Earliest Direct Evidence for Broomcorn Millet and Wheat in the Central

∗ 2 1 eevd 2Mrh21;Acpe:1 a 00 eie:2 a 2010 May 28 Revised: 2010; May 16 Accepted: 2010; March 12 Received: h ol.I oecss oee,vs asi vdnehv etu ihfwdt owork to data few with us left have around evidence societies toward in of pathways gaps strategies vast indirect however, economic and cases, the some direct In differentiate neighbouring the world. or with about the connect comparison arguments that to in production led evidence food often that has dating This domesticated securely for datasets. evidence and direct of finding animals; requisites: and key dynamics two plants by the this defined essential, been and most has its study At them, production. of food line acquired of spread forms different who who of frontiers people those the of along interaction practices, identities regional major agricultural the a and around revolution’, revolved domesticates ‘agricultural has the inquiry scientific with of preoccupation arena archaeological earliest the From Introduction Mar’yashev N. Alexei & Frachetti D. Michael Eurasian region central steppe the in wheat and millet broomcorn for evidence direct Earliest Keywords: ANTIQUITY O610 USA 63130, MO eateto nhoooy ahntnUiest nS oi,1BoknsDieC 14 tLouis, St 1114, Drive-CB Brookings 1 Louis, St in University Washington Anthropology, of Department nttt fAcaooy 4y otk lay 500 eulco Kazakhstan of Republic 050010, Almaty, Dostyk, y. 44 Archaeology, of Institute uhrfrcrepnec Eal [email protected]) (Email: correspondence for Author 4(00:9311 http://antiquity.ac.uk/ant/084/ant0840993.htm 993–1010 (2010): 84 aahtn rneAe atrls,cras ha,mle,arclue cremation agriculture, millet, wheat, cereals, pastoralism, Age, Bronze Kazakhstan, 1 ∗ oetN Spengler N. Robert , 2 993 hs rismgtsga iulrte than commodity. rather subsistence ritual a a that signal suggestion might a grains third the these context, raises later the burial, the cremation Moreover of in BC. presence millet millennium and the wheat revealed both have south- dating in and and Begash recovery precision two, at high Here, Kazakhstan. the east breakthrough in separate a cereals that report early lands for vast The the searching agriculture. world’s are the authors of are nodes these early millet; broomcorn Asia cultivating China western were of societies of and wheat societies cultivating were BC, 3000 Before 1 al .Fritz J. Gayle , 1

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region with, so determinations about the direction of diffusion, chronology, or even independent domestication of crops (and animals) in particular regions, cannot be made with confidence. For example, throughout the history of archaeological study in the central Eurasian steppe zone, a lack of reliable evidence for use of domesticated grains before the first millennium BC has constrained our understanding of the economies and realms of interaction that characterise Eurasian steppe communities and their neighbours before and during the Bronze Age (c. 3000–1000 BC). Within the vast territory of the central Eurasian steppe (Figure 1), direct archaeological evidence of agricultural production, consumption, and regional diffusion of domesticated grains during the Neolithic and Bronze Age is concentrated only in the westernmost regions – essentially the territories north of the Black Sea and further west to central Europe. Yet for decades, archaeologists have argued for the use of domesticated grains in the subsistence economies of pastoralists living throughout the wider steppe region in the late third and second millennia BC (Kuz’mina 2007: 141). This argument has remained largely speculative due to the vast lacuna in directly dated evidence for crops before the first millennium BC outside the westernmost territories of the Eurasian steppe (Lebedeva 2005). To date, a chronological gap of more than 7000 years exists between the first regional domestications of wheat (Triticum spp.) in south-west Asia, broomcorn millet (Panicum miliaceum) in China, and the earliest directly dated evidence of those grains in central Eurasian archaeological contexts. Both geographically and chronologically, the near void in archaeobotanical evidence across the steppe zone, from the Gansu (Hexi) Corridor to the Caucasus, has clouded archaeological models of diffusion of domesticated crops over an enormous territory of the world. However, recent archaeological research suggests that Eurasian mobile pastoralists were key agents for the transmission of numerous technologies and products across the steppe zone, promoting networks of interaction between economies and societies from eastern Asia to south-west Asia and Europe in the third and second millennia BC (Anthony 2007; Frachetti 2008). The earliest cultivation and use of domesticated cereals in Neolithic economies are well documented from at least 8000 BC in south-west Asia (Willcox 2005; Weiss et al. 2006). By the sixth millennium BC, domestic varieties of wheat and barley (Hordeum vulgare)(among other crops) comprise staple foods for Neolithic agriculturalists from south-west Asia to Europe and south Asia (Colledge et al. 2004; Bellwood 2005). Recent studies in eastern Asia at the site of Cishan in north-eastern China document the cultivation of broomcorn millet as early as 8000 cal BC between the Loess Plateau and the North China Plain (Crawford 2009; Lu et al. 2009). Evidence for millets – broomcorn and foxtail (Setaria italica)–is more abundant in later Neolithic sites throughout the Yellow River valley and in more upland regions (i.e. elevated terraces) of central and eastern China by 6000 BC (Zhao 2005; Crawford et al. 2006; Liu et al. 2009). Interestingly, domesticated millet is also found before 5000 BC in western and central Europe, sparking debates about possible pathways across central Eurasia versus scenarios of independent domestication (Lisitsina 1984; Zohary & Hopf 2000). This issue will not be resolved until more is known about genetic relationships between domesticated broomcorn millet and the phytogeography of its wild ancestors. Li et al. (2007) provide the earliest evidence of wheat in China around 2600 BC, at the site of Xishanping (Figure 1). Crawford et al. (2005) present evidence for wheat at the sites

994 ihe .Frachetti D. Michael 995 tal et .

