author contributions: J. Zilho and J. Zapata designed research; J. Zilho, D.E.A., E.B.-G., F.d., F.D., L.D., K.D., T.F.G.H., M.J.M.-S., R.M.-B., S.M.-M., C.P.-S., C.R.-G., M.V., V.V., R.W., and J. Zapata performed research; J. Zilho, D.E.A., E.B.-G., F.d., F.D., L.D., K.D., T.F.G.H., M.J.M.-S., S.M.-M., C.P.-S., C.R.-G., M.V., V.V., R.W., and J. Zapata analyzed data; and J. Zilho, D.E.A., E.B.-G., F.d., F.D., L.D., K.D., T.F.G.H., M.J.M.-S., S.M.-M., C.P.-S., C.R.-G., M.V., V.V., R.W., and J. Zapata wrote the paper. The authors declare no conict of interest.
1

To whom correspondence should be addressed. E-mail: Joao.Zilhao@bristol.ac.uk.

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0914088107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0914088107

PNAS Early Edition | 1 of 6

ANTHROPOLOGY

association with marine mollusk shells and bone remains of horse, deer, ibex, rabbit, and tortoise. Antn (22) is a large rockshelter excavated by the Mula river, which runs through an interior basin located approximately 60 km from the present shore line. The MP occupations are preserved in a more than 4-m-thick sequence of low-energy ood deposits overlain by silts accumulated during recent episodes of inundation by a dam reservoir (SI Appendix, Section II, Table S3, Figs. S11 and S12). The levels are artifact-rich and organized around hearth features in the lower part of the sequence, but occupation remains are scarce toward the top, reecting sporadic and ephemeral incursions. The uppermost fertile units are levels II-l and I-k, for which the ages indicated by radiocarbon dating are, respectively, greater than 43.5 and approximately 37.4 ka cal B.P. (SI Appendix, Section III, Table S5, Figs. S13 and S14).
The Aviones Finds. Three categories of species are represented in the Aviones marine shell assemblage (Table 1). Most (95.7%) are of edible intertidal genera (Cerastoderma, Monodonta, Mytilus, and Patella) and have well preserved surfaces, indicating collection of living animals. A residual proportion (0.2%) are accidentally introduced Nassarius incrassatus and Gibbula sp. As these taxa live or feed on algae, their presence is likely to reect packaging of the edible species in water-soaked algae, a traditional, preindustrial technique of maintaining harvested mollusks alive (they rot very rapidly and must be eaten or cooked extremely fresh) during transportation from the point of collection to the point of consumption. Indeed, remains of Jania rubens var. corniculata, an intertidal epiphytic alga whose calcareous articulated frond allows preservation and recognition in carbonated cave lls, were recovered in levels I and II (20, 21). The remaining 4.0% are infra- and circalittoral genera with variably abraded and bioeroded surfaces, reecting beach collection of the shells of dead animals (SI Appendix, Section I, Figs. S8 S10). The Chama, Laevicardium, Charonia, and Thais material is too broken to assess original condition, and the one Trunculariopsis trunculus is a complete specimen with an irregular, probably postdepositional perforation of the body whorl in an area where the shell is very thin. Of the three Acanthocardia tuberculata, one is complete, another bears an excavation break with loss of the umbo area, the third is a conjoin of two fragments with an ancient, postdepositional break, and all must have been collected and introduced to the site as whole valves. The smaller, from level II, is umbo-perforated (Fig. 1, 1). Of the Glycymeris insubrica, four are complete, ten are ventral margin fragments, and four are small fragments of the middle of the shell; although excavation breaks are apparent in a few, most feature ancient breaks only. Two complete specimens from level II are umbo-perforated (Fig. 1, 23), and residues of a red colorant, identied as hematite (SI Appendix, Section IV Figs. S15 and S16), were found while cleaning the carbonate coating around the perforation of the larger shell. As argued for the comparable modern human-associated material from Near Eastern sites of the MP and early Upper Paleolithic (UP) (Qafzeh, Ksar Akil, aizli), the parsimonious interpretation of Glycymeris shells, even in the absence of pigment residues and irrespective of the origin of the perforation, is that they are personal ornaments (23, 24). The three Spondylus gaederopus shells were originally complete. The inner side of a level II upper valve (Fig. 2) bears residues of a pigmentatious mass composed of a red lepidocrocite base mixed with ground particles of charcoal, dolomite, hematite, and pyrite (SI Appendix, Section IV, Table S6, Figs. S17S19), suggesting use as a container for the storage of colorants or as a kind of paint cup for their preparation. Although examination of their surfaces for pigment residues is hindered by extensive carbonate coating, a similar function can be envisaged for the other valves of Spondylus
2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.0914088107

Table 1. Cueva de los Aviones mollusks: taxa and provenience

I Food taxa Cerastoderma edule Monodonta turbinata Mytilus edulis Patella sp.* Nonfood taxa Acanthocardia tuberculata Callista chione Chama gryphoides Charonia lampas Gibbula cf varia Glycymeris insubrica Laevicardium oblongum Nassarius incrassatus Pecten maximus Spondylus gaederopus Thais hemastoma Trunculariopsis trunculus Unclassied bivalve Total 12 36 10 1 1 10 1 1 72 II 8 14 11 61 1 6 1 1 103 III 1 234 31 96 1 2 2 1 368 IV 5 151 30 69 1 1 1 1 1 260 V 16 1 17 Total 14 427 108 236 3 1 1 1 1 18 1 1 2 3 1 1 1 820

*Mostly P. ferruginea; a few are P. aspera and P. lusitanica

as well as for those of taxa with a similar concave morphologyan excavation-broken Callista chione and two lower valve fragments of Pecten maximus (SI Appendix, Section I, Figs. S9 and S10). Orange pigment was also observed on the tip of an unmodied ancillary metatarsal of Equus sp. from level III, but its composition could not be established, due to the extensive carbonate coating and attendant high uorescence of the specimen (Fig. 3; SI Appendix, Section IV, Fig. S20). This naturally pointed bone may have been used as a stiletto for the preparation or application of mineral dyes or as a pin or awl to perforate soft materials (e.g., hides) that were themselves colored with such dyes. Lumps of red and yellow colorants were noted and sampled at the time of excavation (25) (SI Appendix, Section V, Table S7, Figs. S21S23). X-ray diffraction showed the reddish material to be siderite, goethite, hematite, and nontronite, and the yellow to be siderite and natrojarosite. These pigments can only be manuports because both the cave and the catchment of its sedimentary ll are in Mesozoic limestone and dolomite bedrock, where the identied minerals either do not exist or do not occur as masses of macroscopic size (in the region, such masses formed as a result of volcanism and hydrothermalism processes of Neogene age; SI Appendix, Section V). Sources for the red colorants can be found 35 km to the northwest, in the mining district of La Unin, exploited for gold, silver and base metals since antiquity. For the yellow natrojarosite, the purity of the level III samplea large (23 g) and homogeneous accumulation of prisms, granules, and powder, probably the contents of a small purse made of perishable materialconstrains potential sources to a limited number of localities, the closest lying approximately 7 km to the east (SI Appendix, Section V, Figs. S22 and S24).
The Antn Finds. Level I-k of Cueva Antn, which caps the MP archeological succession and is directly overlain by the reservoir inundation silts, yielded half of an upper, at valve of P. maximus featuring a circular perforation approximately 6 mm in diameter (Fig. 4). The shells numerous sponge holes indicate that it was beached after laying on the sea bottom for a signicant amount of time. Whether the circular perforation relates to the same bioerosion processes (i.e., is natural) or was made after collection (i. e., is anthropogenic) cannot be assessed due to postdepositional
Zilho et al.

Fig. 1. The perforated shells from level II of Cueva de los Aviones (after cleaning): (1) Acanthocardia tuberculata; (23). Glycymeris insubrica (maximum internal diameter of the perforations: 4.2, 9.5, and 6.8 mm, respectively). See also SI Appendix, Sections I and IV, Figs. S6, S7, S15, and S16.

degradation of the holes margins (e.g., by exfoliation and rubbing against the angular clast matrix), which caused the extant sharpness of its paper-thin edges. But we can exclude a postdepositional origin for the perforation itself because the presence of orange pigment residues on the interior surfaces of the exposed structural hollows of the shell implies that the perforation preexisted the application of the colorant. The pigmenta mix of yellow goethite and red hematite for which potential sources exist within approximately 5 km from the siteis ubiquitously preserved but on the shells external, discolored side only, suggesting that it may have been deliberately painted, either to regain the original appearance or to make it the same color as the internal side, which remained its natural red (SI Appendix, Section VI, Figs. S25 and S26). The ephemeral nature of the occupation of level I-k argues against on-site tool production or tool maintenance tasks and strengthens the case for the interpretation of this shell as an item of body decoration. The

alternativesuse as a container or palettecould hold for a lower valve color-stained on the inner, concave surface or if the perforation was postdepositional, but neither is the case.
Comparative Study. It has been argued that symbolic interpretations of the African and Near Eastern MSA/MP marine shell beads are questionable because the anthropogenic nature of their perforations is uncertain, whereas those in the marine shell beads found among modern human-associated cultures of the African Late Stone Age (LSA) and the European UP are clearly manmade (26). By extension, this argument applies to the material presented here too, but is it valid? To assess this issue, we collected from regional beach thanatocenoses a modern reference sample of shells of the bivalve genera that are most common in ornamental assemblages of prehistoric IberiaAcanthocardia, Cerastoderma, and Glycymeris (SI Dataset and SI Appendix, Section VII, Tables S8S10, Figs. S27S30). We then compared the

Fig. 2. (Left) Excavation-broken upper valve of Spondylus gaederopus from Cueva de los Aviones (the pigment residue is indicated); (Right) close-up view of the pigment residue. See also SI Appendix, Section IV, Figs. S17S19.

Zilho et al.

PNAS Early Edition | 3 of 6

ANTHROPOLOGY

Fig. 3. Different views of an ancillary metatarsal of horse with an excavation break from Cueva de los Aviones (Left) and binocular microscope close-ups of its pigment-dotted tip (Center and Right). See also SI Appendix, Section IV, Fig. S20.

Murcia MP specimens with that collection and with (i ) the early modern human-associated MP material from the Near East, and (ii ) the relevant components of ornament assemblages from the UP and the Neolithic of Mediterranean Spain (SI Appendix, Section VIII, Table S11, Figs. S31S42). In the natural collection, umbo-perforated shells occur frequently (45.4%) in the case of Glycymeris, less so in the case of Acanthocardia (2.7%). Both the occurrence and the size of the perforations correlate positively with discrete indicators of weathering such as decalcication/exfoliation, sponge holes, and abrasion (Table 2). Thus, in the absence of production toolmarks, perforations can be presumed anthropogenic only if the umbo holes are large but the shell is fresh or only lightly weathered. Conversely, as umbo perforations occur in 90% of the modern Glycymeris that feature advanced weathering, a natural origin must be presumed for archeological Glycymeris in similar condition. Based on these uniformitarian criteria, the perforations in the MP Glycymeris from Qafzeh and Aviones, as well as in those from the UP and the Early Neolithic of Mediterranean Spain that we examined, are natural, and so too are those in all of the Acanthocardia and Cerastoderma shells from the same sites. The archeological Glycymeris, however, were not gathered completely at random. When our three collections (MP, UP, and Neolithic) are considered together, a pattern of selection based on perforation size emergesirrespective of height, shells with holes in the 4.56.5 mm interval were targeted (SI Appendix, Section VIII, Fig. S31). These observations suggest that the pattern is cross-cultural and determined by threading constraints, with color, shine, and overall surface appearance being of secondary importanceexplaining why all of this material features advanced weathering, despite lightly weathered shells (that preserve more of the original color but lack a perforation or only have a very small one) being rather easy to nd in nature too. Although the perforations on the Nassarius kraussianus from the MSA of Blombos are anthropogenic (1), only one out of the 41 Nassarius gibbosulus beads from Skhul and the North African MP sites featured striations made by a stone toolmost are weathered specimens with types of holes whose size and placement on the shell can also be found in natural thanatocenoses, albeit in different frequencies, indicating selection but not necessarily modication. The use wear, however, is distinctive, and shows they were used as beads even when their perforations are natural (2). In the Murcia MP shellsas is the case, due to the high water content of depositional environments, in most ornamental shell assemblages of the European Paleolithic (27)chemical microweathering (SI Appendix, Section I, Fig. S7) would have long erased
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.0914088107

any perforation-associated use wear evidence. Shallow notches observed on the internal contour of the perforations of two Glycymeris from Qafzeh, where they were interpreted as caused by friction from a string (23), are also apparent in the Aviones specimens (Fig. 1 and SI Appendix, Section I, Fig. S7). However, because similar notches exist in naturally umbo-perforated material (SI Appendix, Section VII, Fig. S28), this evidence is inconclusive. Such irregular internal contours may be caused by continued abrasion in sandy, high-energy burial environments, eventually eliminating the shell margin side of the contour and creating an open hole perforation. Although not uncommon in shells from natural assemblages (SI Appendix, Section VII, Table S10), we found no such open holes among the archeological specimensas should be expected, given their unsuitability for suspension, and strengthening a link between perforation and suspension even when preservation factors prevent assessment of use wear. The origin of perforation argument is therefore not valid. Our ndings suggest that naturally perforated marine shells were extensively used at the time of emergence of the European MP and African MP/MSA traditions of body decoration and continued to be very much in fashion even as the repertoire of ornament types expanded to include more elaborated items. Discussion Natural surface processes and animal behavior can explain inland marine shell accumulations but, as discussed for the South African and Near Eastern beads (2, 23), distance to shore and difference in elevation preclude involvement of such agents in the Murcia sites. The use of the robust shell of C. chione as a replacement for int is documented in the coastal MP of Italy (28), and the taxon is represented at Aviones by one specimen (SI Appendix, Section I, Fig. S10). Conceivably, this sites thicker ventral margin and middle-ofthe-shell Glycymeris fragments (SI Appendix, Section I, Fig. S8) could represent broken-and-discarded instances of such tools and the complete specimens unmodied blanks for their manufacture. However, neither use wear nor any kind of retouch, even incipient, were observed in the specimens intact edges, so the parsimonious interpretation of this Glycymeris material is that it reects postdepositional breakage or on-site discard of broken shells that were beach-collected and used as complete valvesfor ornamentation, when perforated, perhaps as containers, when not. Although, at Aviones, mineral colorants were present in the deposits, the residues observed on the nds discussed here are not incidental or postdepositional contamination. If so, one would expect pigment to be randomly distributed across the different nd categories. This expectation was not met for a sample of 50
Zilho et al.

Table 2. Modern reference collection: perforation and weathering

Perforated Cardiidae Exfoliation Sponge through-holes Glycymeris Weathering stages 23 (n = 55) 50.9% 3.6% (n = 132) 99.2% Nonperforated (n = 770) 9.7% 0.6% (n = 150) 50.7%

Moreover, the limited capacity (approximately 5 cm3) of S. gaederopus upper valves argues against their use for the processing or storage of material related to routine tooling activities. The addition to the pigmentatious mass of ground bits of hematite and pyrite (which, when fresh, have a brilliant black, reective appearance) is also inconsistent with that function, and suggests instead the kind of inclusion for effect that one would expect in a cosmetic preparation. The use of Spondylus shells for such a particular purpose may relate to their attention-grabbing crimson, red, or violet color and exuberant sculpture, which have led to their symbolic- or ritualrelated collection in a variety of archeological contexts worldwide, namely Neolithic Europe and pre-Columbian America (30, 31). These inferences are consistent with the fact that, in the recorded history of pigments, all known uses (in Ancient Egypt) of yellow minerals of the jarosite family are in cosmetics or painting (namely, for the rendition of female skin) (32). Decorationof body, clothing, living space, or artistic representations thereof (33)is also the only archeologically and ethnographically documented function of colored, umbo-perforated Glycymeris and Acanthocardia shells. The well known use of P. maximus valves underlying the species common name (pilgrim shell) is the same, and the Antn specimen may be exceptional for the MP only in its degree of preservation, as P. maximus fragments were also found in level III of Aviones (Table 1) and level V of Higueral de Valleja (Cdiz) (34), an Andalucian cave located almost as far inland (more than 50 km) as Antn.
Fig. 4. K19-3, a perforated upper half-valve of Pecten maximus from level Ik at Cueva Antn (height: 120 mm). (Upper) The internal, naturally red side (Left) and the external, whitish side that was painted with an orange colorant made of goethite and hematite (Right). (Lower) binocular microscope close-ups on pigment masses and pigment stains around the perforation. See also SI Appendix, Section VI, Figs. S25 and S26.

shells of different species examined with a 10 hand magnier after cleaning in an ultrasound vat (examination of the totality of the assemblage was not possible due to the extensive carbonate coating of the surfaces): no pigment was found on the food shells, and, of the nonfood ones, only the one Spondylus discussed above (in addition to one of the two perforated Glycymeris) had any. Moreover, all of the lumps of red pigment sampled during excavation feature in their composition siderite, a mineral that was not identied in the pigment residues of either the Spondylus or the Glycymeris (SI Appendix, Sections IV and V, Tables S6 and S7, Figs. S16S19, S22, and S23); therefore, these residues cannot be accidentally derived from pigments present in the shells immediate burial environment. It has been argued that ochre does not necessarily equate with symbolism, as its manipulation may relate to technology, e.g., the preparation of mastics for hafting (29), and we cannot exclude that the Aviones pigments are in part related to such tasks. However, no pigment residues were observed in the Aviones stone tools (20), and no tool production or tool maintenance tasks were carried out at Antn, so such activities cannot provide a functional context for the pigment-stained shells found at both sites.
Zilho et al.

