alib

li

W

R

4

fauthidd

argument quantitatively, using a sample of 133 Mousterian, Chatelperronian, and Aurignacian ungulate assemblages.

We show that only five Aurignacian assemblages, from three sites, stand out in terms of the degree to which theirungulate faunas are dominated by a single taxon. We also show that some Mousterian cave and rock shelter ungulateassemblages are more heavily dominated by large bovids than Aurignacian assemblages are dominated by reindeer, andthat Mellars argument is highly dependent on the exclusion of open sites from the analysis and on the numericalthreshold he has selected to indicate hunting specialization. 2002 Published by Elsevier Science Ltd. All rights reserved.

Keywords: MOUSTERIAN, AURIGNACIAN, CHA|TELPERRONIAN, NEANDERTALS, PLEISTOCENE,REINDEER, FRANCE.

Introduction

P aul Mellars 1973 examination of the nature ofthe Middle-to-Upper Palaeolithic transition insouthwestern France has had a substantial im-pact on archaeological approaches to understandingthis complex period of time. By focusing on a small setof apparently relevant attributesfor instance, chang-ing stone tool morphology, the use of bone, antler andivory for tool manufacture, the appearance of personalornaments, and the long distance transport of rawmaterialshe produced a powerful synthesis thatclosely matched what others were concluding from theanalysis of Mousterian and Upper Palaeolithic ma-terials across Europe (e.g., Klein, 1973). His eorts inthis realm helped set the stage for what was to followand remain at the centre of the current debate on thefate of the Neanderthals (e.g., White, 1982; Klein,1989, 1992, 1995, 1999, 2000; Potts, 1998; McBrearty& Brooks, 2000).

Here, we address one of the issues that Mellars

Palaeolithic in at least this region. Although, asMellars (1973) carefully noted, this was not a novelargument (e.g., Braidwood & Reed, 1957; Binford &Binford, 1966; Binford, 1968), he amassed far moredata to support it than had his predecessors.

It is also an argument that Mellars has continued tomake, and to refine, over the years (e.g., Mellars, 1989,1992, 1993, 1996). Recently, for instance, he hassuggested that Chatelperronian Neandertal goupspracticed a relatively broad spectrum foraging pattern,usually involving substantial exploitation of at leastthree or four dierent species . . . by contrast, most ofthe faunas recovered from early Aurignacian levelsin the same region show a striking specialization onreindeer, with reindeer often comprising more than90% of the documented remains (Mellars, 1998: 500).

To be sure, Mellars (1998) recognizes that thedierences he sees in southwestern France betweenMousterian and Chatelperronian faunal assemblageson the one hand, and those provided by Aurignacianones on the other, may simply reflect changing climaticJournal of Archaeological Science (2002) 29, 14391449doi:10.1006/jasc.2002.0806, available online at http://www.ide

Specialized Early Upper PalaeoFrance?

Donald K. Grayson

Department of Anthropology, Box 353100, University of

Francoise Delpech

Institut de Prehistoire et de Geologie du Quaternaire, UMBordeaux I, 33405 Talence, France

(Received 7 November 2001, revised manuscript accepted

Paul Mellars has long used cave and rockshelter ungulatethe early Upper Palaeolithic people of this region focusedspecialized hunting distinguishes the Upper from the Mraised in 1973, that the highly specialized hunting ofone species of animal was particularly characteristic ofthe upper Palaeolithic period (1973: 261), and thatsuch hunting distinguishes the Upper from the Middle

143903054403/02/$-see front matterrary.com on

thic Hunters in Southwestern

ashington, Seattle WA 98195, U.S.A.

5808 du CNRS, Avenue des Facultes, Universite

January 2002)

nal assemblages from southwestern France to argue thateir hunting on reindeer (Rangifer tarandus), and that suchle Palaeolithic in at least this region. We examine thisconditions. However, he has also argued that regard-less of whether certain Mousterian groups practiced asignificant element of deliberate economic specializ-ation in the exploitation of particular animal species, it

2002 Published by Elsevier Science Ltd. All rights reserved.

is clear that these patterns became more sharplydefined, and more widespread, during the earlieststages of the Upper Palaeolithic (Mellars, 1996: 201).Although Mellars (1998) argues that most earlyAurignacian assemblages show such specialization, heroutinely calls on just four sites to make the case(Mellars, 1989, 1996, 1998): Abri Pataud, Roc deCombe, Le Piage, and La Gravette.

Mellars arguments concerning early Upper Palaeo-lithic subsistence specialization in western Europe havenot been widely accepted. While some whose geo-graphic specialities fall outside of Europe have re-peated the argument (e.g., Potts, 1998), the same is notthe case for those familiar with the details of thearchaeological record of the region. Freeman (1971,1973), Straus (e.g., 1977, 1985, 1987, 1992), Clark (e.g.,1987, 1997; Clark & Yi, 1983; Clark & Lindly, 1989),and others have shown that no such phenomenonexists in northern Spain. Chase (1987, 1989), Simek &Snyder (1988), Rigaud (1989, 1993) and Otte (1990)reject the argument for more northerly parts ofEurope, and Stiner (1994, 2001) rejects it for Italy andother parts of Eurasia.

We revisit this issue with the analyses that follow.This we do not simply because Mellars continues tomake the argument, but because we are able to showwith some precision how and why this argument isincorrect in its southwestern French homeland.

