Journal of Human Evolution 68 (2014) 1e13

Contents lists available at ScienceDirect

Journal of Human Evolution

journal homepage: www.elsevier.com/locate/jhevol

On the chronology of the Uluzzian

Katerina Douka a,*, Thomas F.G. Higham a, Rachel Wood a,b, Paolo Boscato c, Paolo Gambassini c, Panagiotis Karkanas d, Marco Peresani e, Anna Maria Ronchitelli c a Oxford Radiocarbon Accelerator Unit, Research Laboratory for Archaeology and the History of Art, University of Oxford, Dyson Perrins Building, South Parks Road, OX1 3QY Oxford, United Kingdom b Research School of Earth Sciences, Australian National University, Australia c Dipartimento di Scienze Fisiche, della Terra e dell’Ambiente, U.R. Preistoria e Antropologia, Università degli Studi di Siena, Via Laterina 8, 53100 Siena, d Ephoreia of Palaeoanthropology-Speleology of Southern Greece, Ardittou 34b, 11636 Athens, Greece e Dipartimento di Studi Umanistici, Sezione di Scienze Preistoriche e Protostoriche, Università di Ferrara, Corso Ercole I d’Este 32, I-44100 Ferrara, Italy article info abstract

Article history: The Uluzzian, one of Europe’s ‘transitional’ technocomplexes, has gained particular significance over the Received 10 April 2013 past three years when the only human remains associated with it were attributed to modern humans, Accepted 10 December 2013 instead of as previously thought. The position of the Uluzzian at stratified sequences, al- Available online 7 February 2014 ways overlying late Mousterian layers and underlying early Upper Palaeolithic ones, highlights its sig- nificance in understanding the passage from the Middle to Upper Palaeolithic, as well as the replacement Keywords: of Neanderthals by modern humans in southeastern Mediterranean Europe. Despite several studies Modern humans investigating aspects of its lithic techno-typology, taxonomy and material culture, the Uluzzian chro- Neanderthals Radiocarbon dating nology has remained extremely poorly-known, based on a handful of dubious chronometric de- Italy terminations. Here we aim to elucidate the chronological aspect of the technocomplex by presenting an Greece integrated synthesis of new radiocarbon results and a Bayesian statistical approach from four stratified Uluzzian cave sequences in Italy and Greece (Cavallo, Fumane, Castelcivita and Klissoura 1). In addition to building a reliable chronological framework for the Uluzzian, we examine its appearance, tempo-spatial spread and correlation to previous and later Palaeolithic assemblages (Mousterian, Protoaurignacian) at the relevant regions. We conclude that the Uluzzian arrived in Italy and Greece shortly before 45,000 years ago and its final stages are placed at w39,500 years ago, its end synchronous (if not slightly earlier) with the Campanian Ignimbrite eruption. Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

Introduction beads, mineral pigments, as well as two human deciduous teeth. The technocomplex was assigned to the Upper Palaeolithic (‘Lep- In 1964, Arturo Palma di Cesnola coined the term ‘Uluzzian’ to tolithic’) tradition, yet archaic at its lower layers, due to the pres- describe the technocomplex he identified a year earlier at Cavallo ence of particular tool-types (lunates) that do not exist in the Cave (Grotta del Cavallo) in southern Italy (Palma di Cesnola and preceding Mousterian levels. Palma di Cesnola immediately Borzatti von Löwenstern, 1964). The distinct, dark Uluzzian layers recognized parallels with the Franco-Cantabrian Châtelperronian, overlay a long series of Mousterian deposits and were superposed and efforts to elucidate the geographical, techno-typological and by a late Upper Palaeolithic (Late Epigravettian) phase from which chronological span of the Uluzzian, as well as its role in the Middle- they were separated by a thin stalagmitic crust and a tephra layer. to-Upper Palaeolithic transition, began. The Uluzzian layers were rich in lithic elements of Upper In 2011, almost 50 years after the initial discovery of the tech- Palaeolithic character, terrestrial and marine faunal remains and nocomplex, the Uluzzian was brought to the forefront again. The several hearths. They contained bone points, perforated shell two deciduous teeth from Cavallo Cave, the only human remains associated with the Uluzzian so far, were identified as belonging to anatomically modern humans and not Neanderthals, as was the * Corresponding author. consensus until then (Benazzi et al., 2011). The parallel radiocarbon E-mail addresses: [email protected] (K. Douka), rachel.wood@anu. dating of associated shell beads found across the Cavallo Uluzzian edu.au (R. Wood), [email protected] (P. Boscato), [email protected] stratigraphy rendered the teeth the oldest currently-known re- (P. Karkanas), [email protected] (M. Peresani), [email protected] (A. mains of modern humans in Europe (Benazzi et al., 2011). M. Ronchitelli).

0047-2484/$ e see front matter Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jhevol.2013.12.007 2 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13

The geographic span of the Uluzzian has remained almost un- there is low incidence of lamellar débitage. Local raw material changed over the last decades and several authors have described procurement, including thin slabs of siliceous limestone for the the techno-typological features of the assemblages assigned to it. production of retouched pieces, is a further characteristic of the The chronology of the Uluzzian, however, remained poorly under- earliest Uluzzian phases (Gambassini, 1997; Riel-Salvatore and stood. Given the importance of the technocomplex in our under- Negrino, 2009). standing of the appearance and spread of modern humans in Based on the most complete Uluzzian complex known to date, Europe, a chronological synthesis is long due and is the main focus the stratified sequence of Cavallo Cave, and the typological vari- of the present article. ability observed within it, Palma di Cesnola (1965a, b, 1989) divided the Uluzzian into three distinct industrial phases: Archaic (Cavallo Background layer E III), Evolved (Cavallo layers E II-I) and Final (Cavallo layers D II-D Ib). Current techno-typological and functional analyses are Geography underway to clarify issues of origin, internal evolution, external influences and adaptations of the Uluzzian, as well as its relation- The presence of the Uluzzian has been reported in about 20 ship to the succeeding Aurignacian (De Stefani et al., 2012; Ranaldo, caves and open-air localities of peninsular Italy (Palma di Cesnola, In press), broadly contemporaneous Châtelperronian and preceding 1989, 1993; Riel-Salvatore, 2007, 2009; Ronchitelli et al., 2009). Mousterian. What has become clear is that the one-to-one analogy These include sites mainly in (Cavallo, Uluzzo, Parabita, with the Châtelperronian cannot be supported anymore as the Mario Bernardini, Serra Cicora, Torre Testa, Falce del Viaggio, For- technological details for tool production, as well as the final prod- esta Umbra), Basilicata (Atella Basin), (Cala, Castelcivita, ucts and their typology, are indeed very different. Tornola), (San Pietro a Maida, Punta Safò), Tuscany (La Fabbrica, Poggio Calvello, Val Berretta, San Romano, Indicatore, San Bone industry, shell ornaments and pigments Leonardo, Salviano and Maroccone) (Palma di Cesnola, 1989, 1993 and references therein). More recently, the presence of the Uluz- Along with the distinct lithics and microlithics, several Uluzzian zian at Fumane, Verona, was reported and this is currently the sequences (Cavallo, Castelcivita, La Fabbrica) have yielded worked northernmost occurrence of the technocomplex (Peresani et al., bone industry. About 15 bone implements in cylindro-conical 2008; Peresani, 2012). forms, mainly awls, have been discovered and described so far A single occurrence outside Italy, at the Greek cave of Klissoura (Ronchitelli et al., 2009; d’ Errico et al., 2012). When identifiable, 1, Argolis, was described by Koumouzelis et al. (2001). This was these were manufactured on horse and red deer metapodials and later questioned by Papagianni (2009), but has since been re- on horse fibulae. While they incorrectly associated them with Ne- confirmed by the excavators (Kaczanowska et al., 2010). Another anderthals, d’ Errico et al. (2012) suggested that the Uluzzian bone Greek Uluzzian context from Kephalari Cave, also in Argolis, was tools reveal a techno-typological complexity from their earliest reported on anecdotal evidence (Hahn, 1984). This is currently appearance. In addition to these tools, a recent publication reports being re-examined and more data will become available soon (G. on the presence of bone retouchers in the Uluzzian layers of Marshall, Personal communication). Fumane Cave (Jéquier et al., 2012). The use and manufacture of beads is another key behavioural Lithic technology-typology characteristic of the Uluzzian. Perforated marine shells have been discovered at the Uluzzian layers of Cavallo, Klissoura (and prob- The Uluzzian is a flake-dominated industry (Palma di Cesnola, ably Cala, which is not part of the current study). They make their 1993) that brings together a set of technological innovations (e.g., first appearance at the Archaic Uluzzian phase of Cavallo and in- De Stefani et al., 2012). The production scheme is characterized by crease in number, up to a few dozens, at its final phases (Palma di variable methods and techniques and a significant reduction in Cesnola, 1964, 1989). Several shell beads appear in Klissoura predetermined forms, when compared with the final Mousterian. (n ¼ 32; Stiner, 2010) but no ornaments have been discovered in the The increased importance in the launch of various reduction se- Uluzzian of Fumane and Castelcivita. quences cannot be attributed to the qualities of the raw material Beads are almost exclusively manufactured from the shells of only, but must assume behavioural significance too. marine molluscs, mainly scaphopods (Dentalium sp.) and gastro- Blanks are detached using a unipolar method, with a single pods (e.g., Columbella rustica, Cyclope neritea). A small number of striking platform by direct percussion, and from a bipolar knapping bivalves have also been identified in the assemblages and, with the on an anvil, which produces splintered pieces used mainly as wedges exception of a series of perforated Nuculana cf. sp. from Cavallo, the on medium-hard material. The principal objective was the produc- rest are not humanly modified (e.g., Pecten sp. from Castelcivita and tion of average-thickness flakes and laminar flakes, sometimes other unidentified fragments). Just as in the Aurignacian, the naturally backed. Small blades and a small number of bladelets were Uluzzian shells are transformed into beads through perforation also obtained. Few orthogonal, multi-directional and surface cen- with the aid of lithic tools using direct or indirect percussion. In the tripetal cores are also present (Palma di Cesnola, 1965a, 1966a, 1989; case of the tubular Dentalium shell, the apical end is snapped off Peresani, 2008; Moroni et al., 2013; De Stefani et al., 2012). and the shell is sawn to create smaller sections. Several tool-types are identified as part of the technocomplex: Finally, evidence for the use of colourants (ochre and limonite various end-scraper types, side-scrapers, few burins, denticulates, pieces) has been revealed in Cavallo and Mario Bernardini caves retouched blades and bladelets. The geometric crescent-shaped (Borzatti von Löwenstern, 1970; Palma di Cesnola, 1989). At the backed piece, normally referred to as ‘lunate’ (or ‘semi-luna’ in moment it is not possible to tell whether these minerals were used Italian) is the fossile directeur of the industry; an innovation solely for symbolic practices or other activities (e.g., hide prepara- sometimes of microlithic dimensions (Palma di Cesnola, 1993). This tion, as grinding medium or hafting agent). tool is particularly abundant in Cavallo Cave. Lunates do not seem to originate from specific operational chains. Taxonomy The high incidence of splintered pieces (‘pièces esquillées’)is another characteristic feature of the Uluzzian. Small blades and In terms of biological attribution, the makers of the Uluzzian bladelets are irregular and sometimes with traces of cortex, while have remained elusive, until very recently. The only known fossils K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 3

Table 1 Previously available chronometric determinations for the Uluzzian layers at the four studied sites: Fumane, Castelcivita, Cavallo and Klissoura.

