Veget Hist Archaeobot DOI 10.1007/s00334-007-0121-y ORIGINAL ARTICLE Early Holocene cultivation before domestication in northern Syria George Willcox • Sandra Fornite • Linda Herveux Received: 12 October 2006/Accepted: 14 February 2007 © Springer-Verlag 2007 Abstract Charred plant remains from the sites of Tell Qaramel, Jerf el Ahmar, Dja'de and Tell 'Abr situated in northern Syria and dated to the tenth and ninth millennia cal b.c. demonstrate that a wide variety of wild pulses, cereals, fruits and nuts was exploited. Five lines of evidence suggest that cultivation was practised at three of the sites. (1) Wild einkorn, wild rye and lentils occur outside their natural habitats. (2) The founder crops barley, emmer and single-grained einkorn appear at different times. (3) An assemblage of weeds of cultivation was identified. (4) There is a gradual decrease in gathered plants such as small seeded grasses and Polygonum/Rumex. (5) Barley grains increase in breadth and thickness. Morphological domestication did not become established, perhaps because seed stock was regularly collected from wild stands. Charred rodent droppings indicate large-scale grain storage. Keywords Archaeobotany ■ Syria • Neolithic ■ Cultivation • Wild cereals • Early farming Introduction Two recent articles hypothesize that the origins of cultivation of wild cereals and pulses considerably predate their domestication in the Near East (Tanno and Willcox 2006a; Weiss et al. 2006). Pre-ninth millennium (cal b.c.) sedentary sites in the Euphrates region, which have produced Communicated by A. Fairbairn. G. Willcox (El) • S. Fornite • L. Herveux Archeorient, CNRS, Antenne de Jales, Berrias, 07460 St-Paul-le-Jeune, France e-mail: gwillcox@wanadoo.fr wild cereals and pulses include Abu Hureyra (Hillman 2000) , Mureybet, (van Zeist and Bakker-Heeres 1984) and Gobekli (Neef 2003) (for site locations see map, Fig. 1). Hillman (2000) suggested that wild rye and einkorn were cultivated at Abu Hureyra on the Euphrates by about 11,000 cal b.c. One of his arguments was that the harsh climatic conditions of the Younger Dryas would have not allowed wild stands to grow in the region (Hillman et al. 2001) . For Mureybet, Colledge (1998) suggested that there is evidence for pre-domestic cultivation on the basis of weeds of cultivation. Pre-ninth millennium sites situated farther east and north such as Hallan Cemi, Demirkoy, Nemrik, Qermez Dere and M'lefaat (not shown in Fig. 1) also produced wild cereals and pulses, but evidence of cultivation is less forthcoming (Savard et al. 2006). Reliable signs of morphological domestication indicated by the partial loss of the dispersal mechanism in einkorn, barley and emmer do not appear until about 8,500 cal b.c. These have been found in levels commonly referred to as early PPNB at the sites of Nevah Cori, Tell el Kerkh, Cayonu and Cafer Hiiyok, but wild types persist (Tanno and Willcox 2006a, b; Pasternak 1998; van Zeist and de Roller 1994; de Moulins 1997). Later sites dated to the first part of the eighth millennium cal b.c. (middle PPNB) cover a bigger surface area and have evidence of well-established agriculture, demonstrated by morphological domestication of emmer and barley on all sites so far examined (Nesbitt 2002) . In this report we present new results of identifications of charred plant remains from the sites of Qaramel, Jerf el Ahmar, Dja'de and Tell 'Abr dated to the tenth and ninth millennia cal b.c. (for the location and dating see Figs. 1,2). We concentrate on the similarities and differences of plant assemblages between sites, including in some instances Abu Hureyra and Mureybet, in order to trace how plant use Springer Veget Hist Archaeobot altitude isohytes o_so too km 0 3QO6OO9O0 jjj^^^jj^500 AOO'- """" Fig. 1 Site location map giving the principal sites mentioned in the text Dja'de 9100±80BP Dja'de 9300±80BP Jerf 9300±80BP Jerf 9500±80BP Abr 9700±80BP Jerf 9600480BP Jerf 9800±80BP Qaramel 10000*80BP Qaramel 10200±80BP_ j_I_i_i_i_i_I_i_i_i_i_I_i_i_i_i_1_i_i_i_i_I_i_i_ 12000BC 11000BC 10000BC 9000BC 8000BC Calendar date Fig. 2 The hypothetical date ranges for the four sites based on over 75 14C dates. It demonstrates the overall chronological trend of the four sites, but also the chronological overlap caused by the margin of error. Thus levels with similar dates may not be contemporary and particularly the major cereal components changed. No attempt has been made at spatial or specific archaeological interpretations. This will be the subject of future publications for each individual site. Indeed some results have already been published (Willcox 2002a). Many aspects of these ancient plant economies cannot be established with certainty, such as, how important cultivation was compared to gathering, or how far the wild stands were from the habitation sites where the cereals were stored and consumed. For some taxa we cannot be sure of their use, whether for food, fuel, crafting artefacts or whether they were simply unwanted weeds or ruderals. Despite these difficulties, we are coming to a better understanding of the origins of agriculture thanks to a number of recent archaeobotanical studies. Materials and methods Sampling was carried out using a simple 200 1 flotation tank. Charred material, which floated was sieved with a 0.5 mm mesh. This may have resulted in loss of taxa such as Juncus and Phragmites but finer meshes were difficult to use because they clogged easily. Tests showed that finer sieving produced no significant identifiable material. The density of charred plant remains other than charcoal varied from sample to sample. The sampling procedure consisted of testing sediments by flotation and selecting layers where charred remains were concentrated. Certain areas were sampled intensely, for example the spectacular finds from a kitchen at Jerf el Ahmar, which have already been reported (Willcox 2002a). The four sites span the period from the first quarter of the tenth millennium b.c. to the last quarter of the ninth