The AnalysisOur analysis of Mellars argument is based on a largesample of Mousterian and Aurignacian faunal assem-blages, along with an unfortunately small set ofChatelperronian ones. That sample is provided inTable 1, and includes all Mousterian, Chatelperronian,and Aurignacian assemblages known to us that: (1) arefrom rockshelters and caves in southwestern France,since these are the kinds of sites, and the area, on whichMellars focuses; (2) are associated with a Mousterian,Chatelperronian, or Aurignacian lithic assemblage; (3)have specimen counts available (and not just percent-age values or MNI counts); and, (4) have at least 20identified specimens. In addition, we have excluded allMousterian assemblages that predate oxygen isotopestage 5 (thus, for instance, the faunal assemblages fromGrotte Vaufrey strata IV through VIII are not includedin our sample; see Rigaud, 1988). This process hasprovided us with 30 Aurignacian, 6 Chatelperronian,and 95 Mousterian cave and rock shelter assem-blages (Table 1). Table 1 also includes data for twoMousterian open sites, Mauran and La Borde, forwhich NISP counts are available; these sites arediscussed separately below.

Mellars (1996) has used a 50-specimen minimumlimit for his examination of Mousterian ungulate as-

1440 D. K. Grayson and F. Delpechsemblages, but his analysis of the Aurignacian alsoincludes assemblages for which neither specimen norMNI counts are available (for instance, La Gravette).The 20-specimen limit we employ is both low andarbitrary. As will be seen, however, this does notmatter since, ultimately, no sites with specimen countsthis low play a significant role in our conclusions. Wehave not excluded faunas that lack a detailed tapho-nomic assessment, nor have we excluded assemblagesthat some have suggested may to some extent be mixed(e.g., Rigaud, 1996, 2000). As a result, we have no wayof guaranteeing that aspects of our results do notreflect the impact of either multiple depositionalmechanisms or mixing. Including only assemblagesthat had been vetted in detail in this way would nothave provided a suciently large sample to conductthe analyses that we conduct. Mellars approach isidentical to ours in this regard.

Mellars is quite explicit as to what he means byhunting specialization. To be considered as reflectingspecialized hunting, a faunal assemblage must benumerically dominated by one particular species, withthat species selected from a wider range of taxa avail-able to the hunters (Mellars, 1996: 196). Thus, hisargument requires that the faunal assemblages leftbehind by the people who created Aurignacian assem-blages in southwestern France be dominated by asingle taxon, that other taxa were available but notutilized to the same extent, and that this patterncontrast with that found in Chatelperronian andMousterian assemblages. Even though Mellars speci-fies that the target taxon during the early UpperPalaeolithic was reindeer (Rangifer tarandus), the par-ticular taxon involved is immaterial. It matters onlythat Aurignacian assemblages be dominated by onetaxon, that such dominance not be driven by the factthat little else was available on the landscape, and thatearlier assemblages be dierent in this regard.

Quantitative measures appropriate to assessing theseissues now have a lengthy history in zooarchaeology(Grayson, 1984). Here, we have chosen to focus ouranalyses on the evenness of the ungulate assemblagesinvolved. We measure evenness as pi ln pi /ln S,with S taken to be the number of non-overlapping taxain the assemblage and p the proportion of specimens inthe ith species. This measure, E, can vary from 0 to 1;the lower the value, the less even the assemblage(Magurran, 1988).

We have also calculated Simpsons Index for eachassemblage (D= ni[ni1]/N[N1], where ni=thenumber of specimens in the ith species and N=the totalnumber of specimens). We chose this index because itmore sensitively reflects the dominance of an assem-blage by a single species than does E. The values wereport are the inverse of Simpsons Index: the lower theresultant value, the more the assemblage involved isdominated by a single taxon (Magurran, 1988). Be-cause the values of both of these indices may vary withsample size, we present all of our results as scatter-

grams with NISP as the independent, and the indexvalues as the dependent, variable.

Table 1. Faunal assemblages included in the analysis (NISPdom=NISP of most abundant taxon; NTAXA=number of taxa; E=evenness;1/D=dominance; BB=Bos sp. or Bison sp.; BI=Bison sp.; BO=Bos sp.; CP=Capra sp.; CE=Cervus elaphus; CL=Capreolus capreolus;CS=Cervus simplicidens; EQ=Equus caballus; RT=Rangifer tarandus; TIE=two most-abundant taxa have identical NISP values). The twoopen sites, Mauran and La Borde, are marked by bold letters

Assemblage NISP NISPdom-Taxon NTAXA E 1/D Reference(s)

AurignacianAbri Pataud 7 567 395-RT 9 04651 19376 Bouchud, 1975Abri Pataud 8 20 5-TIE 6 09470 65531 Bouchud, 1975Abri Pataud 11 966 599-RT 6 03982 19150 Bouchud, 1975Abri Pataud 12 174 129-RT 4 05119 16832 Bouchud, 1975Abri Pataud 13 224 221-RT 2 01026 10273 Bouchud, 1975Abri Pataud 14 1495 1481-RT 5 00383 10190 Bouchud, 1975Caminade-Est F 22 9-RT 5 08600 38521 Delpech, 1970Grotte XVI AIB 239 154-RT 8 05816 22497 Grayson et al., 2001Grotte XVI ABB 1051 677-RT 9 05629 22543 Grayson et al., 2001La Chevre 3 287 228-RT 5 04011 15239 Bouchud, 1964La Chevre 4 221 108-BB 6 06781 28273 Bouchud, 1964La Ferrassie F 58 37-CE 10 05979 23923 Delpech et al., 2000La Ferrassie G front 87 32-RT 7 07948 40667 Delpech et al., 2000La Ferrassie H 73 25-RT 8 07807 42499 Delpech et al., 2000La Ferrassie I 120 59-BB 10 06670 32938 Delpech et al., 2000La Ferrassie J 56 29-BB 7 07628 32648 Delpech et al., 2000La Ferrassie K1-3 105 53-BB 8 06689 30921 Delpech et al., 2000La Ferrassie K4 24 12-RT 3 08360 24649 Delpech et al., 2000La Ferrassie K5 186 152-RT 5 03869 14590 Delpech et al., 2000Le Flageolet I VIII 461 240-RT 9 06041 28074 Delpech et al., 2000; Grayson &