Figure 1. Key regions and selected sites with archaeological evidence of wheat and broomcorn millet in prehistory: 1. Begash (Frachetti & Mar’yashev 2007); 2. Arkaim (Gadyuchenko 2002); 3. Alandskoe (Gadyuchenko 2002); 4. Cishan (Lu et al. 2009); 5. Chokh (Lisitsina 1984); 6. Kyul-tepe (Lisitsina 1984); 7. Jeitun (Harris & Gosden 1996); 8. Xishanping (wheat at c. 2500 BC) (Li et al. 2007); 9. Shortugai (Willcox 1991); 10. Liangchengzhen (wheat c. 26–1900 BC) (Crawford et al. 2005); 11. Zaojiaozhu (wheat c. 19–1500 BC) (Crawford et al. 2005); 12. Pirak (Costantini 1981); 13. Lop Nor (Xiaohe & Gumugou sites, wheat c. 2000–1500 BC) (Crawford et al. 2005; Flad et al. 2010); 14. Donghuishan (wheat c. 1700 cal BC) (Flad et al. 2010); 15. Tahirbai-tepe (wheat & millet c. 1500 BC) (Hunt et al. 2008); 16. Tuzusai (wheat & millet c. 700 BC) (Chang et al. 2003; Rosen et al. 2000).

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region of Liangchengzhen and Zaojiaoshu in eastern China around 2000 BC. Flad et al. (2010) directly dated wheat remains from Donghuishan in the Hexi Corridor (western Gansu) to c. 1700 cal BC, demonstrating that by the second millennium BC wheat was widely distributed outside its concentrated region(s) of domestication. Yet, the long chronological hiatus between the earliest evidence of domesticated wheat and millet in regional Neolithic agricultural centres and corresponding archaeological evidence of these domesticates in the Late Bronze Age has left the trail cold beyond the borders of south-west and eastern Asia. Thus, new studies documenting domesticated crops from the third millennium BC in the vast intervening territory – the central Eurasian steppe – represent an essential focus for more complete comprehension of the vectors of transmission of domesticated grains between earlier agricultural centres in China, south-west Asia and Europe. This article presents archaeobotanical evidence from the pastoralist campsite Begash (phase 1a, c. 2500–1950 cal BC), located in the piedmont steppes of the Dzhungar Mountains in south-eastern Kazakhstan (Figure 1). At Begash, carbonised seeds of broomcorn millet and wheat were recovered through systematic flotation of soils from a cremation burial cist and from an associated funerary fire-pit. A direct AMS date from the millet and wheat seeds yielded a 2-sigma range of 2460 to 2150 cal BC, while additional AMS dates of associated charcoal samples from the burial cist and related fire-pit fall between 2290 and 2020 cal BC (Table 1). Nearly all the domesticated seeds were recovered from the burial context; flotation samples from the domestic hearths of the same chronological period at Begash contained only two broomcorn millet grains. Currently, the remains from Begash predate by roughly 1500 years any other absolutely dated evidence of millet or wheat in the steppe zone and are the earliest reported anywhere in central Eurasia from the Don River to the Hexi Corridor (China) (cf. Kuz’mina 2007: 141). The documentation of domesticated grains at Begash establishes a key point of reference for the transmission of both wheat and millet along distinct routes – and possibly in different directions – through the mountains of central Asia and into the steppe territory by the late third millennium BC.

Archaeological context and methods The prehistoric settlement of Begash, located in the Semirech’ye region of eastern Kazakhstan, was excavated in 2002, 2005 and 2006 as part of the joint Kazakh-American Dzhungar Mountains Archaeology Project (DMAP) (Frachetti & Mar’yashev 2007). From at least 2500 BC, Begash was occupied by small groups of mobile pastoralists, whose economy was based on vertically transhumant sheep/goat herding in the Dzhungar Mountains (Frachetti & Benecke 2009). The broad chronology of occupation at Begash is derived from 34 AMS samples taken throughout the site’s stratigraphic levels, which date the earliest occupation (phase 1a) to the Early/Middle Bronze Age (c. 2500 BC), with later construction phases in the Middle Iron Age (phase 3, c. 400 cal BC), medieval and historic periods (phase 5/6) (Frachetti & Mar’yashev 2007). The chronology of phase 1a, in particular, is derived from four AMS samples with overlapping 1-sigma ranges from 3100 to 1950 cal BC. However, given the wide error margin for the earliest AMS sample (4220−+220 yrs BP, uncalibrated), a more conservative calibrated range of 2460–1950 cal BC is preferred by the authors.

996 ea265 S4Rta fire Ritual FS44 Beta-266457 ea266 S0Ritual FS50 Beta-266460 ea265 FS47b Beta-266459 a ape#cnetMtra r PclB a BC cal BC cal BP yrs Material context # sample Lab ea265 FS47a Beta-266458 h ape eeflae sn ipebce ehda ecie nPasl 20) A 2). (2000). (Figure Pearsall 1a in phase described in as hearths method bucket domestic simple the a of using each floated from were sampled ash samples and were The Approximately soil flotation. soil the for of of processed litres and litres 30 collected while 2–3 were level, cist lower from burial the taken the as from for were inside taken cist, soil from sieved were of remains and the litres litres 2 floated 9.5 in and fire-pit, were ash level 1a cist upper phase the the the the of alongside of case fire-pit volume the funerary In the remains. contained the macrobotanical Half from and fragments. samples (cist metals soil bone or context as ceramics small burial well of the and devoid with ashes was associated funerary cist all only (93.9%) cremation samples below) percent The three (see Ninety-four fire-pit). from 3). Cerealia funerary & came or and 2 seeds wheat (Tables domestic as grains the millet classified of broomcorn five as of classified total 28 a and volume. in target resulting the was grains, litres domesticated 10 but size, were feature hearths with samples burials, all varied Feature floors, sizes from occupation data. Sample as collected botanical middens. such were and context, baseline anthropogenic samples assess distinct Bulk to every from flotation. site taken the of throughout purpose layers the cultural for Begash at excavations features two these of ranges conclusion. AMS to this identical addition support nearly In further burial. the a proximity, the as and to interpreted relationship parity is above stratigraphic functional which and from their deposit, proximity charcoal itself, its cist and to cist the ash due the pyre circular approach to funerary a and, adjacent stratigraphically includes wall directly also to context balk Located burial as burial. excavation the the the so contaminating of extension or edge trench disturbing north-west without 1m the a along dug cist burial thus, the of border stone burial cremation associated an and al .Clbae M hoooyo ed n rai ape rmpae1 uilcnetat context burial 1a phase from samples organic and seeds of chronology Begash. AMS Calibrated 1. Table ih oainsmlswr ae rmpae1 otxs n v fteecontained these of five and contexts, 1a phase from taken were samples flotation Eight during collected were – samples feature and samples bulk – samples soil of types Two structure domestic single a revealed Begash at occupation 1a phase the of Excavations ape#&Clna g airto -im calibrated prob.) 2-sigma (95% results: Calibration intercept(s) age Calendar uncalibrated archaeological & # Sample level) (upper pit level) (lower fire-pit uilcist Burial uilcist Burial odcaca 3720 charcoal Wood odcaca 3740 charcoal Wood odcaca 3760 charcoal Wood Carbonised ha grains wheat and millet ihe .Frachetti D. Michael c maa Fgr ) xaaosecutrdteouter the encountered Excavators 2). (Figure away 8m . 997 3840 − + − + − + − + 023 2260–2260 2130 40 024 2280–2240 2140 40 02190 40 029 2460–2190 2290 40 tal et . 2150 2170 2210–2020 2240–2030 2100–2040 2290–2110 2170–2150