millennium. Dating was based on over 50 radiocarbon dates. A selected range of dates and their calibration curves are given in Fig. 2. A brief description of each site is given below. For further details the reader is referred to the original archaeological reports. Tell Qaramel (Mazurowski 2004), the earliest site, is situated on the river Quwayq about 25 km north of Aleppo. It is the only site of the four, which is not situated on the Euphrates. The area today has an average annual rainfall of 350 mm, which is higher than that of the Euphrates sites. Samples were only taken from the Khiamian period dated to approximately 10,000 cal b.c. The site consists of round houses, hearths and pits. In one house a large number of bucrania (aurochs skulls) were found. Artefacts include Khiamian arrowheads, a rich ground stone industry with chlorite bowls, shaft straighteners, querns, pounders and "batons poliš". Jerf el Ahmar (Stordeur 2000; Stordeur et al. 2000) is the most southerly site of the four, situated on the banks of the Euphrates in an area, which receives an average annual rainfall of 250 mm. This site was the most extensively sampled and it covers a period of approximately five centuries. The oldest levels contain houses, which are similar to those excavated at Tell Qaramel. Later levels have rectangular houses and sunken round communal buildings. Dja'de (Coqueugniot 2000) has an annual average rainfall of about 300 mm and is situated 50 km upstream of Jerf el Ahmar. The site lies on a Pleistocene terrace overlooking the river. The earliest levels excavated are contemporary with the latest levels at Jerf el Ahmar and the upper levels date to approximately the end of the ninth Springer Veget Hist Archaeobot millennium cal b.c. Much of the site consists of thick ash layers rich in charred remains. A funerary building was excavated and also a building containing wall paintings. Pits are common, but habitation structures, at least in the excavated area, are rare. Tell 'Abr (Yartah 2004, 2005) is situated about 25 km farther north in an area with an annual rainfall similar to that of Dja'de. The site is contemporary with the earlier levels at Jerf el Ahmar. Both round houses and communal buildings have been excavated. Only a small volume of sediment was sampled from this site. Three samples derived from a concentration of charred grain from one of the communal buildings destroyed by fire and appear to represent storage within it. Similar buildings were found at Jerf el Ahmar and Mureybet (Stordeur et al. 2000) but this is the first time that their use for storage has been demonstrated by the finds of charred grain. Charred seeds from the four sites are generally in poor condition, often fragmented and lacking the testa. In addition, phenotypic variation, the high species diversity for the region as a whole and even the possibility of evolutionary changes means that species level identification for the majority of taxa is not possible. Because the testa is often missing, identification may rely on the internal seed structure. It is probable that hard-coated seeds such as Galium and Adonis survive preferentially and are over-represented compared to fragile seeds. Many of the identifications remain at genus level and in some cases represent more than one species. The taxa have then been divided into groups (Tables 2, 3, 4, 5, 6, 7, 8), which allow a more coherent interpretation. Several hundred illustrations of charred items including digital and SEM photographs, drawings and identification criteria are available for viewing at http:// www.perso.orange.fr/g.willcox/. Bio-mineralised seeds of the Boraginaceae family of which Arnebia was particularly common have not been included because they were rarely carbonised, so we could not be sure that they were ancient. Results The total number of samples, the quantity of sediment, the volumes of charred remains and the number of identified items are given in Table 1. The number of samples and the quantity of sediment varied for each site, but were sufficient for making comparisons. Only 30 samples were obtained from Tell 'Abr and most of the identified plant remains were concentrated in a few samples, which represent a burnt storage structure. The identified taxa have been divided into seven groups for the purposes of interpretation. Each group is presented as a table with absolute numbers of identifications and ubiquity, that is the percentage of samples in which a taxon is present regardless of Table 1 Volumes, number of samples and total items identified from the four sites Site Sediment (1) Flot (ml) Flot Samples Total sorted (ml) items Tell 'Abr 1,520 ? ? 30 4,528 Dja'de 6,122 38,052 24,552 227 32,964 Jerf el Ahmar 12,114 25,802 23,104 266 34,067 Tell Qaramel 1,772 5,934 3,500 108 12,247 quantity (Ub%). Each identifiable plant part, whether it was complete or a fragment, has been scored as one item. Wild cereals (Table 2) Absolute identification scores and percent of ubiquity based on the total number of samples are given in Table 2. A selection of cereal items is given in the bar chart (Fig. 3), providing absolute numbers of finds and their relative proportions. No reliable evidence for domestication was found among the cereal remains. A few specimens of domestic-type fused barley spikelets were found, but they would be expected in a wild population where the basal spikelets do not shatter (Kislev 1992). A large proportion of the charred grains were identified as Triticum/Secale, which includes wild rye and two-grained wild einkorn; these two species are difficult to distinguish from their grains, particularly in the charred state. This taxon is common at all four sites presented here (Fig. 3) and also at the earlier Euphrates sites of Mureybet and Abu Hureyra. Fortunately rye and einkorn can be distinguished on the basis of their spikelet bases. Rye spikelet bases were common at Jerf el Ahmar, rare at Dja'de and totally absent at Tell Qaramel, whereas einkorn spikelet bases were rare at Jerf al