Site Lab code Age Context Material Reference

Fumane Cave LTL-1830A 29,602 240 A3 e str. I Charcoal Higham et al., 2009 Fumane Cave LTL-1831A 29,233 350 A3 e str. II Charcoal Higham et al., 2009 Fumane Cave LTL-1796A 29,361 320 A3 e str. IV Charcoal Higham et al., 2009 Fumane Cave LTL-1795A 37,828 430 A3 e str. IV Charcoal Higham et al., 2009 Fumane Cave OxA-8021 33,300 400 A4 II Charcoal Peresani et al., 2008 Fumane Cave OxA-6462 33,150 600 A4 II Charcoal Peresani et al., 2008 Fumane Cave LTL-566A 33,700 350 A4 II Charcoal Peresani et al., 2008 Fumane Cave ESR 44 7 ka A4 Charcoal Peresani et al., 2008

Castelcivita Cave F-71 32,470 650 rsa Burnt bones Gambassini, 1997 Castelcivita Cave F-106 >34,000 ? Burnt bones Gambassini, 1997 Castelcivita Cave F-107 33,220 780 pie Burnt bones Gambassini, 1997 Castelcivita Cave GrN-13985 33,300 430 rpi Charcoal Gambassini, 1997

Cavallo Cave RM-352 >31,000 E II-I Charcoal Palma di Cesnola, 1969 Cavallo Cave AA-66812 34,900 1900 E III Burnt bone Riel-Salvatore, 2007; Ronchitelli et al., 2009 Cavallo Cave AA-66819 36,600 2300 EIII Burnt bone Riel-Salvatore, 2007; Ronchitelli et al., 2009 Cavallo Cave AA-66814 36,510 2300 EIII Burnt bone Riel-Salvatore 2007; Kuhn et al., 2010 Cavallo Cave AA-66813 32,300 2700 EIII Burnt bone Riel-Salvatore, 2007. Kuhn et al., 2010 Cavallo Cave 6 AMS 14C dates w21e31 ka BP EIII Burnt bone Riel-Salvatore, 2007

Klissoura Cave Gd-7878 17,430 100 V Carbonates Koumouzelis et al., 2001 Klissoura Cave Gd-12037 26,250 310 V/hearth 42 Carbonates Koumouzelis et al., 2001 Klissoura Cave Gd-12027 27,100 600 V/hearth 53 Carbonates Koumouzelis et al., 2001 Klissoura Cave Gd-10714 >31,100 V/hearth 42 Soil organics Koumouzelis et al., 2001 Klissoura Cave Gd-10715 >30,800 V/hearth 53 Soil organics Koumouzelis et al., 2001 Klissoura Cave RTT-4790 29,660 360 V upper Charcoal Kuhn et al., 2010 Klissoura Cave RTT-4791 30,774 410 V upper Charcoal Kuhn et al., 2010 Klissoura Cave Gif-99168 40,100 740 V/Sq. B3/H42/175-180 Bone Kuhn et al., 2010

With the exception of an ESR determination from Fumane, all other results are 14C. associated with the technocomplex are two deciduous molars from Cave. Palma di Cesnola obtained this determination from charcoal Cavallo Cave. Cavallo-B was found at a hearth at the base of the he found at the Evolved Uluzzian layer (E II-I) (Palma di Cesnola, Archaic Uluzzian phase (E III), and Cavallo-C, 20 cm above, in the 1969; Alessio et al., 1970). A series of ten radiocarbon dates from Evolved Uluzzian phase (E II-I). They were initially described and the Archaic Uluzzian layer E III were obtained by J. Riel-Salvatore published by Palma di Cesnola and Messeri (1967), who classified (2007). The four oldest determinations, and thus favoured, ranged the earliest, Cavallo-B, as modern human and Cavallo-C as Nean- between 32 and 36 ka BP (thousands of years ago before present) derthal. Churchill and Smith (2000) were unable to link the molars and were published by Ronchitelli et al. (2009) and Kuhn et al. securely to either Neanderthals or modern humans but, unlike (2010) (Table 1). However, the remaining six dates from the Palma di Cesnola and Messeri (1967), they favoured a group, all from layer E III, fall between w21 and 31 ka BP (Riel- attribution for Cavallo-B. This is most likely due to a mistake Salvatore, 2007) and are completely inconsistent with the strati- (reversed values) in the published dimensions these authors used graphic position of the samples (see discussion below). in their analyses (see Benazzi et al., 2011). Four previous determinations from the Uluzzian layers at Cas- The most recent and definite work over the taxonomy of the two telcivita (‘pie’, ‘rpi’, bottom of ‘rsa’), again on burnt bone, clustered deciduous teeth is that of Benazzi et al. (2011). Two independent at 33.3e32.5 ka BP (Gambassini, 1997)(Table 1). These de- morphometric methods were used to assess the taxonomy of terminations are again inconsistent with the position of the sam- Cavallo-B and Cavallo-C, namely geometric morphometric analysis of ples well below the Campanian Ignimbrite that seals the dental crown outlines and 2-D enamel thickness based on micro-CT archaeological series. scanning. Both lines of investigation placed the two specimens in the At Fumane, the presence of Uluzzian was only recently formal- modern human spectrum with certainty and therefore transformed ized in layers A3 and A4 of the site (Peresani, 2008, 2012). These the traditional interpretation for the makers of the Uluzzian, which layers were associated with several radiocarbon determinations are now considered to be modern humans and not Neanderthals. ranging from 29 to 37 ka BP (Table 1) and one on ESR of a tooth from Finally, a deciduous central incisor found in the faunal assem- A4 at 44 7ka(Peresani et al., 2008; Higham et al., 2009). blage of the lower Uluzzian layer EIII, and tentatively reported by In Greece, Uluzzian layer V at Klissoura Cave 1 was associated Riel-Salvatore (2007) and Gambassini et al. (2006) as belonging to a with five determinations on carbonates and organic fractions Neanderthal, is not human (Benazzi, personal communication; ranging between 29 and 31 ka BP (Koumouzelis et al., 2001) and contra Zilhão, 2013). Recent criticisms on the basis of this one bone determination at w40 ka BP (Gif-99168) (Table 1). The misidentification that “the issue of who made [the Uluzzian] cannot latter has no secure context as it was collected from an OSL be answered simply because both groups were implicated” (e.g. dosimeter tube. Given the small thickness of layer V, it may well Zilhão, 2013) is therefore unsupported. have derived from underlying Middle Palaeolithic layers (Kuhn et al., 2010). Previous chronological framework and problems A good number of these previously available determinations are statistically identical, falling around 32.0e33.5 ka BP or, when cali- Prior to our work, only a small number of Uluzzian contexts brated, between 34 and 38 ka cal BP (Fig. 3). Due to this seemingly were dated (Table 1, Fig. 3). internal consistency, prehistorians have accepted them as reliable. An infinite age (RM-352: >31000 years BP [before present]) was Many researchers fail to acknowledge the limitations posed by until 2009 the only published chronometric information for Cavallo sample selection and pretreatment chemistry in the successful 4 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13

Figure 1. Map of southeastern Mediterranean Europe showing the location of the four studied Uluzzian sites. The location of some of the Aurignacian and Mousterian sites mentioned in the text is also shown with open squares: 1e2, Mochi-Bombrini; 3, Mezzena; 4, Paglicci; 5, Serino; 6, Theopetra; 7, Asprochaliko; 8, Franchthi; 9, Lakonis I; 10, Kalamakia. The dashed line indicates the exposed continental shelf at sea level regression of about 100 m; this corresponds approximately to the coastline at w40,000 years ago. application of radiocarbon dating, especially towards the age limit In addition to problems relating to dating burnt bone, many of the technique. Some of the aforementioned Uluzzian de- previous radiocarbon determinations for the Uluzzian were pro- terminations were produced on material known to be very difficult duced on charcoal/organic matter cleaned using the Acid-Base-Acid to date. For example, most measurements from Cavallo and Cas- (ABA) method. This chemical protocol has been shown to under- telcivita were performed on burnt bones, a highly unreliable type of estimate systematically the age of samples older than 30 ka BP sample (van Strydonck et al., 2005, 2009), especially for when compared with more rigorous new methods, such as the Palaeolithic-aged material. This is because in burnt bone the ABOx-SC protocol (Bird et al., 1999; Santos et al., 2003; Higham measured carbon (C) comes from the mineral matrix, a highly et al., 2009; Douka et al., 2010b; Higham, 2011; Wood et al., porous and non-homogenous phase prone to external contamina- 2012). In turn, this means that the ABA determinations must un- tion. In contrast, when fresh (i.e., non-burnt) bone is dated, the derestimate the true age of the Uluzzian. measured C derives from the more resilient organic (collagen) These methodological, 14C-related, limitations are corroborated fraction. by another fact. In at least three stratified Uluzzian sequences

Figure 2. Composite stratigraphic columns of Fumane, Castelcivita, Cavallo and Klissoura 1 (modified from Palma di Cesnola, 1964; Gambassini, 1997; Martini et al., 2001; Peresani et al., 2008; Giaccio et al., 2008; Karkanas, 2010). The location of the dated samples is indicated. Columns not to relative scale. K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 5

