recovery of mammalian dna from middle paleolithic stone tools

Journal of Archaeological Science (1997) 24, 601–611

Recovery of Mammalian DNA from Middle PaleolithicStone Tools

Bruce L. Hardy and Rudolf A. Raff

Indiana Institute for Molecular and Cellular Biology and Department of Biology, Indiana University,Bloomington, IN 47405, U.S.A.

Venu Raman

Department of Oncology, Johns Hopkins School of Medicine, 370 Ross Building, 720 Rutland Avenue, Baltimore,MD 21205, U.S.A.

(Received 17 October 1995, revised manuscript accepted 24 April 1996)

One of the primary goals of archaeology is to understand past human behaviour. Although stone tools comprise thevast majority of cultural artefacts for most of the archaeological record, their uses in prehistory are poorly understood.The application of the polymerase chain reaction (PCR) to amplify DNA molecules can help establish a physical linkto ancient tool use in processing biological material. Modern experimental stone tools and a sample of stone tools andsoils from the Middle Paleolithic site of La Quina, France, dating between approximately 35 and 65 ka, have beenexamined for the presence of ancient DNA. Extractions from the samples were analysed using PCR with primersamplifying a small region of the vertebrate mitochondrial cytochrome b gene. Subsequent sequence analysis allowed theidentification of some DNAs amplified to family or species of origin. DNA sequences were obtained from tools andassociated soil samples. DNA from boar/pig (Sus scrofa), a species represented osteologically at the site, was recoveredfrom one of the tools. Other tools yielded artiodactyl, human, and rabbit-like sequences.? 1997 Academic Press Limited

Keywords: ANCIENT DNA, BIOMOLECULAR ARCHAEOLOGY, CYTOCHROME B, POLYMERASECHAIN REACTION, STONE TOOL FUNCTION.

Introduction

B ecause stone tools are the most abundant cul-tural remains for the majority of prehistory,archaeologists rely on them to interpret past

human behaviour. Archaeologists have attempted toinfer the various aspects of ancient tool use throughethnographic analogy, experimental replication anduse of tools, and use-wear analysis. Ethnographic andethnohistoric accounts have been used to find ana-logues of prehistoric tool function. However, ethno-graphic observations of tool use cannot be generalizedto all tools of similar morphology (Clark & Kurashina,1981) because one tool morphology can have differentuses, or one task can be performed by different toolmorphologies. Experimental replication has been usedto understand the technological processes of toolmanufacture (Crabtree, 1973; Jones, 1980; Toth, 1985)as well as to experiment with the possible range ofuse-materials and use-actions of a particular tool type(Jones, 1980; Keeley, 1980; Toth, 1985; Anderson-Gerfaud, 1990; Hurcombe, 1992; Shea, 1992). Suchexperiments can only define plausibility.

6010305-4403/97/070601+11 $25.00/0/as960144

As a stone tool is used, its surface is altered throughcontact with the worked material. Traces of use includeedge damage in the form of microflake scars, striations,and changes in the texture and reflectivity of the toolsurface (Keeley, 1980; Odell & Odell-Vereecken, 1980;Shea, 1992). Use-wear studies with low-power magni-fication (#10–100) can yield functional informationthrough edge damage and striation patterns. The ori-entation of microflake scars and striations can helpdetermine how a tool was used. For example, striationpatterns can reveal that a tool was used for cutting andmay indicate in which direction a tool was movedduring use (Odell & Odell-Vereecken, 1980). However,edge damage is not a completely reliable predictor ofworked material as different worked materials mayproduce similar damage patterns (Moss, 1983).The study of microwear polishes at high magnifica-

tion (#100–500) held promise for identifying thespecific materials on which tools were used becausedifferent use-materials (bone, wood, antler, dry hide,wet hide, meat, etc.) formed distinct polishes on stonetool surfaces which could be readily identified (Keeley,1980). However, it has become clear that the formation

? 1997 Academic Press Limited

602 B. L. Hardy et al.

of microwear polishes is a complicated process which isnot easy to interpret (Newcomer et al., 1986; Juel-Jensen, 1988). The formation of the polish is related toboth the use-material, the composition of that materialand to the stone being used. The amount of silicapresent in either the stone or the worked materialaffects the rate and amount of polish formation(Fullagar, 1991). Interpretation is problematic whenpolish is weak or poorly developed, leading to similarpolishes from different materials. It is also possible tohave similar polishes from different use-materials if theuse-duration varies. Polish on a tool used for a shorttime on antler may be indistinguishable from polishused for a longer time on wood. Identification ofspecific use-material can therefore be difficult. Furtherwork investigating the mechanisms of polish formationmay help eliminate these potential ambiguities.Physical and chemical traces of use-material can

survive on a stone tool surface and have yielded evenmore direct functional information. Microscopy hasidentified fragmentary remains of the worked materialon stone tools, including blood, wood, variousplant tissues, feathers, and hair (Briuer, 1976; Loy,1983, 1985, 1987, 1993; Anderson-Gerfaud, 1990;Hurcombe, 1992; Loy & Hardy, 1992; Loy et al.,1992). Microscopic observations of the patterning ofresidue on a tool can provide information about use-action (Anderson-Gerfaud, 1990; Loy & Hardy, 1992).Various chemical techniques have been used to test forthe presence of residues. The most widely usedmethods are immunological, which have been reportedto identify the taxon with which the tool is associated(Kooyman et al., 1992; Newman et al., 1993). How-ever, the validity of immunological testing of residueshas recently been questioned. Short term experiments