Conclusion The Pecten from Antn and Higueral de Valleja relate to the very end of the period of late persistence of Neandertals in Iberian regions to the south of the Ebro drainage (in the Andalucian site, the nd comes from undated deposits capping a thick MP sequence, a chronostratigraphic position akin to Antns and suggesting a similar age) (35). Therefore, they cannot be disentangled from the controversies surrounding the interpretation of contact period ornaments found elsewhere in Europe (818). The Aviones material, however, comes from the approximately 50 ka cal B.P. brecciated remnant of a site ll entirely of MP age and where the notion of later intrusions is an oxymoronthe association of this material with the Neandertals is, literally, rock-solid. The symbolic implications of body painting and of the ornamental use of pigment-stained and perforated marine shells are uncontroversial in UP and later prehistoric contexts but, as shown by the evidence from Africa, the Near East and now Iberia, both behaviors rst occur in the MP/MSA. Their emergence in two continents, among two different lineages and, in the time scale of human evolution, at about the same time, is inconsistent with cognitive-genetic explanations and implies that these innovations were fullling a needaiding in the personal or social identication of peoplethat did not exist in the preceding two million years of human evolution. Our ndings therefore support models of the emergence of behavioral modernity as caused by technological progress, demographic increase, and social complexication and show that there is no biunivocal correlation between modern anatomy and modern behavior (13, 3638). This much is also implied by the North African beads, which are from a time when the region was inhabited by the Dar-es-Soltan people, who, anatomically, were
PNAS Early Edition | 5 of 6

ANTHROPOLOGY

nonmodern (39, 40). Where the personal ornaments of the Chtelperronian and coeval transitional cultures of Europe are concerned, a nal corollary of our results is that Neandertal authorship is the null hypothesis. Materials and Methods
Archeological excavation was by natural stratigraphic units and used 3D pieceplotting. Description of thereference prolesfollowedgeoarcheological criteria, accounting for stratigraphic, sedimentary, and diagenetic features to differentiate units and group them, on the basis of the observed discontinuities, into complexes (formally, allostratigraphic units). Dating was carried out by the radiocarbon method, on samples of shells, at Aviones, and charcoal, using both the ABA and ABOx-SC methods, at Antn. Nondestructive elemental analysis of pigment residues on shells from both sites was carried out by X-Ray Fluorescence and Energy Dispersive Spectroscopy and compound analysis by micro-Raman spectroscopy. The mineralogical composition of the colorant lumps from
1. dErrico F, Henshilwood C, Vanhaeren M, van Niekerk K (2005) Nassarius kraussianus shell beads from Blombos Cave: Evidence for symbolic behaviour in the Middle Stone Age. J Hum Evol 48:324. 2. dErrico F, et al. (2009) Additional evidence on the use of personal ornaments in the Middle Paleolithic of North Africa. Proc Natl Acad Sci USA 106:1605116056. 3. Tattersal I (2009) Human origins: Out of Africa. Proc Natl Acad Sci USA 106: 1601816020. 4. Mellars P, Stringer C (1989) The Human Revolution (University Press, Edinburgh). 5. Klein R (2003) Whither the Neanderthals? Science 299:15251527. 6. Peyrony D (1934) La Ferrassie. MoustrienPrigordienAurignacien. Prhistoire 3:192. 7. Mellars P (1996) The Neanderthal Legacy (University Press, Princeton). 8. Bailey S, Hublin JJ (2006) Dental remains from the Grotte du Renne at Arcy-sur-Cure (Yonne). J Hum Evol 50:485508. 9. Stringer C, Gamble C (1993) In Search of the Neanderthals (Thames and Hudson, London). 10. White R (2001) Personal ornaments from the Grotte du Renne at Arcy-sur-Cure. Athena Review 2:4146. 11. Bar-Yosef O (2007) Rethinking the Human Revolution, eds Mellars P, Boyle K, BarYosef O, Stringer C (McDonald Institute, Cambridge, UK), pp 207217. 12. dErrico F, Zilho J, Bafer D, Julien M, Pelegrin J (1998) Neanderthal acculturation in Western Europe? A critical review of the evidence and its interpretation. Curr Anthropol 39:S1S44. 13. dErrico F (2003) The Invisible Frontier. A Multiple Species Model for the Origin of Behavioral Modernity. Evol Anthropol 12:188202. 14. Zilho J (2007) The emergence of ornaments and art: An archaeological perspective on the origins of behavioural modernity. J Archaeol Res 15:154. 15. Gravina B, Mellars P, Ramsey CB (2005) Radiocarbon dating of interstratied Neanderthal and early modern human occupations at the Chatelperronian type-site. Nature 438:5156. 16. Zilho J, et al. (2006) Analysis of Aurignacian interstratication at the Chtelperroniantype site and implications for the behavioral modernity of Neandertals. Proc Natl Acad Sci USA 103:1264312648. 17. Mellars P, Gravina B, Ramsey CB (2007) Conrmation of Neanderthal/modern human interstratication at the Chatelperronian type-site. Proc Natl Acad Sci USA 104:36573662. 18. Zilho J, et al. (2008) Grotte des Fes (Chtelperron): History of research, stratigraphy, dating, and archaeology of the Chtelperronian type-site. PaleoAnthropology 2008:142. 19. Rabineau M, et al. (2006) Paleo sea levels reconsidered from direct observation of paleoshoreline position during Glacial Maxima (for the last 500,000 yr). Earth Planet Sci Lett 252:119137. 20. Montes R (1987) El Paleoltico Medio en la costa de Murcia (University of Murcia, Murcia).

Aviones was established by X-Ray Diffraction. Assessment of the origin and nature of perforations in the bivalve shells from Aviones and later prehistoric sites of the region was based on visual and microscopic inspection of the surfaces, with reference to systematically collected modern beach thanatocenoses of the same species. For additional details on materials and methods, see SI Appendix. ACKNOWLEDGMENTS. Pedro Callapez, Ivan Mulero, Arnaldo Marn, and Mario Snchez advised on the taxonomy and ecology of marine mollusks and recent regional tectonics. We are indebted to the directors and curators of the Cartagena and Valencia museums for facilitating access to their collections, and to the Spectroscopy of Solids Group of the Materials Science Institute (University of Valencia), especially Nuria Garro, for access to their equipment. F. dE. was supported by the Origin of Man, Language and Languages Programme of the European Science Foundation, the Origin II project of the Aquitaine Region, and the CHIMARC Research Group. Fieldwork was supported by grants to J. Zapata from the Seneca Foundation, the University of Murcia, and the Municipality of Mula and dating by ORADS-NERC Grant 2006/1/4.

21. Montes R (1991) La Cueva de los Aviones. Un yacimiento del Paleoltico Medio (Cartagena, Spain). Memorias de Arqueologa de la Regin de Murcia 2:3558. 22. Martnez C (1997) El yacimiento musteriense de Cueva Antn (Mula, Murcia). Memorias de Arqueologa de la Regin de Murcia 6:1847. 23. Bar-Yosef DE, Vandermeersch B, Bar-Yosef O (2009) Shells and ochre in Middle Paleolithic Qafzeh Cave, Israel: indications for modern behavior. J Hum Evol 56:307314. 24. Kuhn SL, Stiner MC, Reese DS, Gle E (2001) Ornaments of the earliest Upper Paleolithic: New insights from the Levant. Proc Natl Acad Sci USA 98:76417646. 25. Montes R (1986) Historia de Cartagena II, ed Mas J (Ediciones Mediterraneo, Cartagena), pp 3492. 26. Klein R (2008) Out of Africa and the evolution of human behavior. Evol Anthropol 17: 267281. 27. dErrico F, Jardn P, Soler B (1993) Traces et fonction: les gestes retrouvs, eds Anderson P, Beyries S, Otte M, Plisson H (Universit, Lige), pp 245254. 28. Milliken S (2007) Neanderthals, anatomically modern humans, and modern human behaviour in Italy. Oxf J Archaeol 26:331358. 29. Wadley L, Williamson BS, Lombard M (2004) Ochre in hafting in Middle Stone Age southern Africa: a practical role. Antiquity 78:661675. 30. Borrello MA, Micheli R (2004) Spondylus gaederopus, gioiello dellEuropa preistorica. Preistoria Alpina 40:7182. 31. Glowacki M (2005) Food of the Gods or mere mortals? Hallucinogenic Spondylus and its interpretive implications for early Andean society. Antiquity 79:257268. 32. Ambers J (2004) Raman analysis of pigments from the Egyptian Old Kingdom. J Raman Spectrosc 35:768773. 33. Dubin LS (1987) The History of Beads. From 30,000 B.C. to the Present (Thames and Hudson, London). 34. Jennings N, et al. (2009) New dates and palaeoenvironmental evidence for the Middle to Upper Palaeolithic occupation of Higueral de Valleja Cave, southern Spain. Quat Sci Rev 28:830839. 35. Zilho J (2006) Chronostratigraphy of the Middle-to-Upper Paleolithic Transition in the Iberian Peninsula. Pyrenae 37:784. 36. Gilman A (1984) Marxist Perspectives in Archaeology, ed Spriggs M (University Press, Cambridge, UK), pp 115126. 37. Zilho J (2001) Anatomically Archaic, Behaviorally Modern: The Last Neanderthals and Their Destiny (Stichting Nederlands Museum voor Anthropologie en Praehistoriae, Amsterdam). 38. Powell A, Shennan S, Thomas MG (2009) Late Pleistocene demography and the appearance of modern human behavior. Science 324:12981301. 39. Klein R (1992) The archaeology of modern human origins. Evol Anthropol 1:514. 40. Trinkaus E (2005) Early modern humans. Annu Rev Anthropol 34:207230.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.0914088107

Zilho et al.

Symbolic Use of Marine Shells and Mineral Pigments by Iberian Neandertals

Joo ZILHO, Diego E. ANGELUCCI, Ernestina BADAL-GARCA, Francesco dERRICO, Floral DANIEL, Laure DAYET, Katerina DOUKA, Thomas F. G. HIGHAM, Mara Jos MARTNEZ-SNCHEZ, Ricardo MONTES-BERNRDEZ, Sonia MURCIA-MASCARS, Carmen PREZ-SIRVENT, Clodoaldo ROLDN-GARCA, Marian VANHAEREN, Valentn VILLAVERDE, Rachel WOOD, Josefina ZAPATA

Supporting Information Appendix

Supporting Information I: The site of Cueva de los Aviones Supporting Information II: The site of Cueva Antn Supporting Information III: Radiocarbon dating Supporting Information IV: Analysis of mineral pigments in finds from Cueva de los Aviones Supporting Information V: Analysis of mineral pigment samples from Cueva de los Aviones Supporting Information VI: Analysis of mineral pigments in a Cueva Antn Pecten Supporting Information VII: Reference collection of Acanthocardia, Cerastoderma and Glycymeris shells from modern beaches of Murcia and Alicante Supporting Information VIII: Perforated bivalves from prehistoric sites of Mediterranean Spain Supporting Information IX: References

Supporting Information I: The site of Cueva de los Aviones

Cueva de los Aviones is a sea cave located at the base of a natural promontory protecting the SW entrance to the Cartagena harbor (SI Figs. 1-2). Local bedrock is made up of carbonate rocks belonging to the Lower Alpujrride unit, of Middle-Upper Triassic age (1-2). The cave walls are composed of moderately deformed light grey and grey limestone (sometimes containing small chert nodules) and dark grey dolomite with intercalations of calcareous breccia and fine layers of yellow sandstone and clay. Locally, bedding of bedrock dips W/20. The Lower Alpujrride structural unit also contains lithotypes such as slate, quartzite and diabase, all of which outcrop at a short distance from the site. The preservation of brecciated deposits and flowstones well outside the roofed area indicates that the cave was much larger in the Pleistocene, extending several meters E/SE of the present drip-line, whose retreat is related to coastal erosion triggered by sea level rise during the Late Glacial and the Holocene. These erosive processes almost entirely removed the original fill, except for a brecciated remnant preserved against the northwestern wall of the cave, ~4 m of which were excavated in 1985 in a two week-long (September 11-26) salvage operation (3-4) (SI Figs. 2-3). The excavation was carried out in 5 cm spits that followed the natural dip of the stratification and with three-dimensional plotting of stone tool and animal bone finds. The extreme cementation of the deposit (explaining its preservation against millennia of sea erosion and repeated impact of strong waves) made it necessary to proceed with chisel and hammer most of the time, inevitably entailing a significant incidence of excavation breakage among the most fragile category of finds, the shells. The extant succession, described below following criteria outlined elsewhere (5), is composed of slope sediments accumulated in two distinct cycles and resting on a possibly Eemian rock-beach. From top to bottom, three stratigraphic complexes have been recognized: (1) US (Upper Slope Sediment), which survives in the upper part of the 1985 profile and as cemented remnants hanging from the cave walls; (2) LS (Lower Slope Sediment), which corresponds to the excavated archeological sequence; (3) CC (Cemented Cobbles), which outcrops slightly above modern sea level on the NE side of the current cave entrance and is composed of cemented cobbles and pebbles. Correlation with the units recognized during the 1985 excavation is given in SI Table 1, and a stratigraphic column is presented in SI Fig. 4. Complex US is made of diamict-like limestone and dolomite forming a heterogeneous cemented breccia that contains rare fragments of bones and lithic artifacts with random orientation. It rests on an erosive surface with evidence of post-depositional deformation, and is separated from underlying complex LS by a discontinuously preserved carbonate flowstone. Complex LS is composed, on average, of coarse material (mainly fragments of carbonate rocks) with a variable degree of post-depositional carbonation (usually intense) and poorly recognizable bedding. It dips SE-E with an inclination of 20-25 and has an erosive base, with angular unconformity. Over a thickness of ~3 m, this complex, originally subdivided in six archeostratigraphic levels, comprises ten different units, all of which contain a variable sand fraction formed of angular and subangular grains of the same lithology as the coarser material, with minor amounts of silt and clay (SI Table 2). Complex CC underlies the archeological sequence, and is shaped as a ridge made up of cemented, clast-supported, rounded cobbles of dolomite and limestone, and occasional quartz pebbles. As the lithology of the fill is identical to that of the cave, a mainly local source is inferred for the bulk of the US and LS complexes. The geometry of these deposits and of the discontinuous flowstones found along the walls indicates that a several meter-thick seawardsloping talus cone originally filled the entrance to the cave. The apex of that cone would have been located at the base of the passage that connects the rear part of the site with the inner
2