Studied sequences

Given the distinct techno-typological features and important stratigraphic position of the Uluzzian and, as of recent, its connection with the earliest modern humans in Europe, a reliable chronology is of uttermost importance in understanding aspects of the initial appearance, expansion and eventual demise of the technocomplex, which have remained unknown thus far. The reli- able dating of the Uluzzian has been one of the major goals of a project initiated by the authors and performed at the Oxford Radiocarbon Accelerator Unit between 2008 and 2011. We decided to date four of the best-known Uluzzian sites (Cavallo Cave, Cas- telcivita Cave, Fumane Cave and Klissoura Cave 1; Fig. 1). Despite the presence of several open-air localities across Italy identified as Uluzzian on the basis of surface collections, here, priority was given to stratified sequences. Cavallo Cave is situated on the rocky coast of the bay of Uluzzo, in Puglia, SE Italy. It was discovered in 1960 and excavations took place during 1963e1966 (Palma di Cesnola, 1963, 1964, 1965b, 1966b) and from 1986 to 2007 (Sarti et al., 1998e2000, 2002). Salvage excavations were conducted in the late 1970s and early 1980s in the process of installing a gate at the entrance of the cave to protect it from looters, who had ravaged most of the standing archaeological deposits. The site contained a long stratigraphic succession comprising about 7 m of archaeological deposits, directly based on a marine interglacial beach conglomerate (layer O) (Romagnoli, 2012). The archaeological sequence of Cavallo (Fig. 2) is dominated by Mous- terian layers (N-F I), followed by a thin layer of green volcanic ash (Fa). The ash is overlain by the Uluzzian layers (E III to D Ib). These are capped by a stalagmitic floor (D Ia) and a further layer of vol- canic ash (C II) altered at the top (C I). This tephra has been iden- tified as deriving from the Campanian Ignimbrite eruption (R. Sulpizio, Personal communication). Above this are layers B IIeBI, with Late Epigravettian (Romanellian) remains, and the sequence is sealed by Neolithic layer A. Castelcivita is situated in the Calore river valley, at the foot of the Mountains (), w100 m above sea level. Excavations were conducted from 1975 to 1988 when a 3 m-long sequence was Figure 3. Previous available determinations for the Uluzzian in Italy and Greece. The conventional measurements are calibrated using the IntCal/Marine 09 curve (Reimer discovered and investigated in detail (Gambassini, 1997). The et al., 2009) on the OxCal platform (Bronk Ramsey, 2009). Infinite determinations, as archaeological succession contained several Mousterian, Uluzzian well as the group of burnt bone dates from Cavallo, were added separately and are andProtoaurignacian levels andwas sealed by thick volcanic deposits w presented here as rectangular bars. Gd-7878 is out of scale, calibrated at 20 ka cal BP with the distinctive two-phase structure of the CI (Giaccio et al., and is not shown. On the top of the graph, the NGRIP d18O record is shown (Andersen et al., 2006; Svensson et al., 2006) and the Greenland Interstadials, corresponding to 2008). Within the stratigraphic sequence, eight human-derived ameliorated climatic conditions, are numbered. The age for the CI ash is indicated on phases were recognized. The three lower phases are attributed to the top of the sequence as a continuous vertical grey line at the sites it occurs, and as a the Mousterian and correspond to levels XIIIeVII (¼ cgr-gar-rsi00). dashed line in Fumane where its presence is not confirmed. They are characterized by blocks of limestone of different size and by cemented layers. An erosional discontinuity separates the lowermost (Cavallo, Castelcivita, Klissoura), a tephra layer identified as the Mousterian from the two middle parts of the sequence (levels VIe Campanian Ignimbrite (CI) on stratigraphic and geochemical IIIb ¼ rsi0-pie-rpi-rsa00), which contain Uluzzian remains. The Uluz- grounds overlies the Uluzzian or, in the case of Castelcivita, the zian levels are characterized by fine red sediment mixed with lime- Protoaurignacian deposits, occurring directly above the Uluzzian stone blocks derived from the collapse of the entrance of the cave. The (Gambassini, 1997; Giaccio et al., 2006, 2008; Fedele et al., 2008; succeeding three phases (levels IIIa-ars ¼ rsa0-gic-ars), contained Lowe et al., 2012)(Fig. 2). The age of the Uluzzian at these sites Protoaurignacian remains within sandy sediment (see also SOM). should be older than w39.3 0.1 ka, the calendar age for the CI Fumane is a dolomitic limestone cave at the southern slope of eruption (De Vivo et al., 2001). Clearly, the dates from these sites are the Venetian Pre-Alps, between the low alluvial plains and the high much younger, often by several millennia when calibrated (Fig. 3). plateau of Mount Lessini. Systematic excavations started in 1988 In fact, of the 18 finite determinations previously available for the and continue to this day (Bartolomei et al., 1992; Cremaschi et al., Uluzzian (Table 1), 12 postdate the CI chronological marker at 39e 2005; Peresani et al., 2008). The excavations have revealed a 40 ka (Fig. 3). Only four of the burnt bone determinations from deep, 12 m thick sedimentary sequence. On the basis of distinct Cavallo (Ronchitelli et al., 2009; Kuhn et al., 2010) are statistically lithological composition, pedological features and the density and similar to the age of the CI and predate it. However, a more careful nature of cultural evidence, the stratigraphy has been divided into examination reveals that these dates come from the lowermost four main macro-units labelled S, BR, A and D (Martini et al., 2001) Uluzzian layer E III, about 80 cm below the tephra layer, for which (Fig. 2). Layers A4eA3 from macro-unit A were recently charac- an older age should be expected. terized as Uluzzian (Peresani, 2008, 2012; Peresani et al., 2008; see 6 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13

Table 2 New radiocarbon determinations for the Uluzzian layers of Cavallo, Fumane, Castelcivita and Klissoura.

Site/sample ID OxA-code 14C age Context Material Genus/species Cal yr BP (95.4%) Analytical parameters

From To

Collagen quality Fumane Cave/RF44 21736 39,100 1000 A3/69a/53 Bone Large mammal 44990 42030 C/N: 3.4, % Coll.: 2.9, d13C: 19.6, d13N: 6.3 Fumane Cave/RF49 X-2295-52 41,300 1300 A3/78-79/R818 Bone Large mammal 48140 43030 C/N: 3.3, % Coll.: 6.2, d13C: 18.8, d13N: 4.7 Fumane Cave/RF28 21733 41,000 1300 A4II/86g/R367 Bone Large mammal 47740 42800 C/N: 3.3, % Coll.: 2.9, d13C: 18.8, d13N: 7.6 Fumane Cave/RF29 21734 42,000 1400 A4II/100dþg/R1120 Bone Large mammal 49100 43610 C/N: 3.4, % Coll.: 3.3, d13C: 20.5, d13N: 3.8 Fumane Cave/RF30 21735 42,000 1700 A4/struct. II/744 Bone Mammal 49490 43430 C/N: 3.2, % Coll.: 0.8 (low yield), d13C: 19.7, d13N: 5.8

Charcoal quality Castelcivita Cave/No. 7 22622 36,120 360 rsa”, spit 11 Charcoal c.f. Ilex aquifolium 41910 40560 % C: 76.7, d13C: 24.0

% Aragonite-calcite / % C yield Cavallo Cave/Cvl 10 21072 19,685 75 D I Marine shell Cyclope neritea 23,380 22,470 100-0 / 9.3 Cavallo Cave/Cvl 10 Dupl. 19,235 75 D I Marine shell Cyclope neritea 23,220 22,060 100-0 / 9.9 Cavallo Cave/Cvl 2 19254 35,080 230 D 1 ¼ DIb Marine shell Nuculana sp. 40,450 38,860 100-0 / 9.8 Cavallo Cave/Cvl 4 19255 36,260 250 D 2 ¼ DIb Marine shell Dentalium sp. 41,570 40,390 100-0 / 9.5 Cavallo Cave/Cvl 11 20631 36,780 310 D II Marine shell Dentalium sp. 42,010 40,880 99.8-0.2 / 9.9 Cavallo Cave/Cvl 6 19257 42,360 400 D 3 ¼ D II Marine shell Bivalve fragment 45,990 44,640 100-0 / 10.2 Cavallo Cave/Cvl 8 19258 36,000 400 D 8 ¼ DII? Marine shell Dentalium sp. 41,570 39,560 100-0 / 9.7 Cavallo Cave/Cvl 5 19256 39,060 310 E 1 ¼ E-D Marine shell Dentalium sp. 43,510 42,350 99.7-0.3 / 10.1 Cavallo Cave/Cvl 5 X-2280-16 38,300 400 E 1 ¼ E-D Marine shell Dentalium sp. 43,380 42,080 100-0 / 6 (low yield) Cavallo Cave/Cvl 3 19242 39,990 340 E 4 ¼ EII-I Marine shell Dentalium sp. 44,300 43,000 50-50 / 8.5

Klissoura Cave 1/Klis 19 19936 27,950 160 V/AA4b/185-190 Marine shell Cyclope sp. 32,100 31,260 100-0 / 6.8 (low yield) Klissoura Cave 1/Klis 21 21068 34,580 220 V/AA4b/180-185 Marine shell Dentalium sp. 39,930 38,540 15-85 / 9