Delpech, 1998Le Flageolet I IX 681 468-RT 11 04854 20239 Delpech et al., 2000; Grayson &

Delpech,1998Le Flageolet I XI 651 511-RT 6 04458 15860 Delpech et al., 2000; Grayson &

Delpech, 1998Le Piage F 363 242-RT 4 06321 20020 Beckouche, 1981Le Piage G-I 636 328-CP 6 06612 27663 Beckouche, 1981Le Piage J 705 666-RT 4 01935 11193 Beckouche, 1981Le Piage K 527 458-RT 6 03024 13139 Beckouche, 1981Maldidier 5 31 12-RT 5 09126 43048 Delpech, 1983Roc de Combe 5 1199 1109-RT 7 01801 11652 Delpech, 1983; Delpech et al., 2000Roc de Combe 6 183 153-RT 9 03361 14227 Delpech, 1983; Delpech et al., 2000Roc de Combe 7* 918 873-RT 7 01346 11050 Delpech, 1983; Delpech et al., 2000ChatelperronianArcy s/Cure Renne Xc 2590 2132-RT 7 03384 14510 David & Poulain, 1990Grotte XVI B 187 88-RT 9 06617 32701 Grayson et al., 2001La Ferrassie L3ab 70 38-RT 6 06507 25994 Delpech et al., 2000Le Piage F1 60 44-RT 5 05329 17787 Beckouche, 1981Roc de Combe 8 166 68-RT 6 06678 29206 Delpech, 1983; Delpech et al., 2000St. Cesaire EJOP 469 236-RT 9 06759 31576 Patou-Mathis, 1993MousterianArcy s/Cure Renne XI 1343 857-RT 7 04914 20799 David & Poulain, 1990Bourgeois-Delaunay 8 48 27-EQ 6 06872 26911 Armand, 1998aBourgeois-Delaunay 8 139 78-EQ 8 06469 27360 Armand, 1998aBourgeois-Delaunay 9 348 196-EQ 12 05684 27442 Armand, 1998aBourgeois-Delaunay 9 55 30-EQ 5 07719 28137 Armand, 1998aBourgeois-Delaunay 10 132 90-EQ 7 05271 20109 Armand, 1998aCombe Grenal 6 55 24-RT 6 07802 34771 Guadelli, 1987Combe Grenal 7 54 30-RT 6 06406 25780 Guadelli, 1987Combe Grenal 8 23 9-TIE 4 08744 33278 Guadelli, 1987Combe Grenal 9 34 24-CE 5 05483 18889 Guadelli, 1987Combe Grenal 10 31 11-BI 4 09082 36036 Guadelli, 1987Combe Grenal 11 158 565-BI 8 08410 48544 Guadelli, 1987Combe Grenal 12 85 25-EQ 7 08623 49900 Guadelli, 1987Combe Grenal 13 171 107-EQ 6 06787 23568 Guadelli, 1987Combe Grenal 14 537 375-EQ 8 04852 19264 Guadelli, 1987Combe Grenal 1516 93 56-EQ 6 06655 24600 Guadelli, 1987Combe Grenal 17 148 79-RT 7 05882 25562 Guadelli, 1987Combe Grenal 18 49 26-RT 5 07987 29762 Guadelli, 1987Combe Grenal 19 103 68-RT 4 06888 20825 Guadelli, 1987Combe Grenal 20 192 92-RT 8 06673 31776 Guadelli, 1987

Specialized Early Upper Palaeolithic Hunters in Southwestern France? 1441Combe Grenal 21 270 132-RT 8 06705 31486 Guadelli, 1987

Table 1. Continued

Assemblage NISP NISPdom-Taxon NTAXA E 1/D Reference(s)

Combe Grenal 22 982 642-RT 7 05131 20458 Guadelli, 1987Combe Grenal 23 1101 782-RT 6 04883 18298 Guadelli, 1987Combe Grenal 24 285 195-RT 6 05169 19600 Guadelli, 1987Combe Grenal 25 284 181-RT 6 05512 21358 Guadelli, 1987Combe Grenal 26 105 69-RT 4 07236 21304 Guadelli, 1987Combe Grenal 27 296 163-RT 8 05440 24679 Guadelli, 1987Combe Grenal 28 122 73-RT 5 06126 22065 Guadelli, 1987Combe Grenal 29 224 81-RT 8 07045 35932 Guadelli, 1987Combe Grenal 30 77 36-RT 6 07048 30647 Guadelli, 1987Combe Grenal 31 85 52-RT 7 06488 24643 Guadelli, 1987Combe Grenal 32 153 54-RT 7 07396 38066 Guadelli, 1987Combe Grenal 33 54 31-BI 6 06985 26652 Guadelli, 1987Combe Grenal 34 20 8-CE 6 08285 43178 Guadelli, 1987Combe Grenal 35 331 165-CE 9 06748 32468 Guadelli, 1987Combe Grenal 36 78 43-CS 5 07998 28498 Laquay, 1981; Delpech 1996Combe Grenal 37 56 30-CS 4 07756 25974 Laquay, 1981; Delpech 1996Combe Grenal 38 80 34-CS 7 08414 41859 Laquay, 1981; Delpech 1996Combe Grenal 40/39 126 89-CS 8 05267 19444 Laquay, 1981; Delpech 1996Combe Grenal 41 59 36-CS 5 07367 24722 Laquay, 1981; Delpech 1996Combe Grenal 43/42 149 115-CS 5 04371 15954 Laquay, 1981; Delpech 1996Combe Grenal 45/44 31 17-CS 5 07757 29053 Laquay, 1981; Delpech 1996Combe Grenal 47/46 71 43-CS 5 07063 24079 Laquay, 1981; Delpech 1996Combe Grenal 49/48 60 31-CS 6 07661 31211 Laquay, 1981; Delpech 1996Combe Grenal 50 818 680-CS 9 03009 14241 Laquay, 1981; Delpech 1996Combe Grenal 50A 342 262-CS 8 03975 16410 Laquay, 1981; Delpech 1996Combe Grenal 51 73 47-CS 7 05353 21418 Laquay, 1981; Delpech 1996Combe Grenal 52 854 655-CS 9 04184 16656 Laquay, 1981; Delpech 1996Combe Grenal 53 41 29-CS 5 06096 19478 Laquay, 1981; Delpech 1996Combe Grenal 54 282 209-CS 9 04308 17596 Laquay, 1981; Delpech 1996Grotte XVI C 431 162-RT 8 08173 43384 Delpech unpublishedLa Borde 430 410-BO 4 01604 10988 Jaubert et al. 1990La Quina 2A 56 33-EQ 3 07267 20838 Armand, 1998bLa Quina 2B 437 302-EQ 4 05861 18762 Armand, 1998bLa Quina 4A 74 45-EQ 4 07138 23026 Armand, 1998bLa Quina 4B 885 508-EQ 5 05895 22936 Armand, 1998bLa Quina 6A 972 611-BB 6 05350 21377 Armand, 1998bLa Quina 6B 22 19-BB 3 04416 13430 Armand, 1998bLa Quina 6C 271 225-BB 4 04596 14318 Armand, 1998bLa Quina 6D 151 70-BB 5 07864 31279 Armand, 1998bLa Quina 8 2560 2304-RT 6 02232 12265 Armand, 1998bLes Fieux J 56 37-BI 8 05208 21008 Champagne et al., 1990; J.-L.