in blood residue preservation have failed to replicateearlier results and question the survivability of proteinin a form which is immunologically recognizable(Manning, 1994; Eisele et al., 1995).The application of molecular biology techniques,

particularly the polymerase chain reaction (PCR), toarchaeological and paleontological materials hasshown that DNA can survive for thousands or evenmillions of years (Paabo, 1989; Cano et al., 1993;Hagelberg et al., 1994). The use of PCR, which iscapable of producing millions of copies from a singlemolecule of DNA, requires strict precautions againstcontaminating DNA (Cano et al., 1993; Soltis & Soltis,1993; Handt et al., 1994). Due to the extreme sensi-tivity of PCR and the small amounts of ancient DNAtemplate, contamination with modern DNA may occurduring excavation, in transport, or in the laboratory(Soltis & Soltis, 1993; Handt et al., 1994). Although itis sometimes possible to distinguish between the con-taminant and ancient DNA, methods to reducemodern contamination must be used.This paper reports the isolation and characterization

of DNA from prehistoric stone tools and soils from theMiddle Paleolithic site of La Quina, Charentes,France. La Quina is a rock shelter site located insouth-west France with archaeological deposits rang-ing in age from approximately 35,000 to 65,000 years.The deposits are characterized by large numbers ofhighly fragmented and well-preserved animal bones(Table 1), Neanderthal skeletal remains, andMousterian stone tools. La Quina has been excavatedsince the early part of the century by Dr H. Henri-Martin and his daughter, G. Henri-Martin, and orig-inally included approximately 100 m of Mousteriandeposits 7 m deep. The early excavations yielded the

Table 1. Mammalian species represented osteologically at La Quina and species whose cytochrome b sequences are in the database

Major mammalian familiesSpecies found at

La QuinaRelated species in

cytochrome b databaseFamily Common name

Bovidae AurochsBison

Bos primigeniusBison priscus

Bos taurus

Equidae Horse Equus caballusEquus germanicusEquus hydruntinus

Equus caballusEquus greyvi

Rhinocerotidae Rhinoceros * Diceros bicornisCervidae Reindeer

Red deerRoe deer

Rangifer tarandusCervus elaphus

Capreolus capreolus

Rangifer tarandusCervus nippon

Odocoileus hemionusTragulidae Pronghorn antelope * Antilocapra americanaCapridae Sheep * Ovis ariesCanidae Wolf

FoxCanis lupus

Alopes lagopusVulpes vulpes

Canis simensisVulpes velox

Suidae Boar/pig Sus scrofa Sus scrofaHyaenidae Hyaena Crocuta crocuta spelea †Lagomorpha Rabbit * Oryctolagus cuniculusCastoridae Beaver Castor sp. Sciurus nigerElephantidae African elephant * Loxodanta africana

(*) Species not represented osteologically at La Quina, but of potential archaeological interest.(†) Species represented osteologically at La Quina but no sequence available (family not yet represented in the DNA database).

Recovery of Mammalian DNA from Middle Paleolithic Stone Tools 603

fragmentary remains of over 25 hominids. In 1986, A.Debenath and A. J. Jelinek began a new series ofexcavations (Debenath & Jelinek, 1990). La Quina waschosen because it was under excavation when theproject began and excavation techniques could bemodified to reduce the possibility of contaminationwith DNA from modern sources. The efficacy of thetechniques was tested on modern experimental stonetools prior to analysis of the archaeological materials.

MethodsArchaeological samples

Stone tool and soil samples were excavated undercontrolled conditions in order to reduce the possibilityof modern DNA contamination to a minimum. Exca-vators wore new disposable surgical gloves when ex-posing the artefacts. As soon as a stone tool destinedfor DNA analysis was uncovered and its positionrecorded, it was placed in a new, clean, self-sealingplastic bag. Soil samples (approximately 3–5 g) weretaken 2–3 cm away from each tool and were handled inthe same manner (Hardy, 1994). The tools are drawnfrom archaeological layers 6A (Mousterian ofAcheulean Tradition), 8 (Typical Mousterian), and M2(Quina Mousterian). Layers 6A and 8 are approxi-mately 35,000 while level M2 is approximately60,000–65,000 (Debenath & Jelinek, 1990). All ofthe tools are found in close association with fragmen-tary animal bones and other stone tools. To date, eightarchaeological samples have been examined (see Figure1). Seven derive from the upper levels (6A and 8) whileone comes from the older level (M2).

Modern experimental toolsA series of modern experimental tools were manufac-tured and used to cut animal tissue. The tools wereallowed to dry and some were buried at a depth ofapproximately 6 inches for up to 3 years. Theanimal tissue was obtained from the ZooarchaeologyLaboratory, Department of Anthropology, IndianaUniversity.

DNA extractionDNA was extracted by soaking the samples in approxi-mately 100 ml of 4 guanidine hydrochloride (HCl).Since the extractions were performed by Hardy in thefield laboratory at La Quina, guanidine HCl waschosen due to its efficacy as a solvent for DNA and dueto the fact that the extracts required no refrigeration.Once transported to the United States, 1–3 ml aliquotsof tool and soil extract were placed in dialysis tubingand dialysed overnight in 1# Tris–EDTA buffer toremove substances which might interfere with the

PCR. The dialysed samples were then subjected toPCR analysis without further treatment.

AmplificationPCR was performed using primers L15684 (5* CTC-CACACATCCAAACAACG 3*) and H15760 (5*TGTTCGACTGGTTGTCCTCC 3*), for a portion ofcytochrome b, which amplify a fragment 116 bp long(including the primers) in mammals between positions15664 and 15780. PCR reactions were performed in25 ìl volumes using 250 ì dNTPs, 10 ì concen-trations of primers, and 0·125 units of Taq polymerasein each reaction. Next, 2 ìl of template were added andthe reaction was carried out in a MJC PTC-100thermocycler (30 cycles, 92)–1 min, 55)–1 min, 72)–1min). A negative (no DNA) control reaction wasincluded in each set of PCR reactions. PCR productswere visualized on 1–1·5% agarose gels or 4% Meta-phor agarose gels stained with ethidium bromide. Theexpected product size was 116 bp.

Cloning and sequencing of PCR productsPCR products were cloned into Invitrogen TA cloningvector pCR= II and cleaned with QIAGEN plasmidkit (QIAGEN, Inc.) prior to double-stranded sequenc-ing with 35S–dATP using the SEQUENASE (U.S.Biochemicals) protocol. Sequencing was also per-formed using Licor automated sequencing protocols.Because we suspected that multiple DNAs might bepresent in a single sample, we cloned the DNA prior tosequencing in order to increase our chances of findingrare DNAs in a sample. Furthermore, direct sequenc-ing of a PCR product containing two or more DNAswould result in multiple banding and ambiguities atsome sites in the sequence data. In order to avoidpossible sequence errors in the cloning, we sequencedmultiple times from different clonings. Phylogeneticanalysis of sequences was conducted using Phylo-genetic Analysis Using Parsimony (PAUP) Version 3.1(Swofford, 1993).

Cytochrome b sequence databaseDNA sequences obtained from cloned PCR productswere compared with a database of known sequence forspecies identification. The database of cytochrome bsequence currently contains more than 100 differentextant vertebrate species for comparison. The databaseis expanded as new sequences are deposited in Gen-bank, or when we sequence new species of archaeologi-cal importance (performed in laboratories separatefrom the archaeological analysis). The large size of thedatabase increases the chances that a DNA will matcha known species or that a closely related species will bepresent. Furthermore, the large database aided in theidentification of potentially chimeric molecules caused


by PCR jumping. Phylogenetic computer programsallow for identification of unknown sequences.

Precautions against contamination by modern DNADue to the extreme sensitivity of PCR, contaminationwith modern DNA may occur in the choice of site,during excavation, in transport, or in the laboratory.Some ancient DNA studies suffer from a lack ofstringency in the handling of specimens resulting in thedetection of DNA of modern origin (e.g. Golenberg etal., 1990). While it is sometimes possible to distinguish

between the contaminant and ancient DNA (Handtet al., 1994), methods to reduce modern contaminationmust be used.Precautions against contamination included: (a)

choice of sites in primary, undisturbed context;(b) excavation with new disposable surgical gloves;(c) immediately placing and sealing excavated speci-mens in new self-sealing plastic bags; (d) use of brusheswith synthetic bristles (if using brushes for excavation);(e) sampling of surrounding soils and non-artefactualremains; (f ) work in a laboratory where related mod-ern DNAs have not been previously studied (to prevent

(h)

(g)

(f)

(e)

(b)

(a)

(c)

(d) 1 cm

Figure 1. Illustrations of La Quina tools used in this study. (a) QM5–4881, (b) QM5–5008, (c) QN5–2708, (d) QN5–2972, (e) QM5–4950,(f ) QM5–5004, (g) QM5–5093, (h) QE5–507.


cross-over contamination); (g) use of positive displace-ment pipettes dedicated to ancient DNA work (UVirradiated periodically); (h) UV irradiating solutions toensure that they are DNA-free; (i) aerosol resistant orpositive displacement tips (to prevent aerosoling ofDNA between reactions); (j) negative (no DNA) reac-tions to detect contaminating DNA; (k) running PCRreactions with different PCR reagents and machines tocheck for possible cross-over from earlier PCR prod-ucts; and (l) limitation of modern control DNAs tospecies which do not occur at the site.

ResultsDNA was successfully extracted and amplified frommodern experimental tools and tissue as well as ar-chaeological stone tools and soil samples. To date,eight stone tools from La Quina and their correspond-ing soil samples have been analysed and five haveyielded DNA. If a DNA sequence was recovered froma tool, but not in the soil sample taken from a few cmaway, it supports the argument that the DNA from thetool is related to its use. In all PCR reactions, anegative (no DNA) reaction was included to check formodern laboratory contamination.In this paper, we report the recovery of animal

DNA. This is not because we assume that the stonetools were used exclusively on animals instead ofplants. We decided to concentrate initially on animalDNA for the following reasons: (1) the range of animalspecies present at the site was known while the range ofplant species was not; (2) a gene (cytochrome b) with alarge database of known sequences was available foranimals; and (3) the primers for this gene wouldamplify a wide range of species. We chose to look atthe cytochrome b gene of animals for several reasons:(a) because it is mitochondrial, the gene would bepresent in many more copies than a nuclear gene, thusincreasing the chance of its survival and subsequentdetection; (b) a large database of known sequence wasavailable for the gene (e.g. Irwin et al., 1991) (pertinentspecies are listed in Table 1); (c) the small size of thefragment is consistent with previous observations ofthe size of ancient DNA fragments (<500 bp); and (d)the primers targeted highly conserved regions sur-rounding a variable region which would allow speciesidentification from sequence data. In the future, wehope to use a similar gene (such as the rbcL gene) toexamine plant DNAs from prehistoric stone tools.