Conventional ages are expressed in 14C years BP and are calibrated in OxCal (Bronk Ramsey, 2009) using the IntCal09/Marine09 curve (Reimer et al., 2009). The right hand column lists analytical parameters for collagen and charcoal quality as well as percentages of aragonite to calcite (based on high-precision XRD analysis) for shell samples. The dates for sample Cvl-10 are too young to be considered Uluzzian and are therefore excluded from the Bayesian models. also SOM). Layers A2-A1 are classified as Protoaurignacian while three types of dated material. Bones were dated only in the case of layers D3d, D3b and D3a are later Aurignacian units (Broglio et al., Fumane. In all other sites tested bones preserved no collagen 2003). D1d has yielded a small number of Gravettian artifacts and is therefore alternative material was used. All determinations were attributed to the Gravettian period (Broglio et al., 1996e1997) (see produced using the latest preparative methods (ultrafiltration for also SOM). bones, ABOx-SC for charcoal, CarDS for shell), which have been Klissoura Cave 1 is a small limestone cave located at the north- described in detail by Brock et al. (2010), Douka et al. (2010a) and eastern edge of the Argive plain in the Peloponnese, southern Wood et al. (2012). For the vast majority of samples, carbon yields Greece. It was investigated between 1994 and 2006 by Greek and and sample size were within the expected standard ranges. In the Polish teams (Koumouzelis et al., 2001; Karkanas, 2010) that case where an unusual value was observed (variable d13C, low %C or revealed an 8 m long sedimentary succession. The cultural sequence high C/N), these determinations got an OxA-X rather than the was divided into two broad phases: the Middle Palaeolithic phase, routine OxA- prefix(Table 2, see with notes on last two columns). about 6.5 m thick, and the Upper Palaeolithic-Mesolithic phase, We obtained ten radiocarbon dates for Cavallo Cave, the refer- about 2.0 m thick, separated from each other by clear erosional ence Uluzzian sequence (Table 2). With one exception (Cvl 6), all contacts. A series of Middle Palaeolithic layers (XXa-geVII) is fol- determinations were produced on marine shell beads (Dentalium lowed by layer VI, a mixof Upper Palaeolithic and Middle Palaeolithic sp., Nuculana sp. and C. neritea), most likely used as personal or- elements since this layer is directly overlain by the Upper Palaeolithic naments. They come from the two upper phases, Evolved Uluzzian sequence in several areas. Layer V is a thin, discontinuous lensoid (E II-I) and Final Uluzzian (D II-Ib) layers. These determinations horizon, its contents ascribed to the Uluzzian and roughly compared were initially reported by Benazzi et al. (2011). with the ‘evolved’ phase of the technocomplex. It is a dark grey, Most dates (OxA-19242, OxA-19256, OxA-19258, OxA-20631, clayey silty layer comprising mainly reworked and locally in situ OxA-19255, OxA-19254) are consistent with the stratigraphic po- burnt remains. A series of Aurignacian layers (early: IV, middle: IIId- sition of the samples and range from 35 to 40 ka BP, from top to g, late: IIIa-c) overlaid layer V, and were followed by Gravettian à bottom. Only OxA-19257 (D II), a small, indeterminate fragment of a pointes à dos indifférencié or Mediterranean backed bladelets (III0), bivalve shell appears too old for its context (42 ka BP). It is not Epigravettian (IIa-d) and Mesolithic layers (3e5a) (Fig. 2) unusual for bivalve shells to be older than their context (see for (Koumouzelis et al., 2001; Kaczanowska et al., 2010; Karkanas, 2010). example discussion in Sivan et al., 2005; Douka, 2011) and we suggest that this is indeed an old valve, either collected on purpose Results or accidentally brought to the site long after the death of the mollusc. On the other hand, OxA-21072 (and its duplicate) obtained New chronological framework from a C. neritea shell from D I, was much younger than expected considering its position in the sequence. According to the excavator, In the current project, 18 radiocarbon measurements were ob- D Ib was the latest Uluzzian layer at the site. This layer was sealed tained for the Uluzzian layers of Cavallo, Castelcivita, Fumane and by D Ia, a stalagmite crust covered by volcanic ash (C), both forming Klissoura 1 (Table 2). Shell, charcoal and bone collagen were the a rather continuous layer across most of the excavated area. What K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 7 makes this determination more curious is that no other archaeo- logical material ever reported from the cave could be assigned to this time period (w20 ka BP). We have no further details to make an informative suggestion or better explain this young determination, but we consider the shell as unrelated to the Uluzzian of Cavallo. The latter clearly ends around the time of the CI eruption when airborne tephra was deposited and covered the top of layer D. We obtained no dates for the lowermost Archaic Uluzzian phase (E III) due to the scarcity of suitable shell for dating, the absence of bone collagen in all bones we tested, and the lack of charcoal in the stored collection. The antiquity of Archaic Uluzzian phase is, therefore, unknown but based on the corpus of the current de- terminations, independent stratigraphic evidence and estimations of sediment deposition rates, E III must predate E II by at least 1e3 millennia. The age of the first Uluzzian occupation at Cavallo can be determined securely through direct dating of E III or, failing that, dating of the latest Mousterian layer, which will provide a useful terminus post quem for the start of the Uluzzian at the site. Our group is currently working towards these possibilities. At Fumane, we dated bone remains from layers A3 and A4, both assigned to the Uluzzian (possibly Archaic and Evolved, respec- tively). The position of the samples and their relation to structures and other cultural remains are discussed in the SOM.Wehave obtained five new determinations on modified (smashed and cut- marked) bone, all treated with the collagen ultrafiltration method (Higham, 2011). The ages range from 39 to 42 ka BP (Table 2). It appears that the phase in Fumane was relatively short and it is chronologically indistinguishable from the late Mousterian at the site for which several radiocarbon determinations also exist (Higham et al., 2009; Higham, 2011). The single determination we obtained from Castelcivita comes 00 from the uppermost Uluzzian deposits at the site (rsa , top of spit 11, Figure 4. Bayesian modelling based on the new radiocarbon determinations from four square H14 I), which is normally associated with the final Uluzzian Uluzzian contexts in Italy and Greece. There appears to be a clear chrono-stratigraphic phase at Cavallo. This layer is found 50e55 cm below the well- progression of the technocomplex, and close correspondence between the Italian sites and the Greek case. The raw determinations are calibrated using IntCal/Marine 09 characterized CI deposits (Giaccio et al., 2008) and therefore its age curve (Reimer et al., 2009) and the OxCal platform (Bronk Ramsey, 2009). On the top of should well predate the CI eruption. The layer contained evidence of the graph, the NGRIP d18O record is shown (Andersen et al., 2006; Svensson et al., hearths and the particular charcoal sample (cf. Ilex aquifolium)was 2006) and the Greenland Interstadials, corresponding to ameliorated climatic condi- dated twice, as part of experimentation with two radiocarbon pre- tions, are numbered. The age for the CI ash is indicated on the top of the sequence as a grey line at the sites it occurs, and as dashed line in Fumane where its presence in not treatment methods (ABA and ABOx-SC; Wood et al., 2012). The less- confirmed. refined ABA method gave an age much younger than expected at 33 w e ka BP (or 37.0 38.8 ka cal BP), which postdates the calendar age for peak acts as a terminus ante quem for the deposition of the Uluzzian the eruption. Pretreatment of the same charcoal sample with ABOx- layer V, for which we obtained two dates on shell beads (Table 2). w e SC gave a reliable age of 36.1 ka BP ( 40.5 42.0 ka cal BP) (Table 2). Both beads were reported as coming from square AA4, outside the This new measurement is around three millennia older than our ABA dripline of the cave. OxA-21068 at 34.6 0.2 ka BP (39.9e result, as well as all other determinations previously obtained for 38.5 ka cal BP) agrees with the position of the sample beneath the CI. fi Uluzzian material from Castelcivita (Table 1). This is the rst time an The fact that this date was produced on a Dentalium sp. shell is all the Uluzzian determination at the site predates the estimate age of the CI more important given the presence of Dentalium beads also in Italian layer that seals the sequence, and agrees well with the stratigraphic Uluzzian contexts, such as in Cavallo Cave. The second determina- position of the sample. tion, OxA-19936, obtained from a shell found 5 cm below the pre- Just as with Cavallo, the start of the Uluzzian in Castelcivita vious one, is much younger (w28 ka BP). When we questioned the cannot be established because there are no reliable chronometric certainty of this level assignation and looked into the excavation 00 results below rsa . Three previous determinations from the plans it became clear that in squares AA1eAA3 (and their extension Mousterian were obtained in the 1990s (GrN-13982: 39100 1300 in AA4), and between 175 and 185 cm below datum, Aurignacian e and GrN-13984: 42,700 900 from level 11/cgr, spit: 29 30; GrN- layer IV truncates Uluzzian layer V at places and lies in direct contact e 13983: 33800 1300 from level 11/cgr, spits 27 28) (Gambassini, with Middle Palaeolithic layers VI and VII. Given this very young age 1997). These, however, were made on burnt bone and, just as with and the rest of the determinations from overlying Aurignacian layers the previous young Uluzzian determinations, they should also be that centre between 33 and 30 ka BP, OxA-19936 originally assigned considered minimum ages for the Middle Palaeolithic at the site. to layer V, is most likely to belong to layer IV or a later occupation In Klissoura Cave, Uluzzian layer V appears locally contaminated (sub-layers IIIeeg). with younger and older material at its contact with other layers (Aurignacian IV and Mousterian VIeVII). It compares best with the later stages, Evolved or Final Uluzzian, of the Cavallo phasing. Bayesian framework for inter-site comparison Recently, microtephra occurring as a sharp peak at the interface of layers V and IV but also spreading further up the stratigraphic col- In an attempt to place the Uluzzian in its most likely calendar umn was identified as the CI (Lowe et al., 2012). This microtephra age and palaeoclimatic context, as well as assess its temporal 8 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 evolution across the dated sites, all new radiocarbon de- precise and the newly produced chronology for the technocomplex terminations were calibrated using the INTCAL09 (atmospheric and is older and more consistent than previous estimates (compare marine) curve (Reimer et al., 2009) and a Bayesian model was built Figs. 3 and 4). using OxCal 4.1.7 (Bronk Ramsey, 2009). The presence of the CI as a chrono-stratigraphic marker is also Discussion incorporated at the sequences where it occurs and is shown in Figs. 4 and 5. The age of the CI is taken as 39,280 110 years (De The new chronology has wider archaeological implications for Vivo et al., 2001), an age estimate that has been produced from our understanding of the Middle to Upper Palaeolithic transition in an average of 36 40Ar/39Ar measurements from 12 proximal de- southern peninsular Europe (limited here to Italy and Greece). posits. No macro- or microtephra has been identified in Fumane. Below we discuss key points arising from this new chronological Sixteen determinations from Cavallo, Castelcivita, Fumane and framework. Klissoura were included in the generated model (the two deter- mination on Cvl-10 were excluded since we hold that they are not Origins and geographic spread of the Uluzzian really related to the Uluzzian occupation of Cavallo). We set the resolution of the model at 20 years, and also ran an outlier detec- Two of the big questions surrounding the appearance of the tion analysis. The modelled output is shown in Fig. 4. In reality, the Uluzzian are its origins and the local pathway(s) of expansion. structure of the model consists of a few more determinations from Given that its makers were traditionally suspected to be Nean- older and younger phases, i.e., Mousterian and Aurignacian in derthals, the origins of the technocomplex were originally thought Fumane and Klissoura, published by Higham et al. (2009), Kuhn to derive from local denticulate Mousterian (e.g., Palma di Cesnola, et al. (2010) and Higham (2011).InFig. 4 we show the relevant 1993). The recent disambiguation of the biological affinities of the Uluzzian phases and determinations therein. We compare the Cavallo teeth and their attribution to modern humans (Benazzi modelling output against the NGRIP d18O record (Andersen et al., et al., 2011) means that an external source of origin should be 2006; Svensson et al., 2006) tuned with respect to the Hulu Cave sought, a wave of modern human dispersal responsible for the U-series chronology (following Weninger and Jöris, 2008)(Fig. 4). appearance of the Uluzzian. Overall, the modelling of the results suggests that the Uluzzian The traditional east to west scenario for the spread of the Upper at all sites started much earlier than previously thought, at or Palaeolithic in Europe, especially that of the Aurignacian culture shortly before 45 ka cal BP, and persisted for as much as 5e6 (e.g., Mellars, 2004, 2006), cannot be sustained in the case of the millennia, until about 39 ka cal BP. The modelled results are more Uluzzian because no similar technocomplexes have been discov- ered in the Levant or further to the east of Europe. Recently, Moroni et al. (2013) suggested that the origins of the Uluzzian should be sought in sub-Saharan Africa on the basis of distinct techno- typological similarities of the Uluzzian with assemblages belonging to the Middle Stone Age to Late Stone Age transition. These authors also suggested that the Uluzzian spread in southern Europe via an Adriatic coastal route (Moroni et al., 2013). The fact that both the Adriatic and the Peloponnese are part of this geographic spread may relate to the emergence of coastal areas of the Ionian-Adriatic during glacial periods, which could have made such a dispersal route available. Ferentinos et al. (2012) discuss the possibility of seafaring around western insular Greece during the Late Pleistocene. The authors suggest that hominids (both Nean- derthals and early modern humans) were capable of seafaring ac- tivities, possibly encouraged by favourable regional coastal configuration during the Middle and Upper Palaeolithic. Unfortu- nately, the absence of reported Uluzzian sites outside the restricted circumference of Italy and southern Greece and in the intermediate regions between Africa and southern Europe, as well as the inun- dation of large tracks of land in the Adriatic and other coastlines during the Holocene, render the East Africa-to-southern Europe a hypothesis only based on techno-typological affinities of the lithic toolkit and other elements of the archaeological record (production of beads, use of colourants, modern human affinities). Only the excavation of additional sites and the identification and study of further Uluzzian assemblages within and outside the tech- nocomplexes current geographical span will elucidate exact dispersal pathways and areas of Uluzzian influence. Even within the limited geographic area of Italy, where the ma- Figure 5. Probability distribution functions (PDFs) for the start and end boundaries of jority of the archaeological evidence stems from, the debate on the Uluzzian at the studied sites, as well for the start of the Proto-Aurignacian at the fi Italian sites mentioned in the text. These were generated after modelling the new whether the Uluzzian rst appeared in the south, centre (Palma di radiocarbon determinations, shown in Fig. 4, using the OxCal platform (Bronk Ramsey, Cesnola, 2004; Ronchitelli et al., 2009)orthenorth(Peresani, 2009). The start of the Uluzzian in Cavallo is given here as the PDF for the start 2008) of the peninsula remains open. Scenarios of delayed coloni- d18 boundary of the Archaic Uluzzian (E III). On the top of the graph, the NGRIP O record sation of the south by northern Uluzzians (e.g., Kaczanowska et al., is shown in blue (Andersen et al., 2006; Svensson et al., 2006) and the Greenland 2010) have been recently put forward. Here, we attempted to use Interstadials, corresponding to ameliorated climatic conditions, are numbered. The age for the CI ash is shown on the top of the sequence as a grey line at the sites it occurs, our new data to answer this question. Ordering of the probability and as dashed line in Fumane where its presence in not confirmed. distribution functions (PDFs) that were generated by the Bayesian K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 9 model for the start of the Uluzzian phases at each site (Fig. 5)allows Aurignacian layers are currently undated, their age is constrained us to estimate the chances for an Uluzzian layer at a site being older by two ‘known-age’ horizons: the uppermost Uluzzian layer rsa00 than that at another site (Table 3). In the absence of independent dated here between 40.6 and 41.9 ka cal BP (Table 2) and the CI stratigraphic markers to confidently identify one context as being volcanic debris of estimated calendar age around w39.3 ka. This contemporaneous, younger or older than another, we require a start suggests that the Aurignacian in Castelcivita arrived soon after 41e boundary to have significantly more chances of being earlier than 42 ka and advanced in two successive forms: initially as Proto- another start boundary. The results of this ordering exercise suggest aurignacian with Dufour bladelets, and later as a local variant that the generated age estimate for the start of the Uluzzian in the incorporating tiny ‘Castelcivita micropoints’ (Gambassini, 1997). Italian south, represented by Cavallo, appears to be older (70% The Aurignacian lasted at the site for the next w2e3 millennia chances) than that of Fumane in the north of Italy. However, this when the sequence was completely sealed by the pumices and grey should be seen only as a tentative hypothesis given that the lower- ash of the CI eruption (Giaccio et al., 2006). At a nearby open-air most Uluzzian levels and final Mousterian levels at Cavallo (and site, Serino (Fig. 1), the CI covers Aurignacian deposits with rare Castelcivita) remain undated and unconstrained in absolute terms, micropoints similar to those of Castelcivita’s Protoaurignacian while those of Fumane are. Improved chronology might change the (Accorsi et al., 1979). Recently, four charcoal samples from a distinct picture. With regards to the other sites, the Uluzzian in Klissoura hearth in Serino were dated using ABOx-SC at w34.5 ka BP (Wood remains the youngest of all. et al., 2012). When calibrated, these determinations correspond Based on the current data, the only valid conclusion one can exactly to the accepted calendar age of the CI eruption at 39.3 ka reach is that the Uluzzian is already present both in southern and (De Vivo et al., 2001). Based on these two archaeological sequences, northern Italy by 39e40 ka BP (43e46 ka cal BP). How much earlier we can suggest that Protoaurignacian groups had spread into before that, is not known. The evidence from Greece is limited so far western and central Campania by 39 ka and even earlier, but not and appears rather late. It is expected that comprehensive dating of before 41e42 ka. more existing sites will help to answer questions revolving around Moving to northern Italy, the Aurignacian has been securely the origins of the Uluzzian technocomplex. dated in two sites: Fumane in Verona and Mochi in western Liguria. From a climatic point of view, the appearance of the Uluzzian as The Uluzzian in Fumane seems to have ended shortly after 42 ka cal we currently define it, falls at the end of Greenland Interstadial 12 BP, and there seems that occupation of the site came to a halt for (GIS 12) and the cold stadial between GIS 12 and 11. While Cavallo E about a millennium, and possibly more, until the arrival of the first III lacks direct determinations, the Bayesian model tentatively Aurignacians. The Aurignacian chronology for Fumane was places its age well within GIS 12. GIS 12 follows immediately after described by Higham et al. (2009) and Higham (2011) who, on the Heinrich Event 5 (HE5), a climatic low at w48 ka. Müller et al. basis of new determinations for ABOx-treated charcoal and ultra- (2011) have suggested, on the basis of environmental proxies, filtered bone collagen, placed the earliest Protoaurignacian layer A2 that the initial arrival of anatomically modern humans is very likely at 40.5e41.2 ka cal BP. This mirrors very well the estimated age for to have occurred post-HE5 and during GIS 12. The independent the spread of the Protoaurignacian in Castelcivita. Unlike Castelci- new evidence from the Uluzzian sites in our study seems to agree vita, where occupation terminates with the CI eruption, Fumane is with this suggestion. The decrease in temperature inferred from the visited by Aurignacians until about 36e38 ka cal BP as evidenced by composition of the faunal assemblage in Cavallo layer E III (Boscato the cluster of dates at around that time (Higham et al., 2009; and Crezzini, 2011) and at levels A4 and A3 in Fumane (Tagliacozzo Higham, 2011). et al., 2013) may therefore correspond either to the onset of HE5 or In Riparo Mochi (Fig. 1), evidence for older Aurignacian presence the stadial conditions between GIS 12 and 11. If the latter is correct was reported by Douka et al. (2012). A series of 15 radiocarbon the start of the Uluzzian would slightly postdate GIS 12. dates, obtained mainly on shell beads and one charcoal sample, were used to date the largest part of the stratigraphic sequence. No The Uluzzian and its relationship to the Aurignacian in Italy Uluzzian is found at the site, and a semi-sterile layer, currently under study, follows the final Mousterian layer. The new radio- In southern Italy, Castelcivita (and Cala, not examined in this carbon determinations place the start of the Protoaurignacian study) is a crucial archaeological sequence because the Uluzzian is occupation at the site (layer G) at about 37 ka BP or between w42.7 directly followed by a series of Aurignacian layers. Although these and 41.6 ka cal BP. Layer G of Riparo Mochi is therefore the oldest,