karst, whence came the geological components of the deposit (complemented by direct falls from the caves roof and walls). The sedimentary dynamics involved consist mainly of the accumulation of single fragments (scree-like), although mechanisms related to the action of surface-running waters, such as overland-flow and grain-flow, are also responsible for the formation of some of the observed facies (e.g., units LS04 and LS09). Despite the abundant archeological remains, neither in situ features nor anthropic microfacies were observed in the extant profiles, although a concentration of burnt bones and charcoal (identified as holm oak by J.-L. Vernet, University of Montpellier) is reported from level IV (3) and may have been part of a disturbed combustion feature. The orientation of the artifacts and bones in the different layers is consistent with the orientation pattern of the rock fragments, indicating that all accumulated together with the sediment and by means of the same sedimentary dynamics. As a result of these processes, stone tools and animal remains underwent lateral displacement, and their horizontal redistribution across the entire surface of the cave means that their place of excavation does not necessarily coincide with the actual area of use and discard. Given the morphology of the cave and the overall geometry of the deposit, human occupation is likely to have taken place towards the inner part of the talus cone, whence the remains recovered in the area of the 1985 excavation were eventually displaced by such low-energy, syndepositional geological processes as discussed above. Although original site reports (3-4) suggested that a transgressive marine level could be observed in level V (=unit LS061), the entire LS complex is in fact of continental origin. Moreover, there is no evidence that any of its units could correspond, even if only in part, to the redeposition of pre-Tyrrhenian marine deposits once extant in the cliff face at a higher elevation. Therefore, the marine shells found in the archeological sequence are non-geogenic, and their association with abundant stone tools and mammal bones (of horse, red deer, ibex, rabbit and tortoise) indicates an anthropic origin. Five Patella shells from levels I-IV of the 1985 excavation (levels V-VI were sterile) have been radiocarbon dated (Supporting Information III). The results for levels I-III (=units LS02LS05) are in stratigraphic order, indicate accumulation during the ~45-50 ka cal BP (calendar years before present) interval, and suggest correlation of the LS01 flowstone with Greenland Interstadial 12. The fact that, at ~45 ka 14C BP, the results for levels II and III are statistically the same is consistent with the fact that, in our reading of the stratigraphy (SI Figs. 3-4), the base of level II groups with level III to form a single geological unit, LS05. One of the two results for level IV (=unit LS06u), OxA-19312, is not inconsistent with this chronology and, if accepted as valid, leads to a Bayesian model (SI Fig. 13) that constrains the deposition of the archeological sequence to the ~45-48 ka cal BP interval, broadly coinciding with the very cold episode Heinrich Event 5. However, the mid-point of the uncertainty interval of this level IV result is younger than that for levels II and III, which, given the challenging nature of radiocarbon dating in this time range, could reflect undetected, residual contamination. In this case, OxA-19312 would be a minimum age only. In fact, the deposition of level IV (=upper part of unit LS06) may well significantly predate that of level III, from which it is separated by a noticeable discontinuity (the base of the overlying LS05 unit features common carbonate concretions and has a clear lower boundary). The other sample from level IV (OxA-20906; ~42.5 ka cal BP) is statistically younger. However, this sample came from square C2, in the seaward row of the excavation grid, where the in situ deposits are covered by a sheet of reworked material derived from the erosion of both the upper part of the LS complex and the lower part of the US complex (SI Fig. 3). As it went unrecognized at the time of excavation, this sheet must have contributed to the collections provenanced to the different archeological levels. Bearing in mind that the OxA-20906 result is younger than the uppermost unit of the LS complex (level I), the simplest explanation for the anomaly is that this C2 sample was in derived position and originally belonged in the US complex. If so, it would place the deposition of the US complex ~42.5 ka cal BP, in the cold
3

stadial comprised between Greenland Interstadials 11 and 12 (SI Fig. 13). Given the topographic location of the US remnants preserved towards the rear of the cave (high along the cave walls), a corollary of this interpretation is that the site would have become entirely filled up by the end of the Middle Paleolithic, in agreement with the fact that no Upper Paleolithic items were observed among the thousands of lithic artifacts recovered in the 1985 excavations. In all levels, the stone tool assemblages are mostly quartz, suggesting that flint is non-local, but the Levallois method was used in the production of blanks made on both kinds of rawmaterials. The absence of flint cores and the high values of the Sidescraper Index (a tool category for which flint was preferentially used) suggest long-distance transport and curation of the finer raw-materials (3-4). The two perforated Glycymeris shells were found in close association with diagnostic lithics, as noted in the excavation diary, which records their discovery on August 16 and August 17, 1985, during the excavation of layer II of square B1 and in the vicinity of two flint tools, a Levallois point and a Mousterian point (SI Fig. 5). No excavation record exists of the provenience of the perforated Acanthocardia, but in the collections of the Museum of Cartagena it was clearly labeled as coming from level II, in agreement with the published information (4). In the framework of this study, we partially cleaned these shells in order to access the carbonate-coated surfaces around the umbo, for use-wear analysis; they are illustrated in their original condition in SI Figs. 6-7, together with close-up views of the corresponding umbo holes before and after cleaning. Besides these three previously reported perforated shells, and a Spondylus upper valve with pigment residues identified among the unclassified material in the Museum of Cartagena (Fig. 2), levels I-IV yielded other beach-collected shells of non-food gastropods and bivalves, for the most part coming from levels I and II. The majority are G. insubrica whole valves or fragments thereof (SI Fig. 8), but there are also complete or excavation-broken shells of A. tuberculata and S. gaederopus (SI Fig. 9). The other taxa are illustrated in SI Fig. 10: Callista chione (no. 1, dorsal fragment with umbo); Chama gryphoides (no. 4, dorsal fragment with umbo); Charonia lampas (no. 2, body whorl fragment); Laevicardium oblongum (no. 7, ventral fragment); Pecten maximus (two fragments of a large right valveno. 3 is a posterior auricle and no. 6 is part of the anterior ventral edge); Thais haemastoma (no. 5, body whorl fragment); and Trunculariopsis trunculus (no. 8, whole shell with a perforation, in all likelihood post-depositional, in the thinner part of the body whorl). This range of taxa includes (and in fact is wider than) that associated with the perforated Nassarius gibbosulus from the early modern human-associated Middle Paleolithic levels of the Near Eastern site of Skhul, which also featured Acanthocardia deshayesii, Laevicardium crassum and Pecten jacobaeus (6). In terms of ornamental shell find densities, the Aviones values are of the same order of magnitude as those observed in the African MP/MSA sites that yielded significant amounts of Nassarius beads. Considering only the perforated material, the round figures are: for the M1 phase of Blombos, 39 shells over 20 m and an average thickness of 50 cm (7), i.e., 2/m or 4/m; for layer E of Taforalt, 19 shells over 26 m and an average thickness of 40 cm (8), i.e., 1/m and 2/m; for unit II of Aviones, 3 shells over 3 m and 1 m, i.e., 1/m and 3/m (and these are minima only, given the many Glycymeris fragments where the umbo area is missing and that originally may have been perforated too).

SI Table 1. Synopsis of Cueva de los Aviones succession (the thicker dotted lines indicate discontinuities).
Unit US LS01 LS02 LS03 LS04 LS05 Short description slope sediment above flowstone flowstone slope sediment, random orientation pattern slope, parallel orientation pattern slope, random orientation pattern slope, random orientation pattern slope sediment, with a discontinuous carbonate crust at the base (LS05k) slope sediment , with few large stones (LS06u) slope sediment, with common large stones (LS06l) slope sediment, strongly cemented slope sediment, some parallel orientation pattern discontinuous sand intercalation slope sediment, no orientation pattern rock-beach Eemian OIS-3 or OIS-4 II III IV V I conglomerado muy cementado de cantos dolomticos y tierra negra estructura de greze lite conglomerado dolomtico y tierra rojiza caliche microconglomerado areniscoso y tierra rojiza (no description) Correlation with Montes 1987, 1991 Age OIS-3

OIS-3

LS06

LS07 LS08 LS09 LS10 CC

VI

tierra rojiza con algunos cantos dolomticos

SI Table 2. Description of the stratigraphic units in complex LS.

Unit LS01 Description Carbonate flowstone with laminar structure formed of well-crystallised calcite crystals, probably microsparite, strongly cemented. This flowstone is visible at several places along the cave wall, always displaying laminar structure. Its thickness is varied, reaching some 15 cm along the N wall. Heterogeneous clast-supported breccia, locally open-work, formed of subangular (with rare subrounded and angular) fragments with random orientation, sometimes parallel to the lower boundary, with clasts from 2 mm to 15 cm (clasts ranging 2-4 cm are the most represented); clasts are composed of white and light gray dolomite, white and light gray limestone, fine calcareous breccia, and yellow or pink weathered sandstone; the matrix is 7.5YR4/2.5 silty-sandy-loam; the sand fraction is fine and formed of angular and subangular grains of dolomite; the carbonation is intense and massive with some laminar structure on top (at the contact with unit LS01); the porosity is low and formed of very fine channels with carbonate hypocoatings; the archaeological material is common and shows random distribution and orientation patterns; the lower boundary is clear, poorly distinct, dipping 25 E; 60 cm thick. Fine breccia with moderate sorting formed of subangular and angular (rare subrounded) fragments of the same lithologies as above with good degree of isorientation of the tabular and platy elements, which are parallel to the lower boundary; the size ranges from 2 mm to 3 cm, with an average of about 1 cm; very rare fine (5 mm) rounded pebbles are observed; the matrix is 7.5YR4/2.5 silty-sand; the other features are the same as in unit LS02 but for the archaeological material, frequent; the lower boundary is clear, poorly distinct, dipping 25 to E; 32 cm thick. Sandy loam with abundant angular and subangular stones from 2 mm to 6 cm with random distribution and orientation pattern; the matrix is the same as in unit LS03 but some richer in sand formed of angular and subangular grains of local dolomite and limestone; the archaeological material is common; the lower boundary is clear, poorly distinct, dipping 25 to E; 30 cm thick. This unit shows the same sedimentary characteristics as the overlying one (unit LS04) except for: the matrix is composed of coarse to very coarse sand with some very fine gravel; some vague stratification, due to the presence of discontinuous fine-gravel partially-open-work intercalations is recognised; common carbonate concretions are found at the base (sub-unit LS05k); the lower boundary is clear, poorly distinct, dipping 25 to E; 25 cm thick. Fine breccia with moderate sorting formed of subangular and angular (rare subrounded) fragments of local lithologies with poor (locally moderate) degree of isorientation of the tabular and platy elements, parallel to the lower boundary; the size ranges from 2 mm to 3 cm, with large (max. 20 cm) subrounded and subangular fragments of dolomite, limestone and limestone breccia particularly common at the base of the unit (sub-unit LS06l); the matrix is 7.5YR4/2.5 silty-sand; the archaeological material is frequent; the lower boundary is clear, poorly distinct, dipping 25 to E; 55 cm thick. Fine-medium poorly sorted breccia formed of angular and subangular stone fragments from 2 mm to 2 cm, with random distribution and orientation pattern; the is matrix 9YR5/6 sandy-silt, with the sand predominantly in the fine sand sub-class; the carbonation is very intense with strong cementation; no archaeological material is present; the lower boundary is diffuse, poorly distinct; 25 cm thick. This unit has the same characteristics as unit LS07 but contains larger stone fragments (max. 10 cm), sometimes with an orientation parallel the lower boundary; no archaeological material is present; the lower boundary is sharp, discontinuous; 10 cm thick. Discontinuous thin layer formed of breccia with same characteristics as unit LS07 with 10YR6/5 sandy matrix; no archaeological material is present; the lower boundary is sharp, discontinuous; 4 cm thick. Fine clast-supported breccia with occasional stone fragments up to 3 cm, random distribution and orientation patterns; the other characteristics are the same as for unit LS07; the lower boundary is sharp irregular to bedrock or to the complex CC; 20 cm thick.

LS02

LS03

LS04

LS05

LS06

LS07

LS08

LS09 LS10

20 km

SI Figure 1. Murcia and its Middle Paleolithic. Top. Location of the Region of Murcia in a topographic map of Iberia. Bottom. Location of the Middle Paleolithic sites of Murcia with a stratigraphic or site formation context in a topographic map of the region (elevations in meters).
7

hic nt olit ale emna P dle ed r Mid cciat e r b

N
A

A 0 m 5

SI Figure 2. Cueva de los Aviones. Top. Clockwise: site setting, April 2008; the 1985 excavations; the remnant, prior to excavation. Middle. Site plan and profile (3-4). Bottom. View of the cave, April 2009.
8

flowstone

US LS02-03 (I) LS04 (IItop) LS05 (IIbase-III)

d ke or w re

LS06 (IV-V)

LS07-10 (VI)

CC

A1

A1

A1

A1

A1

A1

A1

A1

B2

B1

B1

B2

B1

B1

B2

B1

B1

B2

B1

B1

C2

C1

C1

C2

C1

C1

C2

C1

C1

C2

C1

C1

Level I ~0.4 m

Level II ~1.1 m

Level III ~0.9 m

Level IV ~1.7 m

SI Figure 3. Cueva de los Aviones excavation and stratigraphy. Top. View of the remnant in April 2009, with indication of stratigraphic limits (in parenthesis, the archeological level designations used in 1985); the small yellow ruler at the center measures 20 cm. Bottom. The 1985 excavation grid with indication of the areas and volumes affected, per recognized stratigraphic unit.
9

SI Figure 4. Cueva de los Aviones stratigraphic column. A. distribution of archeological finds as observed in the remnant (B=bone fragments, C=charcoal fragments, L=lithic artefacts, S=shells). B. stratigraphic column (the width is proportional to the grain size of the units matrix; C=clay, Si=silt, Sa=sand, G=gravel). C. units described according to this study (US, LS and CC stratigraphic complexes; LS01 to LS 10 units of the LS complex; numbers correspond to the average depth in centimetres recorded in the remnant). D. units defined during excavation (the numbers indicate depth of boundaries in the profile as illustrated in refs. 3-4).
10

SI Figure 5. Cueva de los Aviones finds. Top and Middle. Flint tools (from the excavations photo archive). Clockwise: point (level I), sidescraper (level I), point (level II), sidescraper (level II), sidescraper (level III), sidescraper (level IV). Bottom. Pages of the excavation diary noting the discovery of the perforated Glycymeris.
11

1 cm

SI Figure 6. Cueva de los Aviones level II perforated shells, prior to cleaning for this study. Top. Acanthocardia tuberculata. Middle and Bottom. Glycymeris insubrica.
12

SI Figure 7. Cueva de los Aviones perforated shells from level II. The perforations prior to (left) and after (right) mechanical cleaning (preceded by removal of the brecciated clutters in an ultrasound vat). Top. Acanthocardia tuberculata. Middle and Bottom. Glycymeris insubrica. Note, especially in the bottom specimen, the heavy chemical micro-weathering around the perforations, which relates to the formation of the carbonate coatings and prevented the presevation of any use wear evidence.
13

1 cm

1 cm

1 cm

1 cm

SI Figure 8. Cueva de los Aviones non-perforated Glycymeris insubrica shells. Top. Fragments from level I. Middle. Whole shell from level I (from excavations photo archive, missing in the Museum of Cartagena) (left) and fragments from level II (right). Bottom. Whole shell and fragment from level III.
14

1 cm

SI Figure 9. Cueva de los Aviones complete non-food bivalve shells. Top. Upper valve of Spondylus gaederopus (level IV). Middle. Lower valve of Spondylus gaederopus (level I). Bottom. Acanthocardia tuberculata shells from level III (left) and level IV (right; excavation-broken).
15

3 2

8 7

1 cm

SI Figure 10. Cueva de los Aviones non-food shells. 1. Callista chione (level I). 2. Charonia lampas (level IV). 3. Pecten maximus (level III). 4. Chama gryphoides (level I). 5. Thais haemastoma (level II). 6. Pecten maximus (level III). 7. Laevicardium oblongum (level IV). 8. Trunculariopsis trunculus (level III).
16