Ahmar, more common at Dja'de and made up all the finds at Tell Qaramel. Concerning rye at the earliest site on the Euphrates, no spikelet material was present at Abu Hureyra, but Triticum/Secale grains were common and according to Gordon Hillman (personal communication) the vast majority of the indeterminate grains were rye. At Mureybet (van Zeist and Bakker-Heeres 1984) rye spikelet bases were identified from impressions in pise (Willcox and Fornite 1999). No einkorn was identified. Some wild einkorn produces spikelets with a single grain (Triticum 1 g in Table 2 and Fig. 3). These are readily identifiable by their convex ventral face and so can be distinguished from Triticum/Secale. The changes in frequencies of single-grained einkorn are revealing from a chronological point of view. At the earliest sites of Mureybet and Abu Hureyra no einkorn was present and it was rare at Tell 'Abr and in the earliest levels at Jerf el Ahmar; finds become increasingly more frequent in the £) Springer Veget Hist Archaeobot Table 2 Scores for grains and spikelet bases of wild cereals from the four sites Abr Dja'de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% H. spontaneum gr 190 57 3,763 79 9,639 91 217 36 H. spontaneum base 153 11 3,325 58 H. domestic rachis 2 1 8 2 Triticum/Secale gr 2,999 86 1,120 64 2,539 81 1,170 46 Secale spk base 16 3 144 11 Triticum base 16 5 5 2 292 9 T. boeoticum 1 gr 90 40 302 32 67 11 1,108 68 T. dicoccoides gr 192 21 4 2 Note that 10% of domestic barley type rachis fragments can occur in wild populations (Kislev 1992) later levels (Fig. 4) and at Dja'de. These einkorn grains, like the einkorn spikelet bases, increase through time on the Euphrates sites. Tell Qaramel is situated about 75 km west of the Euphrates, and single-grained einkorn and Triticum/Secale are present there in equal quantities, which is similar to proportions in modern wild populations in southeast Turkey where both two-grained and single-grained spik-elets can be found on the same ear. These proportions combined with the absence of rye spikelet bases demonstrates that Tell Qaramel differs from the Euphrates sites because einkorn was commonly used and rye was absent. Emmer was absent at the earlier Euphrates sites including Tell 'Abr and Jerf el Ahmar. It appears for the first time at Dja'de (Table 2). Three or four centuries later in the middle PPNB emmer became the principal wheat in 16 Spikelet bases 144 292 .190 Cereal qrains 302 67 3763 9639 11 ?0 2530 Dja'de Jert Qaramel Secale spikelet base ■ T. boeoticum spik.base Abr Djade Jerf T. boeoticum 1 grain ■ H. spontaneum grain ■ Triticum/Secale grain 1108 217 1170 Qaramel Fig. 3 Left absolute scores and proportions of Secale and T. boeoticum spikelet bases, Qaramel has only T. boeoticum, Jerf el Ahmar has predominantly Secale and Dja'de has equal proportions of each. Right absolute scores of single-grained T. boeoticum, Hordeum spontaneum and Secale/Triticum types. Taking into account grain identification and spikelet bases at Qaramel, T. boeoticum dominates, Secale is absent and H. spontaneum rare. At -Jerf el Ahmar and Dja'de, H. spontaneum dominates and Secale is common while T. boeoticum is rare but increases with time (see Fig. 4 ) the region, while rye disappears completely and einkorn remains are rare. Wild barley, absent at Tell Abu Hureyra, the earliest site in the region, appears at Mureybet at the opening of the Holocene but is rare (van Zeist and Bakker-Heeres 1984). At the later Euphrates sites of Jerf el Ahmar and Dja'de it becomes increasingly frequent. This species is not common at Tell Qaramel. Of the wild cereals, it is the most resistant to high temperatures and drought Unlike rye and einkorn, barley grows today on the poor chalk soils of the middle Euphrates. Its appearance is probably linked to the increase in temperature following the Younger Dry as. Being better adapted than rye to climatic and soil conditions there, barley was adopted as a crop. On the basis of other evidence it would seem probable that a combination of climate change and incipient cereal cultivation led to this species becoming increasingly frequent. At Jerf el Ahmar there is evidence based on both ubiquity and absolute 100 n 50 Lower Upper Jerf el Ahmar • Triticum/Secale -—Af— Secale spk base T. boeoticum grain ■ H. spontaneum grain Fig. 4 Changes in the ubiquity frequencies of cereal items between early and late periods at Jerf el Ahmar (see also Tables 9, 10). These trends where barley and einkorn increase and rye diminishes follow an intra-site trend on Euphrates sites Springer Veget Hist Archaeobot Table 3 Scores for the ., ,, T „ „ . ., . ... Abr Dja de Jerf Qaramel identification of wild grasses _ __ _ __ excluding the cereals T Ub% T Ub% T Ub% T Ub% H. murinum/bulbosum gr 632 40 2,100 68 63 16 H.murinum/bulbosum base 1 1 H. bulbosum bulb 5 1 Aegilops base 10 3 144 6 Aegilops gr 13 5 7 3 1 1 Bromus 1 3 56 12 106 15 2 2 Stipa gr 11 13 272 12 118 5 526 11 Sripa twisted awn 14 4 Taeniatherum base 69 8 70 12 Taeniatherum gr 1,816 34 126 23 5 3 Poa bulbosum bract 14 2 17 3 36 1 Poaceae 87 23 3,488 63 2,380 45 267 39 Poaceae base 12 5 52 8 Poaceae culm base 27 7 6 2 Panicoid type 20 4 1 1 Phalaris 1 1 Echinaria 1 0.5 1 1 Eremopyron 42 11 9 4 Lolium 7 3 4 1 3 3 frequencies that wild barley increases progressively from early to late levels (Fig. 4 and see also Tables 9, 10). At Tell 'Abr barley is not common, probably because the samples came from a storage structure which mainly contained Triticum/Secale. Grasses (Table 3) The small wild barleys (H. murinum/bulbosum) which make up a complex of four related species including H. murinum (Nesbitt 2006) are the most common wild grasses found on the sites. H. bulbosum, despite being a perennial with bigger grains and bulbs (swollen basal culms), has also been included in this group because in the charred state it is not easy to distinguish it from the other four species. This species was identified from the bulbs. Taeniatherum caput-medusae, the only