Table 3 Likelihoods generated after ordering the posterior probability distributions (pdfs) of the start and end Uluzzian boundaries for the four studied sites as well as the start boundaries for three Italian Aurignacian sites mentioned in the text.

Probability ULUZZIAN AURIGNACIAN t1 < t2 Start and end boundaries Start boundaries

t1 t2

Fumane Fumane Cavallo Cavallo Castelcivita Castelcivita Klissoura Klissoura Mochi Fumane Serino start end start end start end start end start start start

U Fumane start 0.0000 0.9994 0.2748 1.0000 0.8125 0.9907 0.9937 1.0000 0.9732 0.9995 0.9999 L Fumane end 0.0006 0.0000 0.0019 0.9960 0.2532 0.7330 0.7816 0.9996 0.1534 0.4733 0.9857 U Cavallo start 0.7252 0.9981 0.0000 1.0000 0.8765 0.9942 0.9944 1.0000 0.9806 0.9983 0.9999 Z Cavallo end 0.0000 0.0040 0.0000 0.0000 0.0011 0.0481 0.2538 0.8380 0.0000 0.0075 0.4146 Z Castelcivita start 0.1875 0.7468 0.1235 0.9989 0.0000 0.8987 0.8676 0.9998 0.4328 0.7273 0.9947 I Castelcivita end 0.0093 0.2670 0.0058 0.9519 0.1013 0.0000 0.6775 0.9950 0.0317 0.2458 0.9257 A Klissoura start 0.0063 0.2184 0.0056 0.7462 0.1324 0.3225 0.0000 0.9385 0.1138 0.2138 0.6885 N Klissoura end 0.0000 0.0004 0.0000 0.1620 0.0002 0.0050 0.0615 0.0000 0.0000 0.0044 0.1033 A Mochi start 0.0268 0.8466 0.0194 1.0000 0.5672 0.9683 0.8863 1.0000 0.0000 0.8299 0.9980 U Fumane start 0.0005 0.5267 0.0017 0.9925 0.2728 0.7542 0.7862 0.9956 0.1702 0.0000 0.9830 R Serino start 0.0001 0.0143 0.0001 0.5854 0.0053 0.0743 0.3115 0.8967 0.0020 0.0170 0.0000

The pdfs corresponding to this ordering are shown in Fig. 5. 10 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 directly-dated Aurignacian assemblage in Italy. A second group of It is now thought that Neanderthals may have survived in wider younger dates relates to an Early Aurignacian facies at around 32 ka Europe until about 40 ka (e.g., Pinhasi et al., 2011), several millennia BP or between 37.3 and 36.4 ka cal BP (Douka et al., 2012). The after the appearance of the Uluzzians in Italy and Greece, placed at synchronous shell and charcoal radiocarbon determinations from w 43e45 ka. The claim of insufficient time for co-existence/ Riparo Mochi suggest strongly the reliability of marine gastropods acculturation therefore can no longer be sustained without as an appropriate substrate for 14C dating, and dismiss large vari- further examination of the record for Neanderthal presence syn- ations of the local marine reservoir during the period at issue chronous to the Uluzzian occupation of southern Europe. (contra Banks et al., 2013). Sadly, both in Italy and Greece final/terminal Mousterian con- All other Protoaurignacian or Early Aurignacian industries in texts, thought to represent the latest Neanderthal populations, are Italy either lack secure absolute determinations or have been very badly dated. In Italy, only Mochi and Bombrini in Liguria shown to be younger by several hundreds to a few thousand years (w36e38 ka BP/41e43 ka cal BP) (Douka et al., 2012), Fumane Cave (e.g., Bombrini, Paglicci, Fumane). (41e42 ka BP/44e46 ka cal BP) (Higham et al., 2009) and, less There is a clear chronological and stratigraphic separation be- securely, Mezzena rockshelter (35 ka BP/w40 ka cal BP) (Longo tween Aurignacian and Uluzzian assemblages at the sites where the et al., 2012) have been dated using the latest preparative two co-occur. At no site is there evidence of interstratification be- methods. In the case of Mezzena rockshelter, claims for late tween the two technocomplexes and only the presence of some Neanderthal survival were made on the basis of a single 14C mea- ‘aurignacoid’ tools in the final phase of the Uluzzian offers limited surement on mammal bone. However, a single measurement support that the two populations (Uluzzian and Proto-/Aurignacian cannot provide the precision required for drawing strong conclu- groups) could have met. however, on the basis of the observed sions and more extensive work is needed. In all other instances spatio-temporal distinction we observe on our results we may (Broion, S. Bernardino, Rio Secco, all’Onda, Buca della Iena, Breuil, suggest that the transition from the Uluzzian to the Aurignacian Sant’Agostino, Reali, Castelcivita, Poggio shelter, Oscurusciuto) could have its roots in a secondary modern human population (Hedges et al., 1994; Gambassini, 1997; Caramia and Gambassini, movement, which ultimately spread across Europe. The processes 2006; Milliken, 2007; Peresani and Gurioli, 2007; Riel-Salvatore, involved in this succession, whether those of population replace- 2007; Boscato et al., 2009; Ronchitelli et al., 2009; Peresani, 2011; ment, elimination or incorporation/absorption, remain unknown. Peretto, 2012), Mousterian determinations should be considered at With regards to the advance of the Aurignacian, and therefore best minimum ages. its relationship to the final Uluzzian in the studied regions, the In Greece, the situation is worse since the record is patchy and older determinations from Mochi and the well-established con- discontinuous, and only a handful of poorly-dated sites have been nections of its Protoaurignacian inhabitants with both Italy and assigned to the Middle Palaeolithic period (Fig. 1). Mousterian sites France, may hint at a northwest, possibly to northeast, and finally with absolute ages include Theopetra, dated to the last Interglacial southern Aurignacian spread (i.e., from southern France to north- or before (end of OIS 6/beginning OIS 5; Valladas et al., 2007), ern Italy and finally to southern Italy), from w37 ka BP Asprochaliko imprecisely dated at w40 ka BP (Bailey et al., 1992), (w42.5 ka cal BP) onwards. Although there is little evidence from and Kalamakia at >40 ka (Darlas, 2007). Some later ages for find- intermediate regions, this hypothesis corroborates the later ages spots close to River Peneios, between 28 and 44 ka BP (32e that exist for the Aurignacian in southern Italy (Paglicci, Serino, 48 ka cal BP), have been attributed to a late persistence of Nean- Castelcivita, Cala), where the industry does not appear before derthals in the area (Harvati et al., 2009). Caution is advised, w35 ka BP (39e40 ka cal BP). however, since determinations of such antiquity, produced using The very small number of excavated and dated Aurignacian sites conventional radiocarbon methods and less refined chemical pre- in Greece does not permit a broad discussion on the chronological treatment protocols, must represent minimum ages at best. Klis- position of the technocomplex, neither on its origins. Similarly to soura 1 and Lakonis I both contain final Mousterian assemblages the Italian south, a relatively late arrival for the Aurignacian may be and they are currently the best-dated sites for this period in Greece. suggested, since nowhere does the technocomplex predate 35 ka In Lakonis, late Middle Palaeolithic Unit Ib was dated between 39 BP (e.g., Franchthi, Klissoura) (Kuhn et al., 2010; Douka et al., 2011). and 43 ka BP (43e48 ka cal BP) (Panagopoulou et al., 2004; Elefanti In terms of origins, a northern (Balkan) route is the most prevalent et al., 2008), although none of the determinations were treated model, however, western/north-western influences from across the with modern preparation methods for charcoal (ABOx) and there- Adriatic cannot be excluded either. fore may also correspond to minimum ages. At the same site, a ‘transitional’/Initial Upper Palaeolithic (IUP) complex with evidence Scenarios for co-existence between Uluzzians (modern humans) and of Upper Palaeolithic tool types (Unit Ia) is associated with statis- late Mousterians (Neanderthals) tically identical determinations (38e42 ka BP/42e48 ka cal BP). The excavators have suggested that the industry of Unit Ia was the We wanted to investigate how the older age estimates for the product of Neanderthals on the basis of a Neanderthal tooth found appearance of the Uluzzian in Italy and Greece may influence there (Harvati et al., 2003). The determinations from Unit Ia predate contact scenarios between modern humans and Neanderthals. the Uluzzian evidence of Klissoura and there appears no obvious Reviews on the topic of Neanderthal and modern human contact connection between the IUP of Lakonis with the Uluzzian of Klis- have focused almost exclusively on spatio-temporal overlap to soura in terms of material culture or lithic techno-typology. verify or negate population contacts. Long periods of regional co- Finally in Klissoura, some recently obtained radiocarbon de- existence are bound to facilitate population contacts. This, in terminations place the end of the Mousterian at about 41e40 ka BP turn, would give rise to uni-/bidirectional flow of knowledge (AA-73819: 40920 580, AA-73818: 41480 810; Kuhn et al., (acculturation) and/or of genetic material (interbreeding) leading 2010)oratw44e45 ka cal BP. Further down that stratigraphy, to the development of industries possibly with both Middle layer VII is dated with one measurement (AA-73820: Palaeolithic and Upper Palaeolithic characteristics, as well as hy- 48990 1770 BP) at w47e51 ka cal BP. A chronological gap of bridized human forms. One of the main points of refuting accul- several millennia is identified in the determinations available for turation scenarios between Neanderthals and early modern the terminal Mousterian and the Uluzzian in Klissoura, which humans has been the claim that there was no significant period of corresponds well with the stratigraphic discontinuities observed at coexistence between the two populations (e.g., Zilhão et al., 2008). the interface of Middle Palaeolithic layers VIIeVI and the Uluzzian K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 11 layer V. Again, Klissoura refutes scenarios of population overlap in populations (Neanderthals) living in the same region. The secure the Peloponnese. We conclude that no evidence to support modern chronology presented here for the Uluzzian elucidates the least- human and Neanderthal co-existence/interaction can be found in known aspect of the technocomplex and attempts to place the Greece. arrival of early modern humans in Europe at the right period. This Overall, at all sites where a succession of Mousterian-Uluzzian- we hope will enable convincing future comparisons of the Uluzzian Aurignacian occupations exists, this always follows the indicated technocomplex with other transitional and early Upper Palaeolithic sequence with no interstratifications. In addition, at most sites industries in wider Eurasia. (including the sequences studied here (Cavallo, Castelcivita, Klis- soura)) the Mousterian is separated from the Uluzzian by a sedi- Acknowledgements mentary break, a layer of volcanic ash or other stratigraphic hiatus, which may suggest that a period of time has elapsed between the We would like to thank R.E.M. Hedges for his support and two phases. guidance throughout the period that this work was undertaken. Taken together, while the antiquity of modern human presence Several colleagues were instrumental to our study, and are specially in southern peninsular Europe (Italy and Greece) has been thanked: A. Palma di Cesnola, H. Klempererova, M. Koumouzelis, A. fi extended signi cantly with the attribution of the Uluzzian to Moroni, F. Ranaldo and J. Riel-Salvatore. We would also like to modern populations, neither the available chronology nor the acknowledge the continuing support of the Puglia and Campania stratigraphic evidence of late Mousterian contexts may validate Archaeological Superintendence branches, for facilitating excava- parallel existence and overlap between late Neanderthals and early tion and study of the material. The dating work has been funded by modern humans in the region concerned (contra Longo et al., 2012; a NERC-AHRC NRCF grant, as well as a major NERC grant (NE/ Zilhão, 2013). In turn, this may lead us to suggest that the Uluzzians D014077/1) to TH. KD was funded by IKY-Greece, the Leventis ‘ ’ arrived at virgin lands where Neanderthal numbers were very low, Foundation, the Leverhulme Trust and the European Research if at all present. This is in agreement with the hypothesis expressed Council Grant “PALAECHRON” (ERC-2012-AdG–324139) during the by Moroni et al. (2013). course of this work. Research at Fumane is coordinated by the fi A further implication of this scenario, which (i) identi es the Ferrara University in the framework of a project supported by the Uluzzian as a product of modern humans, and (ii) allows its Italian Ministry of Culture e Veneto Archaeological Superintend- development far from and independently of Neanderthal in- ence, public institutions (Lessinia Mountain Community e Regional fl uences, is the potential support it offers to the suggestion that Natural Park, Fumane Municipality, Veneto Region e Department other transitional industries reported from Europe and the Levant for Cultural Heritage), and private associations and companies may indeed be the product of modern humans and not Neander- (Cariverona Foundation, Banca di Credito Cooperativo della thals (Bar-Yosef and Bordes, 2010; Bordes and Teyssandier, 2011; Valpolicella). Douka et al., 2013). We need, however, to remain cautious and suggest that further detailed work is required if we were to un- derstand aspects of other transitional technocomplexes exhibiting Appendix A. Supplementary data similar characteristics to the Uluzzian, such as the Balkan Bach- okirian (e.g., level 11 of the Bacho Kiro) or the (or Supplementary data related to this article can be found at http:// Emiro-Bohunician) of Central Europe. A good start would be a fresh dx.doi.org/10.1016/j.jhevol.2013.12.007. and reliable absolute dating of these contexts. References Conclusions Accorsi, C.A., Aiello, E., Bartolini, C., Castelletti, L., Rodolfi, G., Ronchitelli, A., 1979. Il fi We present here the first integrated chronological synthesis for giacimento paleolitico di Serino (Avellino): Stratigra a, ambienti e paletnologia. Atti della Societa Toscana di Scienze Naturali, Memorie A86, 435e487. the Uluzzian, currently assumed to be the oldest modern human Alessio, M., Bella, F., Importa, S., Belluomini, G., Cortesi, C., Turi, B., 1970. University technocomplex in Europe. By using a series of new radiocarbon of Rome Carbon-14 dates VIII. Radiocarbon 12, 599e616. determinations on shell, bone and charcoal substrate, all produced Andersen, K.K., Svensson, A., Johnsen, S.J., Rasmussen, S.O., Bigler, M., Röthlisberger, R., Ruth, U., Siggaard-Andersen, M.-L., Steffensen, J.P., Dahl- with the latest preparative methods at the Oxford Radiocarbon Jensen, D., Vinther, B.M., Claussen, H.B., 2006. The Greenland ice core chro- laboratory, and a purpose-built Bayesian statistical framework, we nology 2005, 15e42 ka. Part 1: constructing the time scale. Quatern. Sci. Rev. 25, constrained the age of the Uluzzian and compared it with that for 3246e3257. Bailey, G., Papakonstantinou, V., Sturdy, D., 1992. Asprochaliko and Kokkinopilos: TL previous and succeeding technocomplexes (Mousterian, Aurigna- dating and reinterpretation of Middle Palaeolithic sites in Epirus, northwest cian). We conclude that the Uluzzian arrived in Italy and Greece, Greece. Cambridge Archaeol. J. 2, 136e144. where its only occurrences are identified so far, shortly before 45 ka Banks, W.E., d’Errico, F., Zilhão, J., 2013. Revisiting the chronology of the Proto- ’ fi w Aurignacian and the Early Aurignacian in Europe: A reply to Higham et al. s and continued evolving until its nal stages 39 ka. Issues of origin comments on Banks, et al. (2013). J. Hum. Evol. 65, 810e817. are not easy to disentangle at the moment, but similarities of the Bar-Yosef, O., Bordes, J.-P., 2010. Who were the makers of the Châtelperronian Uluzzian tool techno-typology with assemblages in East Africa have culture? J. Hum. Evol. 59, 586e593. Bartolomei, G., Broglio, A., Cassoli, P., Castelletti, L., Cremaschi, M., Giacobini, G., been drawn. Yet, given the lack of intermediate assemblages be- Malerba, G., Maspero, A., Peresani, M., Sartorelli, A., Tagliacozzo, A., 1992. La tween Africa and southern Europe, as well as the absence of the Grotte-Abri de Fumane. Un site Aurignacien au Sud des Alps. Preistoria Alpina Uluzzian outside the limited geographical circumference of Italy 28, 131e179. and Greece, any quest for roots will remain challenging. Benazzi, S., Douka, K., Fornai, C., Bauer, C.C., Kullmer, O., Svoboda, J., Pap, I., Mallegni, F., Bayle, P., Coquerelle, M., Condemi, S., Ronchitelli, A., Harvati, K., Regardless of where the Uluzzians arrived from, in addition to Weber, G.W., 2011. Early dispersal of modern humans in Europe and implica- modern human anatomy they also convey fully-modern behaviour tions for Neanderthal behaviour. Nature 479, 525e528. fi expressed through techno-typologically complex lithic toolkits, Bird, M., Ayliffe, L.K., Fi eld, K., Cresswell, R., Turney, C., 1999. Radiocarbon dating of ‘old’ charcoal using a wet oxidation-stepped combustion procedure. Radio- highly-adaptive subsistence strategies, bone tool industry, use of carbon 41, 127e140. pigments and productions of ornaments. These are part of the Bordes, J.-G., Teyssandier, N., 2011. The nature of the Châ- ‘Uluzzian package’ since its earliest appearance in southeastern telperronian in south-western France: Archeostratigraphic and lithic evidence. Quatern. Int. 246, 382e388. Europe and many of these features, such as bone tools and shell Borzatti von Löwenstern, E., 1970. Prima di scavi nella Grotta ‘Mario beads, do not belong to the behavioural package of previous Bernardini’ (Nardò, Lecce). Riv. Sci. Preist. 25, 89e125. 12 K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13

