«Abstract Sedentism is usually regarded as a pre-condition for the development of crop husbandry in Southwest Asia and, consequently, sedentary ...»
The presence of canids, brown bear, cape hare, fallow deer, stone marten, wild cat, beaver and European hedgehog has also been noted. Turtle remains were the main nonmammalian remains found, followed by bird, ﬁsh and lizard. Rosenberg et al. (1998: 32–3) suggested the possibility of early pig husbandry, along with the hunting of wild boar.
Though more analyses need to be conducted, a recent study found no conclusive evidence of pig domestication (Starkovich 2005: 34).
material and comparisons with that of Hallan Cemi show a signiﬁcant decrease in the use ¸ of obsidian at Demirkoy, and an increase in the intensity of the use of individual obsidian ¨ pieces. Obsidian may have become scarcer when Demirkoy was occupied, indicating a ¨ possible collapse in the obsidian trade (Peasnall and Rosenberg 2001: 382–4).
The artefact assemblage includes stone bowls and sculpted pestles similar to those found at Hallan Cemi. Clay was used to make ceramic ﬁgurines and vessels. The presence of a ¸ hole along the rim of a clay vessel, similar to those found on the stone bowls, suggests that it was deliberately ﬁred, as these holes, probably used for suspension, would not be functional on unﬁred vessels (Rosenberg and Peasnall 1998: 199).
Qermez Dere Qermez Dere is located in Iraq near the town of Tell Afar, about 50km from the Jebel Sinjar range, overlooking the Jezirah plain and a deeply cut wadi (Watkins and Baird 1987; Watkins et al. 1989). It lies within the present-day potential vegetation zone of the moist steppe. The site had already been damaged by the construction of a road, a pipeline and a communication cable trench, and was further endangered by a modern housing development when rescue excavations were undertaken by Watkins and his team. Initial soundings were conducted in 1986 and excavation took place in 1987, 1989 and 1990.
Three stages were identiﬁed, based in large part on the lithic assemblage composed almost exclusively of ﬂint (Watkins et al. 1991): 1) a ﬁrst Epipalaeolithic-Neolithic transitional stage characterized by an assemblage comprising both microliths and Kiam points; 2) an early aceramic Neolithic stage with no residual Epipalaeolithic traits; 3) a later stage characterized by Nemrik points progressively increasing in number, and by truncated structural remains and their deliberate ﬁll. The ground stone tools and equipment include mortars and crude pestles made of local limestone, and querns and rubbers made from imported basalt (Watkins and Baird 1987: 10).
The occupation seems to have been continuous from c. 10,100 to 9700 BP (uncal.) (Watkins 1995). Seven subterranean buildings have been found; all, except one small stone structure, were made of clay. The structures seem to have been deliberately destroyed after use and back-ﬁlled. A sequence of three buildings was identiﬁed, and human cranial remains from both adults and children were found in the lower ﬁll of the latest building of the sequence.
Remains of gazelle, a small canid, sheep/goat and hare dominate the bone assemblage. Wild cat, badger, pole cat, a variety of bird species, reptiles and amphibians are also represented, along with a very small number of aurochs, onager and boar remains (Watkins et al. 1991).
M’lefaat M’lefaat is located in northern Iraq, about 35km east of Mosul, within the present-day potential vegetation zone of the moist steppe. The small tell is ﬂanked by two small valleys, nowadays dry, and was ﬁrst identiﬁed and sounded by Braidwood in 1954. The site had already been damaged by a Second World War concrete machine-gun emplacement and by a tank trench (Dittemore 1983: 671). A rescue excavation was conducted in 1984 during the construction of the Mosul-Erbil road. Kozłowski led two seasons of ﬁeldwork in 1989 186 Manon Savard et al.
and 1990 on the remaining part of the tell and made a synthesis of all the results available (Kozłowski 1998; Kozłowski et al. 1991).
The village dates to the beginning of the tenth millennium BP (uncal.). Eleven round-tooval buildings, mostly subterranean, were identiﬁed; they belong to two diﬀerent phases of occupation, not far apart in time, and some of these structures are superimposed. The buildings are organized around a central courtyard made of two or three superimposed clay surfaces (Kozłowski 1998: 187).
A local ﬂint was the main material used for the chipped stone industry, as well as a small quantity of obsidian (Kozłowski 1998: 196). The assemblage resembles those of the nearby sites of Nemrik, Qermez Dere and Jarmo but does not include Nemrik-type points (Kozłowski 1998: 197).