species of its genus, was identified from both grain and spikelet material. It was recovered from three of the sites. Aegilops is present both in the form of spikelet bases and grains, the former having thick glumes which probably survive charring more readily than wild cereals. In a single archaeological context at Jerf el Ahmar, 118 spikelet bases were found concentrated as though they represented the residue of threshing. These spikelet bases compared well with those of A. speltoides. Another common taxon was Stipa, also represented by its twisted awns. These grass taxa as an assemblage could represent weeds of cultivation or part of a gathered assemblage, which included wild cereals. If they were consumed, their generally smaller size compared to the larger grained wild cereals and lower frequencies indicate they would have represented a small part of the food resources at these sites compared to wild cereals. This was not the case further north in south-east Turkey where Savard et al. (2006) established that small seeded grasses were the most common at Hallan £emi and Dernirkoy. These sites are considered to have been inhabited primarily by gatherers. Weiss et al. (2004) demonstrated a gradual decline in the use of small seeded grasses between the Epipalaeolithic and the PPNB. Seeds from wild and/or weed taxa (Table 4) This group of 54 taxa includes 18 (marked with * in Table 4), which have no apparent use for food, fibres, dyes etc. This assemblage in relation to its interpretation as weeds of cultivation is discussed in more detail below. Other wild/weed taxa could have been used as food plants, for example Alyssum/Lepidium, a Crucifer type, and Polygonum/Rumex were found in high concentrations in a limited number of samples suggesting intensive use. Taxa such as Atriplex and Verbascum may have been used as fuel. Others such as Androsace maxima, Hyoscyamus, Peganum hamala and Tribulus terrestris were ruderals probably growing near the habitations. Finally, high concentrations of Ziziphora at Tell Qaramel suggest that this aromatic plant had some particular use. Similarly high levels were found at Hallan femi (Savard et al. 2006). Springer Veget Hist Archaeobot Table 4 Scores for taxa of wild plants Abr Dja'de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% Adonis* 1 3 48 13 55 11 7 6 Ajuga/Teucrium* 4 1 Alyssum/Lepidium* 488 15 2 1 Androsace maxima 8 3 32 6 2 Apiaceae 1 1 5 2 5 4 Apiaceae type 1 158 9 46 5 Astragalus 4 13 290 27 224 32 7 5 Atriplex 4,554 69 1 1 Bellevalia* 27 7 201 8 30 15 Brassicaceae 806 60 18 6 Brassicaceae type 1 532 10 Brassicaceae type Brassica 60 5 20 6 Bupleurum* 91 13 Centaurea* 338 31 20 4 Chenopodiaceae 1 3 28 4 103 5 2 2 Chenopodiaceae type 2 294 27 10 3 Compositae 1 3 5 2 27 7 Compositae type 1 105 4 Coronilla* 4 2 3 1 4 4 Crucianella 1 1 3 2 Cucurbitaceae type Bryonia 2 1 28 6 Cyperaceae 8 3 32 9 242 44 Erodium* 3 7 66 7 33 9 Erodium spiral beak* 7 1 Fabaceae type 1 932 35 203 23 3 2 Fumaria* 24 9 8 2 Galium* 67 16 195 31 2 2 Glaucium* 288 16 14 3 8 4 Heliotropium* 19 8 18 4 7 6 Hyoscyamus 31 8 19 6 insect 6 1 3 1 12 4 Lamiaceae 1 3 8 3 53 7 Malva 1 1 8 1 4 4 Medicago 1 1 Medicago pod 1 1 Medicago radiata 6 2 4 2 4 4 Nigella* 1 1 Onobrychis* 107 15 6 2 Ornitkogalum* 2 1 14 4 2 Papaver* 2 Papaveraceae 2 1 Peganum harmala 1 1 Plantago 2 1 1 0 Polygonum/Rumex 37 7 497 38 11 4 rhizome 4 2 1 1 Rubiaceae 4 0 Silene/Gypsophila* 8 1 148 19 519 40 13 8 Springer Veget Hist Archaeobot Table 4 continued Abr Dja'de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% Solanaceae 1 0 Thymelaea* 5 2 31 6 1 1 Tragopogon* 2 1 Tribulus terrestris 2 1 1 1 Trifolieae 317 23 232 30 75 19 Trigonella 7 2 7 2 3 3 Trigonella astroites 10 1 15 4 5 3 Vaccaria 1 0 2 1 Valerianella 2 1 5 4 Verbascum 1 0 2 1 5 2 Ziziphora 27 2 959 53 Indeterminate 3,014 57 2,867 72 520 45 Those marked with * are part of a proposed weed assemblage. The numbers marked in bold are taxa which have high concentrations in some samples and low ubiquity which suggests that they were gathered for a specific use Large seeded legumes (Table 5) early finds come from Tell Kerkh (Tanno and Willcox 2006b). This group is common on all sites, both in abundance and ubiquity. Fragmentation and the absence of the testa lim- Fruits of trees and shrubs (Table 6) ited the precision of identifications. A few well-preserved specimens with the testa intact were identified as Pisum Pistacia (assumed to be P. atlantica) fruits were very elatius, the progenitor of the domestic pea. The low num- abundant as they are on most early agriculture sites in the bers of this species are due to poor preservation. Many of Near East. Wood charcoal was also abundant indicating the pulses identified as Pisum/Vicia/Lathyrus may have that that P. atlantica was far more widespread at this time. been Pisum elatius. The nuts of this species are still gathered and consumed in The most abundant pulse was Lens. No species identi- Turkey and Syria today. Oil is extracted from the nuts and fication was possible but this taxon closely resembles a drink similar to coffee is still produced on a commercial L. orientalis. This species is not a common component of scale in Turkey. At Tayyibeh in central Syria farmers Near Eastern vegetation and is often restricted to small cultivate Pistacia atlantica for its nuts. In recent years it stands, making gathering an arduous task. Its abundance on has been planted in government reforestation programs in the sites, combined with the fact that the middle Euphrates the arid zones of northern Syria. This species appears to is not an ideal habitat for this species suggest that it may have been economically important during the tenth and have been cultivated. Vicia ervilia, Lathyrus and Vicia may ninth millennia b.c. have been weeds of cultivation. Cicer and Vicia faba were Fragments of Amygdalus cf. communis stones are com- new cultivars that appear for the first time at Dja'de; other mon at Tell Qaramel while at Jerf el Ahmar another taxon, Table 5 Scores for Fabaceae ., , r . r • Abr Dja'de Jerf Qaramel seeds from the four sites T Ub% T Ub% T Ub% T Ub% Pisum elatius 13 3 3 1 1 1 PJVicia/Lathyrus 37 26 1,952 68 482 55 682 63 Lathyrus 9 4 2 1 8 6 Uns 230 70 5,850 84 1,820 77 1,113 78 Vicia 19 8 Vicia ervilia 34 9 62 16 11 5 Vicia faba 2 1 Cicer 3 1 Springer Veget Hist Archaeobot Table 6 Scores for fruits and ., , ,___ , . ... Shrubs and trees edible nuts from tree and shrub taxa Abr Dja'de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% Pistacia whole fruit 23 4 4 1 64v 12 Pistacia frags 2 7 911 47 2,650 63 1,705 58 Amydalus cf communis 2,214 48 Amygdalus webbü/orientalis 1,588 56 Vitex agnus-castus 34 4 Vir« sylvestris . 