Boscato, P., Crezzini, J., 2011. Middle-Upper Palaeolithic transition in southern Italy: Harvati, K., Panagopoulou, E., Runnels, C., 2009. The paleoanthropology of Greece. Uluzzian macromammals from Grotta del Cavallo (Apulia). Quatern. Int. 252, Evol. Anthropol. 18, 131e143. 90e98. Hedges, R.E.M., Housley, R., Bronk Ramsey, C., van Klinken, J., 1994. Radiocarbon Boscato, P., Boschian, G., Caramia, F., Gambassini, P., 2009. Il Riparo del Poggio a dates from the Oxford AMS System: Archaeometry Datelist 18. Archaeometry Marina di (Salerno): culture e ambiente. Riv. Sci. Preist. 59, 5e40. 36, 337e374. Brock, F., Higham, T.F.G., Ditchfield, P., Bronk Ramsey, C., 2010. Current pre- Higham, T.F.G., 2011. European Middle and Upper Palaeolithic radiocarbon dates are treatment methods for AMS radiocarbon dating at the Oxford Radiocarbon often older than they look: problems with previous dates and some remedies. Accelerator Unit (ORAU). Radiocarbon 52, 103e112. Antiquity 85, 235e249. Broglio, A., 1996e1997. L’estinzione dell’Uomo di Neandertal e la comparsa dell’Uomo Higham, T.F.G., Brock, F., Peresani, M., Broglio, A., Wood, R., Douka, K., 2009. Prob- moderno in Europa. Le evidenze della Grotta di Fumane nei Monti Lessini. Atti Ist. lems with radiocarbon dating the Middle to Upper Palaeolithic transition in Veneto SS. LL. AA., Classe Scienze fisiche, matematiche e naturali, t.CLV, 1e44. Italy. Quatern. Sci. Rev. 28, 1257e1267. Broglio, A., Cremaschi, M., Peresani, M., Bertola, S., Bolognesi, L., De Stefani, M., Jéquier, C.A., Romandini, M., Peresani, M., 2012. Les retouchoirs en matières dures Fiocchi, C., Gurioli, F., Marini, D., 2003. L’Aurignacien dans le territoire préalpin: animales: une comparaison entre Moustérien final et Uluzzien. C.R. Palevol. 11, la Grotte de Fumane. In: Vasil’ev, S.A., Soffer, O., Kozlowski, J. (Eds.), Perceived 283e292. Landscapes and Built Environments. The Cultural Geography of Late Palaeolithic Kaczanowska, M., Koz1owski, J.K., Sobczyk, K., 2010. Upper Palaeolithic human oc- Eurasia. Atti XIV Congresso Unione Società Preistoriche e Protostoriche, British cupations and material culture at Klisoura Cave 1. Eurasian Prehist. 7, 133e286. Archaeological Reports, International Series 1122, Oxford, pp. 93e104. Karkanas, P., 2010. Geology, stratigraphy and site formation processes of the Upper Bronk Ramsey, C., 2009. Bayesian analysis of radiocarbon dates. Radiocarbon 51, Palaeolithic and later sequence in Klisoura Cave 1. Eurasian Prehist. 7, 15e36. 337e360. Koumouzelis, M., Ginter, B., Koz1owski, J.K., Pawlikowski, M., Bar-Yosef, O., Caramia, F., Gambassini, P., 2006. Il Musteriano evoluto del Riparo del Poggio (strati Albert, R.M., Litynska-Zajac, M., Stworzewicz, E., Wojtal, P., Lipecki, G., Tomek, T., 9 e 10) a Marina di Camerota e Salerno. Rassegna di Archeologia 22, 67e95. Bochenski, Z.M., Pazdur, A., 2001. The early Upper Palaeolithic in Greece: the Churchill, S.E., Smith, F.H., 2000. Makers of the early Aurignacian of Europe. Yearb. excavations in Klisoura cave. J. Archaeol. Sci. 28, 515e539. Phys. Anthropol. 43, 61e115. Kuhn, S.L., Pigati, J., Karkanas, P., Koumouzelis, M., Kozlowski, J.K., Ntinou, M., Cremaschi, M., Ferraro, F., Peresani, M., Tagliacozzo, A., 2005. Il sito: nuovi contributi Stiner, M.C., 2010. Radiocarbon dating results for the early Upper Palaeolithic of sulla stratigrafia, la cronologia, le faune a macromammiferi e le industrie del Klissoura 1 Cave. Eurasian Prehist. 7, 37e46. paleolitico antico. In: Broglio, A., Dalmeri, G. (Eds.), Pitture paleolitiche nelle Longo, L., Boaretto, E., Caramelli, D., Giunti, P., Lari, M., Milani, L., Mannino, M.A., Prealpi Venete: Grotta di Fumane e Riparo Dalmeri. Memorie Museo Civico Sala, B., Thun Hohenstein, U., Condemi, S., 2012. Did Neandertals and Storia Naturale di Verona 9, Verona, pp. 12e22. anatomically modern humans coexist in northern Italy during the late MIS 3? d’Errico, F., Borgia, V., Ronchitelli, A., 2012. Uluzzian bone technology and its im- Quatern. Int. 259, 102e112. plications for the origin of behavioural modernity. Quatern. Int. 259, 59e71. Lowe, J., Barton, N., Blockley, S., Bronk Ramsey, C., Cullen, V.L., Davies, W., Darlas, A., 2007. Le Moustérien de Grèce à la lumière des récentes recherches. Gamble, C., Grant, K., Hardiman, M., Housley, R., Lane, C.S., Lee, S., Lewis, M., L’Anthropologie 111, 346e366. MacLeod, A., Menzies, M.A., Mueller, W., Pollard, M., Price, C., Roberts, A.P., De Stefani, M., Dini, M., Klempererova, H., Peresani, M., Ranaldo, F., Ronchitelli, A., Rohling, E.J., Satow, C., Smith, V.C., Stringer, C., Tomlinson, E.L., White, D., Ziggiotti, S., 2012. Continuity and replacement in flake production across the Albert, P.G., Arienzo, I., Barker, G., Boric, D., Carendente, A., Civetta, L., Ferrier, C., Middle-Upper Palaeolithic transition: a view over the Italian Peninsula. In: Guadelli, J.-L., Karkanas, P., Koumouzelis, M., Müller, U.C., Orsi, G., Pross, J., Pastoors, A., Peresani, M. (Eds.), Flakes Not Blades: the Role of Flake Production Rosi, M., Shalamanov-Korobar, L., Sirakov, N., Tzedakis, P.C., 2012. Volcanic ash at the Onset of the Upper Palaeolithic in Europe. Wissenschaftliche Schriftendes layers illuminate the resilience of Neanderthals and early modern humans to Neanderthal Museums 5, Mettmann, pp. 135e151. natural hazards. Proc. Natl. Acad. Sci. 109, 13532e13537. De Vivo, B., Rolandi, G., Gans, P.B., Calvert, A., Bohrson, W.A., Spera, F.J., Belkin, H.E., Martini, M., Sibilia, M., Croci, S., Cremaschi, M., 2001. Thermoluminescence (TL) 2001. New constraints on the pyroclastic eruptive history of the Campanian dating of burnt flints: problems, perspectives and some example of application. volcanic Plain (Italy). Miner. Petrol. 73, 47e65. J. Cult. Herit. 2, 179e190. Douka, K., 2011. An Upper Palaeolithic shell scraper from Ksar Akil (Lebanon). Mellars, P., 2004. Neanderthals and the modern human colonization of Europe. J. Archaeol. Sci. 38, 429e437. Nature 432, 461e465. Douka, K., Hedges, R.E.M., Higham, T.F.G., 2010a. Improved AMS 14C dating of shell Mellars, P., 2006. Archeology and the dispersal of modern humans in Europe: carbonates using high-precision X-Ray Diffraction (XRD) and a novel density deconstructing the ‘Aurignacian’. Evol. Anthropol. 15, 167e182. separation protocol (CarDS). Radiocarbon 52, 735e751. Milliken, S., 2007. Neanderthals, anatomically modern humans and ‘modern human Douka, K., Sinitsyn, A., Higham, T., 2010b. The influence of pretreatment chemistry behaviour’ in Italy. Oxford J. Archaeol. 26, 331e358. on the radiocarbon dating of Campanian Ignimbrite-aged charcoal from Kos- Moroni, A., Boscato, P., Ronchitelli, A., 2013. What roots for the Uluzzian? Modern tenki 14 (Russia). Quatern. Res. 73, 583e587. behaviour in Central-Southern Italy and hypotheses on AMH dispersal routes. Douka, K., Perlès, C., Valladas, H., Vanhaeren, M., Hedges, R.E.M., 2011. Franchthi Quatern. Int. 316, 27e44. Cave revisited: the age of the Aurignacian in south-eastern Europe. Antiquity Müller, U.C., Pross, J., Tzedakis, P.C., Gamble, C., Kotthoff, U., Schmiedl, G., Wulf, S., 85, 1131e1150. Christanis, K., 2011. The role of climate in the spread of modern humans into Douka, K., Grimaldi, S., Boschian, G., del Lucchese, A., Higham, T.F.G., 2012. A new Europe. Quatern. Sci. Rev. 30, 273e279. chronostratigraphic framework for the Upper Palaeolithic of Riparo Mochi Palma di Cesnola, A., 1963. Prima campagna scavi a Grotta del Cavallo presso Santa (Italy). J. Hum. Evol. 62, 286e299. Caterina (Lecce). Riv. Sci. Preist. 18, 41e74. Douka, K., Bergman, C.A., Hedges, R.E.M., Wesselingh, F.P., Higham, T.F.G., 2013. Palma di Cesnola, A., 1964. Seconda campagna di scavo nella grotta del Cavallo. Riv. Chronology of Ksar Akil (Lebanon) and implications for the colonization of Sci. Preist. 19, 23e39. Europe by anatomically modern humans. Plos One 8, e72931. Palma di Cesnola, A., 1965a. Il Paleolitico superiore arcaico (facies uluzziana) della Elefanti, E., Panagopoulou, E., Karkanas, P., 2008. The transition from the Middle to Grotta del Cavallo, Lecce. Riv. Sci. Preist. 20, 33e62. Upper Palaeolithic in the southern Balkans: The evidence from the Lakonis I Palma di Cesnola, A., 1965b. Notizie preliminari sulla terza campagna di scavi nella Cave, Greece. Eurasian Prehist. 5, 77e87. Grotta del Cavallo (Lecce). Riv. Sci. Preist. 20, 291e302. Fedele, G., Giaccio, B., Hadjas, I., 2008. Timescales and cultural processes at Palma di Cesnola, A., 1966a. Il Paleolitico superiore arcaico (facies uluzziana) della 40,000 BP in the light of the Campanian Ignimbrite eruption, Western Eurasia. Grotta del Cavallo, Lecce (continuazione). Riv. Sci. Preist. 21, 3e59. J. Hum. Evol. 55, 834e857. Palma di Cesnola, A., 1966b. Gli scavi nella Grotta del Cavallo (Lecce) durante il 1966. Ferentinos, G., Gkioni, M., Geraga, M., Papatheodorou, G., 2012. Early seafaring ac- Riv. Sci. Preist. 21, 289e302. tivity in the southern Ionian Islands, Mediterranean Sea. J. Archaeol. Sci. 39, Palma di Cesnola, A., 1969. Datazione dell’Uluzziano col Metodo del Ce14. Riv. Sci. 2167e2176. Preist. 24, 341e348. Gambassini, P. (Ed.), 1997. Il Paleolitico di Castelcivita, Culture e Ambiente. Electa Palma di Cesnola, A., 1989. L’Uluzzien: faciès italien du leptolithique archaique. L’ Napoli, Naples. Anthropologie 93, 783e812. Gambassini, P., Mallegni, F., Ricci, S., 2006. Grotta del Cavallo: the finding of a Palma di Cesnola, A., 1993. Il Paleolitico Superiore in Italia. Garlatti e Razzai, Firenze. Neanderthal deciduous tooth associated to Uluzzian industry. Abstracts of the Palma di Cesnola, A., 2004. Rapports entre l’Aurignacien et l’Uluzzien en Italie. In: XV UISPP Congress, Lisbon, p. 739. Actes du XIVème congrès UISPP, Université de Liège. British Archaeological Giaccio, B., Hajdas, I., Peresani, M., Fedele, F.G., Isaia, R., 2006. The Campanian Reports, International Series 1240, Oxford, pp. 7e12. Ignimbrite tephra and its relevance for the timing of the Middle to Upper Palma di Cesnola, A., Borzatti von Löwenstern, E., 1964. Gli scavi dell’Istituto Italiano Palaeolithic shift. In: Conard, N.J. (Ed.), When Neanderthals and Modern di Preistoria e Protostoria nel Salento durante l’ultimo triennio. In: Atti VIII e IX Humans Met. Kerns Verlag, Tübingen, pp. 343e375. Riunione Scientifica IIPP, Firenze, pp. 27e44. Giaccio, B., Isaia, R., Fedele, F.F., Di Canzio, E., Hoffecker, J.F., Ronchitelli, A., Palma di Cesnola, A., Messeri, P., 1967. Quatre dents humaines paléolitiques trou- Sinitsyn, A.A., Anikovich, M.A., Lisitsyn, S.N., Popov, V.V., 2008. The Campanian vées dans des cavernes de l’Italie méridionale. L’Anthropologie 71, 249e262. Ignimbrite and Codola tephra layers: two temporal/stratigraphic markers for Panagopoulou, E., Karkanas, P., Tsartsidou, G., Kotjabopoulou, E., Harvati, K., the Early Upper Palaeolithic in southern Italy and eastern Europe. J. Volcanol. Ntinou, M., 2004. Late Pleistocene archaeological and fossil human evidence Geotherm. Res. 177, 208e226. from Lakonis Cave, Southern Greece. J. Field Archaeol. 29, 323e349. Hahn, J., 1984. Südeuropa und Nordafrika. Neue Forschungen Zur Altsteinzeit. Papagianni, D., 2009. Mediterranean southeastern Europe in the Late Middle and Verlag C.H. Beck, München, pp. 1e231. Early Upper Palaeolithic: modern human route to Europe or Neanderthal refu- Harvati, K., Panagopoulou, E., Karkanas, P., 2003. First Nearderthal remains from gium? In: Camps, M., Szmidt, C. (Eds.), The Mediterranean from 50 000 to 25 000 Greece: the evidence from Lakonis. J. Hum. Evol. 45, 465e473. BP. Turning points and New directions. Oxbow Books, Oxford, pp. 115e136. K. Douka et al. / Journal of Human Evolution 68 (2014) 1e13 13

