«Abstract Sedentism is usually regarded as a pre-condition for the development of crop husbandry in Southwest Asia and, consequently, sedentary ...»
The role of wild grasses in subsistence
and sedentism: new evidence
from the northern Fertile Crescent
Manon Savard, Mark Nesbitt and Martin K. Jones
Sedentism is usually regarded as a pre-condition for the development of crop husbandry in
Southwest Asia and, consequently, sedentary pre-agrarian sites are an important focus of research
on the origins of agriculture. It is often assumed that wild grasses were as important for hunter-
gatherers as domesticated cereals were for early farmers, and that wild grass exploitation may therefore have had a critical role in enabling sedentism. Results from the analysis of archaeo- botanical assemblages from Hallan Cemi, Demirkoy, Qermez Dere and M’lefaat, and comparison ¸ ¨ with those of other sedentary pre-agrarian sites in Southwest Asia, challenge the role often attributed to the exploitation of grasses at this time. Archaeobotanical and ethnographical evidence instead suggests that hunter-gatherers took an opportunistic approach to the resources available and their subsistence strategies were not necessarily centred on grasses and ‘wild cereals’.
Keywords Neolithic; Epipalaeolithic; Natuﬁan; domestication; foraging; hunter-gatherers.
Sedentism, grasses and the origins of agriculture Sedentism is widely seen as an essential precursor to the earliest agriculture in the Fertile Crescent of Southwest Asia. Three perspectives support this view: ﬁrst, there is extensive archaeological and bioarchaeological evidence for year-round occupation of many settle- ments in the Epipalaeolithic period in the form of solidly built architecture, abundant immovable goods, such as ground stone, and evidence from bones, seeds and molluscs of harvesting of food resources at diﬀerent times of the year. The second perspective is based on ethnographic evidence that links sedentism to increasing population size. As many explanations for the development of food production invoke increased population density during the Epipalaeolithic, there is a natural tendency to seek evidence for sedentism during this period. Third, common sense suggests that once hunter-gatherers started any World Archaeology Vol. 38(2): 179–196 Sedentism in Non-Agricultural Societies ª 2006 Taylor & Francis ISSN 0043-8243 print/1470-1375 online DOI: 10.1080/00438240600689016 180 Manon Savard et al.
form of cultivation of wild plants, the demands of cultivation and of crop storage would favour sedentism.
Both ethnographic and archaeological evidence show that sedentism in hunter-gatherer societies depends on access to abundant food resources, whether in the form of acorns or salmon in ethnographically documented societies in western North America (Keeley 1988) or archaeological examples such as the ﬁsh-dependent society of Mesolithic Lepenski Vir (Chapman 1996). Since the 1960s, grasses (particularly the wild progenitors of cereals) have been identiﬁed as the abundant wild food resource that enabled widespread sedentism in the Epipalaeolithic of Southwest Asia. Although archaeological evidence for Epipalaeolithic diet has, until recently, been scanty, the important (if not the dominant) role of domesticated cereals in early farming sites has often led to the assumption that their wild progenitors had the same importance to hunter-gatherers. Grasses are indeed an important component of several Epipalaeolithic/early Neolithic archaeobotanical assemblages (Table 1) and are useful to archaeobotanists because they can, unlike legumes, show clear evidence of domestication. As a result, a substantial body of research has focussed on wild grasses (e.g. Anderson 1999; Colledge et al. 2004; Harlan 1999; Kislev et al. 2004; Nesbitt 2002; Willcox 1999a, 1999b, 2004). However, the importance of grasses has slipped into the theoretical domain and may have unduly inﬂuenced concepts and vocabulary: for example, the practice of referring to the wild progenitors of cereals as ‘wild cereals’.
The focus on wild grasses in pre-agrarian subsistence can be traced back to the 1950s excavations at Eynan-Mallaha (Perrot 1966). The presence of storage pits at Natuﬁan Eynan and the apparent concentration of Natuﬁan sites in areas with dense stands of grasses (including the wild progenitors of cereals) led to the Natuﬁan being considered ‘not only as the ﬁrst sedentary communities in the Levant, but as ‘‘harvesters of cereals’’’ (Cauvin 2000: 15).
However, Cauvin also writes: ‘we concluded too rapidly, in the absence of any botanical studies, that they [the Natuﬁans] were specialized gatherers of cereals who were somehow preparing, through this choice, the future cultivation of these plants’ (1999: 180). Nonetheless, the role of ‘wild cereals’ remained deeply rooted in the theoretical domain. For instance, Henry (1989: 19,
35) states that the most important change in subsistence strategies during the Natuﬁan was the intense harvesting of ‘wild cereals’ and nuts, which both allowed and demanded sedentism.
Other work incorporates a similar approach: see, for example, Flannery (1969: 80–1), McCorriston and Hole (1991), Bar-Yosef and Meadow (1995: 69) and Smith (1998: 70).