Guadelli, pers. comm.Les Fieux K 79 69-BI 4 03680 13057 Champagne et al., 1990; J.-L.

Guadelli, pers. comm.Mauran** 4192 4150-BI 3 00543 10202 Farizy et al., 1994Pech de lAze IB 4 331 201-CE 8 05802 24155 Laparra, 2000Pech de lAze II 2E 66 35-RT 5 06986 25687 Laquay, 1981; Delpech, 1997Pech de lAze II 2G 88 54-CE 7 06756 24931 Laquay, 1981; Delpech, 1997Pech de lAze II 2G 36 17-CE 7 07869 37495 Laquay, 1981; Delpech, 1997Pech de lAze II 3 173 78-CE 11 07504 40128 Laquay, 1981; Delpech, 1997Pech de lAze II 4A 92 69-EQ 5 05242 17194 Laquay, 1981; Delpech, 1997Pech de lAze II 4A2 33 12-CE 6 08286 42248 Laquay, 1981; Delpech, 1997Pech de lAze II 4B4 315 155-CE 10 05368 26434 Laquay, 1981; Delpech, 1997Pech de lAze II 4C1 38 26-CE 5 06087 20321 Laquay, 1981; Delpech, 1997Pech de lAze II 4C2 492 387-CE 7 04057 15770 Laquay, 1981; Delpech, 1997Pech de lAze II 4E 31 10-CE 8 08118 49456 Laquay, 1981; Delpech, 1997Pech de lAze II 5 21 10-EQ 5 08102 33864 Laquay, 1981; Delpech, 1997Pech de lAze IV G 61 57-RT 4 02238 11460 Laquay, 1981Pech de lAze IV H1 57 47-RT 4 04553 14537 Laquay, 1981Pech de lAze IV H2 102 85-RT 5 03961 14229 Laquay, 1981Pech de lAze IV I1 247 216-RT 5 03299 13006 Laquay, 1981Pech de lAze IV I2 539 332-RT 7 05535 22655 Laquay, 1981Pech de lAze IV J1 63 29-RT 4 08981 32175 Laquay, 1981Pech de lAze IV J2 75 31-CE 5 08304 33190 Laquay, 1981Pech de lAze IV J3 1283 659-CE 12 05329 27886 Laquay, 1981

1442 D. K. Grayson and F. Delpech

the remainder to Bison. This procedure is identical to v

A

03994 15244 Laquay, 198105189 15758 Laquay, 198105490 21636 Delpech, 1996

even(D

nnomp

cent

Eathat which we have used in the past, except for the factthat we have not rounded the results to the nearestwhole number. Thus, in Table 1, the abundance ofBison is reported as 565 specimens (15+47[15/17]). Allrelevant data are presented in Table 1.

Evenness and dominanceFigure 1 presents the relationship between evenness (E)and sample size for the Aurignacian assemblages in oursample. The correlation between these two variables isquite significant (r=077, P

0.008.50

1.00

LN NISP

r = 0.62, p < 0.001

Eve

nn

ess

2.50

0.80

0.60

0.40

0.20

4.00 5.50 7.00

PIVGBRD

MAU

LQ8

FIK

LQ6B

XVIC

Figure 2. Mousterian ungulate assemblage evenness values(BRD=La Borde; FI=Les Fieux; LQ=La Quina; MAU=Mauran;

1444 D. K. Grayson and F. Delpech8 not far behind. The faunas of all seven assemblagesare dominated by reindeer. We observe that if the

PIV=Pech de lAze IV; XVI=Grotte XVI; Mauran and La Borde,0.008.50

1.00

LN NISP

Eve

nn

ess

2.50

0.80

0.60

0.40

0.20

4.00 5.50 7.00

PIVG

BRD

LQ6

AP13RC7

AP14

RC5

LPJ

FIK

LQ8

MAU

XVIC

Figure 3. Composite Aurignacian, Chatelperronian, and Moust-erian ungulate assemblage evenness values (=Aurignacian;=Chatelperronian; =Mousterian; see Figures 1 and 2 for site

abbreviations).more than two standard deviations beneath the pre-dicted value (Les Fieux K, La Quina 6B and Pech delAze IV G); one, Grotte XVI C, has an evenness valuesignificantly greater than expected. Mellars argu-ments, however, focus not on deviations between pre-dicted and expected values, but instead on lack ofevenness per se. Two assemblages stand out in thisregard: Pech de lAze IV G and La Quina 8.