4 2 Quercus acorn 3 2 /-icks can'ca 42 8 11 3 4 4 Fruit frag 3 1 11 5 2 Ce/rw 1 1 400 35 Capparis 166 11 339 22 A. orientalisAvebbii, occurs. It is not known whether these almonds were toxic. If so, they could have been consumed because the toxicity can easily be removed by roasting. The difference in almond taxa between Tell Qaramel and Jerf el Ahmar (none were found at Dja'de), the occurrence of Celtis and acorns only at Tell Qaramel, and the presence of Vitex agnus-castus only at Jerf el Ahmar would appear to reflect the differences in the plant associations found near the sites. Like other sites dated to the end of the Younger Dryas and the beginning of the Holocene, the vegetation consisted of plants which could grow in the area today if human pressure was removed. This suggests that the climate of the Younger Dryas did not greatly affect the vegetation in this part of the Near East (Bottema 1995; Roitel and Willcox 2000). Ficus carica seeds do not occur on the early Euphrates sites of Abu Hureyra and Mureybet phases 1 and 2. At Jerf el Ahmar and Qaramel they are insignificant, but are more common at Dja'de. In the southern Levant where figs were in frequent use at an earlier date, it has been proposed that they were cultivated and even domesticated during the ninth millennium cal b.c. (Kislev et al. 2006). Processed plant remains (Table 7) Charred clusters of fused seed fragments resembling seed cakes, and amorphous charred masses were commonly encountered. These appear to be the remains of prepared food and could potentially provide information about diet and food processing. At the time of writing only the complete seed cake found at Jerf el Ahmar has been studied (Willcox 2002a). The remaining samples will be studied at a later date. Charred animal droppings (Table 8) This is the first time charred rodent droppings have been observed on early farming sites. The regular occurrence of small rodent droppings, which correspond in size approximately to those of the domestic house mouse, is highly significant because it suggests that storage of grain was widely practised at these sites. Six archaeozoological finds of domestic mice (Mus musculus domesticus) were identified at Dja'de and one at Jerf el Ahmar (Cucchi et al. 2005). Other early village sites where domestic house mice have been found include, Hyonim B and Netiv Hugdud in the southern Levant, and Mureybet and Cafer Hoyiik in the north. This rodent was also introduced to Cyprus by about 8,500 cal b.c. (Cucchi et al. 2005). Recovery of small rodent remains requires special techniques, which are not always applied, so this species may have been more widely distributed than the archaeozoological finds suggest. Table 7 Scores for charred finds of processed food and flower bases Abr Dja' de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% Seed cake 11 4 9 1 Amorphous frags 56 40 116 48 119 43 43 19 Flower base 17 3 2 1 17 9 Asteraceae capitulum 33 3 1 1 Table 8 Scores for charred animal droppings found at three of the sites Charred coprolites Abr Dja'de Jerf Qaramel T Ub% T Ub% T Ub% T Ub% Ungulate 2 1 Small rodent 221 25 51 11 49 25 Archaeozoological analyses suggest that the small rodent droppings were probably, of Mus musculus domesticus, the common house mouse. Those of the ungulates are more problematical Springer Veget Hist Archaeobot These finds (including Tell Qaramel) demonstrate that the house mouse was taking advantage of stored grain at many sites in the Near East at the beginning of the Holo-cene. This suggests that we are dealing with a very common wild rodent that was pre-adapted to human habitation, which it regularly colonised. However, the mouse's successful introduction to Cyprus and other parts of the world resulted from regular large-scale movement of grain. Whether transport of grain could have helped mice to colonise areas outside their natural habitats in the Near East at an earlier date is not known. Discussion of the evidence for cultivation The location of sites beyond wild cereal habitats A good sign of pre-domestic cultivation is the presence of wild cereals outside their natural habitats. The sites of Jerf el Ahmar, Cheik Hassan, Mureybet and Abu Hureyra are almost 200 km south of current-day wild rye habitats and between 100 and 150 km south of wild einkorn habitats. Today the climate of the middle Euphrates is too hot and dry for wild rye and wild einkorn. Willcox (2005) suggests that wild rye and wild einkorn habitats were limited in the middle Euphrates by the poor chalk soils, which are not suitable for these species. Even at higher latitudes, with cooler temperatures and a higher rainfall, wild rye does not grow on these soils. The frequency of wild rye on middle Euphrates sites suggests that it was at its maximum during the cool Younger Dryas and that it diminished with the onset of global warming as habitats retreated north, following the regional climatic gradient, in response to increased temperatures. Given lower temperatures during the Younger Dryas could these cereals have grown nearer the sites? As stated in the introduction we cannot know the distances that wild stands were from the sites. The distance would have varied depending on the climatic conditions and the relative positions of the sites (Willcox 2005). Rye would have been collected from areas with more acidic soils, such as those found on basalt. The nearest basalt outcrops are located a few days' walk north-east of Dj'ade and Tell 'Abr. Given cooler climatic conditions than today, these are the most likely habitats for rye and it is possible that a few smaller stands extended further south. The southernmost Euphrates sites lay beyond, or were in a marginal position in relation to wild rye stands during the Younger Dryas. If they were far from wild stands this would have created an incentive to cultivate at an early date as suggested by Hillman (2000). Alternatively, if they were near, then over-exploitation and climate warming would have reduced availability, creating an incentive to cultivate at the beginning of the Holocene. Transport of grain over varying distances is an inevitable factor whether we are dealing with 5 km or 150. When transport of grain for consumption becomes impractical for one reason or another, then the import of seed for cultivation, of which only a tenth of the grain weight would be needed compared with that for consumption, becomes an option. At a later date seed from cultivation would be put aside for planting; in the long term this would have been the preferred option, providing a local, secure and dependable supply of grain. There is, of course, no evidence of pack animals. The rivers Euphrates and Balikh were probably not easily navigable. However, taking into account the extraordinary architectural finds at the site of Gobekli, which is included in the same cultural area as Jerf el Ahmar and Tell 'Abr, we should not underestimate the technical capabilities of the inhabitants of these sites. Wild einkorn, like wild rye, is ill adapted to the edaphic and Holocene climatic conditions of the middle Euphrates. For this reason one would expect it to decline like rye; the fact that it increases strongly suggests that it was being brought into cultivation from elsewhere. The same can be argued for the appearance of emmer. Under cultivation, these cereals would have been able to thrive in adverse climatic and edaphic conditions, because fields would have been situated in favourable microhabitats where competition had been removed (Valkoun et al. 1998; Willcox 2005). Tell Qaramel is situated in a different region with a higher rainfall. The absence of rye at this site compared to the Euphrates sites can be explained by the absence of basalt in the area and the proximity of potential einkorn habitats on decalcified soils, which form on the hard limestone soils not far from the site. We cannot argue that Tell Qaramel was beyond the area of wild cereals. A reduction in small gathered seeds from non-founder plants During the tenth millennium in the northern Levant the gathering of small seeded grasses, rye, Polygonaceae and Cyperaceae show a progressive decline (Weiss et al. 2004; Willcox 2005). Figure 5 illustrates this decrease on Euphrates sites. At Jerf el Ahmar this gradual reduction in the use of small-seeded taxa such as Polygonum/Rumex and small-seeded grasses can be seen by comparing the early and late levels (Tables 9, 10). We suggest that this is evidence that the gathering of wild plants gradually gave way to the cultivation of founder crops. Springer Veget Hist Archaeobot MureybetI&v / / ./ Dja'de ■ ^1 ■ Jerf el Ahmar n ■ ■ JerfelAhmarl H ■ H Mureybct ffl ■ ■ ■ Mureybetl,0 Abu Hureyra 2,3 ^^^^ ^^^1 ■■: Abu Hureyra 1 H| H^HH. ^^^^H' 8500-8000 9000-8500 9500-9000 9500-9000 10500-9500 11000-10500 ___|l1300-11000 Absent Present < 20% of samples Present >20% and <80 of samples Present > 80 of samples Table 9 Selected taxa from Jerf el Ahmar, which show a significant increase in frequencies in later levels ^■Present ■■1 Fig. 5 Ubiquity frequencies of non-founder small seeded taxa from four sites on the Euphrates. They become less frequent as the gathering of wild plants diminishes and cultivation of founder crops increases The gradual adoption of founder crops in the middle Euphrates Table 11 gives cereal finds from the Euphrates sites and demonstrates on a presence/absence basis the progressive increase of barley, single-grained einkorn and emmer. As we have already discussed, these changes may result from several factors, such as proximity of wild stands or changes in climate, but the overriding factor was that these founder species were being progressively brought into cultivation. Barley is absent at Abu Hureyra, but it appears at Mureybet and then increases in frequency. Wild barley, unlike wild rye, is adapted to an arid, warmer climate. It would have tended to expand at the begirining of the Holocene when it came into contact with cultivation and was adopted as a cultivar. Finally at Dja'de, the most recent site, emmer, Viciafaba and Cicer all appear for first time far from their natural habitats. Weeds of cultivation Taxa marked with * and given in Table 4 resemble an assemblage of weeds of cultivation, to which can be added some of the pulses and wild grasses. The relative ubiquity percentages are given in Fig. 6, the figures being taken from Table 4. Although these taxa can also occur in low frequencies in wild cereal stands, they proliferate under cultivation. At Bronze Age sites weed assemblages are comparable (L. Herveux, in preparation). In contrast these taxa are less frequent on Epipalaeolithic sites. Of the 16 selected taxa from Jerf el Ahmar only one occurs at Ohalo II (Kislev et al. 1992), eight at Abu Hureyra and 12 at Tell Qaramel (Table 12). This comparison suggests that weeds of cultivation increase with time. Periods: Early Late Early Late Phases: 4 + 5 1 +2 3+4 + 5 1+2 + 3 Total samples: 81 76 159 154 Total items: 12,097 10,638 17,431 15,972 T. boeoticum base T 3 1 4 T. boeoticum grain T 18 41 26 49 Ub% 11 18 10 14 H. spontaneum gr T 2,353 5,234 3,593 6,474 Ub% 52 90 93 65 H. spontaneum base T 1,546 1,622 1,670 1,746 Ub% 64 84 50 59 Ornithogalum T 1 13 1 13 Ub% 1 6 1 6 Bellevalia T 34 106 73 145 Ub% 25 47 23 34 Vicia ervilia T 10 31 24 45 Ub% 20 28 12 21 Two combinations of phasing are