Modern controlsDNA from modern controls was first amplified andsequenced in order to test the reliability of our tech-nique and to confirm that we could identify speciesbased on sequence data (see Figure 2). Four controlDNAs were used: (1) human; (2) pronghorn antelope(Antilocapra americana); (3) wallaby (Macropus rufog-riseus); and (4) raccoon (Procyon lotor). Except for

human, which was unavoidable in the laboratory, thesecontrol species were chosen for their availability andbecause they do not occur osteologically at La Quina.Contamination between reactions by the controlDNAs could therefore be easily recognized. We suc-cessfully matched the published sequences of humanand pronghorn antelope to DNA that we amplifiedfrom modern tissue. Modern stone tools were used tocut wallaby tissue and were then dried. DNA wasextracted from the surfaces of two experimental toolsand yielded the same sequence as that obtained directlyfrom wallaby tissue. Several modern stone tools wereused to cut raccoon tissue and allowed to dry for 1 h.The tools were then buried behind the CRAFT HumanOrigins Research Center in Bloomington, Indiana, at adepth of approximately 6 inches. The tools remainedburied for 3 years prior to extraction, during whichtime they were subjected to the precipitation andtemperature fluctuations of the area. Mean averageannual temperature ranges from a high of 29·6)C inJuly to a low of "9·3)C in January with an averageannual precipitation of 1351 mm. DNA was success-fully amplified from three of four tools tested. DNAsequence obtained from one buried tool showed one bpdifference from the fresh modern tissue (Figure 2). Thebp difference may be due to the fact that the moderntissue and the tissue cut by the tools were from twodifferent individuals. It could also be an error resultingfrom PCR amplification, possibly due to a damagedsite in the DNA. Two soil samples collected 2–3 cmaway from the raccoon tools failed to yield DNA.

Eliminating contaminationAs stated previously, a negative (no-DNA) control wasincluded in each set of PCR reactions. Occasionally, aband of the appropriate size was observed in theno-DNA lane. These contaminated negative reactionswere cloned and sequenced and proved to be humanand wallaby DNA. The wallaby contamination waseventually traced back to a 100 m stock of dNTPs.The contamination of the stock most likely came froman aerosol from a pipettor. The human DNA mostlikely derives from individuals in the laboratory. Theuse of new reagents eliminated this problem.

Avoiding inhibitionSeveral modern experimental tools used to cut raccoontissue which were buried for several years appeared tocontain an inhibitory substance. Extracts from themdid not amplify with either one or two rounds of PCR.Furthermore, addition of a small aliquot of one ofthese samples into a positive control (wallaby tissueDNA), inhibited amplification. It was necessary todilute the samples 1:100 in water in order for amplifi-cation to occur. At this dilution, the amount of inhibi-tory substance was insufficient to interfere with thereaction. The extraction of the La Quina tools in a


Wallaby tissue

Modern controls

Wallaby tool

Pronghorn

Pronghorn 2

Raccoon tissue

Raccoon tool

Pig (boar)

Archaeological samples

Sequence #1

Sheep

Sequence #2

Rabbit

Sequence #3

Human

Sequence #4

(3)

(2)

(2)

(4)

(9)

(4)

(4)

# Clones

Sequenced

Figure 2. Comparison of cytochrome b sequences from controls and archaeological samples used in this study.


large volume of guanidine HCl (100 ml) may be re-sponsible for the lack of inhibition observed in thesesamples.

La Quina sequencesDNA analysis of archaeological samples comprisedeight stone tools and their corresponding soil samples.DNA was obtained from both tool and soil samples(Table 2). The non-artefactual limestone listed in Table2 was taken from the rock shelter wall at La Quina.The limestone was tested to determine if DNA waspresent on non-artefactual lithic material. No DNAwas recovered from these samples. The identity of thesequences recovered is discussed in terms of percentageidentity (Table 2) and PAUP (Figure 3).Tool QM5–4950, a percoir, has yielded two different

mammalian sequences. Sequence #1 matches 100% toSus scrofa (boar or pig), a species represented osteo-logically at the site (Figure 2). Boar/pig sequence hasonly been found in tool sample #1 (see Table 2). Itdoes not occur in the corresponding soil sample or inthe other tool or soil samples. It has not been found inany modern sample. A large number of clones fromboth tool and soil samples were digested with theenzyme Spe I to test for the presence of boar/pig(Figure 4). An Spe I restriction site is present in onlytwo of the sequences in our database; boar/pig andrhinoceros. Rhinoceros were present in PleistoceneEurope, but are not represented osteologically at thesite. Of the 30 clones tested from tool QM5–4950, 50%were cut by Spe I and were taken to be boar/pigsequence along with the four clones that were se-quenced. The 30 clones from tool QM5–4950 weretaken from two separate sets of dialysis, PCR, and

cloning. None of the clones from the other samples(Soil #1, 20 clones; Tool #QM5-5093, 20 clones; Soil#2, 20 clones) were cut by Spe I.Sequence #2 is present in all of the archaeological

samples (soil and tool) analysed so far that show DNAamplification. It does not match any of the fauna inour database but appears to be mammalian in origin.Table 3 shows the species with the closest match tosequence #2. The closest match is to Ovis aries (seeFigure 2). For this region of cytochrome b, species ofthe same genus typically exhibit 90–95% identity,within the same family usually share 85–95% identity,and within the same order usually share between 80and 90% identity. Below 80% identity, the diversityincludes different orders. While these percentages arenot strict divisions, the list in Table 3 suggests thatsequence #2 is most likely an artiodactyl, possibly a