Figure 2 also shows the position of two Mousteriansites that do not play a role in Mellars argumentsconcerning the advent of specialized hunting during theUpper Palaeolithic: Mauran, an open-air site markedby a substantial fauna composed almost entirely ofBison priscus (Farizy et al., 1994), and La Borde, whosefauna, excavated from an aven, is dominated by Bosprimigenius (Jaubert et al., 1990). We discuss these sitesseparately below; unless otherwise indicated, they havebeen excluded from the calculation of the compositestatistics we provide.

The Aurignacian and Mousterian data are com-bined, and the Chatelperronian data added, in Figure3. In this figure, we have also identified all assemblageswith evenness values beneath 025. Among the cavesand rock shelters, Abri Pataud 13 and 14, Le Piage J,and Roc de Combe 5 and 7 emerge as the least evenassemblages in our set, with Pech IV G and La Quina

indicated by crosses, were not used to calculate the regressionstatistics).entire set of cave and rockshelter assemblages is treatedas a composite, regression analysis shows Abri Pataud13 and 14, Roc de Combe 7, and Pech IV G to be twostandard deviation outliers in the low evenness group;among more even assemblages, Les Fieux K, La Quina6B and Grotte XVI C emerge as two standard devi-ation outliers.

We dwell at less length on the results of the domi-nance (1/D) analysis. As Figure 4 shows, the sameseven cave and rock shelter assemblages emergeas those most dominated by a single taxon. Beyondthis, and as with the evenness analysis, there isalmost complete overlap between Mousterian,Chatelperronian, and Aurignacian assemblages.

Percentage-based analysesMellars based his arguments for early Upper Palaeo-lithic hunting specialization on the analysis of taxo-nomic relative abundances expressed as percentages,and we follow his lead here. Figure 5 replicates hisapproach with our data set, arraying sample sizeagainst the percentage of each assemblage that isprovided by the most abundant taxon. The sameassemblages emerge here as emerged from theexamination of evenness and dominance.

Finally, Figure 6 provides a histogram of the assem-

blages in our sample according to the degree to whichthey are dominated by a single taxon, using the same

Ea7

Specialized201) has observed that the dominant species inMousterian cave and rock shelter faunal assemblages0

16

% Dominant species

No.

of

asse

mbl

ages

14

12

10

8

6

4

2

Moust

Aurig

212

526

30

313

536

40

414

546

50

515

556

60

616

566

70

717

576

80

818

586

90

919

596

100

Figure 6. The distribution of percentage dominance values by 5%

rly Upper Palaeolithic Hunters in Southwestern France? 1445intervals for Mousterian and Aurignacian ungulate assemblages.5% intervals used by Mellars (1996). Mellars (1996:

204.00

Log NISP

% D

omin

ant

spec

ies

1.00

90

80

50

30

1.50 2.00 3.00

PIVGBRD

RC7

RC5LPJ LQ8

MAU

2.50 3.50

70

40

60

Figure 5. The percentage by which ungulate assemblages are domi-nated by a single taxon (see Figures 1 and 2 for site abbreviations).100AP13 AP14

04.00

Log NISP

1/S

imps

on's

inde

x

1.00

6

5

3

1

1.50 2.00 3.00

PIVG

BRDAP13

RC7AP14

RC5

LPJ

LQ8

MAU

2.50 3.50

4

2

Figure 4. Inverse Simpsons Index values for all ungulate assem-blages (=Aurignacian; =Chatelperronian; =Mousterian; seeFigures 1 and 2 for site abbreviations).usually accounts for 6070% of those assemblages, andour data are in accord with this observation. However,the Aurignacian modal decile value falls in exactly thesame interval. In fact, chi-square analysis of thesedistributions shows that they are not significantlydierent from one another (chi-square=1939, P>020). Analysis of single-cell residuals (Everitt, 1977)shows that Mousterian values dier significantly fromAurignacian ones in two instances: there are moreAurignacian assemblages in the 9195% and 96100%classes than can be accounted for by chance (adjustedresiduals=254 and 243, respectively; P020), but there are no significant (i.e.,P

Aurignacian levels have reindeer often comprisingmore than 90% of the documented remains (Mellars,1998: 500). Clearly, most early Aurignacian faunas donot meet this requirement, but it is nonetheless true thatfive do. Even if we include Mauran and La Borde, thereare significantly more Aurignacian assemblages whosemost abundant taxon comprises more than 90% of thedocumented remains (5/30) than is the case in ourMousterian sample (3/97; chi-square=715, P005; see Table 2).Discussion

Our analyses question two important aspects ofMellars argument, and we discuss each of theseaspects in turn.

1. All the analyses we have presented show that thedierences in cave and rock shelter assemblages tar-geted by Mellars reside in a very small subset ofAurignacian assemblages that have specimen counts:Abri Pataud 13 and 14, Le Piage J, and Roc de Combe5 and 7. It is, in fact, simple to make the distinct natureof these assemblages even more obvious.

Of the 30 Aurignacian assemblages in our sample, 23have reindeer as the most abundant taxon. Of the11,114 specimens in these 23 assemblages, 8846, or796%, are of reindeer (see Tables 1 and 3A). Of the 95Mousterian cave and rock shelter assemblages in oursample, 31 have reindeer as the most abundant taxon.These 31 have a total of 10,714 specimens, of which7334, or 685%, are reindeer. This dierence is highlysignificant (chi-square=35302, P050).