given with absolute totals (T) and ubiquity (Ub%) Increase in grain size An increase in grain size is often cited as a sign of domestication (Hillman et al. 2001). A study of barley grain size from several sites including Jerf el Ahmar and Dja'de demonstrated that there is a small increase in thickness and breadth between the lower and upper levels at Jerf el Ahmar but that in the same geographical area there is no further increase for several millennia (Willcox 2004). It is highly improbable that the small increase represents selection within a population for larger grains, particularly when spikelet bases are all of wild type. It is more likely that it represents either the adoption of a plump-grained population from elsewhere, probably the case for single-grained einkorn, or that favourable growing conditions under cultivation produced a higher proportion of fully developed grains than would occur in a wild population. There is no evidence that grain length increased. This change in size may thus be the result of increased reliance on cultivation. Springer Veget Hist Archaeobot Table 10 Selected taxa from Jerf el Ahmar, which show a significant decrease in frequencies in later levels Periods: Early Late Early Late Phases: 4 + 5 1 +2 3 + 4 + 5 1+2 + 3 Total samples: 81 76 159 154 Total items: 12,097 10,638 17,431 15,972 Triticum/Secale gr T 1,382 396 2,041 1,055 Ub% 92 81 83 77 Secale base T 121 18 125 22 Ub% 23 10 13 6 Polygonum/Rumex T 359 73 412 126 Ub% 57 34 44 32 Panicoid T 19 0 20 1 H. murinum/bulbosum T 614 116 724 226 Ub% 66 44 49 38 Tiziphora T 26 0 27 1 Two combinations of phasing are given with absolute totals (7) and ubiquity (Ub%) Cultivation without domestication There is no evidence at these four sites for plant domestication seen in the loss of dispersal mechanisms for either cereals or pulses. Theoretically the domestication process could have been rapid (Hillman and Davies 1990). Why then, if we propose cultivation at these sites, did domestication not appear earlier? Seed stock may have been regularly replenished from wild stands to counter poor Probable weed assemblage Fig. 6 Each bar compares the ubiquity percentages between sites for each potential weed taxon. For the ubiquity values see Table 4. We interpret the overall pattern as being suggestive of an assemblage of weeds of cultivation harvests, which may have been frequent in the arid middle Euphrates. In addition, if harvests occurred early in the season before the ears shattered, then the probability of selection for the rare mutants that had lost their dispersal mechanism would be extremely slim. Finally, recent evidence from a study of spikelet bases from a number of sites suggests that domestication was slow to become established (Tanno and Willcox 2006a). Conclusions Gordon Hillman suggested that the inhabitants of Abu Hureyra were cultivators by about 11,000 cal b.c. Sue Colledge (1998) identified a weed flora at Mureybet. At Jerf el Ahmar and Dja'de we have demonstrated five lines of evidence which suggest cultivation of wild cereals and Table 11 Finds of wild cereals from Euphrates sites Abu Hureyra 1 Mureybet I & 2 Mureybet HJ Tell' Abr Jerf el Ahmar Göbekli Dja'de Nevali Triticum/Secale >5,000 19 >500 2,999 2,539 5 1,120 ? Triticum base ? ? ? ? 5 ? 16 >3,000 Secale sp. Base ? P P ? 145 ? 16 0 H. spontaneum gr 0 5 164 190 9,639 16 3,763 89 H. spontaneum base 0 0 6 ? 3,325 ? 153 P Einkorn 1 gr 0 0 0 90 67 ? 302 661 T. dicoccoides 0 0 0 0 0 9 192 129 Sediment (1) >10,000 ca. 310 ca. 310 4,520 16,600 ? 6,122 ? Number of samples 21 31 31 30 430 1 229 267 Approx. Date 11,000 cal b.c. 9,250 cal b.c. 8,500 cal b.c. 0 = absence significant. ? = absence not significant. Presence/absence demonstrates that new founder crops were brought into cultivation at different times between 11,000 and 8500 b.c Factors effecting why cereals were brought into cultivation at different times may include climate change, varying proximity to wild stands, and most importantly the adoption of cultivation Springer Veget Hist Archaeobot Table 12 Presence/absence of a selected weed assemblage (first identified at Jerf el Ahmar) showing that while these taxa are almost absent from the Kebaran site, the Natufian, Kiamian and PPNA sites have progressively more weeds of cultivation which may reflect the increasing adoption of cultivation at these sites Jerf Qaramel Abu Hureyra Ohaloll Adonis P P A A Alyssum/Lepidium P A P A Bellevalia P P P A Centaurea P A A A Coronilla P P A A Erodium P A P P Fumaria P A A A Galium P P A P Glaucium P P P A Heliotropium P P P A Onobrychis P A P A Silene / Gypsophilia P P P A Thymelaea P P A A Trifoliae P P P A Trigonella P P A A pulses. Tell 'Abr has less information but falls into the same pattern as the Euphrates sites. Tell Qaramel has fewer signs of cultivation. This site occurs in a moister area where wild einkorn and barley could have grown naturally. The similar quantities of charred rodent droppings compared to Jerf el Ahmar and Dja'de are evidence for the intensity of cereal use, including storage. There is no evidence that new cultivars were introduced, but the high frequencies of lentils and the potential weed assemblage suggest cultivation despite the probability that wild einkorn would have been readily available in the area. On a more general level, gathering and cultivation of wild cereals probably occurred simultaneously over a long period, which is why we see no sharp division from one economy to another, the transition being extremely gradual. It is possible that with sedentary life, societies, which collected and stored seeds from annual plants practised small-scale cultivation from an early date, but this would leave little evidence. The incentive to cultivate could have been due to multiple factors, but the overriding factors would be difficulty in finding wild stands or lack of sufficient quantities. Wild lentils, which are ubiquitous on early sites, may have been cultivated at an early date because of their rarity. Even when