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region Figure 2. Plan of excavations of Begash, phase 1a (after Frachetti & Mar’yashev 2007).

998 S52460– FS45 S82460– FS48 S02280– FS50 S42260– FS44 S72460– FS47 S91950– FS19 S 390–50 FS6 ape#(irs#o ffa.Lnt it ctlu Scutellum/seed Scutellum ratio length Width (mm) Length height frag. of (mm) # (mm) of # puffed or whole Total L) floated AD FS2 (litres age & # Sample aei mrcia oueasalrseesz o h ev fraction. heavy the for was size 1.00mm sieve smaller and clots, a samples clay use specifically fraction to material, light inorganic impractical of it for concentration made occupation. high used of The was fraction. phases heavy 0.355mm all for representing of used site size Begash sieve the minimum from floated A were samples 32 of total with Begash from samples Flotation 2. Table hl h oslsd srudadpordn ihasne mronth(iue3). (Figure notch embryo sunken a deep, with be protruding to and tend round which The is furrows, form. ventral side wheat recognised compact dorsal easily a the indicates have while (1.21) Begash free-threshing ratio from a width grains to from cereal length is the wheat therefore, Begash third width; late The in the cist. in burial identified (either the were variety grain from wheat samples complete BC one millennium and fragments cereal large Four evidence archaeobotanical The BC cal 1950 BC cal 1950 BC cal 2030 BC cal 2020 BC cal 2040 BC cal 1700 a BC cal 1420 1220– Domestic Domestic Funerary Funerary uilcist, Burial Domestic Domestic Domestic rhelgclMaueet fwoeseeds whole of Measurements context Archaeological rtcmaestivum Triticum (3.1L) hearth (3.0L) hearth (2.0L) level) (lower fire-pit (9.5L) level) (upper fire-pit (30.8L) cremation human from ash erh(5L) hearth (9.5L) hearth erh(6L) hearth . . . 0.38 0.5 1.6 1.7 1 1 . . . 0.60 0.9 1.4 1.5 1 1 . . . 0.63 1.0 1.5 1.6 3 1 4 0461615050.31 0.5 1.5 1.6 6 4 10 221 . . . 0.40 0.6 1.4 1.6 10 2 12 11 41 13 11 24 583 (See 37 8 45 or ihe .Frachetti D. Michael .turgidum T. aiu miliaceum. Panicum 999 ,mauig52mi eghad4.3mm and length in 5.2mm measuring ), . . . 0.56 0.50 0.47 1.0 1.1 0.9 1.5 2.1 1.6 1.8 2.2 1.9 . . . 0.60 0.6 1.5 1.5 tal et ale4 e Tabl . o esrmnso ae samples) later of measurements for

Research

Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region Worked fibres (unidentifed) (unidentifed) fibres Worked

5 Awn

1 Unidentifiable seed frags. frags. seed Unidentifiable 9 1 3 2 7 1 3 1 6 4 8 1 4 6 6 8

4 4 1 1 1 3 1 1 5

1 Unidentified seed seed Unidentified

1 1 1 1 2 6

1 Asteraceae type type Asteraceae

1 1 2 2 6 4 1

Lithospermum arvens Lithospermum e

1 1 1

sp. sp. m Hypericu

type type a Mentha/Nepat

type type Tribulus

Brassicaceae Brassicaceae

sp. Polygonum

1 6 1 5 1 1 2 1 2

Polygonaceae Polygonaceae

) sylvestris (c.f. sp. a Malv

1 Hyoscyamus niger niger Hyoscyamus

9 1 1 2 1 4 3 2 5 2

Galiu sp. sp. m 9 1 4 2 5 6 5 8 8 8 3 5 4 4

2 3 7 2 1 1 7 1 2

1 1

Chen -ams -ams o 0 2 1 4 1 4 7 4 3 4 5 7 2 6 2

2 5 2 1 1 2 2 1 7

1 1

Chenopodiu spp. spp. m 8 7 8 9 1 3 1 2 7 5 9 8 4 2 5

3 5 2 4 1 6

1 1 Panicoid type type Panicoid

3 1 1 1 2

type Stipa

4 2 1 2 0 1 8 1 1 1 2 9

1 3 1

Panicum miliaceu Panicum

m 2 4 0 1 1 1

1 1 Cerealia Cerealia

Triticum aestivum/turigidu Triticum m 1

Wood (>2.oomm) Wt. Wt. (>2.oomm) Wood 3 3 7 3 5 9 3 2 3 1 3 8 3 3 2 6

1 2 7 5 5 0 0 0 2 6 4 0 0 1 0 0 2 ...... 9 6 4 3 0 6 0 1 0 2 0 0 0 2 7 1 1

1 1 1

Wood (>2.oomm) Ct. Ct. (>2.oomm) Wood

4 4 5 6 9 0 8 6 5 4

9 1 4 C C C C C

5 2 5

5 8 1

1 4 N

N N

Vol. litres litres Vol. 2 5 8 1 1 4 7 8 2 8

. . . .

. . .

. .

2 9 9 5 0 3 0 1 0 0 3 3 6 1 9 1 9

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 3 2 4 5 0 5 5 0 0 0 0 0 0 5 0

9 9 7 0 7 9 7 7 7 7 0 0 9 7 9 7 7 Date range cal BC BC cal range Date 1 1 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 – – – – – – – – – – – – – – – – – 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6 5 5 6 5 8 5 5 6 6 5 5 5 5 5 6 5 4 2 4 9 2 4 4 9 9 9 4 9 9 9 4 9 9 2 2 2 1 1 1 1 1 2 2 1 1 1 2 2 2 1

Flotation sample # # sample Flotation

0 9 8 5 9 7 6 4 3 1 0 2 6 0 8 9 7 5 4 4 4 1 4 Totals 105.2 NC 76.62 176.62 4NC 29 Totals 105.2 92 9 529 514322 5602 91 4 40 2 1 3 51 31 465 2 5 4 4 4

1 4

12 1625–1000 9.5 67 0.48 10 1 40 32 28 11 39 2 12 2 4 2 4 9 85 1 4 3 4 3 3 3 Table 3. Prevalent taxa recovered at Begash in Brone Age contexts (2460–1000 cal BC).

1000 omo ha.Tepuprudse hp fteBgs ha eeal conforms generally wheat Begash the also of trait shape similar a a seed rounded, represent round grains (2n slightly cereal plump only hexaploid the The and to all grains. flat think wheat. barley we rather of in Thus, is common form grains. grain more fragmented feature whole the a in the – common on humped be side to ventral appear The not does side dorsal The view. dorsal b) view; ventral a) (calibrated): BC 2460–2150 (FS47) Begash from seed wheat Carbonised 3. Figure 99 03.Frhrnrhaogtewsenfig ftePmrMutis additional Mountains, Pamir the of fringe western the along north Further 2003). sites, 1999, Harappan fifth later mid at the and least 2000) at Hopf by Mehrgarh & at Zohary region 1984; c Valley Indus (Costantini the BC in millennium known are forms wheat assemblage, free-threshing a Begash from the are of wheats aestivum/turgidum. size T. Begash the sample no that limited (2n possibility researchers tetraploid the the most Given acknowledge associated present. characteristics, authors when not the wheats morphological free-threshing are and of parts size varieties spikelet between in differentiation overlaps practice to longer due including fact, taxa, different In into placed been sometimes s nsuhr eta sab 0020 C n ieyerir(Moore earlier likely and BC, 3000–2000 in by already Asia was central wheat southern free-threshing in round, that use documenting Tepe, Gonur and South Anau 5020 a C(ee 91.Compact 1991). (Weber BC cal 2500–2000 . h dnicto fteBoz g ha rmBgs sipratbcuecompact because important is Begash from wheat Age Bronze the of identification The = 28), = 2 ha species, wheat 42) .turgidum T. hrfr,w osraieylm hmudrtecategory the under them lump conservatively we Therefore, . ihe .Frachetti D. Michael .aestivum T. 1001 .aestivum T. u hr on om fwethave wheat of forms round short but , tal et .sphaerococcum T. . sas dnie tstssc as such sites at identified also is and tal. et .compactum T. 94 Miller 1994; .

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region

Table 4. Measurements of whole seeds of P. miliaceum from Begash, post-1000 BC. Sample # & age Length (mm) Width (mm) Scutellum (mm) Scutellum/length ratio

FS2 AD 1220–1420 2.2 2.1 1.2 0.55 2.3 2.2 1.3 0.57 2.1 2.0 0.9 0.43 2.0 1.8 0.9 0.45 2.3 2.2 1.0 0.43 2.2 2.1 1.0 0.45 2.0 1.8 0.9 0.45 2.3 2.3 1.1 0.48 FS6 390–50 cal BC 2.4 2.4 0.9 0.38 2.3 2.3 0.7 0.30 2.3 2.3 0.8 0.35 2.4 2.3 1.0 0.42 2.5 2.3 0.6 0.24 2.0 1.9 0.7 0.35 2.0 2.0 0.7 0.35 1.9 2.0 1.0 0.53 2.1 2.1 0.7 0.33 2.2 2.0 0.6 0.27 2.3 2.2 1.0 0.43 evidence of free-threshing wheat is documented in phase III levels (c. 2600–2000 BC) at the site of Sarazm in western Tajikistan (Willcox n.d.; Razzokov 2008). Crawford (1992) notes that in eastern Asia, wheat forms are predominantly hexaploid and he suggests this is the case for the earliest wheat in China (c. 2600 cal BC) and for later archaeobotanical wheat from Korea (c. 1000 cal BC) and Japan (beginning of the first millennium AD) (see also Crawford & Lee 2003). Recently published wheat from Donghuishan (c. 1700–1500 cal BC) in the Hexi Corridor (Gansu) conforms to a compact morphotype similar to the wheat recovered from Begash (cf. Flad et al. 2010). Of course, more archaeobotanical evidence and more detailed comparisons are necessary before confident statements can be made about the possible spread of wheat from southern central Asia or the northern Indus Valley through Semirech’ye into China. Nonetheless, chronologically and geographically, the Begash wheat lends support to the hypothesis that a likely trajectory for wheat into China was north through the mountains from southern central Asia and east along the foothills of the Tian Shan and Dzhungar Mountains, spread by mountain pastoralists in the mid to late third millennium BC. Carbonised remains of broomcorn millet exist throughout the 4000 year chronology of habitation at Begash (Table 2), whereas foxtail millet is recovered only in samples dating to the first millennium BC and later. Although the morphology of the third millennium BC broomcorn millet is consistent with that of later periods, the later millets are larger (Table 4). Nevertheless, the measurements of the earliest broomcorn millets (see Table 2) are well within the range of published sizes for domesticated Panicum miliaceum across Eurasia (Renfrew 1973; Fuller 2006). All of the caryopses are round to oval in broad view (Fig- ure 4) and scutellum length is less than two-thirds total caryopsis length with broad width.

1002 rmtebra itwr umte o ietASdtn omr rcsl aeteplant (post-treatment) the carbon date of precisely 0.6mg more only to yielded late dating fragments the these AMS Because direct of Begash. for seeds at contexts submitted evidence millet were undisturbed broomcorn cist AMS from fragmentary burial were seven came the – Subsequently, from ﬁre-pit all 5). associated (Figure that the BC establishing from millennium two third of Analytic, and samples Beta cist Three burial at contexts. the dated obtained corresponding from were their one dates for – AMS and charcoal direct wood remains years, seed 500 domesticated nearly the spans for archaeologically and region the in iue4 abnsdmle ( millet Carbonised 4. Figure ic h aletocpto hs tBgs rdtsayohrpsoaitsettlement pastoralist other any predates Begash at phase occupation earliest the Since .miliaceum u e c a i l i m P. ed rmBgs,pae1:a S7 )F4;d FS50. d) FS44; c) & b FS47; a) 1a: phase Begash, from seeds ) ihe .Frachetti D. Michael 1003 tal et .