Supporting Information II: The site of Cueva Antn

Cueva Antn is a rockshelter excavated into the base of an E-W, 25-30 m high, Eocene limestone escarpment located towards the tail end of the La Cierva reservoir, on the Mula river (SI Fig. 11). Salvage work undertaken in 1991 exposed a >4-m-thick sequence, at the base of which Middle Paleolithic levels were identified (9). After two seasons (2007-08), new excavation work has also exposed Middle Paleolithic occupations in the uppermost units. In the areas currently under investigation, the Antn deposits can be divided in four main geoarcheological complexes: DD (Dam Deposits), made up of fine material with some coarser intercalations, and accumulated over the last decades at times when the level of the La Cierva reservoir raised above the base of the shelters overhang, fully submerging the site; TL (Transitional Layers), formed of disturbed layers of uncertain age found to the north of row 18 of the grid in intermediate position between the DD deposits and the underlying Pleistocene sediments; AS (Archaeological Succession), formed of a number of superposed alluvial sequences featuring significant lateral variation and distinct sedimentary facies, and including intercalations of slope material, particularly against the back wall and especially so in its upper part (SI Fig. 12); FP (Fine Palustrine), a fine organic sediment forming the base of the exposed succession. Geomorphologically, the AS complex can best be described as a fluvial terrace created from an alluvial sequence whose sheltering under an overhang limited the impact of postdepositional modifications and soil formation processes. Burrowing features created by rabbits and small carnivores, a major cause of disturbance in all cave and rockshelter sites of Mediterranean and west Atlantic Iberia, are conspicuously absent from the 1991 profiles and have not been observed in current excavation work. Where the uppermost levels of the AS complex are concerned, these observations are consistent with the fact that the abundant rabbit bone component of the deposits is almost exclusively formed of juvenile remains bearing traces indicative of accumulation by eagle-owls (and individual, intact regurgitation pellets have even been recovered in the finer, silty-loamy units). From a taphonomic perspective, the sequence is therefore pristine, with no evidence for vertical post-depositional dislocations and with only limited, horizontal syndepositional displacement of finds having occurred in the coarser levels (those made up of fine alluvial sands or of cm-size cryoclastic wall debris). Documenting the shifting location of a diverse range of valley bottom environments over the time of accumulation (river channels, gravel bars, sand beaches, bogs, etc.), 48 different units, grouped in five subcomplexes, are currently recognized within the 2-3 m spanned by the AS complex (SI Table 3). Attempts at radiocarbon dating its base have so far failed due to the insufficient yields of the charcoal samples used, but a consistent chronology exists for the upper part, above an erosive unconformity at the base of level II-m. This level corresponds to a gravel/sand bar accumulated at a time when the Mula river ran through the shelter, and the overlying sand beach (level II-l) yielded characteristic sidescrapers and cores for which we have a terminus ante quem of ~43.5 ka cal BPthe age obtained for overlying level II-h/i, an archeologically sterile massive silt deposit accumulated by decantation in the framework of a subsequent episode of low-energy inundation of the site. Carbonate-incrusted level II-d separates the similarly sterile alluvial units that overlie II-h/i from the uppermost archeological occupation, which is contained in level I-k. This unit is a clast-supported breccia formed of small angular limestone debris with a clayey silt matrix and has been dated to ~32.9 14C BP (~37.4 ka cal BP) by a pine charcoal sample pretreated with the ABOx-SC technique (Supporting Information III). In the basal units of the AS complex, the archeological levels correspond to well-preserved, single-event Mousterian occupations organized around hearth features; stone tools are
17

abundant (SI Fig. 12), and the bones, often cut-marked, are of large herbivores. In the uppermost units (II-l and I-k), however, finds are scarce, suggesting sporadic and highly ephemeral visits only, but, technologically and typologically, the lithics are of Mousterian affinities (SI Fig. 12). Where level I-k is concerned, the handful of artifacts (including the Pecten shell) was recovered in close association with the dated charcoal samples (Supporting Information III), and these dates place the human occupation represented therein towards the end of the period of late persistence, in Iberian regions located to the south of the Ebro drainage, of a Neandertalassociated Middle Paleolithic (10), which, at Gorhams Cave (Gibraltar) and Gruta da Oliveira (Portugal), is well documented until ~32 ka 14C BP. Assignment of the level to the latest Middle Paleolithic is further supported by the fact that its characteristics fall outside what would be expected in a context of the regional Upper Paleolithic (for which, moreover, the earliest reliable dates are, at present, <30.5 ka 14C BP). In fact, sporadic occupations of the kind seen at Antn are characterized, in the Upper Paleolithic of SW Iberia, and this from its earliest manifestations, by the discard of endproducts of the chane opratoire (bone and stone points, microlithic barbs, and unretouched blade and bladelet blanks) that remain conspicuously lacking in level I-k, despite >3 m of it having already been excavated. SI Table 3. Stratigraphic layout of complex AS of Cueva Antn.
Stratigraphic span I-g (and unnamed units above) to II-c II-d to II-m

Subcomplex AS1

Short description upper part, with slope/ wall inputs and fine alluvial mainly alluvial sediment, forming at least three minor cycles mainly alluvial sediment, forming at least three minor cycles sand layers preserved in West profile of squares L21-L22 mainly alluvial sediment with a poorly developed buried soil on top, forming at least four minor cycles

Base paraconformity

Archeology I-k (artifacts) II-l (artifacts)

Age I-k: ~31.1 ka 14C BP II-i: ~39.6 ka 14C BP

AS2

erosive unconformity

AS3

erosive unconformity

II- to II-t

OIS-3

AS4

paraconformity

II-

II-u/y (artifacts) III-b/d, III-i/j (artifacts, features)

OIS-3

AS5

paraconformity

II-u to III-n

OIS-3

18

Elevation of exposure of the 1991 unit III deposits 5m

+4

-6

-4 -2 0

+2

1991 Trench 2
+3 +5 +7

1991 Trench 1

Nm
20 21 22 L K J

2007 geological trench 2007-08 archeological trenches Extant profiles

SI Figure 11. Cueva Antn. Top. The La Cierva dam and reservoir, April 2009 (the circle indicates the location of the rockshelter). Middle. View of the excavation at the end of the September 2008 field season. Bottom. Site plan and grid, with indication of the excavated areas.
19

[from Martnez et al., 1991: Fig. 8]

J>I17

J>I18

J>I19

I-k

~37.4 ka cal BP

II-e II-i II-k II-l 218 cm II-m ~43.5 ka cal BP

II- II-r II-q II-u

II-p

SI Figure 12. Cueva Antn. Top. Flint artifacts from level I-k; left to right, centripetal core on flake blank, denticulate and splintered piece. Middle. Flint sidescrapers from level III-f. Bottom. East profile of the J/17-19 squares, with indication of the stratigraphic units recognized in the AS complex and of the age of dated levels.
20

Supporting Information III: Radiocarbon dating

All radiocarbon dates in this paper were produced at the Oxford Radiocarbon Accelerator Unit (ORAU), University of Oxford, and were funded by the ORADS-NERC programme. Two different materials were dated: marine shell from Cueva de los Aviones and charcoal from Cueva Antn. The charcoal dates from Cueva Antn reported here come from samples pieceplotted at the time of excavation. They were identified to tree species or wood typeusing reflected light microscopy, scanning electron microscopy (SEM), and a reference collection at the Department of Prehistory and Archeology, University of Valencia. Cueva de los Aviones Five samples of the limpet Patella ferruginea were selected from the Museum of Cartagena as suitable for dating. Three samples (AVI-1 to AVI-3) come from the three uppermost archeostratigraphic units, I, II and III, and two (AVI-4 and AVI-5) from unit IV. All shell samples were cleaned using an air-abrasive system with aluminium oxide until the surface was removed and the inner parts of the exoskeleton were exposed. A small fragment of the carbonate was sawn off and crushed in an agate mortar and pestle to a fine powder. Since limpets are mostly composed of calcite, the stable polymorph of CaCO3, X-Ray Diffraction for the investigation of mineralogical changes in the carbonate matrix was not considered necessary. Approximately 30 mg of powdered sample was placed in a side-arm (Rittenberg) glass tube and was reacted with 5 ml of 80 % phosphoric acid (H3PO4) for 12 h at 60 C, under vacuum. The CO2 evolved was extracted through a manifold, cryogenically purified and transferred into a glass ampoule. The ampoule was cracked, and the gas passed through an automated elemental analyzer connected to a continuous flow isotope-ratio-monitoring mass spectrometer, where it was further purified. The CO2 was then reduced to graphite with H2 at 560 C for 6 h, in the presence of 2 mg of a Fe+ catalyst. The graphite was pressed into a target holder prior to accelerator mass spectrometry (AMS). The raw determinations and sample details are given in SI Table 4. In SI Fig. 13, the radiocarbon results are incorporated into a Bayesian model built with OxCal 4.1.1 (11) and are compared against the Cariaco-Hulu record (12-13) and the GISP2 18O record. This is a tentative comparison, rather than a calibration, since there is no accepted international calibration curve for this period as yet. Until the release of INTCAL09 calibration curve, the model here is offered in the interim. The dates from the lowermost level IV do not agree with the overall model: OxA-19312 is in poor agreement with the rest of the sequence (agreement index = 14%), as is OxA-20906, which we treated as an outlier and excluded from the Bayesian analysis. The reasons behind this level IV anomaly in are unclear, possibly reflecting post-depositional mixing (Supporting Information I). However, given the difficulties in dating very old samples (>40 ka), the results from Aviones show an acceptable agreement and reflect an accumulation of material around the time of the Greenland Interstadial (GIS) 13 and the subsequent Heinrich Event 5. Interestingly, OxA-20906 seems to fall in the stadial between GIS 12 (Hengelo) and GIS 11, also indicating prevalent cold conditions. Cueva Antn The identification of the dated specimens is given in SI Table 5. Cold ecology pinesPinus nigra (salgareo pine) or Pinus sylvestris (Scots pine)have a very similar anatomy and can be easily differentiated from other pines when presenting the following features:
21

 in the transverse section of wood, vertical resin canals located in the latewood or in the transition zone between early- and latewood and growth ring boundaries always distinct; in radial section, heterocellular rays, parenchyma cells with one, rarely two, large pinoid pits per cross-field, and horizontal tracheids with dentate walls. Distinguishing between P. nigra or P. sylvestris is often not possible (14), although a useful criterion is that, in Pinus nigra, the early- to latewood transition frequently appears more abrupt. In the case of the K19-5 charcoal from level I-k (OxA-20882), a more precise botanical classification was not possible, despite the use of SEM (SI Fig. 14), due to the significant alterationfusion of cell walls, high degree of mineralization, and fissures and fractures of the plant tissue. Given the biogeographic situation of Cueva Antn, this charcoal is probably from P. nigra, but we cannot exclude P. sylvestris because nowadays, in Eastern Iberia, the two taxa occasionally co-occur. Two radiocarbon pre-treatment methodologies, ABA (Acid-Base-Acid) and ABOx-SC (Acid-Base-Oxidation-Stepped Combustion), were used to clean the charcoal fragments prior to dating. The former is the standard technique used in most laboratories (15-17). Whilst this is sufficient to remove contamination from the majority of young samples, it has been shown to produce inaccurate ages for some samples dating to >30,000 years ago. ABOx-SC more effectively removes contaminants in samples of this age, but this more rigorous technique requires around 100 mg of well preserved charcoal (18-19). At Cueva Antn, however, much of the charcoal is small and poorly preserved: only three of the 11 samples treated with ABA survived the treatment to produce a date. Therefore, with the exception of sample E20-1, which was treated only with ABOx-SC due to its excellent preservation and large size, all charcoal fragments were first treated with ABA. Subsequently, one sample, J19-7, which was large enough, was also dated with ABOx-SC to examine whether the ABA treatment was able to effectively remove contaminants from charcoal at this site. During the ABA treatment, the charcoal was sequentially washed at 80 C in 0.5 M HCl (30 minutes), 0.2 M NaOH (30 minutes, replacing the alkali until excess humics were no longer removed), and 0.5 M HCl (up to 1 hour) (15). ABOx-SC involved washing the charcoal at room temperature in concentrated HCl (1 hour), 1M NaOH (30 minutes, replacing the alkali until excess humics were no longer removed), 0.1M K2Cr2O7/2M H2SO4 (20 hours at 60 oC). After freeze drying, the charcoal was combusted in a sealed evacuated tube with CuO at 630 oC for 2 hours. Between each step in the wet chemistry of both techniques the charcoal was rinsed three times in ultrapure MilliQ water or until humics were no longer removed. The products were then combusted, graphitized and measured in an AMS (15-17). The carbon content and 13C values of the three ABA samples are within the expected limits for charcoal, although these indicators are relatively insensitive to contaminants. The ABOxSC date for the sterile deposits capping the Mousterian occupation (level I-i) is modern in age, while that for one of the level I-k samples (J19-7) shows that the ABA result for that same sample is rejuvenated by a couple of millennia (SI Table 5). The other ABA result for level I-k (sample K19-5) should therefore be treated as a minimum age only. The same applies to the ABA result for level II-h/i (sample N20-2)since we cannot exclude that it is similarly rejuvenated, the ~39.6 ka 14C BP date obtained for this level should be treated with caution until confirmed by ABOx-SC. The ABOx-SC result for level I-k has been compared to the two high resolution calibration records available for this periodCariaco-Hulu (12) and Fairbanks et al.s (20)within a Bayesian model in OxCal (SI Fig. 13). Given the plateau that exists in both curves ca.33-35 ka 14 C BP, the comparison places the final Mousterian at the site within a wide interval centered around 37.4 ka cal BP, during GIS-8 or in the cold stadials on either side of it. A date within one of these stadials, or within the minor cold oscillations recorded towards the end of GIS-8, is supported by the texture of the deposit, which includes abundant cryoclastic wall debris,
22

indicating deposition under cold and dry conditionsin agreement with the presence of Pinus nigra/sylvestris in a charcoal assemblage where the predominance of Juniperus sp. (associated with Ephedra sp. in levels II-l and II-u, further down in the stratigraphy) suggests an open, steppic landscape with scattered mountain pine stands. SI Table 4. Radiocarbon dates on limpet shells (Patella ferruginea) from Cueva de los Aviones.
Sample AVI-1 AVI-2 AVI-3 AVI-4 AVI-5 Level I II III IV IV Square A1 B1 B1 A1 C2 OxA 19310 19311 19390 19312 20906 Date BP 43290 45000 45150 42200 38150 +/390 450 650 360 350 13C 3.01 2.69 0.16 3.27 1.61

SI Table 5. Radiocarbon dates on wood charcoal from Cueva Antn.

Sample E20-1 J19-7 Species Pinus halepensis Pinus sp. Level I-i I-k OxA 20115 20881 21244 K19-5 N20-2 Pinus nigra type Juniperus sp. I-k II-h/i 20882 18672 Date (BP) 98 31150 32890 31070 39650 +/23 170 200 170 550 %Yld 12.6 26.9 12.1 (*) 13.8 21.1 %C 75.4 63.4 88.4 61.3 60.1 13C -22.5 -24.0 -22.3 -23.4 -22.3 Treatment ABOx-SC ABA ABOx-SC ABA ABA

(*) this value is estimated, as only approximately half of the sample that remained after the wet chemistry was precombusted.

23

SI Figure 13. Bayesian modeling of radiocarbon dates. Top. Cueva de los Aviones, on Patella ferruginea shells. Bottom. Cueva Antn level I-k, on charcoal from Pinus sp.
24

Transverse and radial section (x500)

Tangential section with a ray and vertical tracheids (x1000)

Radial section with one large pinoid pit per cross-field; ray with tracheid (x1000) SI Figure 14. SEM images of the Cueva Antn charcoal sample K19-5 (level I-k). The different views show features that are diagnostic of the wood anatomy of Pinus nigra and Pinus sylvestris (photos: E. Badal-Garca).
25

Supporting Information IV: Analysis of mineral pigments in finds from Cueva de los Aviones
Pigment residues identified in two shells and a bone fragment from Cueva de los Aviones were analyzed for identification of the minerals and properties of the pigmentatious masses, using SEM-EDS (Scanning Electronic Microscopy with an Energy Dispersive Spectrometer) and Raman spectroscopy, at the Research Center in Archeological Applications of Physics, University of Bordeaux 3. The pigment in a Glycymeris shell The perforation in one of the umbo-perforated Glycymeris insubrica shells from Cueva de los Aviones kept at the Museum of Cartagena was cluttered by carbonate-cemented sediments (Supporting Information I; SI Figs. 6-7). In the process of removing the carbonate coating directly adhering to the inner surface of the shell, near the umbo, red particles were observed, collected and submitted for analysis. Elemental analyses were carried out by SEM/EDS. The microscope was a JEOL 6460 LV. The low vacuum system allows the imagery and analysis without specific preparation (coating) of the sample. An EDS Oxford INCA 300 spectrometer is coupled to the SEM. The powder sample is placed on adhesive carbon disks. Due to the porosity and low conductivity of the Cueva de los Aviones powder sample, the qualitative analyses were made in low vacuum mode (15Pa), with an acceleration voltage of 20 kV. The data were processed with the INCA software (Oxford instruments). Raman spectroscopy was carried out with a Renishaw RM 2000 Raman spectrometer coupled with a confocal Leica DMLM microscope and a CCD detector. The analytical conditions were as follows: wavelength of the laser, 632.8 nm; maximum power, 50 mW; microscope magnification, 50x. The calibration of the spectrometer was done with a Si standard (main band: 520.5 cm-1). The spectra were processed with the GRAMS 32 software. The SEM/EDS elemental analysis (SI Fig. 15) showed that, globally, the major elements are Si and Ca, although lower concentrations of Al and Fe were also detected. Inside the matrix, some minerals are composed of Si, Al, Na and O, while others are mainly composed of Si and O (probably feldspar and quartz, respectively). More rarely, and locally, we found two different mineral species: some flakes composed mainly of Fe and Ti, and some barium sulphate crystals. The provenience of these species (from the original archeological or geological context, or artifacts of the sampling and cleaning procedure) could not be assessed. Under natural light, the observation of the powder collected while cleaning the shell showed some red particles mixed in a white or pink/white sediment. The Raman analysis of several of these red particles produced the characteristic spectra of hematite, and calcite was detected on the white translucent particles of the sample (SI Fig. 16). No other component was found because the intense fluorescence of some minerals occulted the Raman signals. In conclusion: the powder sample analyzed is mainly composed of calcite, probably mixed with siliceous materials (e.g., feldspar and quartz) whose identification by Raman spectrometry was not possible under the experimental conditions (fluorescence of the sample); several dark red particles inside the calcitic matrix are hematite. The pigment in a Spondylus shell After cleaning of the specimen with an ultra-sound cleaner, a reddish accretion associated with black particles could be observed under the naked eye adhering to the bottom of the inner,
26