Although most Aurignacian assemblages in oursample are dominated by reindeer, some have red deer(Cervus elaphus), Capra sp., or large bovids (Bos and/orBison) as the most abundant taxon. Mousterian andAurignacian assemblages dominated by the first two ofthese taxa do not dier significantly from one anotherin the relative abundances of the most abundant taxon(for red deer, chi-square=072, P>050; for Capra sp.,chi-square=177, P>010; see Table 3C and D).

This is not the case, however, for assemblages domi-nated by large bovids (Table 3E). Large bovids are themost abundant taxon in 11 Mousterian cave and rockshelter assemblages, compared to four such Aurigna-cian assemblages. The former have significantly morelarge bovids than do the latter (chi-square=1103,P

the

ei

EaRoc de Combe 5 and 7) revealed by our analysis stemsfrom Mellars decision to include only cave and rockshelter faunas in his analysis and from his designationof 91% single-taxon dominance as tagging a special-ized fauna.

Once open sites are included in the sample, verysimple quantitative analysis reveals that there are

same 5% intervals used by Mellars (1996) in his analy-sis. If we divide these assemblages into two dominance-based percentage classes, those at or above 91% andthose beneath this figure, there are more Aurignacianassemblages in the 91100% interval than can beaccounted for by chance (three, to be exact). However,since the 91% figure was inductively selected to empha-

B. Reindeer dominated assemblages:Abri Pataud 13 and 14, Le Piage J,and Roc de Combe 5 and 7 excluded Reindeer NISP All other taxa NISP

Aurignacian (18) 4496 2077 6573Mousterian (31) 7334 3380 10,714Chi-square=001 (P>050)

C. Red deer dominated assemblages Red deer All other taxa NISP

Aurignacian (1) 37 21 58Mousterian (17) 2187 1567 3754Chi-square=072 (P>050)

D. Capra sp. dominated assemblages Capra sp. NISP All other taxa NISP

Aurignacian (1) 328 308 636Mousterian (2) 102 118 220Chi-square=177 (P>010)

E. Large bovid dominated assemblages Large bovid NISP All other taxa NISP

Aurignacian (4) 249 253 502Mousterian (11) 17815 13155 3097Chi-square=1103 (P

species on the landscape in the vicinity of the sites thathe targets as critical to his argument.AcknowledgementsOur sincere thanks to Kristine M. Bovy, Richard G.Klein, Jean-Philippe Rigaud, and Lawrence G. Strausfor help along the way, including insightful commentson an earlier version of this manuscript. The data fromGrotte XVI reported in this manuscript result fromwork supported by the National Science Foundation(SBR98-04692) and the Sous Direction delArcheologie et Direction des Musees de France(Ministere de la Culture).

ReferencesArmand, D. (1998a). La faune de la grotte Bourgeois-Delaunay,

Commune de La Chaise de Vouthon (Charente): Resultats pre-liminaires. Paleo 10, 7786.

Armand, D. (1998b). Sur la presence dEquus caballus gallicus dansles niveaux superieurs de la station Amont de La Quina(Charente). Quaternaire 9, 345353.

Beckouche, S. (1981). Les grands mammife`res du Paleolithiquesuperieur du Piage (Lot). In (F. Champagne & R. Espitalie, Eds)Le Piage: Site prehistorique du Lot. Memoires de la SocietePrehistorique Francaise 15, pp. 165191.

Binford, L. R. & Binford, S. R. (1966). The predatory revolution: Aconsideration of the evidence for a new subsistence level. AmericanAnthropologist 68, 508512.

Binford, S. R. (1968). Early Upper Pleistocene adaptations in theLevant. American Anthropologist 70, 707717.We do not address this issue here. We see no reasonto, since we see no compelling evidence in Mellarsanalysis or in the data that we have amassed to suggestthat specialized hunting was being practiced to anygreater extent during the Aurignacian than it wasduring the Mousterian. In concluding this, of course,we are saying no more than what many other analystshave already concluded (see the discussion above). Wehope, however, that the approach we have taken hereclarifies how and why Mellars has been led astray.arguments that equate specialized hunting with taxo-nomic abundances of 91% or more, even this weakpositive result does not provide much support forMellars position. As we have noted, if 81% is selectedas the threshold and open sites are included in theanalysis, no significant dierences emerge between theAurignacian and Mousterian in terms of taxonomicdominance.

Mellars (1996) very correctly observes that the termspecialized hunting implies, among other things, theselection of a subset of taxa as a hunting target from abroader array of potential prey taxa on the landscape.Mellars, we note, has not shown that during the timeshis specialized faunas accumulated, there was abroader array of acceptable, yet relatively ignored,

1448 D. K. Grayson and F. DelpechBouchud, J. (1964). Etude sommaire de la faune de gisement de laChe`vre. In (R. Arambourou & P. E. Jude, Eds) Le Gisement de laChe`vre a` Bourdeilles (Dordogne). Perigueux: Magne, pp. 115120.Bouchud, J. (1975). Etude de la faune de lAbri Pataud. In (H. L.Movius, Jr, Ed.) Excavation of the Abri Pataud, Les Eyzies(Dordogne). American School of Prehistoric Research Bulletin 30,pp. 69153.

Braidwood, R. J. & Reed, C. A. (1957). The achievement andconsequences of food-production: A consideration of the archaeo-logical and natural-historical evidence. Cold Spring Harbor Sym-posia on Quantitative Biology 22, 1931.

Champagne, F., Champagne, C., Jauzon, F. & Novel, P. (1990). Lesite prehistorique des Fieux a` Miers (Lot). Gallia Prehistoire 32,128.

Chase, P. G. (1987). Specialisation de la chasse et transition vers lePaleolithique superieur. LAnthropologie 91, 175178.

Chase, P. G. (1989). How dierent was Middle Paleolithic subsist-ence? A zooarchaeological perspective on the Middle to UpperPaleolithic transition. In (P. Mellars & C. Stringer, Eds) TheHuman Revolution: Behavioral and Biological Perspectives on theOrigins of Modern Humans. Princeton: Princeton University Press,pp. 321337.