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region

Figure 5. Calibrated ranges (BC) of AMS dates from flotation samples containing domesticated seeds at Begash. and after consultation with the AMS technicians at Beta Analytic, a fragmentary wheat grain from the same burial-cist flotation sample was added to the sample of millet seeds to produce the Beta-266458 AMS date. The grains are absolutely dated between 2460 and 2150 cal BC (95.4% confidence), with the most probable calibration intercept of 2290 cal BC (Table 1). Given the archaeological context, a weighted average of all the samples was used to yield a 2-sigma range between 2280 and 2060 cal BC with the highest probability density (79%) falling between 2230 and 2130 cal BC (see Figure 5). Wild herbaceous seeds were the predominant plant remains recovered in the phase 1a assemblage and were prevalent throughout the assemblages from all phases at Begash (Table 3). The presence of wild herbaceous seeds at Begash could result from a number of taphonomic processes, including seed rain as well as human and animal foraging. Wild herbaceous seeds could also have been introduced in domestic contexts through the burning of dung laden with seeds, a practice prevalent among ethnographically documented pastoralist communities of the Eurasian steppe. A variety of wild taxa were identified at Begash, with Chenopodium album, Hyoscyamus sp., Galium sp., and Stipa-type being most abundant from phase 1a. Chenopodium is the most common seed type documented throughout all occupation phases at Begash. The smooth testa and relatively large size of these seeds conforms to C. album, a species commonly found in archaeobotanical assemblages across Eurasia. Hyoscyamus seeds are also identified in many of the Begash samples. Only two species from this genus grow in this part of Eurasia, H. niger and H. pusillus (Wu et al. 2006: 306), with H. niger being the more common species found in the area today. The Galium seeds from Begash are large and were likely setose (bristly). A number of caryopsis fragments from a long-seeded species of grass were also identified; these caryopses are similar in morphology to those of Stipa. In addition, they are often associated with twisted awns in the assemblage. While a definitive identification as Stipa is not possible, this is one of the more prevalent grasses in the region today.

1004 endcmne ntorgos ntewsenrgos ot fteBakSa farming Sea, Black the of north regions, western only ( the has einkorn In BC cultivated millennium regions. communities evidence second two the archaeobotanical in before periphery, grains documented its domesticated been of and production Eurasia or use central the of for territory steppe the Across Discussion rmacaooaia tde ferytidmlenu Cbra ugn ntenorth the in kurgans burial BC third-millennium western early the of in studies their BC augment archaeobotanical to from millennia cultigens, than second rather (Anthony to plants, economies wild pastoralist third on primarily the Samara relied Russia’s of region in forest-steppe Krasnosamarskoe communities as that such settlements, illustrate and Valley, burials Age Bronze from soils AMS contexts Age Iron (Rosen from BC barley, cal in and 700 located wheat around to also millet, Tuzusai, dated of as phytoliths such yielded BC, there (Chang millennium analysis Kazakhstan a first south-eastern the of of region of Semirech’ye existence settlements the the from steppe as comes few steppe confirm such Eurasian Hanks a the to (B. in only research, use region Stepnoe, plant trans-Ural future domesticated as the for evidence of in such well-dated Currently, grains results sites domesticated of the nearby distribution at wider await underway we currently Thus, reports that (2002) information. Gadyuchenko morphological full: or in published details not archaeobotanical Triticum are the and samples dated directly the of not of sites are grains Age reported Bronze the Middle However, the 2002). from comes ( Alandskoe steppe and Eurasian Arkaim central or the west in BC to millennium east transmission, of steppe vectors Eurasian possible central understanding the in otherwise. for lacuna impediment the major making China, a north-eastern explain. from to comes difficult more millet far Hunt been 1984; by have (Lisitsina diffusion noted of BC its occurrence As 5000 of The pathways before the 1999). documented although (Miller also 2006), Asia is Austin (Price Europe Asian central Europe central to south-west into in Plateau north millet Iranian of and domesticated the diffusion west across Moore – east slow BC 377; and millennium of 2000) 1996: sixth trajectories the Gosden before distinct & starting with domesticates (Harris associated einkorn BC commonly Dag of 6000 is Kopet cultivation around the the barley for of al. and evidence et piedmont provide wheat northern also emmer the Turkmenistan, central and along day southern Jeitun, present In as 2003). in (Pashkevich range such BC villages millennium Neolithic sixth the Asia, least at from economies ssp rhebtnclaayi eoesadr ehd mn rhelgclpoet nthe in projects archaeological (Anthony and among zone flotation methods steppe have standard recently Only become methodology. analysis archaeological of archaeobotanical artefact an partly is steppe h nypeiu eot facaooaia vdnefrwetadmle ftethird the of millet and wheat for evidence archaeobotanical of reports previous only The h erasneo oetctdpateiec rmayhr ntecnrlEurasian central the in anywhere from evidence plant domesticated of absence near The dicoccum . 94 ilr19;Heet20) nbt ein,tepeec fwetadbarley and wheat of presence the regions, both In 2003). Hiebert 1999; Miller 1994; p rmAki n lnso ihu pce dnicto,drc chronology, direct identification, species without Alandskoe and Arkaim from sp. ha,bre n rocr ilta opnnso hi eetr village sedentary their of components as millet broomcorn and barley wheat, ) tal. et tal. et c 2010 C,lctdi h rn-rlrgo (Gadyuchenko region trans-Ural the in located BC), 2200–1800 . 20)addsusdaoe h aleteiec o broomcorn for evidence earliest the above, discussed and (2008) 05.Yt eetrsac sn opeesv oainof flotation comprehensive using research recent Yet, 2005). tal. et .monococcum T. ihe .Frachetti D. Michael tal. et 05.Lkws,ol idpat aebe recorded been have plants wild only Likewise, 2005). 2000). 1005 ssp monococcum . tal et . tal. et n me ( emmer and ) 03.Smln n soil and Sampling 2003). Panicum es comm pers. .turgidum T. p and sp. .).

