Establishing the validity of domestication genes usingDNA from ancient chickensLinus Girdland Flinka,b,1,2, Richard Allena, Ross Barnetta, Helena Malmströmc, Joris Petersd,e, Jonas Erikssonf,Leif Anderssonf,g, Keith Dobneyh, and Greger Larsona,3

aDurham Evolution and Ancient DNA, Department of Archaeology, bSchool of Biological and Biomedical Sciences, Durham University, Durham DH1 3LE,United Kingdom; cDepartment of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, 75236 Uppsala, Sweden; dDepartment of VeterinarySciences, Institute of Palaeoanatomy, Domestication Research and the History of Veterinary Medicine, Ludwig Maximilian University Munich, 80539 Munich,Germany; eBavarian State Collection of Anthropology and Palaeoanatomy, 80333 Munich, Germany; fDepartment of Medical Biochemistry and Microbiology,Uppsala University, SE-75123 Uppsala, Sweden; gDepartment of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, SE-75007 Uppsala,Sweden; and hDepartment of Archaeology, University of Aberdeen, St. Mary’s, Aberdeen AB24 3UF, Scotland

Edited by Dolores R. Piperno, Smithsonian National Museum of Natural History and Smithsonian Tropical Research Institute, Fairfax, Washington, DC,and approved October 29, 2013 (received for review August 23, 2013)

Modern domestic plants and animals are subject to human-drivenselection for desired phenotypic traits and behavior. Large-scalegenetic studies of modern domestic populations and their wildrelatives have revealed not only the genetic mechanisms underly-ing specific phenotypic traits, but also allowed for the identifica-tion of candidate domestication genes. Our understanding of theimportance of these genes during the initial stages of the domes-tication process traditionally rests on the assumption that robustinferences about the past can be made on the basis of moderngenetic datasets. A growing body of evidence from ancient DNAstudies, however, has revealed that ancient and even historic pop-ulations often bear little resemblance to their modern counter-parts. Here, we test the temporal context of selection on specificgenetic loci known to differentiate modern domestic chickensfrom their extant wild ancestors. We extracted DNA from 80 an-cient chickens excavated from 12 European archaeological sites,dated from ∼280 B.C. to the 18th century A.D. We targeted threeunlinked genetic loci: the mitochondrial control region, a gene as-sociated with yellow skin color (β-carotene dioxygenase 2), anda putative domestication gene thought to be linked to photope-riod and reproduction (thyroid-stimulating hormone receptor,TSHR). Our results reveal significant variability in both nucleargenes, suggesting that the commonality of yellow skin in Westernbreeds and the near fixation of TSHR in all modern chickens tookplace only in the past 500 y. In addition, mitochondrial variationhas increased as a result of recent admixture with exotic breeds.We conclude by emphasizing the perils of inferring the past frommodern genetic data alone.

selective sweep | breed formation | animal domestication | Gallus gallus |cultural history

The resolution afforded by multiple genetic loci and—morerecently—complete genomes has led to an increased un-

derstanding of the pattern and process of plant and animal do-mestication (1, 2). More specifically, genetic analyses haveuncovered selective sweeps, quantitative trait loci, and evencausative mutations underlying a wide range of behavioral andmorphological traits, some of which define specific breeds, andothers that differentiate domestic plants and animals from theirwild ancestors (1, 3, 4).Because many of these traits are present in either single or

relatively few closely related modern breeds, the earliest occur-rences of specific phenotypes (and the underlying causativemutations) are presumed to have occurred well after the initialdomestication process. These phenotypes are referred to (atleast in the plant genetic literature) as “improvement genes” (2).In animals, these traits include hairlessness in Mexican andPeruvian dogs (5), dorsal hair ridges in Vietnamese, Thai, andRhodesian Ridgebacks (6), excessive skin folds in western Shar-Peis (7), double muscling in two cattle breeds (8), and a curly

coat mutation found in Selkirk Rex cats (9), none of which arethought to have been present during early domestication.Some causative mutations, however, underlie traits found in

numerous, distantly related breeds. Alleles that are fixed in do-mestic variants—and often presumed to have been under se-lection at the outset of domestication—are referred to in boththe plant (2) and animal (3) domestication literature as “do-mestication loci” (or domestication genes). In some cases, in-cluding gray coloring (10) and altered gaits in horse breeds (11),brachycephaly in dogs (12), and muscle growth in pigs (13), nohypotheses have been proposed for the time-frame of first ap-pearance of these traits. In others, however, the commonality ofboth small size (14, 15) and chondrodysplasia (16) across moderndog breeds and the widespread occurrence of pea-combs inchickens (17), led the authors of these studies to suggest that thegenetic mutations underlying these characteristics were selectedfor during the early stages of the domestication process. Morerecently, a whole-genome resequencing study that comparedvariation in 14 unrelated dog breeds and wolves identified 36regions potentially targeted during early domestication and in-cluded 10 genes that allowed dogs to better digest starches (18).Because increased amylase activity was ubiquitous in dogs but

Significance

Recent studies have identified the genetic basis of numeroustraits that differentiate modern domestic species from theirwild counterparts. In both plants and animals, traits (and thegenes underlying them) found ubiquitously in modern breedsare often presumed to have been selected early during thedomestication process. Here, by determining genetic variabilityin ancient European chickens over the past 2,000 years, weshow that a mutation thought to be crucial during chickendomestication was not subjected to strong human-mediatedselection until much later in time. This result demonstrates thatthe ubiquity of mutations, which differentiate modern wildand domestic taxa, does not necessarily imply ancient origins.

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

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper has been deposited in the GenBankdatabase (accession nos. KF753251–KF753289).1Present address: Department of Archaeology, University of Aberdeen, St. Mary’s, Aber-deen AB24 3UF, Scotland.

2Centre for Genome-Enabled Biology and Medicine, University of Aberdeen, AberdeenAB24 3RY, Scotland.

3To whom correspondence should be addressed. E-mail: greger.larson@durham.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1308939110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1308939110 PNAS Early Edition | 1 of 6

ANTH

ROPO

LOGY

SPEC

IALFEATU

RE

absent in wolves, the authors concluded that this change musthave occurred when early dogs began adapting to a starch-richdiet provided by early farmers.Recent genetic and archaeological research has also shed light

on domestic chickens and their primary ancestor, the Red Jun-glefowl (Gallus gallus) (19). Based on archaeological bonesidentified from Neolithic sites in the Yellow River basin, chick-ens were thought to have been domesticated as early as 6000 B.C. (20). This conclusion has recently been questioned, however,because bones presumed to originate from chickens in theoriginal faunal analysis (21, 22) have since been shown to bepheasants (23, 24). As a result, a reevaluation of all of the earlyfinds is necessary to establish the true chronology and geographyof chicken domestication.Genes that differentiate modern domestic chickens from Red

Junglefowl include those that underlie the yellow skin phenotypepresent in the vastmajority ofWestern, commercial chicken breeds,as well as numerous geographically restricted and fancy breeds.Yellow skin is caused by a recessive allele of the BCDO2 (β-car-otene dioxygenase 2) gene (25). BCDO2 encodes the β-carotenedioxygenase 2 enzyme that cleaves colorful carotenoids into col-orless apocarotenoids (26). Although the expression of the dom-inant allele in skin tissue results in white skin color, the recessiveallele possesses one or more cis-acting and tissue-specific regula-tory mutations that inhibit expression of BCDO2 in skin tissue.Provided that sufficient carotenoids are available in the diet,the recessive allele reduces carotenoid cleavage and allows themto be deposited in skin tissue, leading to yellow skin (25). Thisrecessive BCDO2 allele is thought to have been acquired throughhybridization with the Gray Junglefowl (Gallus sonneratii) inSouth Asia (25). Red and Gray Junglefowl are known to hybridizein contact zones in the Indian subcontinent (27, 28), and it ispossible that domestic poultry engaged in the same behavior afterthey were introduced from Southeast Asia. Given the ubiquityand genomic signatures of strong human-driven selection of theyellow skin trait in modern, Western commercial chickens (29),Eriksson et al. (25) suggested that this trait was favored by humansafter chickens acquired the trait in South Asia, but before thefirst wave of domesticated chickens arrived inEurope between 900and 700 B.C. (30, 31).In addition, a recent analysis of pooled wild and domestic

chicken samples revealed strong selection signatures acrossa number of loci, as well as a missense mutation in the thyroid-stimulating hormone receptor (TSHR), a locus possibly linked toshifts in seasonal mating (29). Given its ubiquity in domesticbreeds (264 of 271 birds representing 36 global populations werehomozygous for the sweep allele; the remaining 7 were heterozy-gous) and the general absence of the derived allele in Red Jun-glefowl, the authors of that study concluded that the TSHR locusmay have played a crucial role during chicken domestication (29).Here, we investigate whether the TSHR gene was selected for

during the early stages of chicken domestication (29), and if earlypoultry keepers favored the BCDO2 gene that underlies yellowskin in chickens soon after it was acquired from the Gray Jun-glefowl (25, 29). To do so, we genotyped SNPs linked with thesweep alleles in both TSHR and BCDO2 in 80 ancient Europeanchickens dating from ∼280 B.C. to the 18th century A.D. (TableS1 and SI Materials and Methods). If TSHR played a critical roleduring the domestication process, all of the samples analyzedhere should have been fixed for the derived TSHR allele, as hasbeen demonstrated in worldwide modern chicken populations(29). Similarly, if BCDO2 and the yellow skin phenotype wasfavored and maintained soon after its introgression from GrayJunglefowl, a significant proportion of the ancient Europeanindividuals should also possess this phenotype. Finally, we assessthe hypothesis that the presence of mitochondrial DNA (mtDNA)control region (CR) haplogroups A–D has resulted from therecent introduction of East Asian chickens into the European

gene pool, and that haplogroup E is historically associated withEuropean chickens (32).

ResultsFor each ancient individual, we attempted to amplify a 58-bpfragment surrounding the candidate missense (Gly > Arg) SNPin the TSHR gene (29), a 51-bp fragment surrounding a SNP inthe BCDO2 gene associated with the yellow skin allele (SNP B intable 1 of ref. 25), and a 201-bp fragment of the mtDNA CR(33). Overall, 55 of 80 (69%) ancient chicken remains providedreproducible results for at least one of three loci (Fig. 1, Fig. S1,and Table S1). We observed allelic drop out in a number ofheterozygous specimens for both TSHR and BCDO2. However,we estimated the probability of falsely assigning a true hetero-zygous individual as a homozygote to <0.01. In addition, externalreplication on 12 samples yielded identical mtDNA CR andTSHR sequences (SI Materials and Methods).Among the 44 specimens from whom the TSHR locus was

successfully genotyped, 8 individuals were homozygous for thederived (domestic) sweep allele, 14 were homozygous for thewild-type allele, and 22 specimens were heterozygous (Fig. 1 andTable S1). The results of a Fisher’s exact test revealed that thesweep allele was significantly less frequent in the ancient samplethan in modern chickens (P < 0.0001). A binomial probabilitytest demonstrated that observing these frequencies among theancient samples, assuming the frequencies of the modern sam-ples (Fig. S2 and Table S2) (29), is very unlikely (P < 0.0001). Inaddition, a χ2 test on TSHR genotype frequencies derived fromthe second to third century A.D. Quintana/Künzing Romanpopulation (Fig. 1 and Tables S1 and S3) revealed that the ob-served frequencies are consistent with Hardy–Weinberg equi-librium (P > 0.95, χ2 = 0.004, df = 1), although one group hadfewer than the recommended minimum number of expectedgenotypes/individuals.Of the 25 ancient specimens successfully genotyped for the

BCDO2 locus, 20 were homozygous for the white skin allele(found in Red Junglefowl) and five individuals were heterozy-gous (Fig. 1 and Table S2). Because only one functional copy ofthe BCDO2 gene is necessary to effectively cleave carotenoids,the yellow skin phenotype can only be expressed in chickens thatare homozygous for the yellow skin allele and consume sufficientcarotenoids in their diet. Of the 25 successfully genotypedchicken samples, none could express the yellow skin phenotype.Because genotype/phenotype frequencies reported previouslywere selected on the basis of their phenotypes (nonrandomsampling) (25), we did not carry out statistical comparisons ofallele frequencies between ancient and modern populations.The targeted mtDNA CR fragment was successfully se-

quenced in 38 individuals (Table S1). The topology of a maxi-mum-likelihood tree constructed from an alignment of 201-bphaplotypes matched the neighbor-joining tree generated by Liuet al. (19), confirming previous observations that this specific201-bp fragment is sufficient for recovering the major cladespresent in the chicken mitochondrial tree (Fig. 2 and Fig. S3)(33, 34). We identified a total of three haplotypes among theancient specimens, all of which clustered within the E clade onthe chicken mitochondrial tree (19, 35) (Fig. 2 and Table S4).The E3 (n = 1) and E6 (n = 2) haplotypes (19) were present onlyin Medieval and post-Medieval chickens from England (TablesS1 and S4), whereas the remaining 35 individuals possesseda 201-bp haplotype corresponding to haplotypes E1, E5, E12,E15, or E16 described using a 519-bp fragment (Table S4) (19).We find a significant difference in haplogroup frequencies(pooled into two groups of chickens: those belonging to hap-logroup E and those belonging to haplogroup A–D) between theancient and modern datasets (Fisher’s exact test, P < 0.002)(Table S5). Assuming the frequency reported for modern Eu-ropean chickens (Table S5) (i.e., ∼15% of modern European

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1308939110 Girdland Flink et al.

chickens possess haplotypes from clades A–D), a binomial testrevealed that the probability of observing only the E haplogroupin 43 ancient specimens (the unique 38 sequences combined withpreviously published data) (Table S5) is <0.001.

DiscussionTSHR Domestication Locus. The locus encoding TSHR on chro-mosome 5 in domestic chickens has recently been shown to haveundergone a massive selective sweep (29). A nonconservativeamino acid substitution (a missense mutation Gly558Arg) wasidentified as a potential causal and target mutation for the se-lective sweep. Although the function (and corresponding phe-notype) associated with the derived allele remains unknown, it ispossible that this gene variant affects photoperiod control andthe absence of strict seasonal reproduction, a trait commonlyfound in domestic animals but rare or absent among their wildrelatives (29). Because 264 of 271 modern birds representing 36globally distributed populations were homozygous for the de-rived sweep allele (the seven remaining were heterozygous), and

because the Red Junglefowl individuals that also possessed themutation were thought have acquired it from domestic chickens,Rubin et al. (29) suggested that TSHR was a domestication locus.If this selective sweep occurred during the early phase of do-mestication, all ancient chickens that postdate this event (and arelocated outside the natural distribution range of Red Junglefowl,thus eliminating the potential for backcrossing with wild birds)should also possess the derived allele.The results presented here, however, demonstrate that al-

though the derived allele was present in European chickensdated to ∼280 B.C. to the 16th–18th century A.D., it was onlyfound on 43% of the typed chromosomes and the wild-type allelepersisted at intermediate frequencies until at least the 16th–18thcentury AD (Fig. 1 and Tables S1 and S2). These results suggestthat the TSHRmutation was neither a prerequisite, nor critical inthe immediate aftermath of chicken domestication. Although thestrength of the selection pressure that drove the sweep haplotypein modern populations is not in doubt (29), the data presentedhere suggest that the fixation of the derived allele in Europeanchickens was likely much closer to the present-day and certainlywithin the last 500 y, possibly commensurate with the improve-ment of farmyard animals that began during the industrial rev-olution (27).In one possible scenario, the sweep allele was fixed in Western

Europe and spread across the world with newly synthesizedcommercial breeds that emerged during the second half of the19th century (27). This hypothesis seems overly simplistic, how-ever, given that the TSHR allele was found to be fixed (or nearlyso) in modern Egyptian Fayoumi chickens and Silkie, Cochin,and Hua-Tung chickens from China (29) that are unlikely tohave been strongly influenced by European commercial breeds.It is therefore possible that the TSHR sweep allele became fixedin populations originating outside Europe well before the crea-tion of modern breeds, and that these birds in turn replaced theinitial populations introduced into Europe. This narrative issupported by the observation that the Classical Greek chickens(Kassope) were the only population analyzed in this study thatwas fixed for the sweep allele (Fig. 1 and Table S1). Although theearliest Central European chickens that arrived north of theAlps ∼600–400 B.C. likely descended from founder populations

Fig. 1. (A) A map showing the locations and chronology of archaeological sites (black dots) from three different European regions where ancient chickenremains analyzed in this study were excavated (see also SI Materials and Methods and Fig. S1). The pairs of colored boxes adjacent to each region show whichof two alleles of two nuclear genes (TSHR and BCDO2) were present in each genotyped individual (Table S1). (B) Pie charts showing the differing allelefrequencies of all (pooled) ancient samples (from 290 B.C. to the 18th century A.D.) (Table S2). Allele frequencies found in modern chicken populations at theTSHR locus were derived from ref. 29 and at the BCDO2 locus from ref. 25.

Fig. 2. (A) Two pie charts showing differences in mtDNA haplogroup fre-quencies between ancient and modern European chickens. The ancientsample consists of specimens sequenced in this study (n = 38) and froma previous publication (n = 5) (34) and the modern data was compiled from(19, 32, 39) (Table S5). The colors in A correspond to haplogroups depicted ina phylogenetic tree in B. (B) A maximum-likelihood phylogenetic tree con-structed using 519-bp haplotypes defined in ref. 19. White (or colorless)clades were not identified in either modern or ancient European chickens. Amore detailed maximum-likelihood tree, including detailed description ofmethods and results (including nodal support values), is described in the SIMaterials and Methods (see also Fig. S3).

Girdland Flink et al. PNAS Early Edition | 3 of 6

ANTH

ROPO

LOGY

SPEC

IALFEATU

RE

present in eighth to fifth century B.C. Greek settlements on theMediterranean coast, the first century B.C. chicken populationfrom Kassope could have arrived as part of a secondary in-troduction of chickens. During the Achaemenid (550–330 B.C.)and Hellenistic (323–31 B.C.) periods, the Aegean region wasa commercial hub that drew trade from most of the Near andMiddle East, including the Indus Valley (36). Such trade net-works could have been linked to the introduction of chickenpopulations that had undergone a selective sweep at the TSHRlocus. Crucially, however, none of the Greek individuals pos-sessed the introgressed Gray Junglefowl BCDO2 allele (Fig. 1),suggesting these two loci were selected for and fixed at differenttimes and places.

The BCDO2 Yellow Skin Locus. The study that discovered the TSHRsweep (29) also confirmed a selective sweep encompassing theBCDO2 locus. Unlike TSHR, however, BCDO2 is only fixed ina limited number of (often commercial and geographicallywidespread) breeds (25). The yellow skin phenotype was pre-viously shown to result from the presence of a BCDO2 allele thatdomestic chickens acquired not from their primary ancestor—the Red Junglefowl—but from introgression between domesticchickens and Gray Junglefowl indigenous to South Asia (25).The data presented here reveal that although the yellow skin

allele was present on ∼10% of chromosomes in ancient Euro-pean chickens, not a single ancient bird was homozygous for theGray Junglefowl allele (and therefore capable of expressing theyellow skin phenotype) (Tables S1 and S2). This finding is con-sistent with 17th–19th century records suggesting that a numberof widespread, prolific, and economically important breeds raisedin western and southern Europe, including the Dorking, Houdan,Sultan, Spanish, or B/W Bantams, clearly had white legs, but theHamburgh, Polish, Turkish, and Crève Cœur breeds had leg colorsranging between slate blue and dark leaden-blue (27, 37). How-ever, some 17th century European breeds, including the Padua,did possess yellow legs (37) and the trait was explicitly mentionedin relation to heavy, fast-growing types, including Cochin, Brahma,and Malay breeds imported by sea (and therefore named Cap-tain’s birds) from the Far East into Europe and the United Statesduring the 1820s–1850s (27). It is therefore possible that the rise infrequency of the yellow skin phenotype occurred only after recentintroductions of foreign birds to Europe and the breed formationprocess that consequently led to the creation of modern, wide-spread, commercial broiler, and egg-laying breeds. Tegetmeier (27),for example, noted the remarkable rapid growth and great sizeattained by crossbred birds produced by mating Cochins with thelarge traditional French Crève Cœur, La Flèche, and Houdanbreeds. The ubiquity of the yellow skin phenotype in commercial,modern, and some rare, geographically restricted breeds can,therefore, be explained by rapid worldwide spread of newly syn-thesized commercial chickens. This suggestion is further supportedby the fact that the yellow skin phenotype appears infrequently oris completely absent in rare breeds, such as the Friesian Fowl,Houdan, and Westfälischer Totleger (25).

The Mitochondrial Control Region. The mtDNA control region isa widely used locus in chicken genetic studies and has frequentlybeen used to investigate domestication, admixture, and migra-tion (19, 34, 35, 38). Although previous studies have concludedthat modern European chickens (and Western commercialbreeds) primarily possess mitochondrial haplotypes belonging tothe E clade (19, 35), additional studies (32, 39) have demon-strated that haplogroups A–D are also present in a wide varietyof breeds, most likely as a result of the intentional importationand hybridization of East Asian breeds into Europe during breedformation and breed improvement over the past 500 y (32).Although a recent study of complete mtDNA chicken genomes

revealed the existence of 14 mitochondrial clades found worldwide

(35), all 38 ancient chickens sequenced in this study, and five ar-chaeological Spanish chickens typed in a previous study (34),possessed exclusively E-clade haplotypes (which represent 3 out ofthe 14 major clades in ref. 35) (Tables S1 and S5). These resultsfirst demonstrate that chickens initially introduced to Europepossessed a small fraction of the variability present in SoutheastAsia, where chickens were originally domesticated. Second, theseresults demonstrate that all haplotypes belonging to other hap-logroups therefore represent recent introductions. Interestingly,because all but three ancient European chickens possessed a sin-gle haplotype (probably corresponding to the common E1 hap-lotype) (19) (Table S4), the presence of haplotypes E3 and E6 inMedieval and post-Medieval contexts from the United Kingdommay imply secondary introductions.Finally, although the nuclear loci typed in this study show

a dramatic reduction in variability between ancient and modernpopulations, the mitochondrial signatures reveal an initial uni-formity followed by an increase in haplogroup diversity inmodern birds. These seemingly divergent genetic patterns reflectdifferent human goals over the last two centuries, including notonly intensive selection for traits associated with behavior, pro-duction (TSHR), and skin color (BCDO2), but also for pheno-typic variability in fancy breeds achieved through the importationof East Asian varieties (32).

ConclusionsA variety of genetic techniques and analytical approaches haveled to the identification of selective sweeps and causativemutations that differentiate populations of domestic plants andanimals from their modern wild counterparts. The ubiquity ofsweep alleles across numerous breeds has often been used as anargument to infer their ancient origins, and as an argument thatthe SNPs and associated traits were selected for during the earlyphases of domestication (4, 14, 17, 18, 40). The patterns of allelicdiversity presented here, however, challenge the hypothesis thatmodern variation can be directly mapped onto the past. Instead,the allelic variability within TSHR in ancient chickens demon-strates that this locus was not crucial for the early developmentof domestic chickens, and that the fixation of the derived allele(at least in European chickens) took place only in the past fewhundred years. Similarly, although the yellow skin allele waspresent in ancient European chickens, the phenotype was veryrare in the past. In addition, the fixation of the BCDO2 allele innumerous modern chicken breeds was likely also a recent oc-currence, possibly coincident with recent breed formation in-stigated during the industrial revolution (27), and the subsequentdevelopment of modern commercial chicken breeds. This pat-tern demonstrates the ease with which we can underestimate thepotential of alleles to become ubiquitous through rapid strongselection leading to fixation, followed by geographic proliferationthrough human-assisted migration.This result is consistent with most ancient DNA studies that

have revealed the hazards of assuming that modern genetic datacan be used to draw accurate conclusions about past populationvariability. For example, a mutation in the NAM-B1 gene, asso-ciated with increased grain size in wheat, was thought (on thebasis of fixation in modern cultivars) to have been selected forduring the early phases of domestication (40). A genetic surveyof 19th century historical seeds, however, revealed that fixationof the modern, sweep allele occurred only recently during cropimprovement (41). A similar study of ancient maize concludedthat, although two genes (tb1 and pbf) had been fixed for thedomestic variant by ca. 4,000 y ago, one gene (su1) still possessedsignificant variability as long ago as ca. 2,000 y ago (42). Simi-larly, a number of studies of ancient mtDNA have shown thatlineage replacement (often rapid and geographically widespread)was common among both domestic (43, 44) and wild (45, 46)populations.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1308939110 Girdland Flink et al.

The processes of recent breed formation, coupled with strongdirected selection and admixture with exotic breeds, have rapidlyand radically shaped the gene pools of modern domestic plantsand animals, often distorting their deeper genetic history (47).As a result, hypotheses regarding past population dynamicsdrawn solely from modern datasets require verification throughdirect observation. This need is especially true where zooarch-eological or historical records suggest recent origins for sometraits. As mentioned at the outset, conclusions drawn from thegenetic variability in modern samples suggest that small size (14,15) and chondrodysplasia (16) in dogs, and pea-combs in chickens(17), occurred early during domestication. However, small dogs(<30-cm shoulder height) do not appear in the Eurasian ar-chaeological record until after the first millennium B.C. (30), theearliest reported evidence for chondrodysplasia is found inEgyptian tomb art dating to the late third millennium B.C. (48),and the peacomb trait in chickens was still rare in mid-19th cen-tury English fowl (27).These lines of evidence alone do not undermine claims for

much earlier appearances of these and other domestication-related traits, but they do suggest that modern genetic data shouldbe considered cautiously and within a broader context thatincludes an appreciation for the potential of rapid and wide-spread demographic shifts. Future studies that investigate an-cient genetic variability in loci known to underlie behavioral,dietary, and phenotypic differences between wild and domesticplants and animals will help to reveal the first appearance andtiming of selection pressures. These data will lead to the estab-lishment of a significantly more robust geographic and temporalinterpretative framework to more fully understand the earlypatterns and processes of domestication.

Materials and MethodsAncient chicken specimens were selected to represent three geographicallocations and two major time bins (first, Late Iron Age, Late Hellenistic andRoman contexts, dating to ∼280 B.C. to the fifth century A.D., and second,Medieval and post-Medieval contexts, dating to approximately the 10th to18th century A.D.) (Table S1). The majority of chicken bones come fromclosed contexts that have been dated either through stratigraphic (cultural)association or direct radiocarbon dating (SI Materials and Methods).

DNA extraction was performed in dedicated ancient DNA laboratories atDurham University and Uppsala University following stringent laboratoryprocedures according to commonly applied guidelines (49, 50). The work wascarried out by laboratory personal wearing protective laboratory coats andovershoes, or coveralls and dedicated laboratory clogs, and double pairs ofgloves (with the outer pairs of gloves changed in between every step of thepreparation/extraction procedure). All equipment and work surfaces areroutinely cleaned before and after each use with a dilute solution of bleach[5–10% (wt/vol) active sodium hypochlorite] followed by rinsing with ddH2Oand ethanol [70–99% (vol/vol)]. A strict one-way system for entering thelaboratories is in use to avoid carry-over of post-PCR contaminants.

DNA extractions were performed as previously described (44) (see SIMaterials and Methods for a detailed account of experimental procedures).To authenticate the results, 12 chickens were independently replicated forthe CR and TSHR SNP at Uppsala University (Table S1). CR sequences wereamplified using previously published PCR primers GG144F-GG387R (33), andPCR and sequencing primers for TSHR and BCDO2 were designed in PSQAssay Design (Qiagen) (Table S6). CR sequences were Sanger-sequenced inboth directions at the DNA sequencing facility at Durham University. TSHRand BCDO2 PCR amplicons were genotyped on the Q24 (Qiagen) pyrose-quencer in the Department of Archaeology at Durham University, UnitedKingdom, or on a PSQ 96MA (Biotage) pyrosequencer at Uppsala University.Full PCR cycling conditions and sequencing methods are presented in detailin the SI Materials and Methods.

MtDNA CR sequence data were analyzed and assembled in Geneious v.5.4(51) and manually edited in Se-Al (http://tree.bio.ed.ac.uk/software/seal).Reference sequences were compiled from previous publications (19, 32) andaligned with the unique sequences deposited into GenBank (KF753251-KF753289).

ACKNOWLEDGMENTS. We thank Anders Götherström for assistance withdata replication; and the London Archaeological Archive and Research Cen-tre, Museum of London, Alex Croom (Arbeia Roman Fort and Museum), andDeborah Jaques for sample material and discussion. Sample material fromKassope and Central European sites is courtesy of Bavarian State Collectionof Anthropology and Palaeoanatomy, Munich; material from Beverley iscourtesy of Humber Field Archaeology, Hull; and material from York is cour-tesy of Deborah Jaques. This work was supported by a Research Councils UKAcademic Fellowship (to G.L.); Grant NE/F003382/1 from the Natural Envi-ronment Research Council; and Grant F/00 128/AX from the LeverhulmeTrust. This manuscript resulted from a catalysis meeting entitled “Domesti-cation as an Evolutionary Phenomenon: Expanding the Synthesis” that wasawarded and hosted by the National Evolutionary Synthesis Centre, NationalScience Foundation #EF-0905606 in 2011.

1. Andersson L (2013) Molecular consequences of animal breeding. Curr Opin Genet Dev

23(3):295–301.2. Olsen K-M, Wendel J-F (2013) A bountiful harvest: Genomic insights into crop do-

mestication phenotypes. Annu Rev Plant Biol 64:47–70.3. Andersson L (2012) How selective sweeps in domestic animals provide new insight

into biological mechanisms. J Intern Med 271(1):1–14.4. Wayne RK, vonHoldt BM (2012) Evolutionary genomics of dog domestication. Mamm

Genome 23(1–2):3–18.5. Drögemüller C, et al. (2008) A mutation in hairless dogs implicates FOXI3 in ecto-

dermal development. Science 321(5895):1462.6. Salmon Hillbertz NHC, et al. (2007) Duplication of FGF3, FGF4, FGF19 and ORAOV1

causes hair ridge and predisposition to dermoid sinus in Ridgeback dogs. Nat Genet

39(11):1318–1320.7. Olsson M, et al. (2011) A novel unstable duplication upstream of HAS2 predisposes to

a breed-defining skin phenotype and a periodic fever syndrome in Chinese Shar-Pei

dogs. PLoS Genet 7(3):e1001332.8. McPherron AC, Lee S-J (1997) Double muscling in cattle due to mutations in the

myostatin gene. Proc Natl Acad Sci USA 94(23):12457–12461.9. Filler S, et al. (2012) Selkirk Rex: Morphological and genetic characterization of a new

cat breed. J Hered 103(5):727–733.10. Rosengren Pielberg G, et al. (2008) A cis-acting regulatory mutation causes premature

hair graying and susceptibility to melanoma in the horse. Nat Genet 40(8):1004–1009.11. Andersson LS, et al. (2012) Mutations in DMRT3 affect locomotion in horses and spinal

circuit function in mice. Nature 488(7413):642–646.12. Bannasch D, et al. (2010) Localization of canine brachycephaly using an across breed

mapping approach. PLoS ONE 5(3):e9632.13. Van Laere A-S, et al. (2003) A regulatory mutation in IGF2 causes a major QTL effect

on muscle growth in the pig. Nature 425(6960):832–836.14. Gray MM, Sutter NB, Ostrander EA, Wayne RK (2010) The IGF1 small dog haplotype is

derived from Middle Eastern grey wolves. BMC Biol 8(1):16.15. Sutter NB, et al. (2007) A single IGF1 allele is a major determinant of small size in

dogs. Science 316(5821):112–115.

16. Parker HG, et al. (2009) An expressed fgf4 retrogene is associated with breed-definingchondrodysplasia in domestic dogs. Science 325(5943):995–998.

17. Wright D, et al. (2009) Copy number variation in intron 1 of SOX5 causes the Pea-comb phenotype in chickens. PLoS Genet 5(6):e1000512.

18. Axelsson E, et al. (2013) The genomic signature of dog domestication reveals adap-tation to a starch-rich diet. Nature 495(7441):360–364.

19. Liu Y-P, et al. (2006) Multiple maternal origins of chickens: Out of the Asian jungles.Mol Phylogenet Evol 38(1):12–19.

20. West B, Zhou B-X (1988) Did chickens go north? New evidence for domestication.J Archaeol Sci 15(5):515–533.

21. Chow BS (1981) The animal remains discovered at Cishan village, Wu’an, HebeiProvince. Kaogu Xuebao 62(3):339–348. Chinese.

22. Chow BS (1983) The animal remains discovered at Baiyingsite, Tangyin, Henan Prov-ince. Papers on Chinese Archaeology 3:48–50. Chinese.

23. Deng H, Yuan J, Song G-D, Wang C-S, Masaki E (2013) The re-examination of theDomestic Chicken in Ancient China. Archaeology 6(6):83–96. Chinese.

24. Peters J (1998) Römische Tierhaltung und Tierzucht. Eine Synthese aus arch-äozoologischer Untersuchung und schriftlich-bildlicher Überlieferung. Passauer Uni-versitätsschriften zur Archäologie 5. (Marie Leidorf, Rahden, Westfalen, Germany).

25. Eriksson J, et al. (2008) Identification of the yellow skin gene reveals a hybrid origin ofthe domestic chicken. PLoS Genet 4(2):e1000010.

26. Kiefer C, et al. (2001) Identification and characterization of a mammalian enzymecatalyzing the asymmetric oxidative cleavage of provitamin A. J Biol Chem 276(17):14110–14116.

27. Tegetmeier WB (1867) The Poultry Book: Comprising the Breeding and Managementof Profitable and Ornamental Poultry, Their Qualities and Characteristics; To Which isadded “The Standard of Excellence in Exhibition Birds,” Authorized by the Poultry(G. Routledge and Sons, London).

28. McGowan PJK (1994) Phasianidae (Pheasants and Partridges). Handbook of the Birdsof the World, eds del Hoyo J, Elliott A, Sargatal J (Lynx Edicions, Barcelona, Spain), pp434–552.

29. Rubin C-J, et al. (2010) Whole-genome resequencing reveals loci under selectionduring chicken domestication. Nature 464(7288):587–591.

Girdland Flink et al. PNAS Early Edition | 5 of 6

ANTH

ROPO

LOGY

SPEC

IALFEATU

RE

30. Benecke N (1994) Archäozoologische Studien zur Entwicklung der Haustierhaltung in

Mitteleuropa und Südskandinavien von den Anfängen bis zum ausgehenden Mitte-

lalter (Akademie, Berlin).31. Carrasquila Hernández F (1992) Some comments on the introduction of domestic fowl

in Iberia. Archaeofauna 1:45–93.32. Dana N, et al. (2010) East Asian contributions to Dutch traditional and western

commercial chickens inferred from mtDNA analysis. Anim Genet 42(2):125–133.33. Storey AA, et al. (2007) Radiocarbon and DNA evidence for a pre-Columbian in-

troduction of Polynesian chickens to Chile. Proc Natl Acad Sci USA 104(25):

10335–10339.34. Storey AA, et al. (2012) Investigating the global dispersal of chickens in prehistory

using ancient mitochondrial DNA signatures. PLoS ONE 7(7):e39171.35. Miao YW, et al. (2013) Chicken domestication: an updated perspective based on

mitochondrial genomes. Heredity (Edinb) 110(3):277–282.36. Roaf M (1990) Cultural Atlas of Mesopotamia and the Ancient Near East (Equinox,

Facts on File, New York).37. Aldrovandi U (1600) Ornithologiae Tomus Alter cum Indice Copiosissimo Variarum

Linguarum (Giovanni Battista Bellagamba, Bologna).38. Gongora J, et al. (2008) Indo-European and Asian origins for Chilean and Pacific

chickens revealed by mtDNA. Proc Natl Acad Sci USA 105(30):10308–10313.39. Revay T, Bodzsar N, Mobegi VE, Hanotte O, Hidas A (2010) Origin of Hungarian in-

digenous chicken breeds inferred from mitochondrial DNA D-loop sequences. Anim

Genet 41(5):548–550.

40. Uauy C, Distelfeld A, Fahima T, Blechl A, Dubcovsky J (2006) A NAC Gene regulatingsenescence improves grain protein, zinc, and iron content in wheat. Science314(5803):1298–1301.

41. Asplund L, Hagenblad J, Leino MW (2010) Re-evaluating the history of the wheat do-mestication gene NAM-B1 using historical plant material. J Archaeol Sci 37(9):2303–2307.

42. Jaenicke-Després V, et al. (2003) Early allelic selection in maize as revealed by ancientDNA. Science 302(5648):1206–1208.

43. Malmström H, et al. (2008) Barking up the wrong tree: Modern northern Europeandogs fail to explain their origin. BMC Evol Biol 8:71.

44. Ottoni C, et al. (2013) Pig domestication and human-mediated dispersal in westernEurasia revealed through ancient DNA and geometric morphometrics. Mol Biol Evol30(4):824–832.

45. Barnes I, Matheus P, Shapiro B, Jensen D, Cooper A (2002) Dynamics of Pleistocenepopulation extinctions in Beringian brown bears. Science 295(5563):2267–2270.

46. Shapiro B, et al. (2004) Rise and fall of the Beringian steppe bison. Science 306(5701):1561–1565.

47. Larson G, et al. (2012) Rethinking dog domestication by integrating genetics, arche-ology, and biogeography. Proc Natl Acad Sci USA 109(23):8878–8883.

48. Boessneck J (1988) Die Tierwelt des Alten Ägypten (C. H. Beck, Munich).49. Cooper A, Poinar HN (2000) Ancient DNA: Do it right or not at all. Science 289(5482):1139.50. Gilbert MT, Bandelt HJ, Hofreiter M, Barnes I (2005) Assessing ancient DNA studies.

Trends Ecol Evol 20(10):541–544.51. Drummond AJ, et al. (2011) Geneious v5.4, Available from www.geneious.com/. Ac-

cessed July 1, 2011.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1308939110 Girdland Flink et al.

Supporting InformationGirdland Flink et al. 10.1073/pnas.1308939110SI Materials and MethodsArchaeological Specimens. Eighty ancient chicken bones from 12archaeological sites, representing three major European locations(United Kingdom, Central Europe, and Greece) and two majortime bins [(i) Late Iron Age La Tène C and D, Late Hellenistic/Early Roman to Roman, and (ii) medieval and postmedievalperiods] were extracted for DNA (Table S1).Although our sampled specimens are relatively modern and

geographically distant relative to the domestication centers insoutheast Asia, which date to as early as 6000 B.C. (1) (although,see the main text for a discussion on issues relating to Gallusfinds in some early East Asian contexts), they represent amongthe earliest, if not the earliest, introductions to Europe. The firstwave of introduction likely arrived around 900–700 B.C. inGreece and on the Iberian Peninsula (by Phoenician sea trade).The central European chicken was likely derived from the earlyGreek founder population, which was distributed throughout thenorthern Mediterranean, and from there introduced to CentralEurope across the Alps around 600–400 B.C. Zooarchaeologicalevidence for the presence of chicken in Central Europe (Ger-many, Austria, and Switzerland) dates to the Hallstatt C–D pe-riod ca. 800–475 B.C. One major mechanism of introductionconsisted of trade with ancient Greek settlement colonies in thecoastal regions of the northwestern Mediterranean. During thefollowing La Tène period (475–30/1 B.C.) Gallus bones remainrare in archaeological contexts, with frequencies increasingslightly during its later phases. After the Roman conquest,poultry-keeping became relatively rapidly and widely establishedin the northwestern Provinces of the Roman Empire, but not soin regions outside the area of influence of Rome, where it onlybecame economically important in medieval times (2). We cantherefore safely assume that these Late Iron Age/Early Romanbirds descend from early Gallus introduced into Europe. For thebroadly contemporaneous birds from Classical Greece, however,this is not necessarily the case, considering the fact that in themeantime “Greece” had been part of the Persian, Greek, andRoman Empire with trade connections to Mesopotamia andIndia (3). Southeast Europe may therefore have witnessedmultiple introductions from the Near East.

Dating of Archaeological Specimens. The majority of the specimensanalyzed in this study were dated using stratigraphic and/orcontextual evidence (Fig. S1 and Table S1). Given the ease withwhich chicken bones can move between occupation phases, weprimarily sampled faunal assemblages from sites with no over-laying building structures or archaeological strata. A singlespecimen from Altenburg, Germany (Ch38 in Table S1) pre-sumed to date to Iron Age La Tène C/D contexts (280–15 B.C.),however, was the only ancient sample that possessed homozy-gous genotypes for both the sweep TSHR allele and yellow skinassociated Gray Junglefowl (GJF) BCDO2 (β-carotene dioxy-genase 2) allele (two genotypes primarily associated with moderncommercial breeds). In addition, although there was no sub-sequent occupation layer over the majority of the site, there isevidence of a medieval occupation at the outskirts of the site,and German troops potentially used the walls surrounding theLate Iron Age settlement during World War II. Given the ge-netic signals and the higher potential that this sample was in-trusive, we directly dated this bone, the result of which showedthat it dated to 150 ± 30 cal. B.P. (2 σ calibration; Beta Analytic,no. 356195).

We also directly radiocarbon dated a specimen from Kassope(Greece) where there was no evidence of subsequent occupationlayers, and this sample produced a date of 2150–1990 cal. B.P.(2 σ calibration; Beta Analytic, no. 351382), thus confirming thecontextual dating (Table S1). None of the other stratigraphicallydated samples come from sites with intrusive or overlaying oc-cupation layers, and the genetic results strongly suggest the ar-chaeologically derived dates are accurate.

Description of Key Archaeological Sites. Kassope, Greece. Kassope,Greece was an ancient Greek city founded 350 B.C. and aban-doned 30 B.C. (3). Friedl (4) presented an analysis of the fauna.The chicken remains studied here originate from house 3, oc-cupied during the second and first century B.C.Magdalensberg, Austria. Magdalensberg, Austria, was a tradingsettlement founded during the second third of the first centuryB.C. and inhabited until about A.D. 50 (5). The bird remains,including numerous chicken bones excavated at this site, havebeen identified by Dräger (6).Altenburg-Rheinau, Germany. Altenburg-Rheinau, Germany was aCeltic oppidum occupied between ca. 150 and 15 B.C. (7). Ex-cavations on the eastern peninsula during the 1970s producedthe chicken remains, which have been analyzed by Moser (8).Abodiacum/Epfach, Germany. Abodiacum/Epfach, Germany was aRoman military station and settlement, occupied with someinterruptions from the first until the beginning of the fifth centuryA.D. (9). Chicken remains have been found throughout the se-quence (10).Quintana/Künzing, Germany. The Gallus finds originate from theMithras sanctuary (Mithraeum) near the Roman settlement ofQuintana/Künzing, Germany. This sanctuary was in use from thesecond century A.D. until the second half of the third centuryA.D. (11). The birds represent offerings to the God Mithras.Excavations produced an assemblage of >7,500 chicken remains(12).Manching, Germany. Manching, Germany was a Celtic oppiduminhabited between ca. 300 B.C. until 50/30 B.C., probably thelargest city north of the Alps during this period. After aban-donment, some small-scale Roman occupation took place,probably in form of a mansio or perhaps vicus, lasting until themiddle of the third century A.D. (13). The chicken remains formpart of a huge faunal assemblage (>350,000 finds) constitutinga most useful source of information about late La Tène animalhusbandry in Central Europe (14).

United Kingdom: Arbeia, South Shields Roman Fort, England. Arbeiawas a Roman fort built at the mouth of the River Tyne in the latesecond century A.D., which was converted into a supply base forfood in the early second century A.D. After a large fire in thelate third or early fourth century A.D., the fort was redesignedand rebuilt, and continued in use into the early fifth centuryA.D. (15).The remaining chicken remains, from the north of England,

which were used in this study, came from two archaeological sitesin York (Spurriergate and St. Saviourgate) (16), located withinthe core of the medieval city; from a site in Beverley, East Ridingof Yorkshire (17); and from South England and an excavation inEast London (recovered from excavations in preparation for theconstruction of the Docklands Light Railways) (18).The vertebrate material from Spurriergate was recovered from

excavations in a former car park and beneath several 1960s buildingsthat were subsequently demolished before the archaeological

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 1 of 9

interventions. The chicken bones were retrieved mainly from fills ofrubbish pits and dump layers associated with occupation of Anglian/Anglo-Scandinavian, medieval and postmedieval dates. The exca-vations at St. Saviourgate largely revealed pits containing refuse ofa mixed nature, from primary butchery waste to household rubbish.The chicken bones were recovered from pit fills of late medievaldate.Chicken remains from Beverley, East Riding of Yorkshire,

were recovered from excavations at the site of the former PicturePlayhouse and Swimming Pool in the heart of the medieval townat the north side of SaturdayMarket. This had been a market areaof Beverley since the 12th century, becoming known as the CornMarket by the 14th century and as Saturday Market by the 16th.The site was the location of a meat market by the 18th century,and quite probably much earlier, with an arcaded butchers’shambles built in 1753 and a fish shambles built behind thebutchers’ market in 1777. All of the contexts from which the fowlbones came were pit fills or ground raising deposits of medievaland postmedieval date (16, 17).

Ancient DNA Laboratories and Experimental Set-Up. DNA extrac-tions and PCR amplifications were performed in a dedicatedancient DNA laboratory in the Department of Archaeology(Durham Evolution and Ancient DNA) at Durham University,United Kingdom. We followed strict laboratory proceduresaccording commonly used guidelines (19, 20). All equipment andwork surfaces were cleaned before and after each use with a di-lute solution of bleach [5–10% (wt/vol) active sodium hypochlo-rite] followed by ddH2O and ethanol [99% (vol/vol)]. Pipettes andplastic racks were UV-irradiated in a dedicated cross-linker (at<15 cm for at least 30 min at 254-nm wavelength) before andafter use. Pre- and post-PCR laboratories are physically isolated;access to the pre-PCR laboratories is restricted to Ancient DNAlaboratory users only and access is also prohibited if the labo-ratory user had entered post-PCR areas the same day. AncientDNA laboratory users wear clean laboratory coats, double layerof gloves (nitrile and latex), and overshoes to avoid introducingcontaminants from post-PCR areas.Independent replication of 12 ancient specimens was per-

formed in a dedicated ancient DNA laboratory at the Evolu-tionary Biology Centre, Uppsala University, Sweden. Laboratorywork follows commonly used guidelines (19, 20). Laboratoryusers wear coveralls, facemasks, dedicated laboratory clogs, anddouble layers of gloves. The laboratory is equipped with positiveair pressure and UV lamps and is routinely deep cleaned withbleach [1–5% (wt/vol) active sodium hypochlorite], ddH2O, anddilute ethanol [70% (vol/vol)].

Ancient DNA Extraction. The ancient chicken bones were preparedfor DNA extraction (oneDNA extract per specimen) by removingan approximately 1-mm layer of outer bone surface by abrasionusing a Dremel drill with clean, one-time use cut-off wheels(Dremel no. 409). A subsection of the bone was subsequentlyisolated and pulverized in a Micro-Dismembrator (Sartorious-Stedim Biotech), followed by collection in 15-mL Grainer tubes.Milling containers and grinding balls were subsequently sus-pended and cleaned in 1% virkon, and rinsed in absolute [99%(vol/vol)] ethanol.Next, 50- to 100-mg bone powder/specimen was digested in

0.425M EDTA (pH 8), 0.05% SDS, 0.05M Tris·HCI and 400 μgProteinase K, and incubated overnight on a rotator at 50 °C untilfully dissolved. The DNA extraction master mix, excluding Pro-teinase K, was UV-irradiated at 254 nm for 1 h in a cross-linkerbefore use. Once dissolved overnight, 2 mL of solution wasconcentrated in a Millipore Amicon Ultra-4 30 kDa molecularweight cut-off to a final volume of 100 μL. The concentratedDNA extract was purified using the QIAquick PCR PurificationKit (Qiagen) following manufacturers recommendations, except

that the final elution step was performed twice (2 × 50 μL) toproduce a final volume of 100 μL. One in five DNA extractionswere blank negative controls containing only extraction bufferand Proteinase K.During replication inUppsala, the ancient specimens wereUV-

irradiated at 1 J/cm2 per side (254-nm wave length) and 1 mm ofthe surface was removed before powderization. DNA was ex-tracted as in Svensson et al. (21). Next, 40- to 80-mg bonepowder was incubated in 1 mL of 0.5M EDTA, pH 8, 1 M Urea,and 100 μg Proteinase K for 22 h at 38 °C together with fournegative, blank controls. An additional 100 μg Proteinase K wasthen added and the samples were incubated for 3 more hours at55 °C. DNA was further extracted using Qiagen PCR Purifica-tion Kit and finally eluted in 100 μL Elution Buffer (Qiagen).

PCR Amplification. A 201-bp mitochondrial control region (CR)fragment (22, 23) and one SNP in each of two autosomal nuclearloci (BCDO2 and thyroid-stimulating hormone receptor, TSHR)were targeted for PCR amplification (Table S6). PCR set-up wasperformed in a fume hood in a dedicated facility adjacent to thededicated ancient DNA extraction facility. The PCR set-up fa-cility is subject to positive air pressure that minimizes the risk ofintroducing contaminant DNA. One in eight PCR reactions werenegative controls. In addition, one positive control (a modernGJF) was included for each round of PCR amplifications. Toavoid contaminating the ancient DNA PCR reactions withmodern GJF DNA, the modern positive control was stored inthe dedicated PCR/post-PCR room and added to the reactiontubes once placed on the thermal cycler. PCRs were visualizedon a 1–2% (wt/vol) agarose gel using GelRed and UV illumi-nation. PCR products were purified using ExoSAP-IT (USBAffymetrix) and stored at −20 °C before sequencing.Independent replication of CR sequences and the TSHR SNP

was performed in Uppsala with slight modifications to the PCRprotocol: addition of RSA (rabbit serum albumine) was usedinstead of BSA (bovine serum albumine) and Smart Taq (Naxo)was used instead of Taq Gold (see below). Apart from followingthe PCR cycling conditions described below, replicate PCRs forthe CR sequence were also performed following Storey et al.(23).

Mitochondrial DNA CR (201 bp). PCRs were setup in 25-μL reactionsusing 1 U Taq GOLD (Applied Biosystems), 1× Gold buffer(Applied Biosystems), 2.5 mM MgCl2, 0.5 μg/μL BSA, 200 μM ofeach dNTP, 0.4 μM of each forward and reverse primers, and2–5 μL of aDNA extract. One in eight PCR reactions werenegative blank controls. PCR cycling conditions were 95 °C for 5min, 50 cycles of 94 °C for 30 s, 54 °C for 30 s, and 72 °C for 30 s,followed by 72 °C for 10 min.

BCDO2.PCRs were set up in 25-μL reactions using 1.0–1.25 U TaqGOLD (Applied Biosystems), 1× Gold buffer (Applied Bio-systems), 2.5 mMMgCl2, 0.5 μg/μL BSA, 1 M betaine, 200 μM ofeach dNTP, 0.4 μM of each primer, and 2–5 μL of ancient DNAextract. One in eight PCR reactions were negative blank con-trols. PCR cycling conditions were 95 °C for 5 min, 50 cycles of94 °C for 30 s, 57 °C for 30 s, and 72 °C for 30 s, followed by 72 °Cfor 10 min.

TSHR. PCRs were set up in 25 μL reactions using 1.0–1.25 U TaqGOLD (Applied Biosystems), 1× Gold buffer (Applied Bio-systems), 2.5 mM MgCl2, 0.5 μg/μL BSA, 200 μM of each dNTP,0.6 μM of the biotinylated forward primer, and 0.8 μM of thereverse primer, and 2–5 μL of aDNA extract. One in eight PCRreactions were negative blank controls. PCR cycling conditionswere 95 °C for 5 min, 50 cycles of 94 °C for 30 s, 56 °C for 30 s,and 72 °C for 30 s, followed by 72 °C for 10 min.

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 2 of 9

DNA Sequencing, Genotyping, and Data Analysis. Sanger sequencingwas performed on the Applied Biosystems 3730 DNAAnalyzer atthe DNA Sequencing Service at the School of Biological andBiomedical Sciences (Durham University). Trace files weremanually inspected using 4Peaks (Mekentosj) or Geneious v.5.4(24) and built into contigs by hand in Se-Al (http://tree.bio.ed.ac.uk/software/seal) or automatically in Geneious v.5.4 using theassembly function (default parameters). At least two (but usuallythree) independent PCR replicates/sequences (mitochondrialDNA, mtDNA) per sample were performed to ensure authen-ticity. Sporadic nonreplicable postmortem DNA template dam-age was observed in several sequences (C→T transitions). Thistype of damage often results from deamination through hydro-lysis resulting in the conversion of cytosine bases to uracil (al-ternatively hydroxyuracil) or adenine to hypoxanthine. Uracil isread as thymine and hypoxanthine as guanine by DNA poly-merases during PCR amplification and subsequently induces thecommon type 2 (C/T or A/G) damage (20, 25, 26).Pyrosequencing was performed in-house at the Archaeology

department in Durham using the PyroMark Q24 (Qiagen) fol-lowing manufacturers guidelines, and using Qiagen Q24 se-quencing reagent kits. Results, sequences, and genotypes wereanalyzed in the PyroMark Q24 software (Qiagen) using modifiedsettings: accepted peak deviation andminimumpeak heights wereset to less strict to account for low signal intensity and slightdeviations in peak heights (which, if observed, could be the resultof Type-2 C→U deamination/error). Dispensation order wasautomatically generated using the PyroMark Q24 software(Qiagen) (Table S6).To account for allelic dropout that is common in ancient DNA

studies (21), each SNP/genotype was confirmed by repeatedgenotyping from two to eight independent PCRs (at least twoindependent replications for heterozygous specimens, but up toeight replications in for homozygous specimens and heterozy-

gous specimens for which we repeatedly observed allelic drop-out). The probability of falsely assigning a heterozygousindividual as homozygous was calculated as follows: P(false ho-mozygote) = K × (K/2)n − 1, where n is the number of replicatesand K is the observed number of allelic dropouts divided by thetotal number of genotypings of heterozygous individuals (21, 27).Fisher’s exact test, as implemented in R v.2.15.2 (28), was used

to test for differences in allele (and mtDNA CR haplotype)frequencies between modern and ancient populations. In addi-tion, binomial probability distributions (the pdbinom, dbinom,and gbinom functions in R v.2.15.2) were used to examine dif-ferences in allele frequencies between ancient and modernpopulations and to authenticate the ancient DNA results, as-suming that putative contamination would reflect genotype fre-quencies in modern populations (Fig. S2 and Table S2).In Uppsala, the TSHR SNP was genotyped on a PSQ 96MA

(Biotage), using pyrosequencing technology and the SNP soft-ware and SNP reagent kit (Biotage). Sample preparation wasperformed according to Pyrosequencing instructions using 25 μLof PCR product.

Phylogenetic Reconstruction.An alignment of previously publishedmtDNA CR haplotypes (29, 30) was used as a reference toidentify haplotypes (Table S4 depicts unique E-clade hap-lotypes). Modern reference and ancient DNA consensus se-quences were aligned in Geneious v.5.4 (24) using MAFFT (31).Phylogentic analysis was performed using PhyML v.3.0 (32) asimplemented in Geneious v.5.4 (24). The nucleotide substitutionmodel was estimated in using MrModeltest 2.3 (33) and PAUP*4.0 (34), as implemented in MrMtGui (35). The best-fit modelusing both adjusted likelihood ratio test statistics and AkaikeInformation Criterion was HKY+I+G. Nodal support valueswere estimated through bootstrapping. A Gallus gallus bankivasequence (AB007718) was used as an outgroup.

1. West B, Zhou B-X (1988) Did chickens go north? New evidence for domestication. JArchaeol Sci 15(5):515–533.

2. Peters J (1998) Römische Tierhaltung und Tierzucht. Eine Synthese aus archäozoologischerUntersuchung und schriftlich-bildlicher Überlieferung. Passauer Universitätsschriften zurArchäologie 5. (Marie Leidorf, Rahden, Westfalen, Germany).

3. Hoepfner W, Dakaris S, Gravani K, Schwandner EL (1999) Kassope. Eine spätklassischeStreifenstadt in Nordwestgriechenland. Geschichte des Wohnens, Band I: 5000 v. Chr. –500 n. Chr., Vorgeschichte, Frühgeschichte, Antike, ed Hoepfner W ( Deutsche Verlags-Anstalt, Stuttgart), pp 368–383.

4. Friedl H (1984) Tierknochenfunde aus Kassope/Griechenland (4.-1. Jh. v. Chr.).Dissertation (Ludwig Maximilian Univ of Munich, Munich).

5. Sedlmayer H (2009) Die Fibeln vom Magdalensberg. Funde der Grabungsjahre 1948 –

2002 und Altfunde des 19. Jahrhunderts (Landesmuseums für Kärnten, Klagenfurt a.Wörthersee).

6. Dräger N (1964) Kärntner Museumsschriften 33. Klagenfurt. Dissertation, LudwigMaximilian Univ of Munich. (Landesmuseums für Kärnten, Klagenfurt a. Wörthsee).

7. Fischer F (2004) Das Oppidum Altenburg-Rheinau und sein spätlatènezeitlichesUmfeld. Spätlatènezeit und frühe römische Kaiserzeit zwischen Alpenrand undDonau, Akten Kolloquium Ingolstadt am 11. und 12. Oktober 2001. Kolloquien zurVor- und Frühgeschichte, eds Hüssen C-M, Irlinger W, Zanier W (Habelt, Bonn), Vol 8,pp 123–131.

8. Moser B (1986) Die Tierknochenfunde aus dem latènezeitlichen Oppidum vonAltenburg-Rheinau. I. Charakterisierung des Fundgutes, Pferd, Hund, Hausgeflügelund Wildtiere. Dissertation (Ludwig Maximilian Univ of Munich, Munich).

9. Werner J (1969)Der Lorenzberg bei Epfach. Die spätrömischen und frühmittelalterlichenAnlagen. Münchner Beiträge zur Vor- u. Frühgeschichte 2 (C.H. Beck, Munich).

10. Boessneck J (1964) Die Tierknochen aus den Grabungen 1954–1957 auf demLorenzberg bei Epfach. Studien zu Abodiacum-Epfach, ed Werner J (C.H. Beck,Munich), pp 213–261.

11. Bayer-Niemeier E (2004) Museum Quintana. Archäologie in Künzing (Museum Quintana,Künzing).

12. von den Driesch A, Pöllath N (2000) Tierknochen aus dem Mithrastempel vonKünzing, Lkr. Deggendorf. Vorträge des 18. Niederbayerischen Archäologentages, edSchmotz K (Marie Leidorf, Rahden, Westfalen, Germany), pp 145–162.

13. Sievers S (2003) Manching–Die Keltenstadt. Stuttgart: Thesis.14. Boessneck J, von den Driesch A, Meyer-Lemppenau U, Wechsler-von Ohlen E (1971)

Die Tierknochenfunde aus Manching. Die Ausgrabungen in Manching 6 (Steiner,Wiesbaden).

15. Hodgson N (1994) Period 7: The courtyard house. Excavations at South Shields RomanFort, eds Bidwell P, Speak S (Soc. Antiq. Newcastle upon Tyne, England), Vol 1, pp 35–39.

16. Carrott J, et al. (1998) An assessment of biological remains from excavations at landadjacent to the former Gibraltar Farm, Kingswood, Kingston upon Hull (site codeGIB97). Reports from the Environmental Archaeology Unit, York 98/6, pp 13.

17. Jaques D, et al. (2009) Evaluation of biological remains from an archaeologicalexcavation and watching brief at the former Picture Playhouse and Swimming Pool,Saturday Market, Beverley, East Riding of Yorkshire (site code: BPH2008).Palaeoecology Research Services PRS 2009/54.

18. Schofield J (1998) Archaeology in the City of London, 1907–1991: A guide to recordsof excavations by the Museum of London and its predecessors. The ArchaeologicalGazetteer Series, ed Maloney C (Museum of London, London), Vol 1.

19. Cooper A, Poinar HN (2000) Ancient DNA: Do it right or not at all. Science 289(5482):1139.

20. Gilbert MT, Bandelt HJ, Hofreiter M, Barnes I (2005) Assessing ancient DNA studies.Trends Ecol Evol 20(10):541–544.

21. Svensson EM, et al. (2007) Tracing genetic change over time using nuclear SNPs inancient and modern cattle. Anim Genet 38(4):378–383.

22. Storey AA, et al. (2010) Mitochondrial DNA from 3000-year old chickens at theTeouma Site, Vanuatu. J Archaeol Sci 37(10):2459–2468.

23. Storey AA, et al. (2007) Radiocarbon and DNA evidence for a pre-Columbianintroduction of Polynesian chickens to Chile. Proc Natl Acad Sci USA 104(25):10335–10339.

24. Drummond AJ, et al. (2011) Geneious v5.4, Available from www.geneious.com/.Accessed July 1, 2011.

25. Gilbert MTP, et al. (2007) Recharacterization of ancient DNA miscoding lesions:Insights in the era of sequencing-by-synthesis. Nucleic Acids Res 35(1):1–10.

26. Hofreiter M, Jaenicke V, Serre D, von Haeseler A, Pääbo S (2001) DNA sequences frommultiple amplifications reveal artifacts induced by cytosine deamination in ancientDNA. Nucleic Acids Res 29(23):4793–4799.

27. Gagneux P, Boesch C, Woodruff DS (1997) Microsatellite scoring errors associatedwith noninvasive genotyping based on nuclear DNA amplified from shed hair. MolEcol 6(9):861–868.

28. R Development Core Team (2012) R: A Language and Environment for StatisticalComputing (R Foundation for Statistical Computing, Vienna, Austria).

29. Dana N, et al. (2010) East Asian contributions to Dutch traditional and Westerncommercial chickens inferred from mtDNA analysis. Anim Genet 42(2):125–133.

30. Liu Y-P, et al. (2006) Multiple maternal origins of chickens: Out of the Asian jungles.Mol Phylogenet Evol 38(1):12–19.

31. Katoh K, Misawa K, Kuma K, Miyata T (2002) MAFFT: A novel method for rapidmultiple sequence alignment based on fast Fourier transform. Nucleic Acids Res 30(14):3059–3066.

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 3 of 9

32. Guindon S, et al. (2010) New algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the performance of PhyML 3.0. Syst Biol 59(3):307–321.

33. Nylander JAA (2004) MrModeltest v2. Program distributed by the author (Evolu-tionary Biology Centre, Uppsala Univ, Uppsala, Sweden).

34. Swofford DL (2003) PAUP*. Phylogenetic Analysis Using Parsimony (* and OtherMethods). Version 4. (Sinauer Associates, Sunderland, MA).

35. Nuin P (2010) MrMtgui. Available at www.genedrift.org/software/mrmtgui.html.Accessed July 12, 2011.

Fig. S1. A map depicting archaeological sites, the number of sampled specimens and DNA retrieval success rate (success indicated as at least one reproduciblegenotype/specimen) (Table S1). Black represents success, gray represents failure.

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 4 of 9

Fig. S2. A figure depicting the number of expected TSHR wild type alleles in a sample of 44 (corresponding to the biallelic genotypes obtained from theancient European samples) given frequencies reported previously for modern chickens (1) (Table S2), and the expected number of non–E-clade mtDNAhaplotypes in a sample of 43 (representing the ancient European chickens for which we have available mtDNA CR haplotypes) given the frequencies reportedpreviously for modern European populations (∼15%) (2–4) (Table S5). The green stars depict the empirically observed frequencies in the ancient Europeansamples (Tables S2 and S5).

1. Rubin C-J, et al. (2010) Whole-genome resequencing reveals loci under selection during chicken domestication. Nature 464(7288):587–591.2. Dana N, et al. (2010) East Asian contributions to Dutch traditional and western commercial chickens inferred from mtDNA analysis. Anim Genet 42(2):125–133.3. Liu Y-P, et al. (2006) Multiple maternal origins of chickens: Out of the Asian jungles. Mol Phylogenet Evol 38(1):12–19.4. Revay T, Bodzsar N, Mobegi VE, Hanotte O, Hidas A (2010) Origin of Hungarian indigenous chicken breeds inferred from mitochondrial DNA D-loop sequences. Anim Genet 41(5):

548–550.

Fig. S3. (A) A maximum-likelihood tree constructed using the haplotypes reported by Liu et al. (30). Numbers indicate bootstrap support values. (B) Amaximum-likelihood tree constructed using the 201-bp (23) mtDNA CR haplotypes of Liu et al. (30) amplified for the ancient European specimens (SI Materialsand Methods, and Tables S4 and S6).

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 5 of 9

Table

S1.

Archae

ological

chicke

nbones

inve

stigated

forthis

study

Sample

Referen

ceID

Country

Site

Culturalco

ntext

Dates

Contact

mtD

NA

(Gen

Ban

kaccession)TS

HR

BCDO2

Ch1

NrD55

3771

ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

1)Ch2

NrD55

3771

ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

2)Ch3†

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

3)W/W

GJF/RJF

Ch4†

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

4)S/S

RJF/RJF

Ch5

NrD53

830ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

5)Ch6

NrD53

739ob15

09German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

Ch7†

NrD53

760ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

6)S/S

Ch8

NrD53

760ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

7)Ch9

NrD53

760ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

8)Ch10

†NrD53

771ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5325

9)W/W

RJF/RJF

Ch11

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

0)Ch12

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

1)W/S

GJF/RJF

Ch13

†NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

2)W/S

RJF/RJF

Ch14

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

3)W/W

Ch15

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

4)W/S

RJF/RJF

Ch16

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

5)W/S

RJF/RJF

Ch17

NrD53

772ob15

13B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

6)W/S

Ch18

NrD53

884ob15

09B

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

7)Ch19

NrD53

858ob15

09German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

8)S/S

GJF/RJF

Ch20

NrD53

858ob15

09German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5326

9)W/S

RJF/RJF

Ch21

NrD53

771ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5327

0)W/W

RJF/RJF

Ch22

NrD53

771ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

W/S

Ch23

NrD53

771ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd-thirdcentury

A.D.

J.P.

W/W

Ch24

NrD53

771ob15

13A

German

yQuintana/Künzing

Roman

Seco

nd–thirdcentury

A.D.

J.P.

E*(KF7

5327

1)W/S

Ch25

56/112

1German

yAbodiacu

m,Ep

fach

EarlyRoman

First–fifthcentury

A.D.

J.P.

E*(KF7

5327

2)W/S

Ch26

56/554

German

yAbodiacu

m,Ep

fach

EarlyRoman

First–fifthcentury

A.D.

J.P.

W/W

Ch27

†56

/100

9German

yAbodiacu

m,Ep

fach

EarlyRoman

First–fifthcentury

A.D.

J.P.

E*(KF7

5327

3)W/S

RJF/RJF

Ch28

56/112

1German

yAbodiacu

m,Ep

fach

EarlyRoman

First–fifthcentury

A.D.

J.P.

Ch29

?/36

7German

yMan

ching

IronAgeLa

TèneCan

dD

200–

30B.C.

J.P.

Ch30

62/336

German

yMan

ching

IronAgeLa

TèneCan

dD

200–

30B.C.

J.P.

Ch31

†23

9German

yMan

ching

IronAgeLa

TèneCan

dD

200–

30B.C.

J.P.

E*(KF7

5327

4)W/W

RJF/RJF

Ch32

395

German

yMan

ching

IronAgeLa

TèneCan

dD

200–

30B.C.

J.P.

Ch33

†55

Austria

Mag

dalen

sberg

EarlyRoman

100B.C.–A.D.50

J.P.

E*(KF7

5327

5)W

/SCh34

24Austria

Mag

dalen

sberg

EarlyRoman

100B.C.–A.D.50

J.P.

W/S

Ch35

3Austria

Mag

dalen

sberg

EarlyRoman

100B.C.–A.D.50

J.P.

S/S

Ch36

38Austria

Mag

dalen

sberg

EarlyRoman

100B.C.–

A.D.50

J.P.

W/S

RJF/RJF

Ch37

A14

0-43

1German

yAlten

burg

IronAgeLa

TèneCan

dD

280–

15B.C.

J.P.

W/S

Ch38

‡A11

5-40

6German

yAlten

burg

Postindustrial

150±

30cal.B.P.

J.P.

E*(KF7

5327

6)S/S

GJF/GJF

Ch39

A14

7-43

9German

yAlten

burg

IronAgeLa

TèneCan

dD

280–

15B.C.

J.P.

Ch40

A46

-202

German

yAlten

burg

IronAgeLa

TèneCan

dD

280–

15B.C.

J.P.

W/S

RB36

782

09/558

9En

gland

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

RB36

8†24

13/558

9En

gland

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

E*(KF7

5327

7)W/W

GJF/RJF

RB36

971

2/55

89En

gland

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

E*(KF7

5327

8)W/W

RJF/RJF

RB37

0†35

22/89

England

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

E*(KF7

5327

9)W/W

RJF/RJF

RB37

170

2/55

89En

gland

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

RB37

221

9/55

89En

gland

Arbeia,

South

Shields

Roman

A.D.12

0–40

0A.C.

E*(KF7

5328

0)W

/S

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 6 of 9

Table

S1.

Cont.

Sample

Referen

ceID

Country

Site

Culturalco

ntext

Dates

Contact

mtD

NA

(Gen

Ban

kaccession)TS

HR

BCDO2

RB37

32,

(MFI87

)co

ntext

[156

]En

gland

London,DLR

Roman

,Flav

ianperiod

A.D.69

–96

F.G.

W/W

RB37

4†12

,(M

FI87

)co

ntext

[156

]En

gland

London,DLR

Roman

,Flav

ianperiod

A.D.69

–96

F.G.

W/S

GJF/RJF

RB37

55,

(MFI87

)co

ntext

[156

]En

gland

London,DLR

Roman

,Flav

ianperiod

A.D.69

–96

F.G.

W/S

RJF/RJF

RB37

611

,(M

FI87

)co

ntext

[156

]En

gland

London,DLR

Roman

,Flav

ianperiod

A.D.69

–96

F.G.

W/S

RJF/RJF

RB37

8SLW04

[169

]En

gland

Scott

Lane,

Wetherby

Med

ieva

l11

th/12thcentury

A.D.

D.D.

E*(KF7

5328

1)W/W

RB37

919

95.434

[114

]En

gland

St.Sa

viourgate,

York

Med

ieva

l14

th/15thcentury

A.D.

D.D.

RB38

0†19

95.434

[115

]En

gland

St.Sa

viourgate,

York

Med

ieva

l14

th/15thcentury

A.D.

D.D.

W/S

RJF/RJF

RB38

120

00.584

[302

9]En

gland

Spurriergate,

York

Med

ieva

l11

th/12thcentury

A.D.

D.D.

E6(KF7

5328

7)W/W

RB38

220

00.584

[700

4]En

gland

Spurriergate,

York

Med

ieva

l10

th/14thcentury

A.D.

D.D.

RB38

320

00.584

[308

0]En

gland

Spurriergate,

York

Med

ieva

l10

th/11thcentury

A.D.

D.D.

W/W

RB38

420

00.584

[375

7]En

gland

Spurriergate,

York

Med

ieva

l12

th/14thcentury

A.D.

D.D.

E6(KF7

5328

8)S/S

RB38

5[114

6]En

gland

Play

house,Bev

erley

Postmed

ieva

l16

th-18thcentury

A.D.

D.D.

W/S

RB38

6[121

6]En

gland

Play

house,Bev

erley

Med

ieva

l12

th-15thcentury

A.D.

D.D.

RB38

7[120

9]En

gland

Play

house,Bev

erley

Med

ieva

l12

th-15thcentury

A.D.

D.D.

RB38

8[110

9]En

gland

Play

house,Bev

erley

Postmed

ieva

l16

th-18thcentury

A.D.

D.D.

E3(KF7

5328

9)W

/SRA6

H3-Huhn-sca1

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*(KF7

5328

2)S/S

RJF/RJF

RA7

H3-Huhn-sca2

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA8

H3-Huhn-sca3

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA9

H3-Huhn-sca4

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA10

H3-Huhn-sca5

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*(KF7

5328

3)RA11

H3-Huhn-cor1

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RJF/RJF

RA12

H3-Huhn-cor2

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA13

H3-Huhn-cor3

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA14

H3-Huhn-cor4

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RJF/RJF

RA15

H3-Huhn-cor5

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA16

H3-Huhn-cor6

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA17

H3-Huhn-tmt1

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA18

H3-Huhn-tmt2

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*(KF7

5328

4)S/S

RJF/RJF

RA19

H3-Huhn-tmt3

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA20

H3-Huhn-tmt4

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*(KF7

5328

5)RA21

H3-Huhn-tit1

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA22

§H3-Huhn-tit2

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*(KF7

5328

6)S/S

RJF/RJF

RA23

H3-Huhn-tit3

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA24

H3-Huhn-tit4

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

RA25

H3-Huhn-tit5

Greece

Kassope

Late

Hellenistic/Ea

rlyRoman

200–

40B.C.

J.P.

E*=E1

,E5

,E1

2,E1

5orE1

6acco

rdingto

thehap

lotypenomen

clature

ofLiuet

al.(29).Se

efurther

inform

ationin

Table

S4.So

urces:A.C.,AlexCroom,Ty

nean

dW

earArchives

andMuseums,So

uth

Shields,

United

Kingdom;D.D.,Deb

orahJaques,Pa

laeo

ecologyResea

rchSe

rvices

Ltd,KingstonuponHull,

United

Kingdom;F.G.,Fran

cisGrew,Museum

ofLo

ndon,Lo

ndon;J.P,

JorisPe

ters.Gen

otypenomen

clature:

TSHR:S,

swee

pallele;W

,wild

-typ

eallele;BCDO2:

GJF,GrayJunglefowlallele

(yellow

skin);RJF,Red

Junglefowlallele.

†Indep

enden

tlyreplicated

inUppsala

forTS

HRan

dthemtD

NA

CR.

‡Directlyradiocarbondated

to15

0±

30cal.B.P.(2

σcalib

ration).

§Directlyradiocarbondated

to21

50–19

90cal.B.P.(2

σcalib

ration).

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 7 of 9

Table S2. Genotype frequencies in ancient and modern populations

Populaton type

TSHRBCDO2 [SNP B in Eriksson

et al. (1)]

W/W W/S S/S GJF/GJF GJF/RJF RJF/RJF

Modern [Eriksson et al. (1) and Rubin et al., (2)] 0 7 264 73 1 36Roman England 4 4 0 0 2 4Medieval and postmedieval England 3 3 1 0 0 1Iron Age and Roman Central Europe 7 15 4 0 3 10Late Hellenistic/Early Roman Greece 0 0 3 0 0 5Total (ancient) 14 22 8 0 5 20

1. Eriksson J, et al. (2008) Identification of the yellow skin gene reveals a hybrid origin of the domestic chicken. PLoS Genet 4(2):e1000010.2. Rubin C-J, et al. (2010) Whole-genome resequencing reveals loci under selection during chicken domestication. Nature 464(7288):587–591.

Table S3. Hardy–Weinberg test on TSHR genotypes fromKünzing

Genotypes Observed Expected

Homozygote reference 5 5.1Heterozygote 8 7.9Homozygote variant 3 3.1Variant (sweep) allele frequency 0.44

χ2 = 0.004; χ2 P value with 1 df. = 0.95.

Table S4. E-clade haplotypes from Liu et al. (30) and ancient European haplotypes observed inthis study

Nucleotide position

1 2 2 2 2 2 2 2 2 3 3 3 3 3 3 42 4 9 2 2 4 4 4 5 6 8 3 4 4 5 7 9 42 4 9 2 4 3 6 9 6 1 1 0 2 4 5 2 1 6

Reference (A2) T C T A T T C A T C A C A A T A C C

E* ? ? . . . C . . C T . . . . . ? ? ?E1 (E*) . . . . . C . . C T . . . . . . . TE2 . . . . . C . . C . . T . . . . . TE3 . . . . . C . . C T . T . . . . . TRB388 ? ? . . . C . . C T . T . . . ? ? ?E4 . . . . . . . . C T . . . . . . . TE5 (E*) . . . . . C . . C T . . . . . . . .E6 . . C . . C . . C T . . . . . . . TRB381 ? ? C . . C . . C T . . . . . ? ? ?RB384 ? ? C . . C . . C T . . . . . ? ? ?E7 . . C . . C . . . T . . . . . . . TE8 . . . . . C T . C T . . . . . . . TE9 . . . . . C . . C T . . G . . . . TE10 . . . . . . . . . T . . . . . . . TE11 . . . G . C . . C T . T . . . . . TE12 (E*) . . . . . C . . C T . . . . . . T TE13 . . . . . C T G C T . . . G C . . TE14 G T . . . C T . C T . . . . . . . TE15 (E*) . . . . . C . . C T . . . . . . A TE16 (E*) . . . . . C . . C T . . . . . G . TE17 . . . . C C T . C T . . . . . . . TE18 . . . G . C . G C T G . . . C . . T

The shaded region corresponds to the 201-bp sequence (23) amplified and sequenced for this study. Areas outsidethe shaded region correspond to the remaining 318-bp region presented by Liu et al. (30). E* = ancient 201-bpmtDNA haplotype (Table S1) and corresponds to several 519-bp E-clade haplotypes reported by Liu et al. (30). Boldhaplotypes/specimens correspond to the 201-bp haplotypes amplified and sequenced from ancient European chick-ens in this study (Table S1). Variable nucleotide positions are highlighted in accordance to Table S1 (haplotype A2)from Liu et al. (30) (corresponding to nucleotide positions in, e.g., GenBank accession no. AY392407).

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 8 of 9

Table S5. MtDNA haplogroup frequencies in ancient and modern datasets

Source Hg A Hg B Hg C Hg D Hg E

Dana et al. (1) 30 15 12 6 323Liu et al. (2) 4 1 0 0 53Revay et al. (3) 1 11 0 0 65Total modern 35 27 12 6 441Proportion modern 0.07 0.05 0.02 0.01 0.85(Ancient) Storey et al. (4) 0 0 0 0 5(Ancient) Present study 0 0 0 0 38Ancient total 0 0 0 0 43Proportion ancient 0.00 0.00 0.00 0.00 100.00

1. Dana N, et al. (2010) East Asian contributions to Dutch traditional and western commercial chickens inferred from mtDNA analysis. Anim Genet 42(2):125–133.2. Liu Y-P, et al. (2006) Multiple maternal origins of chickens: Out of the Asian jungles. Mol Phylogenet Evol 38(1):12–19.3. Revay T, Bodzsar N, Mobegi VE, Hanotte O, Hidas A (2010) Origin of Hungarian indigenous chicken breeds inferred from mitochondrial DNA D-loop sequences. Anim Genet 41(5):

548–550.4. Storey AA, et al. (2012) Investigating the global dispersal of chickens in prehistory using ancient mitochondrial DNA signatures. PLoS ONE 7(7):e39171.

Table S6. PCR and sequencing primers for the mtDNA CR, TSHR, and BCDO2

Locus Primer sequence (5′–3′) Primer name Source Sequence to analyze

MtDNA CR ACCCATTATATGTATACGGGCATTAA GG144F Storey et al. (23)MtDNA CR CGAGCATAACCAAATGGGTTAGA GG387R Storey et al. (23)TSHR CTTTCTTCTTGCCCTTTT TSHR-F (biotin) Present studyTSHR GATGCTGACTTTGCTGTA TSHR-R Present studyTSHR TGCTGTAGCTGCTGACTC TSHR-S Present study C/TAACCAGTGG

BCDO2 ACTCTTGCATGGATCTGG BCDO2-F (biotin) Present studyBCDO2 TGTGGTCTCAGAATTTGG BCDO2-R Present studyBCDO2 TCAGAATTTGGGACG BCDO2-S Present study C/TTGGCAATGC

Girdland Flink et al. www.pnas.org/cgi/content/short/1308939110 9 of 9