Peresani, M., 2008. A new cultural frontier for the last Neanderthals: the Uluzzian in Santos, G.M., Bird, M.I., Parenti, F., Fifield, L.K., Guidon, N., Hausladen, P.A., 2003. Northern Italy. Curr. Anthropol. 49, 725e731. A revised chronology of the lowest occupation layer of Pedra Furada Rock Peresani, M., 2011. The end of the Middle Palaeolithic in the Italian Alps. An over- Shelter, Piauí, Brazil: the Pleistocene peopling of the Americas. Quatern. Sci. Rev. view on Neandertal land-use, subsistence and technology. In: Conard, N., 22, 2303e2310. Richter, J. (Eds.), Neanderthal Lifeways, Subsistence and Technology. One Sarti, L., Boscato, P., Lo Monaco, M., 1998e2000. Il Musteriano finale di Grotta del Hundred Fifty Years of Neanderthal Study. Vertebrate Paleobiology and Paleo- Cavallo nel Salento, studio preliminare. Origini 22, 45e109. anthropology Series. Springer, Dordrecht, pp. 249e259. Sarti, L., Boscato, P., Martini, F., Spagnoletti, A.P., 2002. Il Musteriano di Grotta del Peresani, M., 2012. Fifty thousand years of flint knapping and tool shaping across the Cavallo e strati H e I: studio preliminare. Riv. Sci. Preist. 52, 21e110. Mousterian and Uluzzian sequence of Fumane cave. Quatern. Int. 247, 125e150. Sivan, D., Potasman, M., Almogi-Labin, A., Bar-Yosef Mayer, D.E., Spanier, E., Peresani, M., Gurioli, F., 2007. The Rio Secco Cave, a new final Middle Palaeolithic Boaretto, E., 2005. The Glycymeris query along the coasts and shallow shelf of site in North-eastern Italy. Eurasian Prehist. 5, 85e94. Israel, southeast Mediterranean. Palaeogeogr. Palaeoclimatol. Palaeoecol. 233, Peresani, M., Cremaschi, M., Ferraro, F., Falguères, C., Bahain, J.-J., Gruppioni, G., 134e148. Sibilia, E., Quarta, G., Calcagnile, L., Dolo, J.-M., 2008. Age of the final Middle Stiner, M.C., 2010. Shell ornaments from the Upper Paleolithic and Mesolithic layers Palaeolithic and Uluzzian levels at Fumane Cave, Northern Italy, using 14C, ESR, of Klissoura Cave 1 by Prosymna, Greece. Eurasian Prehist. 7, 287e308. 234U/230Th and thermoluminescence methods. J. Archaeol. Sci. 35, 2986e2996. Svensson, A., Andersen, K.K., Bigler, M., Clausen, H.B., Dahl-Jensen, D., Davies, S.M., Peretto, C. (Ed.), 2012. L’Insediamento Musteriano di Grotta Reali e Rocchetta a Johnsen, S.J., Muscheler, R., Rasmussen, S.O., Röthlisberger, R., Peder Volturno, Molise, Italia. Annali dell’Università di Ferrara, Museologia Scientifica Steffensen, J., Vinther, B.M., 2006. The Greenland ice core chronology 2005, 15- e Naturalistica 8. Springer, Milan. 42 ka. Part 2: comparison to other records. Quatern. Sci. Rev. 25, 3258e3267. Pinhasi, R., Higham, T.F.G., Golovanova, L.V., Doronichev, V.B., 2011. Revised age of Tagliacozzo, A., Romandini, M., Fiore, I., Gala, M., Peresani, M., 2013. Animal late Neanderthal occupation and the end of the Middle Paleolithic in the exploitation strategies during the Uluzzian at Grotta di Fumane (Verona). In: northern Caucasus. Proc. Natl. Acad. Sci. 108, 8611e8616. Variability in Human Hunting Behavior During Oxygen Isotope Stages (OIS) 4/3: Ranaldo, F., in press. Considerazioni sull’interpretazione della variabilità dei sistemi Implications for Understanding Modern Human Origins. Vertebrate Paleobi- tecnici nel passaggio dal Paleolitico medio al Paleolitico superiore in Italia ology and Paleoanthropology Series. Springer, pp. 129e150. meridionale. Atti Tavola Rotonda sul Paleolitico Medio d' Italia. www.archeo- Valladas, H., Mercier, N., Froget, L., Joron, J.-L., Reyss, J.-L., Karkanas, P., editions.com. Panagopoulou, E., Kyparissi-Apostolika, N., 2007. TL age-estimates for the Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Middle Palaeolithic layers at Theopetra cave (Greece). Quatern. Geochronol. 2, Ramsey, C., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., 303e308. Guilderson, T.P., Hajdas, I., Heaton, T.J., Hogg, A.G., Hughen, K.A., Kaiser, K.A., van Strydonck, M., Boudin, M., Hoefkens, M., De Mulder, G., 2005. 14C-dating of Kromer, B., Mccormac, F.G., Manning, S.W., Reimer, R.W., Richards, D.A., cremated bones, why does it work? Lunula 16, 61e63. Southon, J.R., Talamo, S., Turney, C.S.M., Van der Plicht, J., Weyhenmeyer, C.E., van Strydonck, M., Boudin, M., De Mulder, G., 2009. 14C dating of cremated bones: 2009. IntCal09 and Marine09 radiocarbon age calibration curves, 0-50,000 the issue of sample contamination. Radiocarbon 51, 553e568. years cal BP. Radiocarbon 59, 1111e1150. Weninger, B., Jöris, O., 2008. A 14C age calibration curve for the last 60 ka: the Riel-Salvatore, J., 2007. The Uluzzian and the Middle-Upper Paleolithic transition in Greenland-Hulu U/Th timescale and its impact on understanding the Middle southern Italy. Ph.D. Dissertation, Arizona State University. to Upper Paleolithic transition in Western Eurasia. J. Hum. Evol. 55, 772e Riel-Salvatore, J., 2009. What is a ‘transitional’ industry? The Uluzzian of southern Italy 781. as a case study. In: Camps, M., Chauhan, P. (Eds.), Sourcebook of Paleolithic Tran- Wood, R.E., Douka, K., Boscato, P., Haesaerts, P., Sinitsyn, A., Higham, T.F.G., 2012. sitions: Methods, Theories, and Interpretations. Springer, New York, pp. 377e396. Testing the ABOx-SC method: dating known age charcoals associated with the Riel-Salvatore, J., Negrino, F., 2009. Early Upper Paleolithic population dynamics and Campanian Ignimbrite. Quatern. Geochronol. 9, 16e26. raw material procurement patterns in Italy. In: Camps, M., Szmidt, C. (Eds.), The Zilhão, J., d’Errico, F., Bordes, J.-G., Lenoble, A., Texier, J.-P., Rigaud, J.-Ph., 2008. Mediterranean Between 50e25,000 BP: Turning Points and New Directions. Grotte des Fées (Châtelperron): history of research, stratigraphy, dating, and Oxbow, Oxford, pp. 205e224. archaeology of the Châtelperronian type-site. Paleoanthropology 2008, 1e42. Romagnoli, F., 2012. Risorse litiche e comportamento tecnico dei Neandertaliani: Zilhão, J., 2013. Neandertal-Modern Human Contact in Western Eurasia: Issues of variabilità culturale e adattamento all’ambiente nel Salento. Grotta del Cavallo, Dating, Taxonomy, and Cultural Associations. In: Akazawa, T., Nishiaki, Y., strati L-N, e Grotta Mario Bernardini, strato D. Ph.D. dissertation. Università Aoki, K. (Eds.), Dynamics of Learning in Neanderthals and Modern Humans. degli Studi di Firenze/Universitat Rovira i Virgili. Cultural Perspectives, Replacement of Neanderthals by Modern Humans Series, Ronchitelli, A., Boscato, P., Gambassini, P., 2009. Gli ultimi neandertaliani in Italia: vol. 1. Springer, Tokyo, pp. 21e57. aspetti culturali. In: Facchini, F., Belcastro, G.M. (Eds.), La Lunga Storia di Neandertal. Biologia e Comportamento. Jaca Book, Milano, pp. 227e257.