Clark, G. A. (1987). From the Mousterian to the Metal ages. In (O.Soer, Ed.) The Pleistocene Old World: Regional Perspectives.New York: Plenum Press, pp. 293316.

Clark, G. A. (1997). The Middle-Upper Paleolithic Transition inEurope: An American Perspective. Norwegian ArchaeologicalReview 30, 2553.

Clark, G. A. & Lindly, J. M. (1989). The case of continuity:Observations on the biocultural transition in Europe and westernAsia. In (P. Mellars & C. Stringer, Eds) The Human Revolution:Behavioral and Biological Perspectives on the Origins of ModernHumans. Princeton: Princeton University Press, pp. 626676.

Clark, G. A. & Yi, S. (1983). Niche-width variation in Cantabrianarchaeofaunas: A diachronic study. In (A. R. Hands & D. R.Walker, Eds) Animals and Archaeology: 1. Hunters and their Prey.BAR International Series S163, pp. 183206.

David, F. & Poulain, T. (1990). La faune de grands mammife`res desniveaux XI and Xc de la Grotte du Renne a` Arcy-sur-Cure(Yonne). In (C. Farizy, Ed.) Paleolithique moyen recent et Paleo-lithique superieur ancien en Europe. Memoires du Musee de Pre-histoire dIle-de-France 3, pp. 319323.

Delpech, F. (1970). Faune aurignacienne de Caminade-Est, com-mune de La Caneda (Dordogne). Quaternaria 13, 137140.

Delpech, F. (1983). Les faunes du Paleolithique superieur dans leSud-Ouest de la France. Cahiers du Quaternaire 6.

Delpech, F. (1988). Les grands mammife`res de la grotte Vaufrey, a`lexception des Ursides. In (J.-Ph. Rigaud, Ed.) La Grotte Vaufreya` Cenac et Saint-Julien (Dordogne): Paleoenvironnements, chro-nologie, et activites humaines. Memoires de la Societe PrehistoriqueFrancaise 19, pp. 213289.

Delpech, F. (1996). Lenvironnement animal des Mousteriens Quinadu Perigord. Paleo 8, 3146.

Delpech, F. (1997). Rapport 1997 sur les travaux de paleontologieeectues sur les faunes du Pech-de-lAze IIb (commune de Carsac,Dordogne). Unpublished manuscript on file at the Institut dePrehistoire et de Geologie du Quaternaire, Universite Bordeaux I.

Delpech, F., Grayson, D. K. & Rigaud, J.-Ph. (2000). Biostratigra-phie et paleoenvironnements du debut du Wurm recent dapre`s lesgrands mammife`res de labri du Flageolet I (Dordogne, France).Paleo 12, 97126.

Everitt, B. S. (1977). The Analysis of Contingency Tables. New York:Wiley and Sons.

Farizy, C., David, F., Jaubert, J., Eisenmann, V., Girard, M., Grun,R., Krier, V., Leclerc, J., Miskovsky, J.-C. & Simmonet, R. (1994).Hommes et bisons du Paleolithique moyen a` Mauran. GalliaPrehistoire, Supplement 30, 1269.

Ferrier, J.-G. (2001). La faune des niveaux Paleolithique moyen deSaint-Cesaire. Paleontologie et observations archeozologiques.Memoire de DEA dAnthropologie, option Paleontologie, Univer-site Bordeaux I.Freeman, L. G. (1971). Significado ecologico de les restos deanimales. In (J. Gonzalez Echegaray, L. G. Freeman, K. W.

Butzer, A. Leroi-Gourhan, J. Altuna, B. Madariaga & J. M.Apellaniz) Cueva Morin: Excavaciones 19661968. Santander:Patronato de las Cuevas Prehistoricas, pp. 419437.

Freeman, L. G. (1973). The significance of mammalian faunas fromPaleolithic occupations in Cantabrian Spain. American Antiquity38, 344.

Grayson, D. K. (1984). Quantitative Zooarchaeology. New York:

lations in Europe. In (T. Akazawa, K. Aoki & O. Bar-Yosef, Eds)Neandertals and Modern Humans in Western Asia. New York:Plenum Press, pp. 493507.

Otte, M. (1990). Les processus de transition du Paleolithique moyenau superieur. In (C. Farizy, Ed.) Paleolithique moyen recent etPaleolithique superieur ancien en Europe. Memoires de Musee dePrehistoire dIle-de-France 3, pp. 145149.

Specialized Early Upper Palaeolithic Hunters in Southwestern France? 1449Academic Press.Grayson, D. K. & Delpech, F. (1998). Changing diet breadth in

the early Upper Paleolithic of southwestern France. Journal ofArchaeological Science 25, 11191130.

Grayson, D. K., Delpech, F., Rigaud, J-Ph. & Simek, J. (2001).Explaining the development of dietary dominance by a singleungulate taxon at Grotte XVI, Dordogne, France. Journal ofArchaeological Science 28, 115125.

Guadelli, J.-L. (1987). Contribution a letude des zoocenoses prehis-toriques en Aquitaine (Wurm ancien et interstade wurmien). The`sede lUniversite Bordeaux I, No. 148.

Jaubert, J., Lorblanchet, M., Laville, H., Slott-Moller, R., Turq, A.& Brugal, J.-Ph. (1990). Les chasseurs daurochs de La Borde.Documents dArcheologie francaise 27. Paris: La Maison desSciences de lHommes.

Klein, R. G. (1973). Ice-Age Hunters of the Ukraine. Chicago:University of Chicago Press.

Klein, R. G. (1989). The Human Career. Chicago: University ofChicago Press.

Klein, R. G. (1992). The archaeology of modern human origins.Evolutionary Anthropology 1(1), 514.