The ground stone assemblage includes grinders and querns made from sandstone, mortars made from calcareous mudstone and from limestone, and mortar pounders made from pebbles of various rocks, as well as celts, adzes, hammerstones, mace heads, bowl vessels, whetstones, choppers, ornaments, various pieces described as plates and trays, etc.
(Mazurowski, 1998: 200–8).
Small ruminants, mainly gazelle, dominate the hand-collected mammal assemblage (Lasota-Moskalewska 1998: 215). Bones from wild goat, wild goat/sheep and wild cattle (Bos primigenius) were also identiﬁed, along with remains of hare, badger, wild boar and rodents. Carnivore remains include fox and wild cat. The bone assemblage also comprises
remains of birds, ﬁsh, tortoise, freshwater crab and molluscs (Lasota-Moskalewska 1998:
216; Rielly 1998: 222–3). Isolated human bones have been found in both levels.
(particularly wild lettuce (Lactuca sp.)). Some taxa that are not proportionally well represented show high ubiquity ﬁgures: almonds (Amygdalus sp.) and terebinth nut (Pistacia cf. khinjuk/atlantica var. kurdica) remains each represent about 1 per cent of the assemblage in proportions but were found in more than half of the samples; Ziziphora tenuior type, Taeniatherum caput-medusae and other indeterminate small-seeded grasses were also relatively frequent, though they were not abundant in terms of proportion (Fig. 3).
The Demirkoy assemblage is largely dominated by sea club-rush in terms of both ¨ proportions and ubiquity (Figs 2 and 3). Other well-represented taxa include large-seeded pulses (Vicieae and Lathyrus/Vicia) and small-seeded legumes (Trifolieae/Astragalus).
Although they were of minor importance in terms of proportion, unidentiﬁed small-seeded grasses and small-seeded barleys (Hordeum murinum complex) are nonetheless present in more than half of the samples (Fig. 3). As at Hallan Cemi, nuts are not well represented in ¸ terms of proportions, but the remains of almonds and terebinth nuts were found in many samples, along with those of barley (Hordeum cf. spontaneum), dock/knotgrass and crucifer (Alyssum/Lepidium type) (Fig. 3). However, with only twelve samples analysed at Demirkoy, the ubiquity ﬁgures must be considered with caution.
¨ In terms of proportion and ubiquity, the Qermez Dere assemblage is dominated by unidentiﬁed small-seeded grasses and large-seeded pulses (Vicieae, Lathyrus/Vicia and Lens cf. orientalis) (Figs 2 and 3). Other proportionally important taxa include small-seeded legumes (Trifolieae/Astragalus), Boraginaceae (Arnebia/Lithospermum and Lithospermum cf. tenuiﬂorum), barleys (Hordeum cf. spontaneum and Hordeum murinum complex) and einkorn/rye (Triticum boeoticum/Secale). Except for Boraginaceae, the same taxa dominate in terms of ubiquity (Fig. 3).
188 Manon Savard et al.
Large-seeded legumes (Vicieae, Lathyrus/Vicia, Vicia ervilia and Lens cf. orientalis) dominate the M’lefaat assemblage in terms of proportion (Fig. 2). They are followed by unidentiﬁed small-seeded grasses, goat-grass (Aegilops cylindrica/tauschii/speltoides) and barley (Hordeum cf. spontaneum) (Savard et al. 2003).
Except for the Boraginaceae, the commonest plants of the four assemblages were most likely deliberately gathered as food. In addition to being well represented in terms of proportions and/or ubiquity, there is much archaeological and ethnographical evidence that suggests these plants have been used as food (e.g. Ertu-Yaras¸ 1997; Hillman 2000;
g Zohary and Hopf 2000). There is also good evidence that the bulk of the sea club-rush seed assemblage and, possibly, that of dock/knotgrass2 were introduced as food rather than with construction material: no building structures were found in the central area of Hallan Cemi, while sea club-rush and dock/knotgrass were an important part of its charred seed ¸ and fruit assemblage; in most samples, they did not occur alone but with other food plants.
Ethnographic records and experiments at Abu Hureyra showed that club-rush and knotgrass seeds can be roasted before being ground to ﬂour to produce mush or griddle cakes (Hillman 2000: 354–8). Roasting would have favoured their preservation. As for Boraginaceae, they are quite commonly found in steppe vegetation (van Zeist and BakkerHeeres 1982: 211). Their relative robustness and their high silica content have favoured their preservation. They were most likely introduced incidentally and the fact that preservation of charred remains was relatively poor at Qermez Dere might explain why they were proportionally well represented.