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region

Caspian steppe region (Shishlina et al. 2008). Although similar wild plant taxa were also prevalent in the phase 1a assemblages at Begash, the site provides a unique archaeological case for the use of domesticates among central Eurasian steppe pastoralists during the middle to late third millennium BC, especially in the eastern region. The presence of both wheat and broomcorn millet at Begash highlights the challenges faced by researchers who seek to understand the role of pastoralists in the spread of crops across central Eurasia and the significance, if any, of crops in pastoralist economies at various places and times. At Begash, only two millet seeds were recovered from domestic hearths in phase 1a, while 26 millet seeds and five wheat/cereal specimens were recovered from the burial and associated funerary fire-pit. Given their overall low abundance across the site and increased concentration in ritual features, broomcorn millet and wheat do not appear to have been an everyday food source among pastoralists at this time. To the extent that Begash is representative of the earliest use of domesticated grains in the steppe region, broomcorn millet and wheat may have initially been sought by Eurasian pastoralists in this region as important ritual commodities for use in burial ceremonies during the late third millennium BC. Interment of wheat grains with the dead is further documented in later cemetery sites of the second millennium BC in Lop Nor (Xinjiang) (Flad et al. 2010). Meanwhile, the consumption or offering of other economically restricted resources, such as horses, has recently been documented through lipid analysis of ceramics from Late Bronze Age burials in the central steppe region (Outram et al. in press). These new data suggest that a complex array of ritual practices involving ideologically important commodities, such as domesticated grains and possibly horses, could have helped fuel the transmission of these innovations across Eurasia. Given the isolation of Begash in relation to other known data, it is premature to specify the pathways or motivations that brought wheat and millet to these steppe pastoralists in the late third millennium BC. Several species of wheat – especially emmer and bread wheats – were widely grown in southern central Asia and along the piedmont of central Asian mountains by the third millennium BC (discussed above, also Moore et al. 1994). Begash’s strategic location along a pass through the Dzhungar Mountains may have situated it along what Lu (quoted in Lawler 2009: 941) calls a ‘wheat road’; part of a mountain corridor along which wheat (and other innovations) may have diffused into China in the third millennium BC. The small, roundish wheat grains from Begash conform to a morphotype some researchers have called ‘Indian dwarf wheat’ (T. aestivum ssp. sphaerococcum; Zohary & Hopf 2000: 52), which is believed to have been common in the northern Indus Valley at the time in question (Weber 1991). Later wheat remains of the second millennium BC from western China share the plump, round morphology of the Begash wheats (Flad et al. 2010), suggesting that the central Asian mountains may have provided a key passage for wheat diffusion into western China in the Middle Bronze Age. Although this explanation provides a tantalising vector for regional interaction, complications in sub-specific identification and the need for more comparisons make direct regional associations premature at this time (cf. Fuller 2001). How broomcorn millet spread into Kazakhstan is an even more difficult question given its earlier presence in both China and south-east Europe. Hunt et al. (2008) summarise the evidence for Old World millets predating 5000 BC, stressing the lack of unified identification criteria, wide regional gaps in archaeobotanical data, and the fact that the

1006 si ie aly,sgetn ht–a es niomnal padvalley/foothill (Zhao upland cultivation millet – limited environmentally supported least have present. well at could as – Begash Crawford contexts elevated around that 2005; ( in that suggesting Xinglonggou effectively like valleys, and grown settings Yuezhuang was river Eurasian China as in the in such as millet of sites broomcorn microenvironments outside that from agro-pastoralists ecological show and reports rich dry-farmers Recent by including scale steppe. areas, small a agricultural on requirements, produced major moisture been low have and investment could sowing it minimal days), (30–45 season growing xtc ihsau,o crecosadaia eore sa xiignwdrcinfor analysis direction residue new and exciting archaeobotany Outram an like (cf. of methods is essential exploitation which resources are ritual for animal archaeology, of Eurasian and nature in crops The study scarce horses. or like perhaps high-status, offerings veneration, exotic, or consumptive ideological consumption other broad Bronze ritual of augment for about possibility grains to facts of the emerging acquisition suggests the of communities behind synthesis steppe use motivations hearths), provocative among earliest rituals A cooking their burial in Begash. (e.g. Age subsistence at features wheat beyond pastoralists domestic and importance steppe from millet held by than of plants recovery rather domesticated The that contexts, BC. suggest ritual Begash millennium at and economy second burial pastoralist the domesticated a in that into predominantly and well was BC, what regionally) in 2200 (and provision around minor region a established this were in and grains an wheat commodities domesticated of that rare though propose BC, represent indicative we millennium Instead, millet third analysed. not and late and excavated to is be mid steppes must evidence the sites protracted more the in current Begash along throughout at The economy products living agricultural Eurasia. productive regional pastoralists central among of of especially diffusion mountains exchange, the of and/or networks production prehistoric agricultural innovation of the concerning map of questions the our on reorient node significantly key and transformation a economic document Begash from remains archaeobotanical The Conclusion of ancestor wild s ncnrlErsamr hn10 er noteps,teacaooaia aafrom data diverse archaeobotanical the the of past, understanding chronological the and into geographic years our 1500 expand than questions significantly Begash millet more resolve plant Eurasia broomcorn domesticated not central of and chronology in do wheat absolute during use the Begash of extending millet, domestications from by However, broomcorn regional Neolithic. remains the and or during seed spread wheat The earliest as BC. the such concerning millennium crops, third grain confluence late of early an the diffusion illustrates and for steppe trajectories Eurasian the of and Asia, south-west China, (Chang western landscape region’s political and the social for their economy in base changes 2002). the clear augmented fuelled increasingly and pastoralists production millet and wheat nsm eahrpeet n fteeris ae rhelgclstsa h rsrasof crossroads the at sites archaeological dated earliest the of one represents Begash sum, In tal. et .miliaceum P. 06.Hwvr hr sltl vdnet upr hscamat claim this support to evidence little is there However, 2006). tal. et sntceryietfid u obomonmle’ short millet’s broomcorn to Due identified. clearly not is npes.B h ideo h rtmlenu C both BC, millennium first the of middle the By press). in ihe .Frachetti D. Michael 1007 tal et . c 00BC) 6000 . tal. et

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region vectors of interaction, ritual exploitation, and diffusion among societies of eastern Asia, south and south-west Asia, and regions further west across the Eurasian landmass.