Klein, R. G. (1995). Anatomy, behavior, and modern human origins.Journal of World Prehistory 9, 167198.

Klein, R. G. (1999). The Human Career. Second edition. Chicago:University of Chicago Press.

Klein, R. G. (2000). Archaeology and the evolution of humanbehavior. Evolutionary Anthropology 9(1), 1736.

Laparra, C. (2000). Etude paleontologique, taphonomique et archeo-zoologique de la couche 4 de Pech de lAze Ib (Dordogne). DEAdAnthropologie biologique. Universite Bordeaux I.

Laquay, G. (1981). Recherches sur les faunes de Wurm I en Perigord.The`se de Doctorat de troisie`me cycle, Universite de Bordeaux I,No. 1596.

Magurran, A. E. (1988). Ecological Diversity and its Measurement.Princeton: Princeton University Press.

McBrearty, S. & Brooks, A. S. (2000). The revolution that wasnt: Anew interpretation of the origin of modern human behavior.Journal of Human Evolution 39, 453563.

Mellars, P. A. (1973). The character of the Middle-Upper Paleolithictransition in south-west France. In (C. Renfrew, Ed.) The Expla-nation of Culture Change: Models in Prehistory. London:Duckworth, pp. 255276.

Mellars, P. (1989). Major issues in the emergence of modern humans.Current Anthropology 30, 349385.

Mellars, P. A. (1992). Archaeology and modern human origins inEurope. Proceedings of the British Academy 82, 135.

Mellars, P. A. (1993). Archaeology and the population-dispersalhypothesis of modern human origins in Europe. In (M. J. Aitken,C. B. Stringer & P. A. Mellars, Eds) The Origin of Modern Humansand the Impact of Chronometric Dating. Princeton: PrincetonUniversity Press, pp. 196216.

Mellars, P. (1996). The Neanderthal Legacy. Princeton: PrincetonUniversity Press.

Mellars, P. (1998). The impact of climatic changes on the demogra-phy of later Neandertal and early anatomically modern popu-Patou-Mathis, M. (1993). A taphonomic and paleoethnographicstudy of the fauna associated with the Neanderthal of Saint-Cesaire. In (F. Leveque, A. M. Backer & M. Guilbaud, Eds)Context of a Late Neandertal. Monographs in World Archaeology16. Madison: Prehistory Press, pp. 79102.

Potts, R. (1998). Environmental hypotheses of hominin evolution.Yearbook of Physical Anthropology 41, 93136.

Rigaud, J.-Ph. (Ed.) (1988). La Grotte Vaufrey a` Cenac et Saint-Julien (Dordogne): Paleoenvironnements, chronologie, et activiteshumaines. Memoires de la Societe Prehistorique Francaise 19.

Rigaud, J.-Ph. (1989). From the Middle to the Upper Paleolithic:Transition or convergence? In (E. Trinkaus, Ed.) The Emer-gence of Modern Humans: Biocultural Adaptations in the LaterPleistocene. Cambridge: Cambridge University Press, pp. 142153.

Rigaud, J.-Ph. (1993). Passages et transitions du Paleolithique moyenau Paleolithique superieur. In (V. Cabrera Valdes, Ed.) El origendel hombre moderno en el suroeste de Europa. Madrid: UNED, pp.117126.

Rigaud, J.-Ph. (1996). Lemergence du Paleolithique superieur enEurope occidentale. Le role du Castelperronien. In (O. Bar-Yosef,L. L. Cavalli-Sforza, R. J. March & M. Piperno, Eds) The Originof Modern Humans. Colloquia of the 13th International Congressof Prehistoric and Protohistoric Sciences X 5, pp. 219223.

Rigaud, J.-Ph. (2000). Human adaptation to the climatic deterior-ation of the last Pleniglacial in southwestern France. In (W.Roebroeks, M. Mussi, J. Svoboda & K. Fennema, Eds) Hunters ofthe Golden Age: The Mid Upper Paleolithic of Eurasia, 30,00020,000 BP. Leiden: University of Leiden, pp. 325336.

Simek, J. F. & Snyder, L. M. (1988). Changing assemblage diversityin Perigord archaeofaunas. In (H. L. Dibble & A. Montet-White, Eds) Upper Pleistocene Prehistory of Western Eurasia.Philadelphia: University of Pennsylvania Museum, pp. 321332.

Stiner, M. C. (1994). Honor among Thieves: A ZooarchaeologicalStudy of Neandertal Ecology. Princeton: Princeton UniversityPress.

Stiner, M. C. (2001). Thirty years on the Broad Spectrum Revol-ution and Paleolithic demography. Proceedings of the NationalAcademy of Sciences 98, 69936996.

Straus, L. G. (1977). Of deerslayers and mountain men: Paleolithicfaunal exploitation in Cantabrian Spain. In (L. R. Binford, Ed.)For Theory Building in Archaeology. New York: Academic Press,pp. 4176.

Straus, L. G. (1985). Stone Age prehistory in northern Spain. Science230, 501507.

Straus, L. G. (1987). Hunting in late Upper Paleolithic westernEurope. In (M. H. Nitecki & D. V. Nitecki, Eds) The Evolution ofHunting. New York: Plenum Press, pp. 147176.

Straus, L. G. (1992). Iberia before the Iberians: The Stone AgePrehistory of Cantabrian Spain. Albuquerque: University of NewMexico Press.

White, R. (1982). Rethinking the Middle/Upper Paleolithic tran-sition. Current Anthropology 23, 169192.

Specialized Early Upper Palaeolithic Hunters in Southwestern France?IntroductionThe AnalysisTable 1Table 1. ContinuedTable 1. ContinuedEvenness and dominanceFigure 1Figure 2Figure 3Percentage-based analysesFigure 4Figure 5Figure 6Table 2

DiscussionTable 3

AcknowledgementsReferences