100

CowSheepGoatBighorn sheepBlacktail deerSequence #2Sika deerReindeerPronghornFallow deerPigSequence #1SquirrelHumanSequence #4FoxEthiopian wolfSequence #3RabbitCatZebraHorseRhinocerosBatElephantMouseRatWallabyFrog

100

100

69

98

878776

64

78

10057

61

100

100100 100

61

Figure 3. Majority rule consenus tree resulting from heuristic searchof 500 replicates. Third positions were ignored. Species included inthe tree are those of archaeological relevance or modern controlsused in the laboratory. Frog (Xenopus laevis) is the outgroup. Thistree is not meant to represent an exact mammalian phylogeny due tothe small number of characters examined. Numbers along the treebranches indicate the frequency of occurrence of the branchingsshown. Cow=Bos taurus, sheep=Ovis aries, goat=Capra hircus, big-horn sheep=Ovis canadensis, blacktail=Odocoileus hemionus, Sikadeer=Cervus nippon, reindeer=Rangifer tarandus, pronghorn=Antilocapra americana, fallow deer=Dama dama, pig=Sus scrofa,squirrel=Sciurus niger, human=Homo sapiens sapiens, arcticfox=Vulpes velox, Ethiopian wolf=Canis simensis, rabbit=Oryctolagus cuniculus, cat=Felis domesticus, zebra=Equus greyvi,horse=Equus caballus, rhinoceros=Diceros bicornis, bat=Urodermabilobatum, elephant=Loxodonta africana, mouse=Mus musculus,rat=Rattus norvegicus, wallaby=Macropus rufogriseus, frog=Xenopus laevis.

Table 2. Summary of DNA recovered from La Quina tools

Tools and corresponding soils DNA recovered

Tool QM5–4950 (percoir) Sequence #1 boar (Sus scrofa)Sequence #2 probable artiodactyl

Soil QM5–4950 Sequence #2 probable artiodactylSequence #3 mammalian

Tool QM5–4881 (flake) Sequence #2 probable artiodactylSoil QM5–4881 Sequence #2 probable artiodactylTool QM5–5004 (flake) Sequence #2 probable artiodactylSoil QM5–5004 No amplificationTool QM5–5008 (flake) No amplificationSoil QM5–5008 No amplificationTool QM5–5093 (flake) Sequence #4 human (ancient?)

Sequence #2 probable artiodactylSoil QM5–5093 Sequence #2 probable artiodactylTool QN5–2708 (flake) No amplificationSoil QN5–2708 No amplificationTool QN5–2972 (flake) No amplificationSoil QN5–2972 No amplificationTool QE5–507 (scraper) Sequence #2 probable artiodactylSoil QE5–507 No amplification

Non-artefactual samplesLimestone 1 No amplificationLimestone 2 No amplification


bovid or a cervid. PAUP (Figure 3) supports ourassignment of this taxa to the family Artiodactylae.Two tools, QE5–507 and QM5–5004, yielded sequence#2, but their corresponding soil samples did not am-plify. This would normally imply that sequence #2 isrelated to the use of these tools. However, due to its

prevalence in other samples at the site, it is more likelythat sequence #2 is present on these tools as part of itsgeneral pattern at the site. The presence of sequence #2in all of the samples amplified suggests that it may be a‘‘background’’ sequence, i.e. one which is scatteredthroughout the site. The fact that three tools and fivesoil samples did not amplify argues against the possi-bility that sequence #2 is a systematic contaminant.Sequence #2 has not been recovered from any modernsamples. Of the species represented osteologically at LaQuina, there is one artiodactyl for which we do nothave sequence, Bison priscus, which is extremely com-mon at the site. Work is currently under way to obtainbison sequence for comparison with sequence #2.Sequence #3 has only been found in the soil sample

SQM5–4950. It does not match any species in thedatabase, but is clearly cytochrome b and appears to bemammalian in origin. The best match is with Canissimensis (Ethiopian wolf) at 80·3% identity. Whilethere are canids present at the site (Canis lupus), anidentity of 80·3% is insufficient to provide an identifi-cation (see Figure 3). PAUP (Figure 3) indicates thatthis sequence is most closely related to Canis simensisor Oryctolagus cuniculus (rabbit). Oryctolagus cuniculushas only 71% identity with sequence #3. We are unableto identify this species positively beyond the mamma-lian level at this time. Further expansion of thesequence database may aid in identification of thissequence.Sequence #4 shows 99% identity with modern hu-

man (see Figure 2). There is one bp difference betweenthe two sequences. The one bp difference has beenobserved in several different clones. Sequence #4 hasonly been found from tool QM5–5093. While it isdifficult to rule out the possibility of modern humancontamination, this sequence was obtained from PCRreactions with clean no-DNA negative control reac-tions. This, along with the fact that longer fragments ofDNA did not amplify (see below) suggests the possi-bility that this DNA may be ancient. However, this isonly a possibility. Further analysis will be necessarybefore any conclusion can be drawn. Primers forportions of the mitochondrial control region have beenused to document population variation among modernhumans and have also been used with ancient samplesto establish ethnicity (Handt et al., 1994). Furtheranalysis with control region and other primers will benecessary before we can confirm that ancient humanDNA is present at La Quina. In summary, a total ofeight tool and corresponding soil samples have beentested from La Quina. Five of the eight tools and threeof the eight soils have yielded DNA. DNA fromboar/pig is only found on one tool and is not present inthe soil. This DNA is most likely use-related. The samemay be true for the human DNA on tool QM5–5093.The remaining DNAs are more difficult to interpret(see below).The results reported here have been repeated several

times in our laboratory from different aliquots of

Mar

kers

300

21 3 4 5 6

Cu

t

Un

cut

Cu

t

Un

cut

Mar

kers

Pronghorn Pig

200

100

bp

Figure 4. Restriction fragment analysis of PCR-generated productsfor the presence of cytochrome b boar/pig DNA. The screeningprocedure for boar/pig was based on the presence of a uniquerestriction enzyme (Spe I) site within the cytochrome b DNA besidesthe one in the polylinker region. Thus the digested boar/pig cyto-chrome b DNA yielded three bands at 96, 100, and 129 bp. Speciesother than pig will generate two bands at 96 and 229 bp. Lanes 1 and6 are DNA markers. Lanes 2 and 4 are the digested pronghornantelope (Antilocapra americana) and boar/pig DNA respectively.Lanes 3 and 5 are the undigested DNA from pronghorn antelope andboar/pig.

Table 3. Best match to sequence #2

Species Identity

Ovis aries (sheep) 84·0%Cervus nippon (Sika deer) 81·6%Rangifer tarandus (reindeer) 80·2%Ovis dalli (dall sheep) 78·9%Capra hircus (goat) 78·9%Ovis candensis (bighorn sheep) 78·9%Odocoileus hemionus (blacktail deer) 78·9%Capricornis crispus (Japanese serow) 78·9%Diceros bicornis (rhinoceros) 78·7%Antilocapra americana (pronghorn) 78·7%Bos taurus (cow) 77·6%Ovibos moschatus (musk ox) 77·6%Bos javanicus (Javanese cow) 76·3%

Control speciesHomo sapiens (human) 72·3%Macropus rufogriseus (wallaby) 69·7%


extracted samples. We will provide aliquots of samplesfor independent corroboration by other researchers.

Size of DNA recoveredIt is generally agreed that DNA from ancient sources isonly present in fragments <500 bp long (e.g. Higuchiet al., 1984; Paabo, 1985, 1989; Rogers & Bendich,1985; DeSalle et al., 1992; Lindahl, 1993; Soltis &Soltis, 1993; Handt et al., 1994). In order to test thishypothesis, primer L15162 (5* GCAAGCTTCTAC-CATGAGGACAAATATC 3*) described in Irwinet al. (1991), was used in combination with primerH15760 from this study to obtain a product of 618 bp.Primer L15162 is also designed to amplify a wide rangeof vertebrate species. The 618 bp fragment wassuccessfully amplified from human genomic DNA pre-pared from modern tissue culture. No 618 bp productswere obtained from any of the archaeological samples.We were also unable to obtain 618 bp fragments fromwallaby and pronghorn antelope DNA prepared frommuscle tissue. The tissue used to prepare this DNA wasobtained after the animals had been dead for anunknown period of time (several hours to severaldays). These results suggest that degradation to smallfragment size occurs very rapidly (see Tuross, 1994).However, the lack of amplification of large productsfrom the archaeological samples suggests that theDNA does not derive from modern lab contaminationand is therefore consistent with an ancient origin.

DiscussionIs the DNA recovered related to tool use?

The recovery of DNA from the surfaces of stone toolsdoes not necessarily mean that it is related to the use ofthe tool. In order to establish that DNA recoveredfrom stone tools was related to their use, we took thefollowing measures; precautions against contaminationby modern DNA (as described above), testing of soilsamples taken in close proximity to the tools to estab-lish that DNA is isolated on the tool, testing ofnon-artefactual materials to see if DNA is scatteredthroughout the site, and attempting to amplify largefragments (¢500 bp) of DNA which would not becharacteristic of ancient material.There are three possible patterns of DNA distribu-

tion to interpret:

(1) DNA is isolated on the tool. In this case, theDNA most probably derives from tool use. Thispattern is found with the boar/pig DNA;

(2) DNA is found on the tool and in the surroundingsoil. The DNA may be present because of tool useor it may be incidental to tool use. This pattern isfound with the probably artiodactyl sequence;

(3) DNA is found throughout the site, including onnon-artefactual materials. The DNA most likelycomes from sources not related to tool use.

The first two of these patterns have been found onthe tools from La Quina. The first pattern shows a highprobability that the DNA is related to tool use. Thesecond pattern has two potential sources. Because ourcurrent understanding of the taphonomy of DNA atarchaeological sites is limited, we cannot say whetherthis DNA is related to or incidental to tool use.Further investigation and experimentation into thesurvival and movement of DNA in an archaeologicalenvironment will be necessary to understand thesource(s) of DNA found on tools and in soil.Loy (1993) has previously reported the identification

of bovine satellite DNA on one 2200-year-old toolfrom British Columbia on the basis of nested PCR.This study has examined a sample of tools and theirsurrounding soils. We have found DNA from severaltools and soil samples and have used sequence data,rather than PCR, to identify species of origin. Thediscovery of boar/pig DNA on tool QM5–4950, butnot from other samples, suggests that the DNA isrelated to the use of the tool. The tool which yieldedboar/pig DNA is classified typologically as a percoir orawl. If the DNA and tool were associated with oneanother through a mechanism other than the use of thetool, we would expect the DNA to be more widespreadthan it is. The same is true for the occurrence of humanDNA, although the possibility that the human DNA isof modern origin is always high and cannot be ruledout here.The preliminary data reported here clearly illustrate

that DNA survives on buried modern tools and canalso be detected on materials from La Quina. Theidentification of species of origin of the DNA can bemade from sequence data. This is a level of identifi-cation previously unavailable to the archaeologist. Thediscovery of DNA in the soil from La Quina indicatesthat each archaeological site may have a distinct DNAprofile. Further analysis of tools and soils from LaQuina taken in combination with evidence from moretraditional archaeological approaches could provideone of the most detailed pictures of Neanderthalbehaviour to date. If Neanderthal DNA is present, itssequence can be contrasted with the range of modernhuman DNAs. Such data would cast light on thedebate concerning the origins of modern humans.

ConclusionsStone tool function at La Quina

Interpretations of stone tool function are often madeon the basis of close physical association between stoneartefacts and faunal remains. However, close physicalassociation does not always imply a functional rela-tionship. Amplification and identification of DNAfrom stone tools helps archaeologists to establish moredirect links to tool function. The data presented hereindicate that one of the tools from La Quina was usedin processing boar/pig. This species is represented


osteologically at the site, but its remains are notcommon. Without the DNA results, boar/pig remainscould have been interpreted as incidental to the activi-ties at the site. The DNA results, however, indicate thatboar/pig was related to Neanderthal behaviour at thesite, at least on one occasion. Although it is clear fromthe faunal remains at the site that Neanderthals at LaQuina were exploiting animal resources, DNA analysiscan provide associations of specific artefacts withspecific animal species.

Archaeological importance of DNA analysis of stonetoolsDNA analysis of residues on stone tools will poten-tially allow archaeologists to: (1) test the assumptionthat associated stone tools and animal bones areassociated due to use and not by chance; (2) testhypotheses linking specific tool types with specific uses;(3) test for multiple uses of tools; (4) detect behaviourwhich occurred away from the site (if tools werereturned to the site, but the material processed wasnot); (5) investigate paleoenvironmental data throughancient biomolecules; (6) investigate the mechanisms ofpreservation of ancient DNA; (7) detect processing ofsmall animals whose remains may be under-represented at a site; (8) identify plants which wereused in prehistory; (9) investigate molecular evolution(including humans); and (10) provide evidence forstone tool function.Further work in ancient DNA analysis of stone tools

will include the sequencing of additional modernspecies to increase the cytochrome b database, bothfrom tissues of modern animals and possibly frombone or tissue of extinct animals. Increasing the samplesize of tools analysed from La Quina will allow us toaddress such archaeological questions as the functionalvalidity of typological categories and whether or notthe association of animal bones and stone tools accu-rately reflects the activities at the site. Furthermore, itmay be possible to reconstruct broader patterns ofsubsistence and possibly even recover NeanderthalDNA. Archaeologists are urged to consider the possi-bility of DNA analysis when excavating sites andto modify their excavation procedures accordinglyto assure the collection of uncontaminated artefactsamples as well as sediment samples for potentialDNA residue analysis.

AcknowledgementsWe thank A. J. Jelinek and A. Debenath for allowingaccess to the La Quina materials; N. Toth and K.Schick of CRAFT, W. R. Adams of the Zooarchaeol-ogy Laboratory at Indiana University, R. Cann of theUniversity of Hawaii at Manoa for their support.Thanks to C. Metcalf for laboratory assistance. P.Jamison, G. Garufi, K. Schick, and M. Sahnouni forcomments on earlier versions of this paper. Partially

supported by the LSB Leakey Foundation and theIndiana Institute for Molecular and Cellular Biology.The La Quina field project was supported by theNational Science Foundation and the Wenner GrenFoundation.

ReferencesAnderson-Gerfaud, P. (1990). Aspects of behavior in the MiddlePaleolithic: functional analysis of stone tools from southwestFrance. In (P. Mellars, Ed.) The Emergence of Modern Humans:An Archaeological Perspective. New York: Cornell Univ. Ithaca,pp. 388–418.

Briuer, F. (1976). New clues to stone tool function: plant and animalresidues. American Antiquity 41, 478–484.

Cano, R. J., Poinar, H. N., Pieniazek, N. J., Acra, A. & Poinar,G. O. Jr. (1993). Amplification and sequencing of DNA froma 120–135 million year-old weevil. Nature 363, 536–538.

Crabtree, D. (1973). An Introduction to Flintworking. Pocatello:Occasional Papers of the Idaho State University Museum 28.

Clark, J. D. & Kurashina, H. (1981). A study of the work of amodern tanner in Ethiopia and its relevance for archaeologicalinterpretation. In (R. A. Gould & M. B. Schiffer, Eds) ModernMaterial Culture: The Archaeology of Us. New York: AcademicPress, pp. 303–320.

Debenath, A. & Jelinek, A. J. (1990). La Quina: Campagnes 1988–1990, Rapport de Synthese. Report presented to the FrenchMinistry of Culture.

DeSalle, R., Gatesy, J., Wheeler, W. & Grimaldi, D. (1992). DNAsequences from a fossil termite in Oligo-Miocene amber and theirphylogenetic implications. Science 257, 1933–1936.

Eisele, J. A., Fowler, D. D. & Lewis, R. A. (1995). Survival anddetection of blood residues on stone tools. Antiquity 69, 36–46.

Fullagar, R. L. K. (1991). The role of silica in polish formation.Journal of Archaeological Science 18, 1–24.

Golenberg, E. M., Giannasi, D. E., Clegg, M. T., Smiley, C. J.,Durbin, M., Henderson, D. & Zurawaski, G. (1990). ChloroplastDNA sequence from a Miocene Magnolia species. Nature 344,656–658.

Hagelberg, E., Quevedo, S., Turbon, D. & Clegg, J. B. (1994). DNAfrom ancient Easter Islanders. Nature 369, 25–26.

Handt, O., Richards, M., Trommsdorff, M., Kilger, C., Simanainen,J., Georgiev, O., Bauer, K., Stone, A., Hedges, R., Schaffner, W.,Utermann, G., Sykes, B. & Paabo, S. (1994). Molecular geneticanalyses of the Tyrolean ice man. Science 264, 1775–1778.

Hardy, B. L. (1994). Investigations of Stone Tool Function ThroughUse-Wear, Residue, and DNA Analyses at the Middle PaleolithicSite of La Quina, France. Ph.D. Thesis, Indiana University.

Higuchi, R., Bowman, B., Friedberger, M., Ryder, O. A. & Wilson,A. C. (1984). DNA sequences from the quagga, an extinct memberof the horse family. Nature 312, 282–284.

Hurcombe, L. (1992). Use-wear Analysis and Obsidian: Theory,Experiments and Results. Sheffield: J. R. Collis.

Irwin, D., Kocher, T. & Wilson, A. (1991). Evolution of thecytochrome b gene in mammals. Journal of Molecular Evolution32, 128–144.

Jones, P. (1980). Experimental butchery with modern stone tools andits relevance for Paleolithic archaeology. World Archaeology 12,153–165.

Juel-Jensen, H. (1988). Functional analysis of prehistoric flint toolsby high-power microscopy: a review of the West Europeanresearch. Journal of World Prehistory 2, 53–88.

Keeley, L. H. (1980). Experimental Determination of Stone ToolUses. Chicago: University of Chicago.

Kooyman, B., Newman, M. E. & Ceri, H. (1992). Verifyingthe reliability of blood residue analysis on archaeological tools.Journal of Archaeological Science 19, 265–269.

Lindahl, T. (1993). Instability and decay of the primary structure ofDNA. Nature 362, 709–715.


Loy, T. H. (1983). Prehistoric blood residues: detection on toolsurfaces and identification of species of origin. Science 220,1269–1271.

Loy, T. H. (1985). Preliminary residue analysis: AMNH specimen20.4/509. In (D. H. Thomas, Ed.) The Archaeology of HiddenCave, Nevada. New York: Museum of Natural History.

Loy, T. H. (1987). Recent advances in blood residue analysis. In (J.Olin & J. Blackman, Eds) Archaeometry: Further AustralasianStudies. Canberra: Australian National University, pp. 57–65.

Loy, T. H. (1993). The artifact as site: an example of the biomolecu-lar analysis of organic residues on prehistoric tools. WorldArchaeology 25, 44–63.

Loy, T. H. & Hardy, B. L. (1992). Blood residue analysis of90,000-year-old stone tools from Tabun Cave, Israel. Antiquity 66,24–35.

Loy, T. H., Spriggs, M. & Wickler, S. (1992). Direct evidence forhuman use of plants 28,000 years ago: starch residues on stoneartifacts from the northern Solomon Islands. Antiquity 66, 898–912.

Manning, A. P. (1994). A cautionary note on the use of hemastix anddot-blot assays for the detection and confirmation of archaeologi-cal blood residues. Journal of Archaeological Science 21, 159–162.

Moss, E. H. (1983). Some comments on edge-damage as a factor ina functional analysis of stone artefacts. Journal of ArchaeologicalScience 10, 231–242.

Newcomer, M., Grace, R. & Unger-Hamilton, R. (1986). Investigat-ing microwear polishes with blind tests. Journal of ArchaeologicalScience 13, 203–217.

Newman, M., Yohe II, R. M., Ceri, H. & Sutton, M. Q. (1993).Immunological protein residue analysis of non-lithic archaeologi-cal materials. Journal of Archaeological Science 20, 93–100.

Odell, G. H. & Odell-Vereecken, F. (1980). Verifying the reliabilityof lithic use-wear assessments by blind tests: the low-powerapproach. Journal of Field Archaeology 13, 195–212.

Paabo, S. (1985). Molecular cloning of ancient Egyptian mummyDNA. Nature 314, 644–645.

Paabo, S. (1989). Ancient DNA: extraction, characterization,molecular cloning, and enzymatic amplification. Proceedings of theNational Academy of Sciences of the USA 86, 1939–1943.

Rogers, S. O. & Bendich, A. J. (1985). Extraction of DNA frommilligram amounts of fresh, herbarium, and mummified planttissues. Plant Molecular Biology 5, 69–76.

Shea, J. J. (1992). Lithic microwear analysis in archaeology. Evolu-tionary Anthropology 1, 143–150.

Soltis, P. S. & Soltis, D. E. (1993). Ancient DNA: prospects andlimitations. New Zealand Journal of Botany 31, 203–209.

Swofford, D. L. (1993). PAUP: Phylogenetic Analysis Using Parsi-mony, Version 3.1. (Illinois Natural History Survey, Champaign,IL).

Toth, N. (1985). The Oldowan reassessed: a close look at early stoneartifacts. Journal of Archaeological Science 12, 101–120.

Tuross, N. (1994). The biochemistry of ancient DNA in bone.Experientia 50, 530–535.

recovery of mammalian dna from middle paleolithic stone tools

Documents