Acknowledgements Funding for this research was provided by the National Science Foundation, grant # 0535341, the Mary Morris- Stein Foundation, and by Washington University in St Louis. We are grateful to Dr Olga Pontes (Washington University in St Louis) for her help with seed imaging. Professors Gary Crawford and Philip Kohl contributed helpful comments and improvements to the ﬁnal manuscript. Finally, we thank Professor Karl Baipakov, director of the Institute of Archaeology, Almaty Kazakhstan, for his support of continued international collaboration.

References CRAWFORD, G.W. 1992. Prehistoric plant domestication in East Asia, in C.W. Cowan & P.J. ANTHONY, D.W. 2007. The horse, the wheel, and Watson (ed.) The origins of agriculture: an language: how Bronze Age riders from the Eurasian international perspective: 7–38. Washington (DC): steppes shaped the modern world. Princeton (NJ): Smithsonian Institution. Princeton University Press. – 2009. Agricultural origins in North China pushed ANTHONY, D.W., D. BROWN,E.BROWN,A. back to the Pleistocene-Holocene boundary. GOODMAN,A.KOKHLOV,P.KOSINTSEV,P. Proceedings of the National Academy of Sciences of the KUZNETSOV &L.POPOVA. 2005. The Samara United States of America 106: 7271–2. Valley Project: Late Bronze Age economy and ritual in the Russian steppes. Eurasia Antiqua 11: CRAWFORD, G.W. & G.A. LEE. 2003. Agricultural 395–417. origins in the Korean peninsula. Antiquity 77: 87–95. AUSTIN, D.F. 2006. Fox-tail millets (Setaria: Poaceae) — abandoned food in two hemispheres. Economic CRAWFORD, G., A. UNDERHILL,ZH.ZHAO, G.A. LEE, Botany 60: 143–58. G. FEINMAN,L.NICHOLAS,F.LUAN,H.YU,H. FANG &F.CAI. 2005. Late Neolithic plant remains BELLWOOD, P. 2005. First farmers: the origins of from northern China: preliminary results from agricultural societies. Malden (MA): Blackwell. Liangchengzhen, Shandong. Current Anthropology CHANG,C.,N.BENECKE,F.P.GRIGORIEV, A.M. ROSEN 46: 309–327. OURTELLOTTE & P.A. T . 2003. Iron Age society and CRAWFORD, G.W., X. CHEN &J.WANG. 2006. Houli chronology in south-east Kazakhstan. Antiquity 77: culture rice from the Yuezhuang site, Jinan. 298–312. Dongfang Kapgu [East Asia Archaeology] 3: 247–51. HANG OURTELLOTTE AIPAKOV C ,C.,P.T ,K.M.B &F.P. FLAD,R.,S.LI,X.WU &ZH.ZHAO. 2010. Early wheat GRIGORIEV. 2002. The evolution of steppe in China: results from new studies at Donghuishan communities from Bronze Age through medieval in the Hexi Corridor. Holocene 17: 555–60. periods in south-eastern Kazakhstan (Zhetysu). Sweet Briar (VA): Sweet Briar Press. FRACHETTI, M.D. 2008. Pastoralist landscapes and social interaction in Bronze Age Eurasia.Berkeley(CA): COLLEDGE,S.,J.CONOLLY &S.SHENNAN. 2004. University of California Press. Archaeobotanical evidence for the spread of farming in the Eastern Mediterranean. Current FRACHETTI,M.&N.BENECKE. 2009. From sheep to Anthropology 45: S35–58. (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash (south-eastern COSTANTINI, L. 1981. Palaeothnobotany at Pirak: a Kazakhstan). Antiquity 83: 1023–37. contribution to the second millennium BC agriculture of the Sibi-Kacchi Plain, Pakistan, in H. FRACHETTI, M.D. & A.N. MAR’YASHEV. 2007. Hartel¨ (ed.) South Asian archaeology 1979: 271–7. Long-term occupation and seasonal settlement of Berlin: Dietrich Reimer. eastern Eurasian pastoralists at Begash, Kazakhstan. Journal of Field Archaeology 32(3): 221–42. – 1984. The beginning of agriculture in the Kachi Plain: the evidence from Mehrgarh, in B. Allchin FULLER, D.Q. 2001. Responses: Harappan seeds and (ed.) South Asian archaeology: 29–33. Cambridge: agriculture: some considerations. Antiquity 75: Cambridge University Press. 410–14. – 2006. A millet atlas: some identiﬁcation guidance. Available at: http://www.homepages.ucl.ac.uk/ ∼tcrndfu/Abot/Millet%20Handout06.pdf (accessed 4 June 2010).