cereals and pulses were regularly cultivated, seed may have been obtained from wild stands for long periods, which would explain why morphological evidence of domestication was slow to appear, only becoming established when intensive cultivation was practised exclusively and on a large scale. Acknowledgments Our thanks to archaeologists E. Coqueugniot (CNRS Lyon), R. Mazurowski (Warsaw University), D. Stordeur (CNRS Jales) and T. Yartah (Antiquities Department, Damascus, Syria) for helping with sampling in the field and entrusting the plant remains to us. We also thank F. Barbot who helped with the sorting. This project was partially funded by a European project (ICA3-CT-2002-10022). References Bottema S (1995) The Younger Dry as in the eastern Mediterranean. Quaternary Sci Rev 14:883-891 Colledge S (1998) Identifying pre-domestication cultivation using multivariate analysis. In: Damania AB, Valkoun J, Willcox G, Qualset C (eds) The origins of agriculture and crop domestication. ICARDA, Aleppo (Syria), pp 121-131 Coqueugniot E (2000) Dja'de (Syrie), un village a la veille de la domestication (seconde moitie du 9e millenaire av. J.C.). In: Guilaine J (ed) Les premiers paysans du monde, naissance des agricultures. Paris, Errance, pp 63-79 Cucchi T, Vigne JD, Auffray JC (2005) First occurrence of the house mouse (Mus Musculus Domesticus, Schwarz and Schwarz 1943) in the Western Mediterranean : a zooarchaeological revision of subfossil occurrences. Biol J Linnaean Soc 84:429-445 de Moulins D (1997) Agricultural changes at Euphrates and steppe sites in the mid-8th to the 6th millennium B.C. BAR International Series 683, Oxford Hillman G (2000) Plant food economy of Abu Hureyra. In: Moore A, HiUman G, Legge T (eds) Village on the Euphrates, from foraging to farming at Abu Hureyra. Oxford University Press, New York, pp 372-392 Hillman G, Davies S (1990) Measured domestication rates in wild wheats and barley under primitive cultivation, and their archaeological implications. J World Prehistory 4:157-219 Hillman GC, Hedges R, Moore A, Colledge S, Pettitt P (2001) New evidence of late glacial cereal cultivation at Abu Hureyra on the Euphrates. Holocene 11:383-393 Kislev M (1992) Agriculture in the near east in the 7th millennium B.C. In: Anderson P (ed) Prehistoire de l'agriculture. Paris: Monographie du CRA n6, Editions CNRS, pp 87-94 Kislev M, Hartmann A, Bar-Yosef O (2006) Early domesticated fig in the Jordan valley. Science 312:1372-1374 Kislev ME, Nadel D, Carmi I (1992) Epipalaeolithic (19,000 B.P.) cereal and fruit diet at Ohalo II, Sea of Galilee, Israel. Rev Palaeobot Palynol 73:161-166 Mazurowski R (2004) Tell Qaramel excavations, 2003. Polish Archaeol Mediterr 15:355-370 Neef R (2003) Overlooking the steppe forest: preliminary report on the botanical remains from early Neolithic Gobekli Tepe (southern Turkey). Neolithics 2/03:13-15 Nesbitt M (2002) When and where did domesticated cereals first occur in southwest Asia? In: Cappers RTJ, Bottema S (eds) The dawn of farming in the Near East. Studies in Near Eastern production, subsistence and environments, ex oriente, Berlin, pp 113-132 Nesbitt M (2006) Identification guide for Near Eastern grass seeds. Institute of Archaeology, UCL, London Pasternak R (1998) Investigations of botanical remains from Nevali Cori PPNB, Turkey. In: Damania A, Valkoun J, Willcox G, Qualset C (eds) The origins of agriculture and crop domestication. ICARDA, Aleppo (Syria), pp 170-177 Roitel V, Willcox G (2000) Analysis of charcoal from Abu Hureyra. In: Moore AMT, Hillman G, Legge T (eds) A village on the Euphrates. Oxford University Press, Oxford, pp 544-547 Springer Veget Hist Archaeobot Savard M, Nesbitt M, Jones MK (2006) The role of wild grasses in subsistence and sedentism: new evidence from the northern fertile crescent. World Archaeol 38(2): 179-196 Stordeur D (2000) New discoveries in architecture and symbolism at Jerf el Ahmar (1997-1999 Syria). Neolithics 1:1-4 Stordeur D, Brenet M, Der Aprahmian G, Roux J-C (2000) Les batiments communautaires de Jerf el Ahmar et Mureybet. Horizon PPNA. Syrie Paleorient 26:29-44 Tanno K, Willcox G (2006a) How fast was wild wheat domesticated? Science 311:1886 Tanno K, Willcox G (2006b) The origins of cultivation of Cicer arietinum L. and Viciafaba L.: Early finds from north west Syria (Tell el-Kerkh, late 10th millennium BJP.). Veget Hist Archaeobot 15:197-204 Valkoun J, Giles J, Waines J, Konopka J (1998) Current distribution and habitat of wild wheats and barley. In: Damania A, Valkoun J, Willcox G, Qualset C (eds) The origins of agriculture and crop domestication. ICARDA, Aleppo (Syria), pp 293-299 van Zeist W, Bakker-Heeres JAH (1984[1986J) Archaeobotanical studies in the Levant 3. Late-Palaeolithic Mureybet Palaeohis-toria 26:171-199 van Zeist W, Roller GJ de (1991/1992[1994]). The plant husbandry of Aceramic Cayonu, SE Turkey. Palaeohistoria 33/34:65-96 Weiss E, Kislev M, Hartmann A (2006) Autonomous cultivation before domestication. Science 312:1608-1610 Weiss E, Wetterstrom W, Nadel D, Bar-Yosef O (2004) The broad spectrum revisited: evidence from plant remains. PNAS 101:9551-9555 Willcox G (2002) Charred plant remains from a 10th Millennium B.P. kitchen at Jerf el Ahmar (Syria). Veg Hist Archaeobot 11:55-60 Willcox G (2004) Measuring grain size and identifying Near Eastern cereal domestication: evidence from the Euphrates valley. J Archaeol Sci 31:145-150 Willcox G (2005) The distribution, natural habitats and availability of wild cereals in relation to their domestication in the Near East: multiple events, multiple centres. Veg Hist Archaeobot 14:534— 541 Willcox G, Fornite S (1999) Impressions of wild cereal chaff in pise from the tenth millennium at Jerf el Ahmar and Mureybet: northern Syria. Veg Hist Archaeobot 8:21-24 Yartah T (2004) Tell 'Abr 3, un village du Neolithique preceramique (PPNA) sur le Moyen-Euphrate. Premiere approche. Paleorient 30:141-158 Yartah T (2005) Les batiments communautaires de Tell 'Abr 3 (PPNA, Syrie). Neolithics 1/05:3-9 Springer