concave side of a Spondylus gaederopus upper valve from Cueva de los Aviones. Analysis under the binocular microscope confirmed that the accretion represented a pigment residue, which was surface-analyzed on the specimen, with neither sampling nor preparation, by SEM/EDS and Raman spectrometry, using the same equipment and operational parameters described above for the Glycymeris shell. Identification of the Raman spectra used published sources (21-23). Under SEM/EDS analysis (back-scattered electrons mode), the red-colored areas did not contrast with the rest of the shell as much as required to allow for precise emplacement, except for the clearly accretionary scale in area a. of SI Fig. 17. In global analysis, this scale is mostly iron. Calcium predominates in the surrounding shell surface (where, however, iron occurs in a significant percentage, some times >10%), and the shape of the crystals, identified as calcite by Raman spectroscopy, suggests recrystalization. Raman spectroscopy also shows that the ironrich red matrix of this scale is lepidocrocite (FeOOH, an iron hydroxide). Under greater magnification, components in several shades of yellow, orange and red can be observed, but no other compound besides the red to brown lepidocrocite could be identified. Black and light inclusions were also observed in the scales matrix by SEM-EDS analysis. Some are mostly iron, calcium and magnesium; given the geological context, these last two elements are probably dolomite (CaMg(CO3)2), but inclusions rich in iron, or iron and sulfur, also exist. Raman spectroscopy shows that most black inclusions present the characteristic spectrum of carbon blacka category of potentially varied origin, e.g., wood charcoal, lampblack, bone or ivory charcoal (24)but hematite and pyrite (FeS2) were also identified. Visually, both the dolomite and the black inclusions seem to constitute an integral part of the pigmentatious mass. Conceivably, the dolomite and charcoal could represent postdepositional additions gained from the burial environment. However, this is unlikely, because, according to the excavation reports, only levels III and IV had any charcoalnone was recorded in level II, whence the shell comes, and detailed inspection of the profile corroborated these field observations (SI Fig. 4). Moreover, the hematite and pyrite inclusions present fresh breaks. Despite the non-homogeneous distribution of the inclusions, these facts suggest that the pigmentatious mass corresponds to a red base primarily made up of lepidocrocite to which were added, for binding or effect, ground elements of different origin and darker color charcoal, pyrite and hematite. Black and dark red inclusions also exist outside the red-colored areas of the shell, some of which could also be analyzed by SEM/EDS and Raman spectroscopy. In area b. of SI Fig. 17, hematite, carbon black and pyrite could be identified (SI Fig. 18; nos. 2 and 5, no. 3 and no. 1, respectively) but the composition of one of the grains (SI Fig. 18; no. 4), rich in iron and featuring poorly defined banding, remains unknown. SEM/EDS analysis of a white streak adjacent to the red-colored accretions (area c. of SI Fig. 17) revealed a very strong concentration in calcium, and Raman spectroscopy identified calcite (SI Fig. 19). Under SEM/EDS, calcium was shown to predominate in the red areas in direct contact with the streak, although iron percentages approached 10% in these areas; however, no component other than calcite could be detected by Raman spectroscopy. In a larger black inclusion analyzed by SEM/EDS (area d. of SI Fig. 17), Si and Al predominate, but Raman spectroscopy further identified carbon black (SI Fig. 19). This presence may explain the color of the inclusion, but the other elements raise the possibility that it corresponds to an unidentified compound, one whose association with the shell may either relate to the objects colorant-related function or represent post-depositional accretion of a substance present in the immediate burial environment. Two inclusions that could only be analyzed by SEM/EDS are rich in calcium and magnesium, in one case, and rich in sulfur and iron in the other. They probably correspond to dolomite and pyrite, respectively.
27

Except for the white streak, which is superimposed on the pigmentatious mass and probably represents a post-depositional precipitate, the carbon black, dolomite, pyrite and hematite accretions scattered around the scale in area a. of SI Fig. 17 are probably degraded residues of the same mass. This interpretation probably pertains as well to the pink/red staining of the areas of the shell adjacent to those where pigment was detected and where Raman spectroscopy obtained an iron signal. In conclusion ( SI Table 6): iron oxides (lepidocrocite and hematite) responsible for the red color of the accretions observed are indeed deposited on the shell; other componentscarbon black, pyrite, dolomiteexist in the pigmentatious mass analyzed, as well as in smaller accretionary clusters dispersed around it. The pigment in an ancillary metatarsal of horse Under the binocular microscope, small orange particles were observed in the tip of a broken ancillary metatarsal of horse (SI Fig. 20). This part of the bone was analyzed by SEM/EDS and Raman spectroscopy, with the equipment and operational parameters described above. Under SEM/EDS, the lighter areas were identified as post-depositional barium sulfate accretions, while the remainder of the bones surface appears almost entirely covered by deposits that are for the most part made of calcium carbonate. Two areas where orange particles are readily apparent under the binocular microscope were the object of a finer analysis. On area a., only carbon-rich (organic matter), silicium-rich (quartz?) and calcium-rich (calcite?) particles could be detected. Besides these components, area b. also showed the presence of copper and zinc particles, probably contaminants resulting from the metal tool used to scrape the carbonate coating in order to access the bone surface. No particles were detected that could correspond to the orange material. Raman spectroscopy was attempted but, because of the small size of the colorant particles and the high fluorescence of the sample, unsuccessfully. SI Table 6. Compounds identified in accretions on the inner side of a Spondylus shell from Cueva de los Aviones.
Shell area Red to yellow pigment scale White/pink areas adjacent to the scale Inclusions light black black dark red to black White streak Red areas adjacent to the streak 17 18, 19 18 18 19 19 Ca, Mg, Fe C, Si, Al Fe, S Fe Ca Ca, Fe Unknown Carbon black + unidentified alumino silicate Pyrite Hematite Calcite Pigment-stained calcite SI Fig. 17 17 Elemental analysis (SEM/EDS) Fe Ca, Fe Compound analysis (Raman spectroscopy) Lepidocrocite Pigment-stained calcite

28

Si

Ca

Al C Fe Na Mg K 3.5 Ca 4 4.5 5 5.5 6 Fe 6.5 Fe 7 7.5 keV

0 0.5 1 1.5 2 2.5 3 Pleine chelle 8684 cps Curseur: 0.015 (2173 cps )

SI Figure 15. SEM/EDS global analysis of the powder collected while cleaning the breccia-cluttered umbo perforation in a Cueva de los Aviones Glycymeris shell. Top. EDS spectrum: major elements are Si and Ca; minor elements are Al, Fe, with very small amounts of Na, K and Mg. Middle. Back-scattered electrons SEM image showing a probable feldspar (in the circled area) and flakes composed of Fe and Ti. Bottom. Magnification of the flakes (left) and the barium sulphate crystals (bright acicular elements) (right).
29

226 295

Red mineral (hematite)

Raman Intensity

666 412 617 1330

506

140

340

540

740

940

1140

1340

Wavenumber (cm-1)

1086

White mineral (calcite) Raman Intensity

284 155

714

140

340

540

740

940

1140

1340

Wavenumber (cm-1)

SI Figure 16. Raman spectroscopic analysis of particles in the powder collected while cleaning the brecciacluttered umbo perforation in a Cueva de los Aviones Glycymeris shell. Top. Microscope image and Raman spectrum of a red particle (laser power: 5mW; acquisition time: 60s); the bands at 1330, 617, 506, 412, 295, and 226 cm-1 correspond to hematite; in the databases, the 666 cm-1 band is attributed to either hematite or magnetite but its intensity and occurrence in all our spectra allows the attribution to hematite in this case. Middle. Microscope image and Raman spectrum of a white particle (laser power: 5mW; acquisition time: 10s); the bands at 1086, 714, 284 and 155 cm-1 correspond to calcite. Bottom. Microscope images under natural light of the sample showing red, translucent white and pink/white mineral elements.
30

d.

c.

b.

a.

2 3

251

Red matrix (lepidocrocite) Raman Intensity

381

647 533

1319

200

400

600

800

1000

1200

1400

Wavenumber

(cm-1)

SI Figure 17. Pigment residues in a Spondylus shell from Cueva de los Aviones. Top. Binocular microscope photo indicating the different areas analyzed. Middle. SEM (back-scattered electrons mode) image of the red scale in area a. (left) and detail of edge of the scale and adjacent shell surface indicating elemental compositions in three areas (1. Ca, Mg, Fe; 2. Fe; 3. Ca) (right). Bottom. Raman spectrum and microscope image of the scales red matrix, showing the characteristic bands of lepidocrocite (main bands: 251, 381, 533, 647 and 1319 cm-1).
31

2 4

340

Black inclusion (pyrite)

375

Raman Intensity

1324 429

180

380

580

780

980

1180

1380

Wavenumber (cm-1)
227 293

Dark red inclusion (hematite)

Raman Intensity

412 1328 616 506 668

180

380

580

780

980

1180

1380

Wavenumber (cm-1)
SI Figure 18. Pigment residues in area b. of a Spondylus shell from Cueva de los Aviones. Top. SEM (backscattered electrons mode) image; the indicated compounds (1. pyrite; 2. hematite; 3. carbon black; 4. unknown; 5. hematite) were identified by Raman spectroscopy. Middle. Raman spectrum and microscope image of a black inclusion identified as pyrite (main bands: 340 and 375 cm-1). Bottom. Raman spectrum and microscope image of a dark red inclusion identified as hematite (main bands: 227, 293, 412, 506, 616, 668 and 1328 cm-1).
32

White mineral (calcite) Raman Intensity

1084

153 206

135

335

535

735

935

1135

1335

Wavenumber

(cm-1)

Black inclusion (carbon black)

1604 1352

Raman Intensity
200

700

1200

1700

Wavenumber

(cm-1)

SI Figure 19. Accretions in areas c. and d. of a Spondylus shell from Cueva de los Aviones. Top. Raman spectrum and microscope image of the white streak in area c., which is calcite (main bands: 153, 206 and 1084 cm-1). Middle. Raman spectrum and microscope image of a large black accretion (area d.) identified as carbon black (main bands: 1352 et 1604 cm-1). Bottom. SEM (back-scattered electrons mode) images of the white streak, area c. (left) and the large black accretion, area d. (right).
33

a.

2 b.

SI Figure 20. The orange-dotted tip of an ancillary metatarsal from Cueva de los Aviones. Top left. Binocular microscope photo with indication of the areas with orange particles that were analyzed. Top right. SEM (back-scattered electrons mode) image of the bone tip, with indication of the different types of coatings identified (1. barium sulphate; 2. calcium carbonate). Bottom. SEM (back-scattered electrons mode) image of the areas with orange particlesarea a. (left) and area b. (right); the dark grey particles look like organic matter and the light ones like calcium or silica; bare bone appears white in the right hand image.
34

Supporting Information V: Analysis of mineral pigment samples from Cueva de los Aviones
During the 1985 excavation of Cueva de los Aviones, several lumps of colorant material were recognized and collected in levels I, II and III of the archeological sequence (25; SI Fig. 21). We found samples thereof alongside the sites faunal and lithic collections kept in the Museum of Cartagena, with clear provenience identification (to site, level and grid unit). From each sample, we extracted two subsamples of identical size. Given the destructive nature of the procedure, the even-numbered subsamples were kept as backup and the odd-numbered ones were submitted to X-Ray Diffraction (XRD) analysis in the Department of Agricultural Chemistry, Geology and Soil Science of the Faculty of Chemistry, University of Murcia. X-Ray Diffraction (XRD) analysis allows a semiquantitative estimation of the mineralogical composition of the samples and was carried out using Cu-K radiation with a PW3040 Philips Diffractometer. The X-powder software (26) was used to analyze the X-Ray diffraction diagrams obtained by the crystalline powder method. The powder diffraction file (PDF2) database was used for peak identification, taking into account that the determination of minerals from soils by XRD analysis is not accurate below a limit of 5% of the total weight in a sample (depending on the crystallinity of individual minerals). The results are given in SI Table 7 and the corresponding spectra in SI Figs. 22-23. The yellow colorants are primarily natrojarosite, while nontronite, hematite and goethite were identified in the red ones, and siderite on both types. Cueva de los Aviones is adjacent to one of the main mining districts of SE Spain, La Unin, where stratabound sulfide deposits are found in carbonate sequences of the Nevado-Filbride and Alpujrride complexes (27-28). These complexes consist of a series of superimposed thrusting mantles of Alpine age, affected by upward metamorphism. After a very strong erosion phase, a transgressive Neogenous series was deposited on these mantles. Subsequently, there was a phase of strong fracturing followed by volcanic phenomena and uplifting, and more recent erosion processes. This Pb-Zn-(Ag-Sn) district is an excellent example of the interplay between basin formation, normal faulting, volcanism, hydrothermal activity and mineral deposition (29), all synergistically combining to create exceptional concentrations of metals. The Neogene magmatic process is represented by calc-alkaline rocks (andesites, pyroxene and amphibolebearing andesites and dacites) and alkaline basalts (basanites and olivine basalts). The first is characterized by domes and dykes, with only rare lava and pyroclastic deposits. The domes are typically made up of fragments of volcanic rocks embedded in a matrix of the same composition. Basaltic volcanism is characterized by pyroclastic rocks and small lava flows. Despite their young age, the erupted materials are highly weathered and the volcanic vents are strongly eroded (30). Where these volcanites push up, they constitute the typical cabezos of the area (including the islands of the Mar Menor, a marine lagoon aligned along 70 N), as well as the embankments and domes (aligned 130 N) that intersect the litho-structural levels, including the Miocene. These volcanites, dacites and andesites (which have been assigned ages of 7 to 11 million years) are affected by strong hydrothermal alteration in the areas where mineral outcrops occur. The processes affect both the volcanic rock and the parent material. Hydrothermal activity related to middle Miocene subvolcanic magmatism was focused along normal faults bounding a graben. Mineralization occurred (i) within a strongly altered zone above the Miocene footwall; (ii) in pebbly mudstone beds where the hydrothermal activity led to dissolution, void formation and mineral deposition; and (iii) in fault breccias along the normal faults that bound the Miocene sediments (27). The primary mineralization of the area was succeeded by a strong secondary transformation process, resulting in the appearance of a large number of secondary
35

minerals, while the supergenic alteration processes affected both the primary mineralization products and the host rocks, as occurs in the nearby mining district of Mazarrn (31). This volcanism process occurred mainly from the Oligocene to the Quaternary along a strip measuring 150x25 km (9000 km2), between Almera (Cabo de Gata) and Cartagena (Manga del Mar Menor). It is known as the Almera-Cartagena Volcanic Belt (ACVB) (32) or Southeast Volcanic Province (SEVP) (33). Groutite is a manganese compound, while the other minerals identified in the colorant material analyzed are iron compounds. A local origin can be excluded for the latter, because the catchment of the Aviones karstic network is confined to Triassic dolomite and limestone terrain that went unaffected by the volcanism and hydrothermalism processes discussed above. The walls of the cave, therefore, cannot themselves be the source for the iron sulfides (e.g., pyrite) and the byproducts of their weathering (e.g., natrojarosite) identified in the Aviones samples simply because, in the region, such minerals did not form until some 200 million years after the laying down of the sediments that make up the sites bedrock. Likewise, the goethite and hematite samples cannot result from the precipitation in the archeological deposits of minerals present in the encasing dolomite and limestone walls because, in these rocks, iron only occurs in trace amounts. Intercalated lenses of red clay can be observed in the dolomite stratification exposed in the cave walls, and it is therefore conceivable that wall degradation processes contributed the mineral components of these clays to the sites fill. There is no evidence, however, that such contributions occurred as masses of macroscopic size, as none were recorded in the profiles or during excavation, and the texture of the archeological deposits is that of a coarse sand (SI Fig. 4). So, although provenance from the clay lenses intercalated in the dolomites is conceivable for any iron oxides and hydroxides (e.g., hematite and goethite) found in the sedimentological fine fraction of the Aviones deposit (i.e., in the clay and silt grain size classes), such a provenance can be excluded for the macroscopic-sized colorant samples. For the lepidocrocite identified in a Spondylus shell from level II (Supporting Information IV), however, this is not even an issue, because this mineral is absent from calcareous soils, and the same applies to natrojarosite, which only occurs in sulfidic soils (34). Since the sampled colorants are not of local origin, they can only represent an anthropic contribution to the Aviones deposits. Potential sources are ubiquitous throughout the ACVB, and the nearest such suitable terrain lies ~3.5 km to the west (around the Cabezo de la Estrella) or ~5 km to the north (across the Sierra de Pelayo) (1-2). However, for jarosite and other minerals of the same family, like the natrojarosite in subsamples 1 and 11, the number of source localities can be narrowed down to a few. These minerals result from the weathering of pyrite and other iron sulfides when and where the latter are atmospherically exposed, which results in the formation of weathering crusts that can also contain secondary byproducts of the process (e.g., goethite and other hydroxides). Thus, the acquisition of pure natrojarosite, as that in subsample 1, requires a source where the parent mineral is of considerable mass as, otherwise, contaminants derived from the weathering of the encasing rock would also be present (35-36). Such quality jarosite can therefore be presumed to come from one of the major outcrops of iron sulfides in the region, whose location is provided in SI Fig. 24. The closest are located in an area ~7 km east of the cave, along the Rambla de Escombreras, across the bay of Cartagena. Goethite and hematite can also be found beyond the boundaries of the ACVB. On Triassic bedrock especially, small, non-mapped outcrops can exist, because, throughout the province of Murcia, the Triassic features varied marly deposits, in general with a high content in gypsum and other soluble salts, that are, therefore, rather plastic. These deposits lie at the base of the numerous diapirs of the interior of the province (Segura valley and the different Tertiary lacustrine basins), where ophite outcrops occur. The latter are intermediate volcanic rocks that can be associated with hydrothermalism processes resulting in the presence of iron minerals
36