In this paper, the role of grasses in sedentary hunter-gatherer subsistence strategies is reexamined, along with its implications for theoretical models of sedentism and the origins of agriculture. New evidence from four broadly contemporary sites from the northern Fertile Crescent (Hallan Cemi, Demirkoy, M’lefaat and Qermez Dere) is presented and ¸ ¨ compared to archaeobotanical results from other Epipalaeolithic, Natuﬁan or early Neolithic sites1 that display some evidence for sedentism, most of them from the Levantine and the Euphrates corridors (Fig. 1).
Some key assumptions and questions
It is not the purpose of this paper to re-evaluate the evidence for sedentism in pre-agrarian Southwest Asia. We agree that most round-house villages of the Epipalaeolithic, Natuﬁan Table 1 Proportions of seed remains from Palaeolithic to PPNA sites in Southwest Asia
Absolute numbers of seeds are given, but where these data are not available, subjective rankings are given on the following scale: p – presence, unknown abundance;
þ infrequent; þþ frequent; þþþ dominant. Some sites from which few seeds have been published are omitted from this table: Hayonim Cave and Terrace, Jericho, Mallaha and Nemrik 9.
Notes on categories: in this table, large-seeded legumes are the Cicer, Lathyrus, Lens, Pisum and Vicia genera, and seeds scored as Vicieae or large/medium unidentiﬁed legumes. Large-seeded grasses are Aegilops, Avena, Hordeum spontaneum, Secale and Triticum, and those scored as unidentiﬁed large-seeded grasses. Nuts are Amygdalus, Pistacia and Quercus. Numbers are for seeds broadly deﬁned (i.e. reproductive propagules), and include nutshell but not legume pod or grass chaﬀ remains.
References: Abu Hureyra: Hillman (2000); Cayonu: van Zeist and de Roller (1991–2, 2003); Iraq ed-Dubb: Colledge (2001); Jerf al Ahmar: Willcox (1996); Kebara Cave: Lev ¸ ¨¨ et al. (2005); Mureybit: van Zeist and Bakker-Heeres (1984); Netiv Hagdud: Kislev (1997); Ohalo II: Weiss et al. (2004a, 2004b); Tell Aswad: van Zeist and Bakker-Heeres The role of wild grasses
Figure 1 Location map (adapted from Aurenche and Kozłowski 1999).
(‘base camps’) or PPNA were probably occupied year-round (though only solid bioarchaeological evidence can attest it) and that their inhabitants depended on food resources obtainable within a few days’ walk of the settlement (c.f. Belfer-Cohen and BarYosef 2000). It is therefore reasonable to assume that the plant and animal remains found at these sites will go some way towards answering the question of how sedentary huntergatherers were able to support a sedentary lifestyle. Was the plant element of diet dependent on the harvesting of wild grasses, or on another dominant resource, or on collection of a wide range of plant foods?
Our second key assumption is that the bulk of the plant remains found at these sites are in some way representative of the seeds consumed. Some plants may have been incidentally introduced to the site, but it is expected that they would represent only a minor portion of the assemblage. Other alternatives have been considered, such as dung burning, and the use of plants as construction material. It is unlikely that dung was an important source of fuel on sites where animals were hunted. Nonetheless, this possibility was closely examined for the four sites we studied: the absence of chaﬀ (except at M’lefaat where the robust chaﬀ of a single taxon, goat-grass, was found), the low seed-charcoal ratio, and the presence of various riparian tree taxa among the charcoals identiﬁed suggest that, if used at all, dung was not an important source of fuel (Savard 2005: 250–3). Similar conclusions were drawn for Abu Hureyra I (Hillman et al. 1997). There is indeed a possibility that some plants could have been used as ﬁbre, for bedding, thatching, etc.
Though it is not excluded that some plants might have had multiple purposes, evidence (presented later in this paper) suggests that the bulk of the seed assemblage is not derived from the use of plants as ﬁbre.
183 The role of wild grasses Interpretation of plant remains is complicated by diﬀerential survival, whereby fragile material, such as leaves or nut cotyledons, are under-represented, by processing practices that result in plant remains being generated oﬀ-site and by cooking practices. Abundance as an indicator of importance is also obscured by the varying size and frequency of seeds produced by diﬀerent species. It is obviously inappropriate to take seed abundance as a direct index of use; at the same time, there is a great deal of variation between the proportions of seeds at hunter-gatherer sites, which requires explanation. There is little evidence of context-related variation in seed composition at the study sites: at Epipalaeolithic and early (pre-Pottery) Neolithic sites, seeds are most often present at low concentration, with relatively little variation between contexts, which suggests that seed material has been subject to secondary deposition and homogenization. Amalgamation of results from individual samples for comparison between sites is (although a crude tool) therefore justiﬁed.
New archaeobotanical evidence from the northern Fertile Crescent
Hallan Cemi, Demirkoy, Qermez Dere and M’lefaat (Fig. 1) belong to the Taurus-Zagros ¸ ¨ Round House Horizon, as deