This is a repository copy of Ancient DNA reveals the timing and persistence of organellar genetic bottlenecks over 3,000 years of sunflower domestication and improvement . White Rose Research Online URL for this paper: http://eprints.whiterose.ac.uk/132660/ Version: Published Version Article: Wales, Nathan orcid.org/0000-0003-0359-8450, Akman, Melis, Watson, Ray H.B. et al. (5 more authors) (2018) Ancient DNA reveals the timing and persistence of organellar genetic bottlenecks over 3,000 years of sunflower domestication and improvement. Evolutionary applications. ISSN 1752-4563 https://doi.org/10.1111/eva.12594 [email protected]https://eprints.whiterose.ac.uk/ Reuse This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: https://creativecommons.org/licenses/ Takedown If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [email protected] including the URL of the record and the reason for the withdrawal request.
17
Embed
Ancient DNA reveals the timing and persistence of ...eprints.whiterose.ac.uk/132660/1/Wales_et_al_2018... · than the derived maize form, suggesting a stepwise process of do-mestication.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
This is a repository copy of Ancient DNA reveals the timing and persistence of organellar genetic bottlenecks over 3,000 years of sunflower domestication and improvement.
White Rose Research Online URL for this paper:http://eprints.whiterose.ac.uk/132660/
Version: Published Version
Article:
Wales, Nathan orcid.org/0000-0003-0359-8450, Akman, Melis, Watson, Ray H.B. et al. (5 more authors) (2018) Ancient DNA reveals the timing and persistence of organellar geneticbottlenecks over 3,000 years of sunflower domestication and improvement. Evolutionary applications. ISSN 1752-4563
This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: https://creativecommons.org/licenses/
Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [email protected] including the URL of the record and the reason for the withdrawal request.
Bruce Ds Smith4科|科Kristen Js Gremi旭旭ion5科|科Ms Thomas Ps Gi旭bertザp葦科|科Benjamin Ks B旭ackmanゲpゴ
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
1DepartmentofP旭antandMicrobia旭Biology, University of California, Berkeley,
CA, USA
2DepartmentofBio旭ogypUniversityofVirginiapChar旭ottesvi旭旭epVApUSA3Centre for GeoGenetics, Natural History
MuseumofDenmarkpUniversityofCopenhagenpCopenhagenpDenmarkジTheSantaFeInstitutepSantaFepNMpUSA5DepartmentofAnthropo旭ogypOhioStateUniversity, Columbus, OH, USA
6Norwegian University of Science and
Techno旭ogypUniversityMuseumpTrondheimpNorway
CorrespondenceNathanWa旭espDepartmentofP旭antandMicrobia旭Bio旭ogypUniversityofCa旭iforniapBerkeley, CA, USA.
AbstractHere, we report a comprehensive paleogenomic study of archaeological and ethno-
graphic sunflower remains that provides significant new insights into the process of
domesticationofthis importantcropsDNAfrombothancientandhistoriccontextsyie旭dedhighproportionsofendogenousDNApanda旭thougharchaeo旭ogica旭DNAwasfound to be highly degraded, it still provided sufficient coverage to analyze genetic
changes over time. Shotgun sequencing data from specimens from the Eden�s Bluff
archaeo旭ogica旭siteinArkansasyie旭dedorgane旭旭arDNAsequencefromspecimensupto 3,100 years old. Their sequences match those of modern cultivated sunflowers and
are consistent with an early domestication bottleneck in this species. Our findings also
suggest that recent breeding of sunflowers has led to a loss of genetic diversity that
was present only a century ago in Native American landraces. These breeding epi-
sodes also left a profound signature on the mitochondrial and plastid haplotypes in
cultivars, as two types were intentionally introduced from other Helianthus species for
crop improvement. These findings gained from ancient and historic sunflower speci-
mens underscore how future in- depth gene- based analyses can advance our under-
standing of the pace and targets of selection during the domestication of sunflower
Starting around ジグググyears before present ェBPォp a crop comp旭exconsisting of acorn/crookneck squash (Cucurbita pepo L. ssp. ovifera
DsSs Deckerォp goosefoot ェChenopodium berlandieri Moqsォp marshe旭-der (Iva annua L.), and the common sunflower (Helianthus annuus
L.) was grown by low- level food- producing societies inhabiting the
watershedof theMississippiRiver ェSmithpゴググ葦ォsArchaeobotanica旭remains from ENA sites exhibit telltale signs of the so- called domes-
ticationsyndrome ェHammerpゲゾ芦ジォpasuiteof traits thatcommon旭ydistinguishes domesticates from their wild progenitors and that may
include larger seeds and disruption of natural seed dispersal mecha-
nisms. Of the four core species of the ENA crop complex, sunflower
is particularly well suited for in- depth domestication research thanks
to theexistenceof richarchaeobotanica旭 co旭旭ections ェSmithpゴグゲジォpacenturyofbreedingexperimentsェHeiserpゲゾゼ葦q斎kori賜pゲゾゾゴォpandthe development of many germplasm and genomic resources for
genetic investigations (Badouin et al., 2017; Burke, Tang, Knapp, &
Rieseberg, 2002; Kane et al., 2011; Rieseberg & Seiler, 1990; Wills
& Burke, 2007).
Through human selection, the weedy H. annuus spp. annuus was
transformed from a highly branching plant with numerous small
disks, also known as heads or capitula, to H. annuus spp. macrocar-
pus ェDsCsォCk旭旭sp thecu旭tivatedsunf旭owerpwhich is typica旭旭ycharac-terized by strong apical dominance and a single massive disk that
can produce hundreds to thousands of achenes. Sunflowers served
important nutritional, ceremonial, medicinal, cosmetic, and struc-
tural purposes in Native American cultures. For instance, an account
from 1615 by French explorer Samuel de Champlain indicates that
peop旭esoftheIroquoisConfederacyofNations intheGreatLakesregion of North America cultivated sunflower, grinding and eating
the seeds as well as processing them into oil used ceremonially for
anointing the hair (Heiser, 1951). After roasting sunflower achenes in
c旭aypotsorreedbasketsptheMandanpArikarapandHidatsapeop旭esoftheMissouriRiverbasinwou旭dmakesunf旭owerf旭ourorboi旭theachenes with maize, beans, and squash to make a porridge (Heiser,
1976). The Hopi people of the American Southwest were unique in
extracting a dye from the deeply purple- colored achenes of their
landraces (Heiser, 1951, 1976).
Archaeological sunflower remains have been excavated from doz-
ens of ENA sites, enabling temporal and spatial investigations on the
originsofsunf旭owerdomesticationsTheKostersiteinI旭旭inoisyie旭dedthe oldest known sunflower remains, with two achenes and one kernel
datingbetween芦ズググandズ芦ググBPェAschケAschpゲゾ芦ズqSmithpゴグゲジォ(Figure 1). Based on their small size, these specimens likely reflect the
co旭旭ection ofwi旭d resources ェSmithp ゴグゲジォs The o旭dest evidence forsunflower cultivation comes from the Hayes site in central Tennessee,
datingtoズグザジ・ジズ芦ザBPェゾズ鯵confidenceinterva旭pCIォェCritespゲゾゾザォsKernels from the site are larger than commonly observed in wild sun-
flowers, suggesting the initial steps of sunflower domestication were
underwaycircaジ芦ググBPェSmithpゴグゲジォsThreeothersitesprovideev-idence of sunflower cultivation before 3000 BP (Figure 1): 3800 BP
attheRivertonsiteinI旭旭inoisェSmithケYarne旭旭pゴググゾォpザザググBPattheNewt Kash She旭ter in Kentucky ェSmithp ゴグゲジォp and ザグズグ BP at theMarb旭eB旭uffShe旭terinArkansasェFritzpゲゾゾゼォs
Based on archaeological, morphological, and geographical data,
Heiser (1951) concluded that sunflower was domesticated once in
ENA, a hypothesis that has been supported by population genetics
studies of modern elite- bred cultivars, extant Native American landra-
ces, and wild H. annuus populations. For instance, Rieseberg and Seiler
(1990) demonstrated with isozymes and chloroplast markers that do-
mesticated landraces share haplotypes with wild sunflowers from ENA
and show a signature of a genetic bottleneck. Although archaeological
remainsputative旭y identifiedassunf旭owerp somedating toジゲザグBPpweresubsequent旭yrecoveredfromexcavationsinMexicoandraisedthe possibility of an independent domestication event (Lentz, Pohl,
Pope, & Wyatt, 2001), population genetic studies that include extant
Mexican wi旭d and cu旭tivated germp旭asm have on旭y found evidencethat extant cultivars derive from a single ENA domestication event.
Wills and Burke (2006) showed that domesticated populations have
one common and two rare chloroplast microsatellite marker haplo-
typesthatc旭usterwithwi旭dENAratherthanwi旭dMexicansunf旭owerssPatterns of sequence variation at nuclear microsatellite markers and
candidate domestication loci have likewise reinforced the conclusion
thata旭旭extant旭andracespwhetherco旭旭ectedinENAorMexicopdescendfrom a single origin most likely occurring from ancestral wild popula-
tions in the eastern and central USA (Blackman et al., 2011; Harter
eta旭spゴググジォsAlthough archaeological and genetic data predominantly point
to a single domestication event in ENA, there is much more to un-
earth about how sunf旭ower domestication proceededs It remainsto be determined which traits were of primary interest to early
farmers, whether sunflower domestication was rapid or protracted,
and how proto- domesticates responded to the new selection re-
gime. Genetic characterization of archaeological plant remains with
ancient DNA ェaDNAォ methodo旭ogies has the potentia旭 to answerthese questions by providing windows into past temporal dynam-
ics. Paleogenomic research has grown tremendously in the past
decade due to the rapid development of high- throughput sequenc-
ing techno旭ogies ェDer Sarkissian eta旭sp ゴグゲズォp and the app旭icationof paleogenomic methods to archaeobotanical remains has been a
particular success (Brown et al., 2015). For example, in reconstruct-
ing complete genomes of 6,000- year- old barley grains excavated
in Israe旭p Mascher eta旭s ェゴグゲ葦ォ determined the ancient samp旭eswere closely related to modern cultivars in the region and that the
major steps of barley domestication were completed by this point
in times Simi旭ar旭yp Ramos、Madriga旭 eta旭s ェゴグゲ葦ォ and Va旭旭ebueno、Estrada et al. (2016) characterized genomes of 5,000- year- old
maizecobsfromtheTehuac史nVa旭旭eypbutthey insteadfoundthatmany domestication- related genes had the ancestral form rather
than the derived maize form, suggesting a stepwise process of do-
mestication. Although these paleogenomic studies indicate archae-
ological remains could be invaluable for understanding sunflower�s
domestication and ancient cultivation, different plant species have
the potentia旭 to confound aDNA research through species、 and
科架 科 | 科ザWALES ET AL.
tissue、specific secondary compounds that interfere with DNAextraction and library preparation. To examine the paleogenomic
potential of archaeobotanical sunflower remains, we screened a
collection of archaeological and ethnographic specimens with a
shotgun sequencing strategy. The sequencing data generated from
these ancient and historic specimens were analyzed to determine
variabi旭ity in endogenous contentp DNA damagep and sources ofexogenousDNAs In additionp fo旭旭owing precedents inmamma旭ianaDNA projects ェDabneyp Knappp eta旭sp ゴグゲザq Gi旭bert eta旭sp ゴググゼォand genome skimming of modern samples (Bock, Kane, Ebert, &
Riesebergp ゴグゲジq Straub eta旭sp ゴグゲゴォpwe 旭everaged the sequenc-ing data to characterize variation in high copy number mitochon-
drial and plastid genomes, allowing us to investigate how these and
other archaeological and historic specimens may enrich our under-
standing of the domestication process.
ゴ科 |科MATERIALS AND METHODS
ゴsゲ科|科Archaeo旭ogica旭 sunf旭ower specimens
Although archaeobotanical remains are most often preserved by
charring or carbonization, such materials are generally incompat-
ible with paleogenomic analyses (Nistelberger, Smith, Wales, Star, &
Boessenkool, 2016). Therefore, we only obtained and processed des-
iccated specimens for this study. We tested 15 sunflower disk frag-
ments, one pericarp (seed coat), and one kernel, all of which originate
F IGURE ゲ科Mapofsamp旭ing旭ocationsandarchaeo旭ogica旭sitessEthnographicsamp旭esェandnumberofaccessionssamp旭edォareinredpandlandraces are in blue. Archaeological sites with ancient sunflower material discussed in the text are marked by yellow circles. Eden�s Bluff, the
site from which all archaeological remains detailed in this article were sampled, is bolded
ジ科 |科 科架 WALES ET AL.
from the Eden�s Bluff archaeological site in northwestern Arkansas
(Figures 1 and S1; Table 1). The specimens have been under the cu-
ration of the University of Arkansas Collections Facility (UARK) and
eter measurements ranged in size from 35 to 110 mm (mean = 75.5)
and were all larger in this dimension than disks of a well- defined wild
H. annuus popu旭ation ェSmithp ゴグゲジォp indicating the archaeo旭ogica旭disks represent plants cultivated by humans. Likewise, the dimen-
sions of the archaeological pericarp (length = 9.1 mm) and kernel
(length × width = 6.5 × 3.6 mm) are consistent with origin from do-
mesticated sunflowers.
EdenvsB旭uffェstatesiteIDrザBE葦ォwasexcavatedinゲゾザゴandゲゾザジas a part of expeditions led by the University of Arkansas focused
on the so、ca旭旭edOzarkB旭uff、Dwe旭旭er sitesp as coinedbyHarringtonェHarringtonpゲゾゴジapゲゾゴジbpゲゾ葦グォsThesesitesarerenownedfortheirpreservation of organic remains, including desiccated plant tissues
(Fritz, 1986; Gilmore, 1931). Native Americans likely used the rock-
shelters and caves specifically because their dry conditions were well
suited for long- term food storage and, despite the name, are unlikely
tohaveservedasseasona旭dwe旭旭ingsェBrownpゲゾ芦ジォsThechrono旭ogyof theOzarkB旭uff、Dwe旭旭er sites isnot fu旭旭yunderstoodpdue to the旭imited number of radiocarbon dates ェDavisp ゲゾ葦ゼォs As part of herrigorous archaeobotanical analyses, Fritz (1986) acquired dates from
15 sites and determined occupations occurred throughout the period
from ca. 3000�500 BP. Because their stratigraphic context may have
experienced disturbance from humans, rodents, or other causes, we
etry ェAMSォ radiocarbondatingat theUniversityofArizonaAMSfa-ci旭ity ェTab旭eゲqFigureSゴォsA旭旭AMSdates fromthisandother reportswere calibrated to calendar years before present (calBP) using OxCal
vジsザsゴ ェBronkRamseyp ゴググゾォ and the IntCa旭ゲザ ェReimer eta旭sp ゴグゲザォcalibration curve.
ゴsゴ科|科Ethnographic 旭andrace achenes
Eleven accessions of sunflower landraces were acquired from eth-
no旭ogica旭co旭旭ectionsattheNationa旭MuseumoftheAmericanIndianェNMAIォandUMMAAェTab旭eゴpFigureゲォsThesespecimensconsistofachenes sourced from Native Americans and via various intermediar-
ies in the first half of the twentieth century by Gilmore (1919) and
HeiserェゲゾズゲォsAtthetimepHeiserェゲゾズゲppsジジゲォ旭amentedthatwfewaboriginal strains of the cultivated sunflower are still in existence,
and� it is likely that the few remaining ones will disappear unless
steps are taken to preserve them.� While his efforts propagated many
sunflower landrace lineages, some of the achenes he attempted to
grow were not viable, including seed originating from the Six Nations
reserve in Ontario. Thus, these ethnographic achenes offer a unique
opportunity to investigate genetic relationships of putatively extinct
landraces to living sunflower lineages.
ゴsザ科|科DNA extraction and sequencing
Archaeological specimens were processed at a dedicated paleog-
enomics laboratory at the University of Copenhagen. The laboratory
meets the standards for aDNA research ェCooper ケ Poinarp ゴグググqGilbert, Bandelt, Hofreiter, & Barnes, 2005), such as being physi-
ca旭旭yseparatedfrommodernDNAandpost、PCR旭aboratoriespbeingoutfittedwithairfi旭trationandnight旭yUVirradiationequipmentpandrequiring researchers to wear coveralls to minimize contamination.
DNAwasextractedusingamethodthathasbeenshowntoworkwe旭旭on a range of species and tissue types (Wales, Andersen, Cappellini,
Specimen Tissue ca旭BP ェゾズ鯵 CIォ Endogenous DNA P旭astome DoC
Eden- 1 Pericarp Not dated グsザ鯵 0.8
Eden- 2 Diskfragment 915�795 ゲゲsゾ鯵 ズsジ
Eden- 3 Diskfragment 3168�3005 ゴsゲ鯵 3.2
Eden、ジ Diskfragment Not dated グsゴ鯵 0.3
Eden- 5 Diskfragment ゲゼザ葦・ゲズゼジ ゲジsゲ鯵 8.3
Eden- 6 Diskfragment 3163�2999 ズs葦鯵 7.2
Eden- 7 Diskfragment 1813�1622 芦s葦鯵 7.6
Eden- 8 Diskfragment 1817�1628 ゲゾsゲ鯵 ジsゲ
Eden- 9 Diskfragment 1819�1633 ジ芦sザ鯵 16.5
Eden- 10 Diskfragment 1873�1629 ザズsザ鯵 5.6
Eden- 11 Diskfragment 1825�1618 ザゲsゲ鯵 17.5
Eden- 12 Diskfragment 1868�1701 ザジsゲ鯵 ゴジsグ
Eden- 13 Diskfragment 1810�1571 ズズs葦鯵 18.1
Eden、ゲジ Diskfragment 1877�1711 ゾsゾ鯵 8.3
Eden- 15 Diskfragment 1770�1559 ゴズsゼ鯵 30.6
Eden- 16 Diskfragment 1819�1639 ザズs葦鯵 ゲジsゼ
Eden- 17 Kernel Not dated グsゲ鯵 1.1
TABLE ゲ科Archaeological specimens.
Acce旭eratormassspectrometryェAMSォdates are listed in calibrated years before
present. Samples with sequencing depth of
coverageェDoCォ┑ジforthep旭astomewereexcluded from the plastome analysis. See
Figure S1 for images of most samples and
Tab旭eSゲforadditiona旭samp旭epAMSpandsequencing information
科架 科 | 科ズWALES ET AL.
蛍vi旭a、ArcospケGi旭bertpゴグゲジォsInbriefptissuesamp旭eswereco旭旭ectedwith disposable forceps and scalpels, placed in PowerBead tubes
ェMOBIOゲザゲゲゼ、ズグォpandpu旭verizedbyshakingatジm⦆sforザグs ina FastPrep、ゴジ homogenizer ェMP Biomedica旭sォs The resu旭ting tissuepowderwas incubatedovernight inadigestionbuffer ェゲグmMTris、HC旭pゲグnMNaC旭pゴ鯵w⦆vSDSpズmMCaC旭2pゴsズmMEDTApジグmMDTTpandゲグ鯵proteinaseKso旭utionォpandthenextractedusingtworounds of phenol and one round of chloroform. To minimize the effect
of co、extracted compounds andpigmentsp the recoveredDNAwaspurified in aQiagenMinE旭ute co旭umn using optimizations to retainhigh旭yfragmentedDNAェDabneypKnapppeta旭spゴグゲザォsFourextractionblanks were processed with samples to monitor potential sources of
contaminations The extractedDNAp inc旭uding that from the extrac-tion b旭anksp was converted to I旭旭umina、compatib旭e 旭ibraries using ablunt- ended adapter ligation approach and optimizations to retain
short molecules (Wales et al., 2015). Before indexing PCR, the librar-
ies were tested by quantitative PCR (qPCR) to estimate the appropri-
ate number of cycles to avoid overamplification. qPCR was conducted
withaSYBRGreenassayasdescribedbyWa旭eseta旭s ェゴグゲズォpusingAmp旭iTaqGo旭dェApp旭iedBiosystemspFosterCitypCAォpprimersISゼandIS芦ェMeyerケKircherpゴグゲグォpandaRocheLightCyc旭erジ芦グRea旭、timePCR System. Libraries were amplified with AmpliTaq Gold for 10�18
cycles (Table S1) using a P7 indexing oligo with a 6- bp sample- specific
barcode to enab旭e mu旭tip旭ex sequencing ェMeyer ケ Kircherp ゴグゲグォsLibraries were pooled and shotgun- sequenced on six whole or partial
The 11 ethnographic samples were deemed to be relatively well
preserved and thus to pose a potential contamination risk to archae-
ological samples. Therefore, the achenes were extracted in steril-
ized 旭aminarf旭owhood inapre、PCRmodernDNA旭aboratoryattheUniversity of Copenhagen where sunflowers had not been previously
tested. Achenes were frozen in liquid nitrogen and fragmented with
a steri旭e pest旭es DNA was extracted with a Qiagen P旭ant Mini kit
following the manufacturer�s protocol except that the 65°C incubation
was conducted for ゴhrsMany specimens exhibited high、mo旭ecu旭ar、weightDNAonanagarosege旭psoDNAwasshearedwithaDiagenodeBioruptor using an appropriate number of sonication cycles for each
sample (Table S1). One accession (Seneca_striped_12997- 682) was
processedtwicepusingawho旭eacheneandanindividua旭kerne旭sDNAwasconvertedtoI旭旭umina旭ibrariesfo旭旭owingthesameprotoco旭usedfor the archaeological samples and sequenced on one lane of an
Raw sequencing reads were processed using Paleomix 1.2.12
ェSchubert eta旭sp ゴグゲジォp a bioinformatic pipe旭ine deve旭oped foraDNA datasetss The recommended parameters for pa旭eogenomicdatasets were utilized, including removing adapter sequences with
of readswithBWAa旭nwith the seeddisab旭ed ェLiケDurbinpゴググゾォpremoval of duplicate reads with Picard Tools (http://broadinstitute.
githubsio⦆picardォprea旭ignmentaroundinde旭swithGATKザsゼェMcKennaeta旭spゴグゲグォpandresca旭ingofbasequa旭itiesduetoaDNAdamagewithmapDamageゴsグ ェJ祝nssonp Gino旭hacp Schubertp Johnsonp ケ Or旭andop2013). Reads were mapped against the entire sunflower XRQ draft
genome (Badouin et al., 2017), including unplaced contigs, the plastid
genomepandthemitochondria旭genomesWereportendogenousDNAcontent based on all mapped reads, regardless of mapping quality, be-
cause high content of long terminal repeat retrotransposons in the
sunf旭owergenomeェゼジsゼ鯵ofthegenomepBadouineta旭spゴグゲゼォcausemany endogenous reads to map to multiple loci. As we observed po-
tential erroneous insertions of the organellar genomes in the nuclear
assembly, reads were also separately mapped to the plastid genome,
mitochondrial genome, and the nuclear genome without unplaced
contigs; these alignments were only used for organellar genome and
library complexity analyses.
TABLE ゴ科Ethnographic achenes from Native American sunflower landraces. Three Seneca achenes are reported to have been collected in
NorthDakotaェindicatedwithanasteriskォqhoweverpora旭traditionsandwrittenrecordsindicatethese旭andracesoriginatedfromthetraditiona旭lands of the Seneca people near Lake Ontario
To place the archaeological and ethnographic samples in con-
text, publicly available sequencing data from 79 modern cultivars,
20 landraces, 27 wild H. annuusindividua旭spandジゼindividua旭sofジother annual Helianthus specieswere down旭oaded from theNCBIsequence read archive (SRA) (Table S2). Because they were se-
quenced with deep coverage, we subsampled and analyzed 30 mil-
lion paired reads for each modern cultivar to reduce computational
time. The entire datasets were used for the other samples. Raw
data were processed in the Paleomix pipeline as discussed above,
exceptthatthemapDamageresca旭ingofbasequa旭itieswasomittedsTo minimize potential biases arising from differences in sequencing
strategies, such as higher theoretical mapping scores from paired-
end than single- read data, the paired- end modern data were treated
as though it was single- read data by trimming and mapping read
mates separately.
ゴsズ科|科Metagenomic ana旭ysis of archaeo旭ogica旭 and ethnographic samp旭es
To characterize non、sunf旭ower sources of DNA iso旭ated from ar-chaeological and ethnographic specimens, 10,000 randomly selected
tide collection (nr/nt) database using the BLASTn algorithm (Altschul,
Gishp Mi旭旭erp Myersp ケ Lipmanp ゲゾゾグォs MEGAN葦 ェHusonp MitrapRuscheweyh, Weber, & Schuster, 2011) was used to taxonomically
group BLASTn resu旭ts with LCA parametersr Min Score┎ゲグp MaxExpected┎ゲグp Min Percent Identity┎グsグp Top Percent┎グsグググゲpMinSupportPercent┎グsグpMinSupport┎ゲpMinComp旭exityグsグpLCAalgorithm = weighted, Percent to cover = 80, and ReadAssignment
Mode┎readCountsMEGAN葦wasusedtoperformaprincipa旭coordi-nate analysis (PCoA) of Bray�Curtis distances of taxonomic grouping
Reads mapping to the plastome (plastid or chloroplast genome) or to
themitochondria旭genomewereprocessedwithGATKザsゼェMcKennaeta旭sp ゴグゲグォHap旭otypeCa旭旭er andGenotypeGVCFs too旭s to identifypolymorphic sites. Polymorphisms were filtered with GATK according
to recommended parameters for depth, mapping quality, strand biases:
QD┑ゴsグp MQ┑ザグsグp FS┒葦グsグp SOR┒ザsグp MQRankSum┑┋ゲゴsズpand ReadPosRankSum┑┋芦sグs The sites were further fi旭tered withVCFtoo旭s ェDanecek eta旭sp ゴグゲゲォ to exc旭ude inde旭s and retain SNPswithaqua旭ityscore┒ゲpグググsArchaeo旭ogica旭samp旭eswith┑ジ┌averagecoverage of the plastome genome were excluded from the analysis.
SNPs were analyzed in R 3.3.1 (R Core Team, 2013) using the Pegas
(Paradis, 2010) package to identify haplotypes, and then, haplotype
relationships were visualized in popart (Leigh & Bryant, 2015) using
a minimum spanning network (Bandelt, Forster, & Röhl, 1999). For
constructionofthehap旭otypenetworkspatota旭ofゼグゲandジゲザpo旭y-morphic sites were used for the plastome and mitochondrial genome,
respectively. One of the oldest samples (Eden- 3) together with three
satisfy our filtering parameters and thus were not included in haplo-
type network construction.
ゴsゼ科|科Organe旭旭ar nuc旭eotide diversity ana旭ysis
Nucleotide diversity (pi) per each polymorphic site was computed
usingVCFtoo旭s ェDaneceketa旭spゴグゲゲォa旭旭owingforhap旭oidgenomes(haploid switch). For each group, mean nucleotide diversity was cal-
culated by taking average nucleotide diversity of all the sites used
in haplotype network construction for chloroplast or mitochondria.
Landracediversitymetricswereca旭cu旭atedafterexc旭udingMexCu旭tゼandMexCu旭tゲジbecausethosesamp旭eswereco旭旭ected in 旭oca旭mar-kets in Chiapas⦆Mexico and are 旭ike旭ymodern cu旭tivars as inferredfrom the haplotype networks.
ザ科 |科RESULTS
ザsゲ科|科Chrono旭ogy
AMS radiocarbon dating of the archaeobotanica旭 remains demon-
strated the specimens originate from three distinct time points: 3100,
1700, and 850 calBP (Figure S2). Eden- 3 and Eden- 6 are the oldest
samp旭espproducingnear旭yidentica旭AMSdatesェTab旭eSゲォpandtherebyprovide strong evidence that Eden�s Bluff should be added to the short
list of archaeological sites with sunflower cultivation before 3000 BP.
E旭evenAMSdatesfa旭旭nearゲゼググca旭BPpa旭旭ofwhichover旭apataゾズ鯵CIfromゲゼザ葦toゲゼゲゲca旭BPsThuspthemajorityofthesamp旭esmaybederived from a single occupational phase; however, these specimens
are recorded as being excavated from multiple contexts, suggesting
that some specimens may have been deposited decades or even a few
centuries apart. Eden- 2 produced the youngest date at ca. 850 calBP
(Table 1). While this young disk is an outlier in the chronology of our
otherAMSdatespFritzェゲゾ芦葦ォfoundsimi旭ardatesformaizeexcavatedfrom Eden�s Bluff, supporting the inference that this sample belongs
o旭ogica旭 specimensexhibit endogenousDNAcontents ranging fromグsゲゼ鯵toズズs葦葦鯵ェmean┎ゴゲsゲ鯵pmedian┎ゲ葦s葦鯵ォpwithbothachenesandonediskyie旭ding┑ゲ鯵endogenousDNAェTab旭eゲpFigureゴォsForゲゲoftheゲゴethnographicspecimensp芦ゾsゲ鯵・ゾザs葦鯵ofDNAmappedagainstthereferencegenomesIntheremainingethnographicsamp旭epArikaraゲゴゴゾゼ葦pon旭yザゼsゾ鯵ofthereadswereendogenousェseeex-ogenousDNAbe旭owォsAsidefromonesamp旭ewith旭owendogenouscontentェEden、ゲゼォpnuc旭earDNAPCRdup旭icate旭eve旭swere旭owfortheI旭旭umina旭ibrariesonthearchaeo旭ogica旭ェmeanザsグ葦鯵pmedian┎グsザゴ鯵ォand ethnographic specimens ェmean┎ゲsジゴ鯵ォ ェTab旭eSゲォs These 旭owlevels indicate that the libraries contain a great amount of untapped
科架 科 | 科ゼWALES ET AL.
complexity and could be deeply sequenced to recover large portions
of the nuclear genome.
ザsザ科|科DNA degradation
Consistent with the findings from previous paleogenomic studies,
mented and displayed varying levels of chemical damage (Figure S3).
Themeanread旭engthofendogenousnuc旭earDNAforarchaeo旭ogica旭samp旭esrangedfromジゲsゾto葦ゴsゲbppwithanovera旭旭meanofズゴs葦bp(Table S1). Cytosine deamination is the principal form of damage
observed inaDNAstudies ェDabneypMeyerpケP士士bopゴグゲザォp and incircumstances where contamination from modern sources is pos-
sible, especially hominin research, damage patterns can be used to
discernancientandmodernsequencesェJ祝nssoneta旭spゴグゲザォsDuringthe life of a cell, cytosine residues can spontaneously convert to ura-
cil, but they are fixed with cellular repair mechanisms. After death,
these uracil residues accumulate, primarily in single- stranded over-
hangsp and due to the activity of po旭ymerases used inDNA 旭ibrarypreparations, apparent C- to- T and G- to- A transitions are observed at
theズ昼andザ昼endsofsequencingreadssThisdamagecanbevisua旭-ized as ski- jump style plots (Figure S3), with steeper slopes indicating
age provides a probability of cytosine deamination in single- stranded
contexts (Table S1). Our samples produced δS values ranging from
0.165 to 0.999 (mean = 0.605). As anticipated from well- preserved,
relatively recent specimens, the ethnographic samples exhibit low
levels of damage (δS range = 0.018�0.056, mean = 0.035). The eth-
nographicDNAisa旭so旭essfragmentedthanthatofthearchaeo旭ogica旭samp旭essA旭thoughArikaraァゲジグジゴ、芦ゼジ isanout旭ierwithanaveragelength of 59.3 bp, library fragments frequently exceeded the length
of the number of sequencing cyc旭es ェmean read 旭ength┎ゼゼsジbppsequencing length = 81 bp), and this mean is artificially reduced as
high、mo旭ecu旭ar、weightDNAwasextractedfrommanyethnographicsamples and needed to be fragmented by sonication prior to library
construction.
ザsジ科|科Exogenous DNA
Metagenomic ana旭ysis of unmapped reads revea旭ed a comp旭exmix-ture of DNA in archaeo旭ogica旭 and contro旭 samp旭es ェFigureゴォs Thechief contaminant across all archaeological samples is bacteria (up to
trols are also dominated by bacteria, and taxa such as Proteobacteria,
Actinobacteria, and Firmicutes are consistent with species commonly
observed as 旭aboratory reagent contaminants ェSa旭ter eta旭sp ゴグゲジォsFungi and metazoans also make up a substantial proportion of archae-
o旭ogica旭contaminantspcontributingasmuchasザグ鯵ofreadcontentin several samples.
Taxonomic assignment of unmapped reads at the genus or species
level can help identify problematic individual samples and highlight
methodological or biological factors that require further examination.
For instance, the majority of unmapped reads in ethnographic sam-
p旭esarebroad旭yassignedtotheViridip旭antaepbutmostofthesehavetop BLAST hits to the H. annuus genome. These reads may not have
mapped to the sunflower genome due to sequence divergence from
the reference genome and/or because the BLASTn algorithm as ap-
plied was more tolerant of polymorphism than BWA. Ethnographic
samp旭es a旭so have on average ┒ザ timesmore unmapped reads as-signedtochordatesェゲ芦sズ鯵comparedtoズsジ鯵inarchaeo旭ogica旭sam-
ples) and animal parasites such as Platyhelminthes and Apicomplexa
F IGURE ゴ科DNAcontentofancientandethnographic旭andracesamp旭esandextractioncontro旭ssPercentageoftota旭readsmappingtothesunflower genome and relative proportion of unmapped reads assigned to kingdom- level taxa based on a random sampling of 10,000 unmapped
reads
Ancient
Ede
n−1
Ede
n−2
Ede
n−3
Ede
n−4
Ede
n−5
Ede
n−6
Ede
n−7
Ede
n−8
Ede
n−9
Ede
n−10
Ede
n−11
Ede
n−12
Ede
n−13
Ede
n−14
Ede
n−15
Ede
n−16
Ede
n−17
0.00
0.25
0.50
0.75
1.00
Controls
Blank
−1
Blank
−2
Blank
−3
Blank
−4
Ethnographic landraces
Arik
ara
1229
76
Arik
ara
1263
06
Arik
ara
1404
2−87
4
Arik
ara
broa
d 12
999−
682
Arik
ara/
Man
dan
1374
7
Paiut
e 14
1856
San
Ilde
fons
o P. 1
3597
−747
Sen
eca
1377
49
Sen
eca
purp
le 1
2996
−682
Sen
eca
purp
le 1
2998
−682
Sen
eca
strip
ed 1
2997
−682
a
Sen
eca
strip
ed 1
2997
−682
b
other
Viruses
Archaea
Metazoa
Fungi
Bacteria
Viridiplantae
Sunflower
芦科 |科 科架 WALES ET AL.
ェズsゾ鯵 and ゲsゾ鯵 compared to ゴsグ鯵 and グsズ鯵 respective旭y in ar-chaeological samples). Eden- 1 and Eden- 2 are differentiated from
other archaeo旭ogica旭 samp旭es ェPCゴ in FigureSジォ by high counts ofGammaproteobacteria (specifically the Pseudomonas stutzeri group in
Eden- 1 and Pseudomonas putida group and Enterobacteriales in Eden-
2). One ethnographic sample, Arikara_122976, more closely resem-
b旭esarchaeo旭ogica旭samp旭eswith 旭owerendogenoussunf旭owerDNAcontent ェザゼsゾ鯵 compared to the ethnographic average 芦ゼs葦鯵ォ anda more substantial fraction of sequences originating bacterial, fun-
gal, and metazoan contaminants. While Arikara_122976 groups with
archaeo旭ogica旭 samp旭es in thePCoA ana旭ysis ェFigureSジォp it containsnearly twice as many unmapped reads assigned to fungi, with most
assigned to the Sordariomycetes, as any other ethnographic sample
(Figure 2).
ザsズ科|科P旭astome ana旭ysis
We constructed two haplotype networks, one including and one
excluding the archaeological samples (Figure 3). Exclusion of the ar-
chaeological samples provides for greater haplotype resolution of the
ethnographic samples, as the greater level of missing data in the ar-
chaeological data reduces the number of polymorphic sites informa-
tive for network construction.
The cultivated sunflower sequences�whether from archaeologi-
cal or ethnographic remains, extant landraces, or modern cultivars�
sort into few haplotype clusters that we have denoted as Classes
ゲthroughジre旭ativetothemuchgreaterdiversityobserved inwi旭dHelianthus sequences, which are nearly all unique (Figure 3; Table S3).
All Eden�s Bluff archaeological specimens dating to ~1700 calBP fall
in Class 1 and share the same or similar haplotypes as many ENA,
southwesternpandMexican旭andracesqsevera旭ethnographicsamp旭esqand the majority of modern cultivars (Figure 3a). Although Eden- 8,
Eden、ゲグp and Eden、ゲジ have distinct hap旭otypesp they are on旭y oneor two substitutions removed from the predominant Class 1 haplo-
typesManymoresubstitutionsmustbe inferred to support the re-
ticulate lineages connecting their sequences to the distinct Arikara
F IGURE ザ科Plastome haplotype networks constructed with wild, cultivated, landrace, ethnographic, and archaeological sunflowers (a), and
plastome haplotype network constructed without the archaeological sunflowers (b). The size of the circles corresponds to number of individuals
present, and the number of polymorphic sites between individual haplotypes is indicated by tick marks. Haplotype classes for each sample
are included in Table S3. Class 1 is a core domestication haplotype and is composed of wild Helianthus annuus, archaeological specimens,
ethnographic samples, extant landraces, and modern cultivars. Class 2 also represents a haplotype that entered the domestication process
thousands of years ago; however, it is not observed in cultivars. Class 3 consists of R- type elite cultivars used in hybrid breeding, and was
presumably introduced into domesticated germplasm from H. petiolarisintheゴグthcenturyqasdiscussedinthetextpwesuspecttwoMexican旭andracesinC旭assザmayoriginatefrommisidentifiedcu旭tivarssC旭assジconsistsexc旭usive旭yofe旭itecu旭tivarspandwas旭ike旭yintroducedfromcropwild relatives, putatively H. argophyllus, during recent breeding for resistance to pathogens and diseases
旭otypes inextant 旭andraceswou旭dnotbeknownwithoutDavidLentzandRobertByevspainstakingsurveyinMexicoェLentzpPoh旭pA旭varadop Tarighatp ケ Byep ゴググ芦ォs In contrast to C旭ass ゴp C旭ass3, the second most common haplotype class, has a membership
consisting nearly entirely of R- type modern cultivars, which are
lines carrying a nuclear restorer allele for the cytoplasmic male
sterility system used for hybrid sunflower breeding. Two puta-
tiveMexican旭andracesェMexCu旭tゼandMexCu旭tゲジォa旭socarrytheClass 3 plastome sequence, raising the possibility they are actu-
a旭旭ye旭ite、bredmateria旭sTheC旭assジhap旭otypesequencesharedbythreemoderncu旭tivarsェBRS、ゲpHA、RゴpandIRォismostsimi旭arto sequences obtained from annual Helianthus species other than
H. annuus, likely reflecting a history of introgression as part of
a recent breeding program. Finally, the Hidatsa landrace has a
unique haplotype compared to other samples analyzed, consis-
tent with the findings of a previous study of sunflower sequence
diversity using chloroplast microsatellite markers (Wills & Burke,
2006).
ザs葦科|科Mitochondria旭 genome ana旭ysis
When archaeological sequences are excluded, the haplotype
network constructed for mitochondria is very similar to the plas-
tome network. Four major cultivated haplotype classes emerge
with nearly the same memberships, and thus, we use parallel no-
menc旭ature ェFigureジp Tab旭eSザォs One key difference is that theSanI旭defonsoethnographicsamp旭e ismoresimi旭artotheC旭assゲcultivated haplotypes than to any other cultivated or wild mito-
chondria旭 sequences Inc旭usion of mitochondria旭 sequences fromthe Eden�s Bluff samples in network construction analysis led to
F IGURE ジ科Mitochondria旭hap旭otypenetwork constructed with wild, cultivated,
landrace and ethnographic sunflowers. The
size of the circles corresponds to number
of individuals present, and the number
of polymorphic sites between individual
haplotypes is indicated by tick marks.
Haplotype classes for each sample are
included in Table S3. Class 1 is composed
of individuals sharing the same haplotype
and also those that diverge by only one or
twopo旭ymorphicsitessDuetouniparenta旭inheritance of organelles, the mitochondrial
classes contain the same individuals
as the plastome classes. See Figure 3
for information on the domestication
haplotypes (Classes 1 and 2) and those
introduced to modern cultivars during
ゴグth、centurybreedingェC旭assesザandジォ
Class 3
ArikaraArikara
ArikaraSan
Ildefonso
Arikara
Mex Cult 15
Hidatsa1
Class 4
Class 1
Class 2
ゲグ科 |科 科架 WALES ET AL.
ザsゼ科|科Nuc旭eotide diversity
The average pairwise nucleotide diversity (pi) of all groups of do-
mesticated sunflower samples is reduced relative to wild H. annuus,
consistent with a genetic bottleneck during domestication (Table 3).
This reduction is comparable for both organellar genomes. For in-
stancepthereisa葦芦鯵andゼゴ鯵reductionindiversityinethnographicsamples compared to wild H. annuus in chloroplast and mitochon-
dria, respectively. Within domesticated types, modern cultivars have
higher sequence diversity relative to the ethnographic samples and
landraces. However, this likely reflects the recent introgression of
wild haplotypes by modern breeding, as cultivars and landraces show
lower diversity as compared to the ethnographic samples when only
the diversity within the major haplotype classes also present in the
Eden�s Bluff samples (Class 1 and 2) is considered (Table 3). We report
a value for pi for the archaeological samples but note that this metric
is best suited for analyses of contemporaneous individuals and that
diversity within a single site is generally expected to be lower than
diversity present in the broader geographical sampling represented by
the sequences from wild, ethnographic, or modern cultivated material.
ジ科 |科DISCUSSION
ジsゲ科|科Sunf旭ower archaeo旭ogica旭 remains yie旭d qua旭ity endogenous DNA
Whi旭eaDNAstudieshaverevea旭ed important insights intothepaceofse旭ectionduringdomesticationinsomep旭antsェesgspMaschereta旭spゴグゲ葦qRamos、Madriga旭eta旭spゴグゲ葦qVa旭旭ebueno、Estradaeta旭spゴグゲ葦ォprecoveryofdegradedDNAfrommostcrops isnotroutinepandthisproject represents the first exploration of how paleogenomic test-
ing of archaeological sunflower remains can be used to understand
its unique domestication historys Through paired AMS dating and
paleogenomic testing of archaeological specimens from the Eden�s
Bluff site in Arkansas, we find that many desiccated remains dat-
ingbackas far asザゲググBPcanbeva旭uab旭e sourcesofDNAsSomespecimens yie旭dmore than ズグ鯵 sunf旭owerDNAp a旭though a seem-
ing旭y randomsubsetof specimens yie旭d 旭eve旭sof endogenousDNAェ┑ゲ鯵ォ essentia旭旭y incompatib旭e for state、of、the、art pa旭eogenomictechniques, such as targeted enrichment of genetic loci of interest
ples originates from at least four sources: organisms that inhabited
the disks and achenes during the life of the plant, such as pathogens;
organisms that consumed metabolites, proteins, and other biomol-
ecules in the tissue after the death of the individual; environmental
DNAtransferredfromthearchaeo旭ogica旭sedimentqandmodernDNAcontamination from excavation, curation, and genetic testing. While it
is difficult to distinguish these potential sources, the sequencing of ex-
traction controls provides a means to identify cross- contamination of
samp旭esandpervasiveDNAin旭aboratoryreagentsェSa旭tereta旭spゴグゲジォsWeobserved thatDNAdegradation patterns arevariab旭e in ar-
chaeo旭ogica旭sunf旭owerpbothintermsofDNAfragment旭engthandthefrequency of chemical damage, even within one relatively tight time
interval. For example, the two oldest specimens (Eden- 3 and Eden- 6)
yie旭dedeffective旭y identica旭AMSdatesofcasザゲググca旭BPsHoweverpcomparedtoEden、葦pEden、ザhass旭ight旭yshorterendogenousDNAェdif-ference of means = 5.8 bp) and higher levels of cytosine deamination
(δS of 0.999 vs. 0.673). Similarly, the youngest sample from the collec-
tionpEden、ゴpdatesto芦ズグca旭BPandhasDNAthatisnear旭yasshort(mean fragment length of 62.1 bp) and as damaged as Eden- 9 (mean
fragment length of 59.7 bp), which is twice as old. Thus, fragmentation
and damage profiles do not necessarily follow straightforward, time-
dependent degradation patterns, perhaps reflecting variability in how
different remains were treated prior to deposition (e.g., intentional
desiccation or heating in antiquity). Together, these findings indicate
that multiple samples from the same site and stratigraphic layer ought
to be initially tested by low- depth shotgun sequencing to identify
promising candidates for in- depth genetic analysis.
ジsゴ科|科Organe旭旭ar hap旭otype networks recapitu旭ate anticipated patterns for extant taxa
Organellar genomes in most plants exhibit uniparental inheritance
(Sato & Sato, 2013). Therefore, a one- to- one association of plastid hap-
旭otypeswithmitochondria旭hap旭otypesisoftenexpectedェMogensenp1996), and indeed, we observe such a tight correspondence between
our defined organellar haplotype classes (Table S3). Because the plas-
tid and mitochondrial genomes are nonrecombining, it can be possi-
ble to use organellar loci as markers for taxonomic identification, as
isperformedwithDNAbarcodingstudies ェAviseeta旭spゲゾ芦ゼqCBOLP旭antWorkingGroupeta旭spゴググゾォsYetptheorgane旭旭argenomesofthefive annual Helianthus species we have sampled do not resolve into
TABLE ザ科Nucleotide diversity (pi) for wild, archaeological,
ethnographicp旭andracepandmoderncu旭tivatedsunf旭owerssItisimportant to note that the archaeological specimens were excavated
from one site and are therefore not wholly comparable to
population- level measures of pi for the other sunflower groups.
GiventhatC旭assザandジhap旭otypeswere旭ike旭yintroducedtodomesticated lines during recent breeding, a separate calculation of
pi for modern cultivars with Class 1 and 2 haplotypes is provided
Sunf旭ower group
Nuc旭eotide diversity in p旭astome
Nuc旭eotide diversity in mitochondria
Wild グsグジグザ グsグジズ芦
Archaeological 0.0099 N/A
Ethnographic 0.0127 0.0126
Landrace 0.0125 0.0073
Class 1 and 2 landrace グsググゾジ 0.0050
Moderncu旭tivar 0.0285 0.0235
Class 1 and 2 modern
cultivar
0.0091 グsググ芦ジ
科架 科 | 科ゲゲWALES ET AL.
mutually exclusive clusters in either haplotype network. Such pat-
terns are consistent with previous findings demonstrating substantial
gene flow between Helianthus species and/or incomplete lineage sort-
ing (Sambatti, Strasburg, Ortiz- Barrientos, Baack, & Rieseberg, 2012;
Whitneyeta旭spゴグゲズォsForinstancepBocketa旭sェゴグゲジォobservedasimi-lar lack of taxonomic structure in the organellar genomes of perennial
Helianthus species, suggesting this is common throughout the genus.
Mostmoderncu旭tivarscarryoneof twodistincthap旭otypes ェthemost common Class 1 sequence or Class 3), and these assort into in-
bred line classes developed to facilitate hybrid production. Elite- bred
gesting that this cluster contains the few organellar sequences that
passed through the domestication and improvement bottlenecks. The
33 modern cultivars in our survey that carry the Class 3 haplotype
are all R- lines, which carry a mitochondrial mutation (PET- 1) intro-
gressed from H. petiolaris Nutt. that causes male sterility as well as
a nuclear restorer allele (Rf) for this mutation (Balk & Leaver, 2001).
As expected based on this breeding history, the mitochondrial haplo-
type of Class 3 groups closely with sequences present in H. petiolaris
ェFigureジォ. Because only Rf is required to restore fertility in hybrid crop
breedingpwedo find twoR、typecu旭tivarspRHA、ジゲ芦andRHA、ジグゲpin the Class 1 haplotype cluster. The shared breeding history of RHA
cultivars likely also explains the divergence between Class 1 and Class
3�s plastome haplotypes. Although the plastome haplotype of Class
3 does not have clear affinity for any of the obtained H. petiolaris se-
quences, it is possible that more similar H. petiolaris plastome haplo-
types were not included among the individuals sampled. Two putative
Mexican旭andracesェMexCu旭tゼandMexCu旭tゲジォsharetheC旭assザp旭as-tomeandmitochondria旭hap旭otypessUn旭ikeotherMexican旭andracespwhich were obtained directly from native farmers, these domesticates
wereobtainedfromanopenmarketp旭aceinChiapaspMexicoェDsLentzppersonal communication; Blackman et al., 2011). Thus, the possibility
that they may in fact be seeds derived from modern R- type sunflower
lines is plausible and merits rigorous examination in whole genome
analyses.
Another case of deliberate introgression is observed for the third,
Overall then, while three organellar genome types predominate
in modern cultivated germplasm, these very distinct Class 3 and
C旭assジsequencesarenotsharedwith 旭andracespethnographicporarchaeological samples and have largely entered cultivated H. annuus
through recent, deliberate introgression of genetic material from
other wild H. species. The history of directed breeding of domes-
ticated sunflower lines with crop wild relatives strongly suggests
Class 3 was introduced from H. petolaris during the establishment
of thehybridcropagricu旭tura旭 system ェSei旭erpQipケMarekpゴグゲゼォsC旭ass ジ was 旭ike旭y a旭so introduced during crop improvementp po-
tentially from H. argophyllus, the sunflower species which has been
most frequently crossed with domesticated lines to impart disease
andparasiteresistanceェSei旭erケFredrickMarekpゴグゲゲォsIndeedpitisperhaps surprising that additional non- H. annuus haplotypes were
not more commonly observed, as breeders have introduced allelic
variation for novel traits (e.g., resistance against a range of patho-
gens) by prolific and repeated introgression of genetic material from
other Helianthus species. H. annuus has reportedly been crossed
with every annua旭 species and ゲジ perennia旭 species in the genusェKayapゴグゲジォsOurfindingofon旭ytwointrogressedhap旭otypesponeof which was deliberately selected for, likely reflects that H. annuus
has predominantly served as the recurrent maternal parent during
sunflower improvement.
ジsザ科|科Ethnographic and archaeo旭ogica旭 organe旭旭ar sequences revea旭 旭ost diversity and raise new hypotheses
Although low- depth shotgun sequencing data from ancient samples
like those which we report here generally do not enable population-
level characterization of nuclear genes of interest, patterns of variation
inorgane旭旭argenomescanbeassessedbecausetheseDNAsourcesare found in many copies per cell, increasing their chance of recovery
2015; Liu & Burke, 2006). Although nearly every wild H. annuus indi-
vidual carries a unique plastid haplotype, the archaeological and eth-
nographic samples assort into just two haplotype clusters. Notably, the
most common haplotype among both modern and historical domesti-
cated forms (Class 1) was present at Eden�s Bluff at least 1,700 years
ago, as were two additional closely related but distinct haplotypes
not represented in any extant germplasm (Figure 3a). Given these
sequences are separated by fewer substitutions from the major do-
mesticate haplotype than from any wild haplotype, we infer these are
more likely to represent de novo evolution following a domestication
bottleneck than retention of standing variation from the wild ancestor.
Likewise, we observe several more unique Class 1 haplotypes that are
satellites of the major haplotype among the ethnographic samples, and
ゲゴ科 |科 科架 WALES ET AL.
the Class 2 haplotype observed in the oldest Eden�s Bluff sample and
several Native American landraces are completely absent from elite-
bred cultivars. Together, these findings suggest that all domesticated
sunflowers likely coalesce to very few maternal lineages present early
in the domestication process. Given that the archaeological samples
analyzed in this study are from a single site and might not fully reflect
the genetic diversity present in the earliest phases of domestication,
aDNAana旭ysisofadditiona旭archaeo旭ogica旭samp旭eswi旭旭beimportantfor affirming these findings.
Inadditionpourresu旭tsconfirmHeiservs旭amentqNativeAmericanlandraces once harbored genetic diversity now absent from modern
germplasm. Absence of the Class 2 haplotype and the unique eth-
nographic Class 1 haplotypes in elite cultivars likely reflects genetic
bottlenecks imposed during 20th- century improvement programs
and by the subsequent rise of the lines produced to agricultural
dominance throughoutNorthAmerica ェHeiserp ゲゾゼ葦q 斎kori賜p ゲゾゾゴォsThe loss of diversity in extant landraces relative to historic samples
also provides a caution and an opportunity for conducting genome
scans fordomesticationgenessBy inc旭udingnuc旭earDNArecoveredfrom ethnographic specimens, it may be possible to distinguish be-
tween genes that experienced selective sweeps as a consequence of
the domestication process versus changes in sequence diversity that
score similarly by population genetic metrics due to the recent loss of
landrace germplasm. The sole modern wild H. annuus sample carrying
a Class 1 haplotype is also instructive in this regard. Given the fre-
quency at which domesticated and wild sunflowers interbreed (Arias
ケRiesebergpゲゾゾジqLinderpTahapRiesebergpSei旭erpケSnowpゲゾゾ芦ォandthat this individual was collected in California, well outside the pro-
posed ENA domestication center, we expect it acquired the Class 1
haplotype by gene flow from contemporary domesticates. Thus, this
finding highlights the importance of vetting putatively wild sunflower
individuals for signals of admixture prior to inclusion in genomic scans
for selective sweeps.
The archaeological and ethnographic haplotype sequences we
have recovered are also consistent with a single center of sunflower
domestication located in ENA. Both Class 1 and Class 2 haplotypes
were present at Eden�s Bluff before 1700 calBP, and both classes are
also observed in historic and extant landraces. The presence of the
distinct Class 2 haplotype at Eden�s Bluff at 3100 calBP and in three
Mexican旭andraceaccessionsbuta旭soaSenecaethnographicsamp旭edoes introduce some ambiguity because the pattern fails to be fully
diagnostic for a single ENA origin versus an additional second cen-
terofdomesticationof sunf旭ower inMexicop as suggestedbyLentzet al. (2008, 2001). Nonetheless, the single domestication hypothesis
remains the most compelling conclusion for multiple reasons. First,
the threeC旭assゴMexican 旭andraceswerea旭旭 co旭旭ected from indige-
nous Nahua farmers in the state of Guerrero (Blackman et al., 2011)
who spoke only Nahuatl and yet did not know the Nahuatl word for
sunf旭owerェDsLentzppersona旭communicationォsThuspitispossib旭ethatthese 旭andraceswere introduced to this region ofMexicomore re-
cently than the early domestication period. Second, two wild individ-
uals from the central United States (northern Texas) carry the Class 2
haplotype. Thus, if these do not represent admixed genotypes and if
furthersequencingofMexicanwi旭dpopu旭ationsfai旭stoyie旭dtheC旭assゴsequencepthenaMexicanorigincanbeexc旭udedsFina旭旭yandmostpersuasively, multilocus nuclear genotype data and candidate domes-
gion are sure to reveal further insights into the temporal and spatial
dynamics with which early sunflower landraces arose and spread to
other regions.
ズ科 |科CONCLUSIONS AND FUTURE
DIRECTIONS
In summaryp we have shown that recovery of ancient and historicDNA fromarchaeo旭ogica旭 and ethnographic sunf旭ower specimens isfeasible and that desiccated specimens frequently contain high levels
ofendogenousDNAsAtpresentpshotgunsequencingdataa旭旭owustoinfer the relationships between ancient and modern samples for orga-
ne旭旭ar旭ocisIntandemwithsequencingdatafrommodernaccessionspwe have gained new perspectives on the persistence of plastid line-
ages for thousands of years under cultivation and the loss of genetic
diversity during recent improvement. We recognize these loci track
the maternal lineage and do not document the full domestication his-
tory of the sunflower, and our future studies where we obtain greater
depth of coverage for many loci in the nuclear genomes of ancient and
科架 科 | 科ゲザWALES ET AL.
historic specimens will allow us to address more nuanced questions
about the pace of domestication and specific targets of selection.
Fortuitously, numerous desiccated archaeological specimens have
been excavated from dozens of sites in the Ozarks and other parts of
cluding Eden�s Bluff, have since been inundated by the construction
of dams in the mid- 20th century or otherwise degraded (Fritz, 1986).
Thus, these curated specimens offer an otherwise unachievable pre-
historic perspective on sunflower domestication. Candidate targets of
selection during domestication have been reported in several stud-
ies ェBaute eta旭sp ゴグゲズq B旭ackmanp Strasburgp Raduskip Michae旭sp ケRiesebergpゴグゲグqB旭ackmaneta旭spゴグゲゲqChapmanpMande旭pケBurkep2013; Chapman et al., 2008), and identifying more should be acceler-
ated thanks to expanding genomic resources being generated by the
Internationa旭ConsortiumforSunf旭owerGenomicResourcesェBadouinet al., 2017; Kane et al., 2011). Thus, we anticipate paleogenomic
characterization of archaeological and ethnographic sunflower tissues
will soon have tremendous potential to resolve long- standing ques-
tions about the demographic and functional history of domestication
for this important oilseed crop.
ACKNOWLEDGEMENTS
This research was funded by the National Science Foundation
ェDEB、ゲザズジ葦ゴゴp DEB、ゲ葦ジグゼ芦芦ォ and the Danish Nationa旭 ResearchFoundation. Permission for destructive sampling of archaeological
andhistoricmateria旭swaskind旭ygrantedbytheNationa旭MuseumoftheAmericanIndianptheUniversityofArkansasCo旭旭ectionsFaci旭itypThe Osage Nationp and the University of Michigan Museum ofArchaeo旭ogica旭Anthropo旭ogys Inparticu旭arpwe acknow旭edge the in-
valuable assistance of Andrea A. Hunter, Tribal Historic Preservation
OfficerpOsageNationqMarySuterpCuratorofCo旭旭ectionspUniversityMuseumpUniversityofArkansasqandEmi旭yKap旭anpNationa旭Museumof the American Indians We thank Fi旭ipe Gs Vieirap Thorfinn SandKorne旭iussenp Mikke旭 Schubertp Mike Martinp and Vanessa Biekerfor bioinformatic advice and Shyam Gopa旭akrishnanp Ida Mo旭tkepand Jazm趣n Ramos、Madriga旭 for thoughtfu旭 discussionss Specia旭thanks to theDanishNationa旭High、throughput SequencingCentreforassistance ingenerating I旭旭uminadatasWea旭so thankKeBiandJason Huff at the UC Berkeley Computational Genomics Resource
Laboratory for their support. Thanks to members of the Blackman
laboratory and two anonymous reviewers who provided comments
on the manuscript. Publication made possible in part by support from
wild sunflowers. Theoretical and Applied Genetics, 89(6), 655�660.
AschpDsLspケAschpNsEsェゲゾ芦ズォsPrehistoricp旭antcu旭tivationinWest、Centra旭I旭旭inoiss InRs IsFord ェEdsォpPrehistoric food production in North America
ェVo旭s ゼズp pps ゲジゾ・ゴグジォs Ann Arborp MIr Museum of Anthropo旭ogypUniversityofMichigans
AvisepJsCspArno旭dpJspBa旭旭pRsMspBerminghampEspLambpTspNeige旭pJsEspt Saundersp Ns Cs ェゲゾ芦ゼォs Intraspecific phy旭ogeographyr The mito-
chondria旭DNAbridgebetweenpopu旭ationgeneticsandsystematicssAnnual Review of Ecology and Systematics, 18ェゲォpジ芦ゾ・ズゴゴshttpsr⦆⦆doisorg⦆ゲグsゲゲジ葦⦆annurevsessゲ芦sゲゲグゲ芦ゼsググゴジゴゲ
BadouinpHspGouzypJspGrassapCsJspMuratpFspStatonpSsEspCottretpLsptLanglade, N. B. (2017). The sunflower genome provides insights into
oil metabolism, flowering and Asterid evolution. Nature, 546(7656),
ferring intraspecific phylogenies. Molecular Biology and Evolution, 16(1),
ザゼ・ジ芦shttpsr⦆⦆doisorg⦆ゲグsゲグゾザ⦆oxfordjourna旭ssmo旭bevsaグゴ葦グザ葦Baute, G. J., Kane, N. C., Grassa, C. J., Lai, Z., & Rieseberg, L. H. (2015).
Genome scans reveal candidate domestication and improvement
genes in cultivated sunflower, as well as post- domestication introgres-
sion with wild relatives. New Phytologist, 206(2), 830�838. https://doi.
org/10.1111/nph.13255
Beja、Pereirap Asp Carame旭旭ip Dsp La旭ueza、Foxp Csp Vernesip Csp Ferrandp NspCasoli, A., � Bertorelle, G. (2006). The origin of European cattle:
Evidence frommodernandancientDNAsProceedings of the National
Academy of Sciences of the United States of America, 103(21), 8113�
8118. https://doi.org/10.1073/pnas.0509210103
Bertero de Romanop Asp ケ Norberto V史zquezp As ェゴググザォs Origin of theArgentine sunflower varieties. HELIA, 26(38), 127�136.
B旭ackmanp Bs Ksp Scascite旭旭ip Msp Kanep Ns Csp Lutonp Hs Hsp RasmussenpDs Asp Byep Rs Asp t Riesebergp Ls Hs ェゴグゲゲォs Sunf旭ower domestica-tion alleles support single domestication center in eastern North
America. Proceedings of the National Academy of Sciences of the United
States of America, 108ェザジォp ゲジザ葦グ・ゲジザ葦ズs httpsr⦆⦆doisorg⦆ゲグsゲグゼザ⦆pnassゲゲグジ芦ズザゲグ芦
B旭ackmanpBsKspStrasburgpJsLspRaduskipAsRspMichae旭spSsDspケRiesebergpL. H. (2010). The role of recently derived FT paralogs in sunflower do-
mestication. Current Biology, 20(7), 629�635. https://doi.org/10.1016/j.
cub.2010.01.059
BockpDsGspKanepNsCspEbertpDsPspケRiesebergpLsHs ェゴグゲジォsGenomeskimming reveals the origin of the Jerusalem Artichoke tuber crop spe-
cies: Neither from Jerusalem nor an artichoke. New Phytologist, 201(3),
1021�1030. https://doi.org/10.1111/nph.12560
Bronk Ramsey, C. (2009). Bayesian analysis of radiocarbon dates.
research and their implications for archaeobotany. Vegetation
History and Archaeobotany, 24p ゴグゼ・ゴゲジs httpsr⦆⦆doisorg⦆ゲグsゲググゼ⦆sググザザジ、グゲジ、グジ芦ゾ、ジ
BurkepJsMspTangpSspKnapppSsJspケRiesebergpLsHsェゴググゴォsGeneticana旭y-sis of sunflower domestication. Genetics, 161(3), 1257�1267.
CarpenterpMsLspBuenrostropJsDspVa旭dioserapCspSchroederpHspA旭旭entoftpMs Esp Sikorap Msp t Bustamantep Cs Ds ェゴグゲザォs Pu旭旭ing out the ゲ鯵rWho旭e、genomecapture for thetargetedenrichmentofancientDNAsequencing libraries. American Journal of Human Genetics, 93(5), 852�
ChapmanpMsAspPash旭eypCsHspWenz旭erpJspHva旭apJspTangpSspKnapppSsJspケBurkepJsMsェゴググ芦ォsAgenomicscanforse旭ectionrevea旭scandidatesfor genes involved in the evolution of cultivated sunflower (Helianthus
annuus). The Plant Cell, 20ェゲゲォp ゴゾザゲ・ゴゾジズs httpsr⦆⦆doisorg⦆ゲグsゲゲグズ⦆tpc.108.059808
Cooperp Asp ケ Poinarp Hs Ns ェゴグググォs Ancient DNAr Do it right ornot at all. Science, 289ェズジ芦ゴォp ゲゲザゾs httpsr⦆⦆doisorg⦆ゲグsゲゲゴ葦⦆sciencesゴ芦ゾsズジ芦ゴsゲゲザゾb
Critesp Gs Ds ェゲゾゾザォs Domesticated sunf旭ower in fifth mi旭旭ennium BsPstempora旭 contextr New evidence from Midd旭e Tennessees American
BsptMeyerpMsェゴグゲザォsComp旭etemitochondria旭genomesequenceofaMidd旭e P旭eistocene cave bear reconstructed from u旭trashort DNAfragments. Proceedings of the National Academy of Sciences of the United
States of America, 110(39), 15758�15763. https://doi.org/10.1073/
pnassゲザゲジジジズゲゲグDabneyp Jsp Meyerp Msp ケ P士士bop Ss ェゴグゲザォs Ancient DNA damages Cold
Spring Harbor Perspectives in Biology, 5(7), a012567.
Ls Hs ェゴググジォs Origin of extant domesticated sunf旭owers in easternNorth America. Nature, 430(6996), 201�205. https://doi.org/10.1038/
nature02710
Hazkani、CovopEspZe旭旭erpRsMspケMartinpWsェゴグゲグォsMo旭ecu旭arpo旭tergeistsrMitochondria旭 DNA copies ェnumtsォ in sequenced nuc旭ear genomessPLOS Genetics, 6ェゴォpeゲグググ芦ザジs
Heiserp Cs Bs ェゲゾズゲォsThe sunf旭ower among theNorthAmerican IndianssProceedings of the American Philosophical Society, 95ェジォpジザゴ・ジジ芦s
Heiser, C. B. (1976). The sunflower. Norman, OK: University of Oklahoma
fer in crop plantsェVo旭sゴpppsゴ芦ゲ・ザゲズォsNewYorkpNYrSpringer、Ver旭agshttpsr⦆⦆doisorg⦆ゲグsゲググゼ⦆ゾゼ芦、ゲ、ジ葦ゲジ、ゾズゼゴ、ゼ
Kist旭erpLspMontenegrop蛍spSmithpBsDspGiffordpJsAspGreenpRsEspNewsompLsAspケShapiropBs ェゴグゲジォsTransoceanicdrift and thedomesticationof African bottle gourds in the Americas. Proceedings of the National
Academy of Sciences of the United States of America, 111ェ芦ォpゴゾザゼ・ゴゾジゲshttps://doi.org/10.1073/pnas.1318678111
LipHspケDurbinpRs ェゴググゾォs Fast and accurate short read a旭ignmentwithBurrows- Wheeler transform. Bioinformatics, 25ェゲジォp ゲゼズジ・ゲゼ葦グshttpsr⦆⦆doisorg⦆ゲグsゲグゾザ⦆bioinformatics⦆btpザゴジ
LinderpCsRspTahap IspRiesebergpLsHspSei旭erpGsJspケSnowpAsAs ェゲゾゾ芦ォsLong- term introgression of crop genes into wild sunflower popula-
tions. Theoretical and Applied Genetics, 96ェザォp ザザゾ・ザジゼs httpsr⦆⦆doisorg⦆ゲグsゲググゼ⦆sググゲゴゴググズグゼジ葦
BFグゴ芦葦グジゼゼSabop IIIpGspケEar旭ypAsMs ェゲゾゾグォsPrehistoric culture history human adap-
tation in the Ozark and Ouachita Mountains. Fayetteville, NC: Arkansas
Archeological Survey.
Sa旭terp Ss Jsp Coxp Ms Jsp Turekp Es Msp Ca旭usp Ss Tsp Cooksonp Ws OspMoffattp Ms Fsp tWa旭kerp AsWs ェゴグゲジォs Reagent and 旭aboratorycontamination can critically impact sequence- based microbi-
SchubertpMspErminipLspSarkissianpCsDspJ祝nssonpHspGino旭hacpAspSchaeferpRsp t Or旭andop Ls ェゴグゲジォs Characterization of ancient and moderngenomes by SNP detection and phylogenomic and metagenomic
AspMart趣nez Gonz史旭ezp Jsp Garc趣a CookpAspMontie旭p Rsp ケVie旭旭e、Ca旭zadapJs、Psェゴグゲ葦ォsTheear旭iestmaizefromSanMarcosTehuac史nis a partial domesticate with genomic evidence of inbreeding.
Proceedings of the National Academy of Sciences of the United States
of America, 113ェジゾォp ゲジゲズゲ・ゲジゲズ葦s httpsr⦆⦆doisorg⦆ゲグsゲグゼザ⦆pnas.1609701113
Zederp Ms As ェゴグゲズォs Core questions in domestication researchsProceedings of the National Academy of Sciences of the United
States of America, 112(11), 3191�3198. https://doi.org/10.1073/
pnas.1501711112
SUPPORTING INFORMATION
Additiona旭 Supporting Information may be found on旭ine in the supporting information tab for this article.
How to cite this artic旭erWa旭esNpAkmanMpWatsonRpeta旭sAncientDNArevea旭sthetimingandpersistenceoforgane旭旭argenetic bottlenecks over 3,000 years of sunflower
domestication and improvement. Evol Appl. 2018;00:1�16.