Food plants that are important both in terms of proportions and ubiquity were probably among the staple foods. Among the various food plants represented at each site, a limited number clearly dominate and may have been stored. Indeed, at Hallan Cemi, ¸ evidence of year-round occupation, along with high ubiquity ﬁgures for taxa available seasonally, suggests storage practices. Storage may have taken place in above-ground structures with poor archaeological visibility.
Results at other sites
Table 1 and Figure 4 present archaeobotanical results at Palaeolithic, Epipalaeolithic and PPNA sites. PPNA sites are included because they are either without evidence of domestication or with only questionable evidence (Nesbitt 2002). They are therefore treated as sites with economies based on foraging, or foraging with some cultivation of wild plants.
It is striking that large-seeded grasses (‘wild cereals’) are dominant (74 per cent of seed remains) at only one site: Mureybit on the Euphrates. However, at Mureybit, grasses, particularly einkorn, become an important part of the assemblage only in phase III; before that, they seem to have played a minor role. For the earliest phases, knotgrass, Boraginaceae, asparagus and small-seeded legume remains dominate the archaeobotanical assemblage (van Zeist and Bakker-Heeres 1984). Large-seeded grasses form between 11 and 16 per cent of the seed remains at M’lefaat and Qermez Dere in northern Iraq, Netiv Hagdud in Israel and Tell Aswad in Syria. Allowing for the fact that large-seeded grasses have robust grains that char well in ﬁres, and that archaeobotanists devote extra eﬀort to identifying them, these frequencies appear low for a staple food resource. Small-grained grasses form between 190 Manon Savard et al.
Figure 4 Proportions of seed remains from Palaeolithic to PPNA sites in Southwest Asia.
17 and 29 per cent of the seeds at Ohalo II, M’lefaat and Qermez Dere, and outnumber large-seeded grasses at eight of the twelve sites for which we have ﬁgures. Even allowing for the greater seed numbers of small-seeded grasses, this supports Weiss et al.’s (2004b) view that small-seeded grasses were an important food resource.
The assemblages of M’lefaat, Kebara Cave and, to a lesser extent, Qermez Dere are dominated by large-seeded legumes (respectively 52.4, 57.8 and 30.1 per cent) (Table 1 and Fig. 4). Small-seeded legumes dominate the assemblage of Tell Aswad (42.1 per cent).
Hallan Cemi and Demirkoy are both dominated by sea club-rush (respectively 32.1 and 70 ¸ ¨ per cent) (along with dock/knotgrass at Hallan Cemi (27.3 per cent)). Of the twelve sites ¸ with seed counts available, strong dominance by a single seed type occurs at M’lefaat, Kebara Cava, Mureybit, and Demirkoy (Figs 2 and 4). The other sites are highly diverse in ¨ their seed remains.
Small-seeded grasses are often an important component of archaeobotanical assemblages; their exploitation seems to be a signiﬁcant part of hunter-gatherer subsistence strategies in the Levant. The remains found at Ohalo II suggest this has been the case since at least 19,500 (uncal.) BP (Kislev et al. 2002; Weiss et al. 2004a). However, contrary to Weiss et al. (2004b: 9553–4), we ﬁnd that the use of small-seeded grasses does not fall sharply after the Epipalaeolithic period. Small-seeded grasses are more abundant than at Ohalo II (17.8 per cent) at the PPNA sites of M’lefaat (24.6 per cent) and Qermez Dere (28.9 per cent), and are important at Abu Hureyra and Iraq ed-Dubb (9.8 per cent).
Overall, sites show much diversity in both time and space. Grasses appear not to have had the same importance everywhere: Hallan Cemi and Demirkoy mark one extreme in ¸ ¨ the low importance of grass exploitation in hunter-gatherer subsistence strategies. The relatively high ubiquity ﬁgures for some of the grasses (including barleys) within the Hallan Cemi and Demirkoy assemblages indicate that they might have been relatively ¸ ¨ common in the vicinity of the site, and frequently gathered and/or maybe even stored, but their extremely low proportions suggest that they were not staple resources. Instead, valley-bottom plants were undoubtedly among the staple foods.
Given the paucity of large-seeded grasses at pre-agrarian sites, it is all the more striking that the Neolithic ‘founder crops’ were to include four cereals: einkorn and emmer wheat, barley and rye. The selection of these, and a small number of large-seeded legumes, is most plausibly explained by their morphological and ecological suitability for cultivation (cf.
Bar-Yosef and Kislev 1989: 636–40), rather than their abundance in pre-agrarian diet.