1008 L G L L L L H H H K L M U I EBEDEVA IU ISITSINA AWLER UZ ADYUCHENKO IEBERT UNT ARRIS ,X.,J.D ILLER ,H.,J.Z ,X.,H.V.H edn Brill. Leiden: etr eta sai h hloihcadBronze and Chalcolithic Ages. the in Asia central western 06 i Russian). (in 50–68 Arkheoligii v Issledovaniya Europe. Mezhdistsiplinarnye Eastern in agriculture Age Bronze 942–3. raeigo giutr nNoihcChina. Holocene Neolithic in agriculture of broadening otra:Balkema. Rotterdam: W.A. & Zeist (ed.) van W. Casparie in Eurasia, in farming ancient Man. Eurasia central beyond palaeogeography; archaeoecology, and Age; geoarchaeology, 1st Iron the the to 2: 3rd Volume the millennium. from complex Eurasia BC central models of global societies of light (ed.) in Zdanovich specifics D.G. Regional & Jones-Bley K. ‘Country in Towns’ the of of necropoli and settlements fortified M M mtsna Institution. Smithsonian Eurasia in pastoralism D.R. Harris in (ed.) Asia, central western in agriculture Archaeobotany World. Old the in sites the of records genera early of context and chronology uemPress. Museum Pennsylvania of University (PA): TurkmenistanPhiladelphia Anau, at excavations civilization, of Antiquity farms. earliest China’s North of perceptions archaeological changing foothills: and valleys h ntdSae fAmerica of States United the ago. miliaceum aletdmsiaino omnmle ( millet common of domestication Earliest ..J M.K. ’ MINA .. .V M. H.V., , ASUTOMOTO OTUZAITE ,D.R.&C.G ..19.Arclua eeomn in development Agricultural 1999. N.F. , .20.Mle ntemove. the on Millet 2009. A. , F.T. 2003. , rceig fteNtoa cdm fSineof Science of Academy National the of Proceedings ..18.TeCuau:acnr of centre a Caucasus: The 1984. G.N. , h rgn n pedo giutr and agriculture of spread and origins The eeainHsoyadArchaeobotany and History Vegetation .20.Acaooayadsuisof studies and Archaeobotany 2005. E. , .2007. E. , ONES ODSON Panicum 0–1.Wsigo D) nttt of Institute (DC): Washington 400–418. : HANG 7 555–60. 17: 3 82–95. 83: nEs saetne o1,0 years 10,000 to extended Asia East in ) ..20.Ogncrmisfrom remains Organic 2002. L.L. , UNT 08 ilt cosEurasia: across Millets 2008. . -M ,K.L ,X.Z lnsadacetman ancient and Plants 7 S5–18. 17: ANDER 07 al utvtdwetand wheat cultivated Early 2007. . ATUZEVICIUTE ..J M.K. & , and eta sa ilg ttedawn the at village Asian central A TheoriginoftheIndo-Iranians OSDEN IU HOU Setaria 7–9 ahntn(DC): Washington 370–89. : ,N.W L INDEN eeainHsoyand History Vegetation ,H.Z 96 h einnsof beginnings The 1996. . ONES 0:7367–72. 106: rmarchaeological from U ,S.C ,X.L ,Y.L HANG 09 River 2009. . Science OLLEDGE I IU , TAL ET &R. 285–92. : ,G. ihe .Frachetti D. Michael :13–19. 8: 325: Panicum 2009. . Opus: & 4: . . 1009 M Hiebert F. in North, Anau at plants of use The 2003. – R P R R W W 91 abnsdpatrmisfo Shortuga from remains plant Carbonised 1991. – W S O P P HISHLINA ASHKEVICH EARSALL RICE OSEN ENFREW AZZOKOV UTRAM OORE ILLCOX EISS EBER nvriyo enyvnaMsu Press. Museum Pennsylvania of University Anau, at Turkmenistan excavations civilization, of dawn the (ed.) Kurbansakhatov K. & akRsulk Tadzhikistan. Respubliki Akademii nauk Donnsha A. etnografiiim. I arekheologii coadIsiueo Archaeology. of Institute McDonald horse the and adaptation steppe M. in Age, Bronze and Levine,C.Renfrew&K.Boyle(ed.) East Neolithic of Late steppe the forest in and Europe steppe the in agriculture of abig nvriyPress University Cambridge odpat fteNa atadEurope Press. and University East Columbia Near the of plants food procedures C) etiwPress. Rojdi Westview from (CO): change and stability of example Science domestication. before cultivation Autonomous efe (ed.) Renfrew CNRS. Orientale, Pr de Institut report, paleoethnobotanical dnug:EibrhUiest Press. University Edinburgh Edinburgh: U aionaLsAngeles. Los California of University Archaeology, of Institute Cotsen New and World nomadism World Old mobility: of (ed.) archaeology Wendrich W. The & Barnard H. and in occupation production, seasonal the of on models Hills steppe: Yerugueni Eurasian South the of sites Nomadic Kazakhstan. south-eastern in agro-pastoralists Age Saka-Wusun Iron of economy and Paleoenvironments rs.Hre o h ed ueayfowy in Kazakhstan. foodways Age funerary Bronze dead: the for Horses press. 418–27. civilization. Oxus the oasis of early settlements in herding and Agriculture 1994. ..2000. T.D. , SMANOVA ,E.,M.E.K tal et .. .C C. A.M., , ..1991. S.A. , ,K,N.M .. ..S N.A. A.K., , ..2000. D.M. , ,G.n.d.UnpublishedSarazm ..1973. J.M. , .. ..G E.I. N.I., , .2008. A. , 1:1608–10. 312: . .20.Pleehooaia evidence Palaeoethnobotanical 2003. G. , a ig C) cdmcPress. Academic (CA): Diego San . ,V.V Antiquity 2–8 hldlha(A : (PA) Philadelphia 127–38. : ILLER e ih neryfarming early on light New ISLEV uoesfis farmers first Europe’s HANG lnsadHrpa ussec:an subsistence: Harappan and Plants 3–9 o nee (CA): Angeles Los 230–49. : ARFOLOMEEV Sarazm aaotnbtn:teprehistoric the Palaeoethnobotany: ,F.H TEAR aeehooay adokof handbook a Paleoethnobotany: AK &A.H &F.P.G 4 611–23. 74: ..B A.V. & ,A.K uhne ntttistorii, Institut Dushanbe: . IEBERT Antiquity eta sa ilg at village Asian central A ARTMANN 8–7 Cambridge: 287–97. : ASPAROV &R.P.E RIGORIEV &R.M ORISOV . Cambridge: . Antiquity Prehistoric 2006. . .NewYork: ,E. 2008. . VERSHED EADOW 2000. . 139–54. : .Boulder ehistoire ´ ¨ ,i J. in ı, 68: . .In

Research Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region

– 2005. The distribution, natural habits and availability ZHAO,ZH. 2005. Flotation results from the of wild cereals in relation to their domestication in Xinglonggou site, North China and the problem of the Near East: multiple events, multiple centers. dryland agricultural origins. Antiquities of Eastern Vegetation History and Archaeobotany 14: 534–41. Asia A: 188–99. WU,Z.,P.RAVEN & Missouri Botanical Gardens. 2006. ZOHARY,D.&M.HOPF. 2000. Domestication of plants Flora of China. St Louis (MO): Missouri Botanical in the Old World: the origin and spread of cultivated Gardens. plants in west Asia, Europe and the Nile Valley. Oxford: Oxford University Press.

1010