(hematite, magnetite) and, with hydrothermal weathering, in the generation of clays (smectite) and such secondary minerals as iron hydroxides (goethite, lepidocrocite, limonite, etc.). Another potential source for hematite are the clays that result from the degradation of carbonate rocks, in which case quartz and calcite in varying amounts will also be present. Such clays can be found as constituents of the red soils found in the Campo de Cartagena. However, the presence of siderite in all the red-colored samples from Aviones indicates that any potential hematite sources in the Campo de Cartgena area can be excluded because siderite cannot form as a result of soil weathering processes. Therefore, the Aviones hematite- or siderite-based red pigments must come from sources related to post-volcanic hydrothermal processes such as those associated with the diapirs in inland Triassic bedrock or with the littoral AlmeraCartagena Volcanic Belt. The latter are those closest to the site. In short, the mineral pigment masses sampled during excavation in levels I-III of Aviones are manuports, not natural components of the deposit. A measure of uncertainty remains concerning the exact location of their sources, but the local and regional geology implies transport over distances >3 km for the red pigments and >5 km for at least the yellow natrojarosite. SI Table 7. Provenience of the pigment samples from Cueva de los Aviones and results of the XRD analysis.
Subsamples 1&2 3&4 5&6 7&8 9 & 10 11 & 12 Square C1 C1 (?) A1 A1 A1 A1 Level III III I II II II Color Yellow Red Red Yellow Red Yellow Subsample analyzed and colorant minerals identified 1 Natrojarosite 3 Siderite 5 Nontronite, Siderite 7 Siderite 9 Goethite, Hematite, Siderite 11 Groutite, Natrojarosite, Parabutlerite

37

SI Figure 21. Cueva de los Aviones pigments. Top. Page of the excavation diary recording the presence of pigments in level III, grid unit B1. Bottom. Pages of the field books from grid units A1 (left) and C1 (right) recording the presence of yellow and red pigments in levels I and III, respectively.
38

681 613 545 477 409 341 272 204 136 68 0

Counts

Natrojarosite

2 mm

Source of subsample 1 [Level III, square C1]

5 670 603 536 469 402 335 268 201 134 67 0 5 660 594 528 462 396

10

15

20

25

30

35

40

45

50

55

60

65

70 75 2-theta

Counts

Siderite

Source of subsample 3 [Level III, square C1]

10

15

20

25

30

35

40

45

50

55

60

65

70 75 2-theta

Counts

264 198 132 66 0 5 10

Nontronite

Siderite
Source of subsample 5 [Level I, square A1]

15

20

25

30

35

40

45

50

55

60

65

70 75 2-theta

SI Figure 22. Cueva de los Aviones. XRD pigment analysis. Top. Subsample 1; all peaks are natrojarosite. Middle. Subsample 3; other peaks are quartz, calcite, dolomite and muscovite. Bottom. Subsample 5; other peaks are anorthite, calcite and muscovite.
39

5 mm

330

646 581 517 452 388 323 258 194 129 65 0

Counts

Siderite

5 mm

Source of subsample 7 [Level II, square A1]

5 631 568 505 442 379 316 252 189

10

15

20

25

30

35

40

45

50

55

60

65

70 75 2-theta

Counts

Goethite

Siderite Hematite
Subsample 9 [Level II, square A1]

126 63 0 5 658 592 526 461 395 329 263 197 132 66 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 2-theta 10 15 20 25 30 35 40 45 50 55 60 65 70 75 2-theta

Counts

Groutite Natrojarosite Parabutlerite

Subsample 11 [Level II, square A1]

SI Figure 23. Cueva de los Aviones. XRD pigment analysis. Top. Subsample 7; other peaks are quartz, calcite and muscovite. Middle. Subsample 9; other peaks are quartz, calcite, muscovite, dolomite and sanidine. Bottom. Subsample 11; other peaks are quartz.
40

SI Figure 24. Cueva de los Aviones. Sourcing of the colorant materials analyzed by XRD. Top. Physiographic map. Bottom. Geological map (same scale). The white triangle represents Cueva de los Aviones, and the black triangle represents Cueva Antn. Two potential source areas exist for the red colorants, the littoral Almera-Cartagena Volcanic Belt and the ophites and associated weathered deposits of the interior diapirs, separated by the Quaternary glacis of the Campos de Cartagena, where siderite cannot occur. The orange dots indicate the major outcrops of iron sulphides whence the pure natrojarosite from level III probably came.
41

Supporting Information VI: Analysis of mineral pigments in a Cueva Antn Pecten

Naked eye and binocular microscopic inspection of an upper valve of Pecten sp. recovered at Cueva Antn suggested that residues of red/yellow pigment were preserved on one of the sides of the shell. In order to verify these observations and identify the minerals present, nondestructive analyses using X-Ray fluorescence and micro-Raman spectroscopy were carried out at the Institute of Materials Science, University of Valencia. X-Ray fluorescence (XRF) We used a portable XRF spectrometer with a silver transmission anode and a Si-PIN semiconductor detector cooled by Peltier effect with a resolution of 170 eV (FWHM @ 5.9 keV). The analyzes were carried out at 30 kV, 3 A, and with an acquisition time of 180 seconds. The measuring geometry was 45, with the incident beam collimated at a diameter of 3 mm and directed perpendicular to the sample. The distance between sample and detector was 18 mm. The internal face of the shell presents a reddish coloration, for which observation under binocular microscope suggested a biogenic cause. The external face presents a whitish coloration and features red/yellow terrigenous deposits that, under the binocular microscope, overlie the surface of the shell. We carried out XRF analysis of one of these accretions (point P01) and compared the results with those obtained by XRF analysis of the shells body in areas where it presents both a whitish (point P02) and a reddish (point P03) coloration. In the fluorescence spectra obtained (SI Fig. 25), the fluorescence peaks related to the constituent materials of the measuring equipment were marked with the label tube. The position of the so-called Ca-escape peak, which relates to the presence of this element, was also signaled. The fluorescence spectra indicate a calcareous matrix (Ca) with impurities of iron (Fe), sulfur (S) and strontium (Sr); the latter is often associated with calcium. The spectra of the calcareous matrix are practically identical in both the whitish and the reddish areas, indicating that the coloration is biogenic. The spectrum of the red/yellow accretion presents an iron fluorescence peak significantly more intense than that of the calcareous matrix. This indicates the presence of an iron oxide compound with an iron concentration higher than that of the shells; RAMAN spectroscopic analysis (see below) indicates that goethite (FeO[OH]) is its major component. The accretionary nature of this mineral deposit is best appreciated in the detail graph at the bottom of SI Fig. 25. This graph represents, on a linear scale, the Ca and Fe fluorescence peaks. In point P01 (the accretion), an increased Fe fluorescence corresponds to a decreased Ca one. This indicates that Fe attenuates the Ca signal in the areas with superficial deposits of iron oxides. Micro-Raman spectroscopy Micro-Raman spectroscopy was carried out at room temperature, in a backscattering configuration and using the 514.5 nm line from an Argon laser for excitation. The scattered light was dispersed by a Jobin Yvon T64000 triple spectrometer and detected with a liquid nitrogen-cooled CCD. The typical system spectral resolution was around 2 cm-1. The laser beam was focused on the samples surface using a Leica microscope 100x objective. In order to prevent local heating and transformation, laser power varied from 3 to 14 mW. Recording times ranged from 30 to 300 seconds, with 2 to 10 accumulation scans depending on the sample. Measurements were carried out directly on the sample with no previous preparation of any kind.
42

The Raman spectra in SI Fig. 26 have been corrected for background fluorescence and the base line is shifted for clarity. The identification of blanks and pigments was carried out by comparison with published spectra and available databases. Red/yellow pigment is visible only on the whitish, external side, where it can be found across the whole surface, with a discontinuous distribution, filling grooves and spilling over adjacent areas. To see whether undetected pigment residues existed also on the internal, reddish side, we obtained Raman spectra from both sides and in different parts of the shell. The areas analyzed are indicated in SI Fig. 26. The Raman spectrum of the internal (reddish) side shows calcite bands (285, 714 and 1088 cm-1) (37) attenuated by the presence of bands related to the pigments of the shell itself (1019, 1134 and 1301 cm-1), which are carotenoids (38-40). No anthropogenic mineral pigments were detected on this side of the shell. The different spectra obtained in the three areas with red/yellow pigment that were analyzed show very clearly the presence of bands that correspond to two iron oxides: hematite (Fe2O3), responsible for the red color (290, 408, 506, 608 and 642 cm-1); and goethite (FeOOH), responsible for the yellow color (215, 272, 386 and 588 cm-1) (22, 41). The final result is a mixture of both pigments, yielding an overall orange hue. Hematite is the most stable iron oxide. Nevertheless, structural variations from pure hematite, due to the replacement of Fe atoms by other atoms (e.g., Al or OH groups), can be found in nature. These atomic substitutions can cause displacement of the Raman bands (by comparison with those for the well crystallized, pure oxide), due to small variations in the crystalline structure. The Raman bands observed in the analyzed oxides fit those of pigments that were naturally modified and subject to ageing and corrosion processes (41). The same applies to goethite. No bands around 660 cm-1 were detected. This band, which indicates the presence of magnetite, is known to occur in some spectra, often as a result of the utilization of too much laser power. Hematite and goethite occur in the La Unin mining district, in the littoral area of the Murcia province, but they can also occur as a result of the weathering of ophites and other volcanics, e.g., fortunites, in diapirs of the interior (see Supporting Information V). Since an outcrop of fortunites is mapped near the town of Puebla de Mula, ~5 km from Cueva Antn, such is the closest potential source for the mineral pigment found on the Pecten shell recovered at the site.

43

PO1

PO3

PO2

10-1

Ca P01 red/yellow accretion P02 white (shells body) P03 red (shells body) Sr
tube

Normalized counts

10-2
Ca-escape tube

Fe Fe S

10-3

10-4

10-5 5 10 15 20

Energy (keV)
0.05

Ca

0.04

Normalized counts

0.03

P01 red/yellow accretion P02 white (shells body) P03 red (shells body)

0.02

0.01

Ca

Fe

0.00 3 4 6 7

Energy (keV)
SI Figure 25. XRF study of the Cueva Antn Pecten. Top. The shell, with indication of the points where the different spectra were obtained. Middle. The fluorescence spectra. Bottom. Detail of the fluorescence spectra in the areas of the calcium and iron peaks.
44

1134

Raman Intensity

Internal (red) side [calcite with carotenes]

1088 285 714 1019 977 1301

150

250

350

450

550

650

750

850

950

1050

1150

1250

1350

Raman Shift
272

(cm-1)

215

External (white) side [mineral pigments] Goethite

408 290 386 506 608 588

Raman Intensity

Hematite
50 m

642

1088 1123

150

300

450

600

750

900

1050

1200

1350

Raman Shift (cm-1)

SI Figure 26. Raman micro-spectroscopic study of the Cueva Antn Pecten. Top. The shell, with indication of the points where the different spectra were obtained. Middle. Results for the internal (reddish) side. Bottom. Results for the external (whitish) side and microphoto (at 100x) of pigment residue.
45

Supporting Information VII: Reference collection of Acanthocardia, Cerastoderma and Glycymeris shells from modern beaches of Murcia and Alicante
In order to assess the nature and significance of the perforated bivalve shells from Cueva de los Aviones and other prehistoric sites of Mediterranean Spain, we assembled a reference collection whose composition and collection criteria are given in SI Table 8. Because of the different purposes involved, we used a set of attributes different and simpler than those in a recent, paleontologically oriented taphonomic study of Glycymeris shells (42). The collection is kept and openly available for analysis at the University of Murcia, Department of Zoology and Physical Anthropology. All measurements were made with a digital caliper, following Claassen 1998 (43); perforation dimensions were measured internally, as the diameters of an ellipse. These measurements and the associated discrete criteria (defined in SI Table 9) are tabulated in a Supporting Data Excel file and summarized in SI Table 10. Samples 4 and 5 reflect a purely marine catchment. Sample 4 is made up of weathered material and was almost entirely collected in a concentration of 5-10 m radius in Cala Arturo, immediately to the north of the Calblanque beach (Murcia), but also contains a few specimens found along the latters dunes. Sample 5 is made up of a very few specimens collected in the intertidal zone of a beach to the west (Playa Parreo), but mostly comes from a back dune exposure and, given small sample size, was not included in the study. These beaches are 23 km due east of Cueva de los Aviones, in a similar geological and geomorphological setting, and yielded bivalve shell assemblages of similar composition to that recovered at the site. All the Glycymeris shells in these samples belong to the G. insubrica species. Samples 1 and 2 are primarily made up of Cerastoderma glaucum and reflect a mixed estuarine/marine catchment, in agreement with the location of the Guardamar beaches (Alicante), north and south of the mouth of the Segura river. Besides a majority of recent material, these samples may also contain shells of earlier ages derived from submerged Pleistocene or Early Holocene deposits. The shells are for the most part weathered, but display a rather varied surface appearance, from erosion limited to the dorsal edges with near intact crenulated ventral edges to heavy corrosion and thinning of the entire shell (as if bathed in acid); also, a high proportion bear a grey color indicative of post-mortem burial in clay environments deprived of oxygen. In contrast, most Acanthocardia tuberculata specimens are in near-pristine condition, especially the complete ones in sample 3. Another species of the same genus, A. paucicostata, is also represented in the Guardamar samples but in low numbers and, hence, was not considered in the analyzes. In the G. insubrica sample (SI Fig. 27), 45.8% of the shells preserving the umbo area are perforated, and, with the exception of a dorsally placed hole in a specimen that is both very small (18.2 mm height) and quite thinned, all the perforations are in the umbo. The presence of a perforation is not determined by size, as the height of perforated (36.56.9 mm) and nonperforated (37.57.2 mm) shells is statistically the same, and there is no correlation (r=0.34) between the size of the shell and the area of the hole (measured as an ellipse), for which the maximum internal diameter is a good predictor (r=0.89) and was therefore used as a proxy (in the preceding as well as in the following discussion, averagestandard deviation values are used). The key factor underlying variation in both presence and size of the perforation is weathering stage. The more weathered the shell is, the higher the probability that the umbo will be perforated (stage 4 is an apparent exception only, as it is defined on the basis of different criteria; SI Table 9), and the size of the perforation increases with weatheringfor shells where the umbo area was preserved and hole size was measurable, only one stage 1 specimen (1.3%) was perforated, and with a small hole (0.9 mm), contra 63.7% of the material in stages 2 and 3 (hole size, 5.32.9 mm) (Supporting Data; SI Table 10, Fig. 28).
46

In the C. glaucum sample (SI Fig. 29), the percentage of perforated shells, mostly below the umbo, is 7%. Only two (0.3%) fall in the full umbo category and, in both cases, the shells are small (with heights of 13.3 and 17.5 mm) and the holes narrow (1 and 2 mm, respectively). Eight shells (1.1%), all also rather small (between 9.8 and 18.7 mm in height), feature an open umbo perforation with a larger diameter (3.91.2 mm). The number of shells with a perforation below the umbo is 40 (5.5%), of which only one is ventrally placed. These shells are larger (20.44.9 mm in height), and so are the holes (8.03.8 mm). Significantly, 57.5% of the dorsally or ventrally perforated shells display visible exfoliation, a percentage that is similar to that observed among the umbo-perforated specimens (50%) and contrasts markedly with the frequency of the attribute (11.0%) among the non-perforated ones. It has been argued that, in Cerastoderma, a dorsal or ventral position of the hole is an indicator of anthropic origin because umbo perforations should be the norm in such shells (44). As the Guardamar collection shows, this argument is not valid. The positive correlation between exfoliation and perforation indicates that, in Cerastoderma, perforation occurs naturally as a byproduct of corrosion/decalcification processes, as the examples in SI Fig. 29 well illustrate. That, possibly in relation to burial environment, corrosion rather than abrasion is the main weathering process affecting Cerastoderma shells may also explain why, in this taxon, perforations below the umbo are four times more common than perforations of the umbo (although factors relating to shell geometry and shell texture may also be involved). In the A. tuberculata sample (SI Fig. 30), the percentage of perforated shells (4.5%) is of the same order of magnitude as that found among C. glaucum, but the number of full umbo (three) and dorsal (two) holes is similar. A significant difference between the two taxa is that exfoliation is a rare occurrence (1.8%) in the A. tuberculata sample, suggesting that, in this case, the main weathering process in operation is abrasion rather than corrosion. That a higher energy burial environment pertains for the A. tuberculata shells from Guardamar is corroborated by the fact that the two dorsal holes occur in shells with rather good surface condition and result from shock breaks that entailed loss of a circular area below the umbo, with no subsequent abrasive smoothing of the edges of the perforation. These shells illustrate how entirely natural processes can mimic direct percussion and imply (contra refs. 45-46) that perforation by direct percussion is not, by itself, a valid criterion upon which to base an anthropogenic diagnosis. In conclusion, the two types of wear evidence that have been invoked in the literature to support an anthropogenic diagnosis for the perforations seen in shell ornamentsabrasion and direct percussionare entirely natural processes too. Therefore, at least where bivalves are concerned, and especially so for the taxa that are most common in archeological assemblages and for which reference material is available (Cerastoderma, Acanthocardia and Glycymeris), our observations imply that natural, as opposed to anthropic, is the null hypothesis for the origin of the perforation, and that a perforation can be deemed anthropogenic only if one of the following conditions applies: evidence is produced that a tool was involved in the perforation, e.g., microwear typical of the use of a drill; the weathering stage and perforation patterns (size and emplacement) deviate from those seen in natural thanatocenoses, e.g., for Glycymeris, if the hole is large but made on a fresh, unweathered shell and/or away from the umbo; the hole is associated with artificial modification of the shells geometry, e.g., gouging of the external surface to prepare a perforation in the middle of the shell, as in the Neolithic specimen from Gruta do Caldeiro illustrated in SI Fig. 28 (47).

47

SI Table 8. Collection data for the modern bivalve reference sample.

Parameters Date Time # of collectors Beach Province Geographical coordinates Target taxa Sample 1 10-01-2009 16:15-17:45 2 Guardamar Alicante 384'58" N to 385'53" N Acanthocardia Cerastoderma Glycymeris Sample 2 02-02-2009 13:15-15:15 2 Guardamar Alicante 386'05" N to 386'37" N Acanthocardia Cerastoderma Glycymeris Sample 3 02-02-2009 16:30-17:30 2 Guardamar Alicante 386'46" N to 387'30" N Acanthocardia Glycymeris Sample 4 03-02-2009 12:00-13:30 2 Cala Arturo Murcia 043'33" W to 043'53" W Glycymeris Sample 5 03-02-2005 14:00-15:00 2 Parreo Murcia 045'16" W to 045'52" W Glycymeris

Collection criteria

Glycymeris, regardless of Any and all, completeness, plus complete or dorsal regardless of Cerastoderma/Acanthocardia completeness (broken ones collected occasionally) Chamelea gallina, Spondylus gaederopus, Arca noae, Mytilus galloprovincialis, Chama (?) sp., Callista chione, Laevicardium sp. (cf. oblongum), Donax trunculus, Modiolus sp. (cf. barbatus), Tapes decussatus, Eastonia rugosa, Tellina incarnata, Lima inflata, Lima lima, Chlamys glabra, Chlamys varia, Mactra stultorum, Venus verrucosa, Ostrea edulis Patella aspera, Patella caerulea, Fissurella nubecula, Semicassis undulata, Thais haemastoma, Turritella communis, Trunculariopsis trunculus, Aporrhais pespelecani

Any and all, regardless of completeness

Other bivalve taxa recorded

Cerastoderma glaucum, Chamelea gallina, Spondylus gaederopus, Acanthocardia sp., Arca noae, Mytilus galloprovincialis, Anomia ephippium, Chama (?) sp.

Gastropod taxa recorded

Patella aspera, Patella caerulea, Patella rustica, Semicassis undulata, Thais haemastoma, Charonia lampas, Cancellaria cancellata, Bulla striata, Acteon tornatilis, Columbella rustica, Calliostoma laugieri, Monodonta turbinata, Nassarius mutabilis, Nassarius sp. (cf. incrassatus or pygmaeus), Nassarius nitidus, Nassarius granum

48

SI Table 9. Definition of discrete attributes recorded in the modern bivalve reference sample.
Attributes Weathering A. tuberculata and C. glaucum G. insubrica

0. fresh Yes/No (Yes, if crenulated edge is significantly abraded on the ventral 1. some weathering; hinge teeth and denticulation side, opposite the umbo; not recorded of ventral margin well apparent for C. glaucum) 2. significant weathering; teeth and/or denticulation of ventral margin partly erased but both still somewhat apparent in part 3. major weathering; teeth and denticulation of ventral margin (or at least one of the two) completely erased 4. fragment with umbo whose breakage edges were fully regularized/rounded (with no angular breaks) by subsequent abrasion

Sponge holes

Through: if, visibly to the naked eye, there are sponge holes that traverse the shell from one side to the other Non-through: if there are sponge holes even if they don't traverse the whole shell Absent Yes/No Yes/No Yes/No

Yes/No (through-holes only)

Exfoliation Predator holes Bioaccretions Intactness

Not recorded Yes/No Yes/No

Complete: height and length measurable despite small mechanical fractures (if any), reduction in size from living due to abrasion only Dorsal: umbo area preserved, losses to mechanical fracture elsewhere, at least one dimension non-measurable Ventral: umbo area lost to mechanical fracture, height non-measurable Edgeless: fragment comprised of flattish area in the middle of the shell, with all edges rounded and no identifiable area of the shell margin present

Perforation

Full umbo: hole with complete circumference preserved Open umbo: hole with break to the inner side Ventral: hole closer to the ventral than to the dorsal edge Dorsal: hole closer to the dorsal than to the ventral edge None: no perforation

49

SI Table 10. Descriptive statistics for the modern bivalve reference sample (all measurements are in mm and given as averagestandard deviation).
Attributes Weathering Stage A. tuberculata (samples 1-3) N = 94 Yes = 22.4% No = 77.6% C. glaucum (samples 1-2) Not recorded G. insubrica (sample 4) N = 705 0 = 0.0% 1 = 17.5% 2 = 27.4% 3 = 50.3% 4 = 4.9% Sponge holes N = 110 Absent = 96.4% Non-through = 0.0% Through = 3.6% Exfoliation N = 110 Yes = 1.8% No = 98.2% Predator holes N = 110 Yes = 1.8% No = 98.2% Bioaccretions N = 110 Yes = 0.9% No = 99.1% Intactness N = 110 Complete = 54.5% Dorsal = 45.5% N = 715 Absent = 98.0% Non-through = 1.5% Through = 0.4% N = 715 Yes = 14.1% No = 85.9% N = 715 Yes = 0.3% No = 99.7% N = 715 Yes = 0.6% No = 99.4% N = 715 Complete = 90.3% Dorsal = 9.7% N = 705 Yes = 0.3% No = 99.7% N = 705 Yes = 0.7% No = 99.3% N = 705 Complete = 29.9% Dorsal = 14.0% Ventral = 37.9% Edgeless = 18.2% Perforation N = 110 Full umbo = 2.7% Open umbo = 0.0% Ventral = 0.0% Dorsal = 1.8% None = 95.5% Height Length Thickness Height (perforated) Maximum internal diameter of the perforation 28.29.5 (N = 92) 25.47.3 (N = 61) 10.93.9 (N = 88) 29.64.6 (N = 6) 15.31.2 (N = 3) N = 715 Full umbo = 0.3% Open umbo = 1.1% Ventral = 0.1% Dorsal = 5.4% None = 93.0% 19.44.3 (N = 695) 20.14.3 (N = 659) 8.52.1 (N = 685) 19.44.9 (N = 50) 4.93.9 (N = 50) N = 310 Full umbo = 42.9% Open umbo = 2.6% Ventral = 0.0% Dorsal = 0.3% None = 54.2% 33.18.6 (N = 242) 33.5.8.8 (N = 222) 11.93.3 (N = 255) 36.96.3 (N = 89) Stage 1: 0.9 (N = 1) Stage 2: 4.52.7 (N = 30) Stage 3: 5.62.9 (N = 58) N = 705 Absent or Non-through = 94.3% Through = 5.7% Not recorded

50

SI Figure 27. Glycymeris insubrica shells from Cala Arturo beach. Examples of the different weathering stages recognized, from stage 1 (at the top) through to stage 4, at the bottom.
51

G. insubrica from Cala Arturo (Calblanque, Murcia)

Stage 4
18 9 10

100%

Non-perforated

75%

Stage 3

67 18

67

79

50%

74 79

Stage 2

25%

Perforated

Stage 1

74

52 10

53

0%

Nonperforated

Perforated

Stage 4

Stage 3

Stage 2

Stage 1

SI Figure 28. Perforated Glycymeris shells Top. Presence of a perforation as a function of weathering stage in the G. insubrica shells from Cala Arturo. Middle. Close-up views of the perforations in two G. insubrica stage 3 shells from Cala Arturo. Bottom. Anthropically gouged and perforated G. glycymeris shell from the Early Neolithic of Gruta do Caldeiro, Portugal.
52

0%

25%

50%

75%

100%

SI Figure 29. Cerastoderma glaucum shells from Guardamar beach. Top. Shells with predator holes. Middle. Shells with incipient dorsal perforation. Bottom. Dorsally perforated shells (note that the holes formed in areas thinned by corrosion and exfoliation).
53

SI Figure 30. Acanthocardia tuberculata shells from Guardamar beach. Top. Umbo-perforated shells; the edges of the holes are smoothed by continued abrasion of the dorsal area of the shell. Bottom. Dorsally perforated shell; the hole was formed by shock break, mimicking direct percussion, with no subsequent smoothing.
54

Supporting Information VIII: Perforated bivalves from prehistoric sites of Mediterranean Spain
In order to assess whether the characteristics of the perforated shells from Cueva de los Aviones deviated from those found in later prehistoric times, we examined, using the same criteria and definitions applied to the modern bivalve reference collection (Supporting Information VII), the perforated Acanthocardia, Cerastoderma and Glycymeris from two major sites of Mediterranean Spain kept in the Prehistory Museum of Valenciathe Upper Paleolithic (Solutrean and Magdalenian) Cova de Parpall (44-45) and the Early Neolithic (Cardial and Epicardial) Cova de lOr (48). Counts and metrics are provided in SI Table 11 and SI Fig. 31, and the studied specimens are illustrated in SI Figs. 32-42. Parpall The Parpall inventory (45) mentions six Cerastoderma shells, five of which are drawn; we were able to examine these five, plus two additional specimens that we could not match to the published description. Where Acanthocardia and Glycymeris are concerned, two and five shells, respectively, had been previously reported and are the same that we examined and illustrate herewith, except for a Glycymeris of uncertain stratigraphic provenience that is not drawn and that we could not find either. Concerning the origin of the perforations, those in the two Acanthocardia had been diagnosed as anthropic and possibly made by a drill; those in three of six Cerastoderma had been diagnosed as anthropic (one of which possibly through the use of a drill), and one as a break produced either by natural processes or by the use of direct percussion; and those in four of the five Glycymeris were also diagnosed as anthropic (one made by abrasion, another by percussion, the other two with a drill). In our examination, however, we detected tool marks in none of the shells. Moreover, their surface condition and weathering stage are consistent with a natural origin of the perforation in all cases. The Acanthocardia (SI Fig. 32) and Cerastoderma (SI Fig. 33-34) are heavily weathered and exfoliated shells, the perforations often coincide with the most degraded area of the shell, and the edges of the perforations are smoothed and pierced by clionid holes whose distribution and placement indicate that the perforation occurred in the sea bottom, prior to beaching. The single exception is a relatively unweathered Cerastoderma that presents a large, irregular hole with sharp, fresh breaks. However, such damage, mimicking direct percussion, can occur naturally, as exemplified by the modern Acanthocardia from our reference collection illustrated in SI Fig. 30. Moreover, the fact that, in contrast with the predominance of umboperforated specimens in the other two taxa, Cerastoderma are principally perforated dorsally, below the umbo (six out of seven), matches the pattern seen in the modern reference collection. Where the Glycymeris are concerned (SI Fig. 35-36), the four shells that we examined all feature advanced weathering (stage 3) and, in our reference collection, 89.8% of the specimens assigned to that stage bear a natural perforation. We conclude that all the Upper Paleolithic perforated bivalves from Parpall that have been classified as ornamental shells correspond to naturally perforated, beach-stranded material. Deliberate collection of such material by the caves inhabitants is independently corroborated by three Donax trunculus shells with typical predator drill-holes (49) (SI Fig. 31). Or The Or inventory of perforated shells (48) mentions six Acanthocardia, 11 Cerastoderma and 33 Glycymeris and considers them all to be anthropogenic. Only four of the Acanthocardia and 16 of the Glycymeris could be made available for analysis but, as this sample corresponds to 50% of the collection, it can be consdiered as representative of the collection as a whole.

55

Like those from Parpall, the Or Acanthocardia (SI Fig. 37) are exfoliated, decalcified and weathered specimens featuring clionid holes on the edges of perforations of smoothed appearance, all of which is consistent with natural perforation in the sea bottom, prior to beaching. The Glycymeris (SI Fig. 38-41) are all in weathering stages 2 or 3 and, in our reference collection, 43.7% of the stage 2 shells and 89.8% of the stage 3 ones bear a natural perforation. Moreover, no tool-marks were apparent on any of these shells. As in Parpall, the Cova de lOr shell ornaments correspond to naturally perforated, beach-collected material. In three Glycymeris shells (F2/c5 and the two labelled H/c7; SI Figs. 38, 41), the umbo area is abrasion-flattened, as can also be seen in H1/c6, a non-perforated specimen (SI Fig. 42). In all four, however, the edges of the flattened surface are round and smooth, not at an angle with the rest of the shell, as one would expect if they were anthropogenic. Moreover, the flat areas remaining around the perforations show the same kind of pitting from bioerosion and corrosion apparent throughout the shells surfaces. Both the abrasion-flattening and the perforations are entirely natural processes (42), and any human involvment in the production of the shells extant morphologies can only have consisted of the modification of a pre-existing condition. Conclusions A significant difference between the two sites is that larger shells were being targeted in the Early Neolithicthe height of the Or material is 44.010.5 mm, as opposed to 19.44.3 mm at Parpall. However, it should be noted that the latter are very small by comparison with the umbo-perforated shells of the reference assemblage, whose height is 36.96.3 mm (SI Tables 10-11, SI Fig. 31) (all values in the preceding are averagestandard deviation). Another, possibly preservation-related difference is that ochre residues are conspicuous in eight (one Acanthocardia and seven Glycymeris) of the 20 perforated Or shells that we could examine. At this site, and in the Early Neolithic in general, the symbolic value of Glycymeris shells is further attested by their use as raw-material for the manufacture of tear-shaped beads, one of the periods most common ornament types (and represented at Cova de lOr by 36 specimens) (48). At Gruta do Caldeiro (Portugal), such tear-shaped beads were found in burials together with perforated whole valves (47). This Neolithic context of mass production of artificially cut, shaped and pierced Glycymeris beads makes it all the more remarkable that the perforations in all the umbo-perforated complete valves from Cova de lOr included in our study are natural rather than anthropic. At Parpall, despite the smaller size of the shells, the maximum diameters of the holes are in the same range (~5 mm) seen at Cova de lOr, and, when the totality of the archeological material that we examined is compared with the modern reference sample, it is apparent that the primary target of the prehistoric collectors consisted of shells that, irrespective of overall size, had umbo perforations in the 4.5-6.5 mm range (SI Fig. 31). In the past as much in the present, it is clear, assuming an ornamental function for the shells, that, at the time of collection, people would have used a categorical classification of the perforations (and calipers did not exist in prehistoric times anyway): good to pass a string through versus not so good to pass a string through. Bearing this in mind, the selection pattern is also shown by the pvalues obtained when chi-square testing the difference in composition of the two samples (archeological versus natural) when thusly (good versus not so good) divided up: depending on the proxy used for good perforation sizes, 0.0245 (using the inter-quartile range of the perforations in the archeological specimens), or 0.0377 (using the modal class of the histogram of archeological perforations), both statistically significant. This featurepossession of a hole of the appropriate sizeseems therefore to have been the main criterion underlying selection, implying that color, shine and overall surface appearance were less important than threading constraints. Such a selection pattern would explain why the archeological Glycymeris are entirely made up of material in weathering stages 2 or 3, despite shells in stage 1 preserving more of the original color and being rather easy to find too, since
56

they can be found in significant numbers in the same beach thanatocenoses as the more weathered material. In our reference collection, stage 1 shells are 17.5% of the total, but only one is perforated, and with a very small, <1 mm wide hole, while 92.1% of the perforated material assigned to stages 2 or 3 features holes in the size range (0.6-9.5 mm) observed in the three archeological collections (Aviones, Parpall and Cova de lOr). These inferences are consistent with the fact that archeological Glycymeris often present residues of ochre whose overall distribution on the shells indicates they once were entirely paintedsomething one would not expect if freshness and original color impacted the meaning attached to the shell in any significant way. Although not considered in this discussion because of uncertainty regarding their representativity, we also examined perforated Acanthocardia (two shells) and Glycymeris (13 shells) from the Early Neolithic cave site of La Sarsa (Valencia) (48) and the Eneolithic open air settlement of Las Amoladeras (Murcia) (50). These samples match point by point the characteristics of those presented above, as do, based on the published descriptions, the perforated Glycymeris from the Upper Paleolithic of Cova Beneito (49) and the perforated Acanthocardia, Cerastoderma and Glycymeris shell ornaments from the Upper Paleolithic and Mesolithic of Cantabria and the Ebro valley (46).

57

SI Table 11. Size and weathering of perforated bivalve shells from the Solutrean and Magdalenian of Cova de Parpall and the Early Neolithic of Cova de lOr (Valencia, Spain) (all measurements are in mm and given as averagestandard deviation).
Attributes Acanthocardia Exfoliated With sponge holes Umbo-perforated Dorsal-perforated Ochred Height Length Perforation size (maximum internal diameter) Cerastoderma Exfoliated With sponge holes Umbo-perforated Dorsal-perforated Ochred Height Length Perforation size (maximum internal diameter) Glycymeris Weathering stages 2-3 With sponge holes Umbo-perforated Dorsal-perforated Ochred Height Length Perforation size (maximum internal diameter) Parpall 2 2 2 1 1 41.28.6 38.96.6 5.70.9 7 6 6 1 6 24.92.7 26.52.9 3.00.8 4 4 4 4 19.44.3 20.23.3 4.62.7 Or 4 4 4 3 1 1 34.46.8 34.15.7 3.40.9 16 16 15 16 7 44.010.5 46.610.7 4.61.5

58

1 3 1. Lower Magdalenian (Centro Oeste, 1.5-1.7 m); 2-3. Basal Upper Magdalenian (1.2-1.5 m and Talud capa 5)

50

maximum internal diameter of perforation (mm) % 50

Modern [N=89] Aviones, Or and Parpall [N=21]

40

40

30

30

20

20

10

10

10.5-12.5

0.5-2.5

2.5-4.5

4.5-6.5

6.5-8.5

8.5-10.5

10.5-12.5

0.5-2.5

mm

8.5-10.5

2.5-4.5

4.5-6.5

6.5-8.5

>12.5

>12.5 60

mm

maximum internal diameter of perforation (mm) 20

Aviones Or

15

Parpall Stage 1, modern Stage 2, modern

10

Stage 3, modern

0 15 20 25 30 35 40 45 shell height (mm) 50 55 65

SI Figure 31. Top. Donax trunculus from Parpall bearing typical predator drill-holes. Middle. Relative frequency histograms of umbo perforation sizes in the archeological and the reference collections. Bottom. Scatter plot of height versus umbo perforation size in the archeological and the reference collections.
59

Sector O (1.2-1.5 m), Lower Magdalenian; height - 35.1 mm

Sector L (5.5-5.75 m), Middle Solutrean; height - 47.2 mm SI Figure 32. Cova de Parpall. Upper Paleolithic perforated shells of Acanthocardia tuberculata. Note the exfoliation and the sponge holes associated with the perforations.
60

P-6332, unknown level; height - 27.3 mm

P-6325 (Sector P, 5.75-6 m), Middle Solutrean; height - 28.0 mm

Centro Oeste, Capa 3, Upper Magdalenian; height - 27.2 mm

SI Figure 33. Cova de Parpall. Upper Paleolithic perforated shells of Cerastoderma sp. Note the exfoliation and the sponge holes associated with the perforations.
61

Talud (3-3.5 m), Lower Magdalenian; height - 23.3 mm

P-6350 (Galeria Este, capa 6), Solutrean or Magdalenian; height - 23.7 mm P-6319 (Sector P, 3.5-3.75 m), Lower Magdalenian; height - 20.5 mm

P-6327 (Sector C, 1.7-2 m), Upper Magdalenian; height - 24.1 mm

SI Figure 34. Cova de Parpall. Upper Paleolithic perforated shells of Cerastoderma sp. Note the exfoliation and the sponge holes associated with the perforations in three of the shells. P-6350 is the only fresh, relatively unweathered specimen in the assemblage; it bears a shock perforation that mimicks direct percussion but is known to occur in natural thanatocenoses (cf. SI Fig. 30).
62

Centro Oeste Capa 3, Upper Magdalenian; height - 15.8 mm

Sector P 2-2.2 m, Upper Magdalenian; height - 20.5 mm

SI Figure 35. Cova de Parpall. Upper Paleolithic umbo-perforated valves of Glycymeris insubrica (both weathering stage 3; note the decalcification of the smaller shell).
63

P-6330 (0.2-0.4 m), Upper Magdalenian; height - 25.1 mm

Sector L (4.5-5.25 m), Upper Solutrean; height - 16.3 mm SI Figure 36. Cueva de Parpall. Upper Paleolithic umbo-perforated valves of Glycymeris insubrica (both weathering stage 3).
64

H2/c6 height: 29.4 mm

H1/c7 height: 28.4 mm

H4/c7 height: 36.8 mm

H1/c7 height: 43.0 mm

SI Figure 37. Cova de lOr (Early Neolithic). Dorsal- (topmost) and umbo-perforated valves of Acanthocardia tuberculata. Note the significant weathering, including exfoliation and sponge holes.
65

H/c7 height: 34.4 mm

H/c7 height: 41.2 mm

H4/c6 height: 28.7 mm

H4/c6 height: 52.2 mm

SI Figure 38. Cova de lOr (Early Neolithic). Umbo-perforated valves of Glycymeris insubrica (weathering stages 2 and 3).
66

H3/c6 height: 45.4 mm

H4/c5 height: 33.5 mm

H2/c6 maximum internal diameter of the perforation: 2.2 mm

H3/c6 height: 51.6 mm

SI Figure 39. Cova de lOr (Early Neolithic). Umbo-perforated valves of Glycymeris insubrica (weathering stages 2 and 3).
67

H2/c6 height: 45.7 mm

H2/c3 height: 49.8 mm

H2/c2 height: 55.1 mm

H1/c7 height: 52.0 mm

SI Figure 40. Cova de lOr (Early Neolithic). Umbo-perforated valves of Glycymeris insubrica (weathering stage 2).
68

H1/c4 height: 44.8 mm

Grieta Norte/c13 height: 40.8 mm

F2/c5 height: 23.2 mm

H1/c6 height: 61.1 mm

SI Figure 41. Cova de lOr (Early Neolithic). Umbo-perforated valves of Glycymeris insubrica (weathering stages 2 and 3).
69

SI Figure 42. Cova de lOr Glycymeris. Close-up views of the abrasion-flattened, perforated umbos of the two H/c7 shells in SI Figure 38 (top and middle) and a non-perforated shell (H1/c6) with a similar weathering pattern (bottom). The smooth edges of the flattened areas and the pitting from bioerosion and corrosionidentical to that seen elsewhere in the shellsshow that these features, including the perforations, are natural. Although the vertical striations apparent in the middle image could be anthropogenic in part, they can represent no more than modification of a pre-existing natural perforation.
70

Supporting Information IX: References

1. 2. 3. 4. 5. Gordillo A, Espinosa JS, Martn JM, Prez A (1973) Mapa Geolgico de Espaa 1:50.000, Cartagena, Hoja n 977 (Ministerio de Industria, Madrid). Espinosa JS et al. (1977) Memoria, Mapa Geolgico de Espaa 1:50.000, Cartagena, Hoja n 977 (ITGE, Madrid). Montes R (1987) El Paleolitico Medio en la costa de Murcia (University of Murcia, Murcia). Montes R (1991) La Cueva de los Aviones. Un yacimiento del Paleolitico Medio (Cartagena, Spain). Memorias de Arqueologa de la Regin de Murcia 2:35-58. Angelucci D (2002) in Portrait of the Artist as a Child. The Gravettian Human Skeleton from the Abrigo do Lagar Velho and its Archeological Context, eds Zilho J, Trinkaus E (Instituto Portugus de Arqueologia, Lisboa) pp 58-91. Vanhaeren M et al. (2006) Middle Paleolithic Shell Beads in Israel and Algeria. Science 312:1785-1787. dErrico F, Henshilwood C, Vanhaeren M, van Niekerk K (2005) Nassarius kraussianus shell beads from Blombos Cave: evidence for symbolic behaviour in the Middle Stone Age. J Hum Evol 48:3-24. dErrico F et al. (2009) Additional evidence on the use of personal ornaments in the Middle Paleolithic of North Africa. Proc Natl Acad Sci 106:16051-16056. Martnez C (1997) El yacimiento musteriense de Cueva Antn (Mula, Murcia). Memorias de Arqueologa de la Regin de Murcia 6:18-47.

6. 7.

8. 9.

10. Zilho J (2006) Chronostratigraphy of the Middle-to-Upper Paleolithic Transition in the Iberian Peninsula. Pyrenae 37:7-84. 11. Ramsey CB (2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51:337-360. 12. Hughen K, Southon J, Lehman S, Bertrand C, Turnbull J (2006) Marine-derived 14C calibration and activity record for the past 50,000 years updated from the Cariaco Basin. Quat Sci Rev 25:3216-3227. 13. Wang YJ et al. (2001) High-resolution absolute-dated Late Pleistocene monsoon record from Hulu Cave, China. Science 294:2345-2348. 14. Schweingruber FH (1990) Anatomy of European Woods (P. Haupt, Bern/Stuttgart). 15. Brock F, Higham TFG, Ditchfield P, Ramsey CB (in press) Current pre-treatment methods for AMS radiocarbon dating at the Oxford Radiocarbon Accelerator Unit (ORAU). Radiocarbon. 16. Dee M, Ramsey CB (2000) Refinement of graphite target production at ORAU. Nucl Instrum Meth B 172:449-453. 17. Ramsey CB, Higham TFG, Leach P (2004) Towards High-Precision AMS: Progress and Limitations. Radiocarbon 46:17-24. 18. Bird MI et al. (1999) Radiocarbon dating of old charcoal using a wet oxidation, steppedcombustion procedure. Radiocarbon 41:127-140. 19. Higham, TFG (2009) Problems with radiocarbon dating the Middle to Upper Palaeolithic transition in Italy. Quat Sci Rev 28:1257-1267.

71

20. Fairbanks RG (2005) Radiocarbon calibration curve spanning 0 to 50,000 years BP based on paired 230Th/234U/238U and 14C dates on pristine corals. Quat Sci Rev 24:1781-1796. 21. Bersani D, Loticci P, Casoli A, Cauzzi D (2008) Pigments and binders in Madonna col Bambino e S. Giovannino by Boticelli investigated by micro-Raman and GC/MS. J Cult Herit 9:97-102. 22. Bouchard M, Smith DC (2003) Catalogue of 45 reference Raman spectra of minerals concerning research in art history or archaeology, especially on corroded metals and coloured glass. Spectrochim Acta A 59:2247-2266. 23. Burgio L, Clark RJH (2001) Library of FT-Raman spectra of pigments, minerals, pigment media and varnishes, and supplement to existing library of Raman spectra of pigments with visible excitation. Spectrochim Acta A 57:1491-1521. 24. Winter J (1983) The characterisation of pigments based on carbon. Stud Conserv 28:48-66. 25. Montes R (1986) in Historia de Cartagena II, ed Mas J (Ediciones Mediterraneo, Cartagena), pp 34-92. 26. Martn JD (2004) XPowder: A Software Package for Powder X-Ray Diffraction Analysis (www.xpowder.com). 27. Lunar R et al. (2002) in The Geology of Spain, eds Gibbons W, Moreno MT (Geological Society, London), pp 473-510. 28. Sainz B, Gonzlez J, Vials J (2003) Secondary Minerals of the Mazarrn-guilas Mining District, Murcia Province, Spain. Mineral Rec 34:315-334. 29. Manteca JI, Ovejero G (1992) in Recursos Minerales de Espaa, eds Garca J, Martnez J (CSIC, Madrid), pp 1085-1110. 30. Lpez J, Cebri JM, Doblas M (2002) in The Geology of Spain, eds Gibbons W, Moreno MT (Geological Society, London), pp 417-438. 31. Ortiz R (1991) Quimismo de los productos de alteracin supergnica en el distrito minero de Mazarrn (Murcia) (Murcia, University of Murcia,). 32. Oyarzun R, Marquez A, Ortega L, Lunar R, Oyarzun J (1995) A late Miocene metallogenic province in southeast Spain: atypical Andean-type processes on a smaller scale. T I Min Metall B 104:197-202. 33. Lpez JA (1992) in Recursos Minerales de Espaa, eds Garca J, Martnez J (CSIC, Madrid), pp 1103-1112. 34. Eastaugh N, Walsh V, Chaplin T, Siddall R (2004) The Pigment Compendium. A Dictionary of Historical Pigments (Oxford, Elsevier Butterworth-Heinemann). 35. Prez C, Martnez MJ, Garca C (1999) in Fate and Transport of Heavy metals in the Vadose Zone, eds Selim HM, Iskandar IK (Lewis Publishers), pp 117-191. 36. Navarro MC et al. (2008) Abandoned mine sites as a source of contamination by heavy metals: A case study in a semi-arid zone. J Geochem Explor 96:183-193. 37. Lecuyer C, Reynard B, Martineau F (2004) Stable isotope fractionation between mollusc shells and marine waters from Martinique Island. Chem Geol 213: 293-305. 38. Barnard W, de Waal D (2006) Raman investigation of pigmentary molecules in the molluscan biogenic matrix. J Raman Spectrosc 37:342-352. 39. Karampelas S et al. (2007) Determination by Raman scattering of the nature of pigments in cultured freshwater pearls from the mollusk Hyriopsis cumingi. J Raman Spectrosc 38:217-230.
72

40. Hedegaard C, Bardeau J-F, Chateigner D (2006) Molluscan Shell Pigments: An In Situ Resonance Raman Study. J Mollus Stud 72:157-162. 41. Parras D, Montejo M, Ramos N, Sanchez A (2006) Analysis of pigments and coverings by X-ray diffraction (XRD) and micro Raman spectroscopy (MRS) in the cemetery of Tutugi (Galera, Granada, Spain) and the settlement convento 2 (Montemayor, Crdoba, Spain). Spectrochim Acta A 64:1133-1141. 42. Rogalla NS, Amler MRW (2007) Statistic approach on taphonomic phenomena in shells of Glycymeris glycymeris (Bivalvia: Glycymerididae) and its significance in the fossil record. Palaeontol Z 81:334-355. 43. Claassen C (1998) Shells (Cambridge, University Press). 44. Soler B (2001) in De neandertales a cromaones. El inicio del poblamiento humano en las tierras valencianas, ed Villaverde V (Valncia, Universitat de Valncia), pp 367-376. 45. Soler B (1990) Estudio de los elementos ornamentales de la Cova del Parpall. Sagvntvm 23:39-59. 46. lvarez E (2006) Los objetos de adorno-colgantes del Paleoltico Superior y del Mesoltico en la Cornisa Cantbrica y en el Valle del Ebro: una visin europea (Salamanca, Universidad de Salamanca). 47. Zilho J (1992) Gruta do Caldeiro. O Neoltico Antigo (Lisboa, Instituto Portugus do Patrimnio Arquitectnico e Arqueolgico). 48. Pascual JL (1998) Utillaje seo, Adornos e dolos Neolticos Valencianos (Valencia, Servicio de Investigacin Prehistrica). 49. dErrico F, Jardn P, Soler B (1993) in Traces et fonction: les gestes retrouvs, eds Anderson P, Beyries S, Otte M, Plisson H (Universit, Lige) pp 245-254. 50. Garca J (1987) El habitat eneoltico de Las Amoladeras (La Manga). Campaa 19811984. Memorias de Arqueologa de la Regin de Murcia 1:65-92.

73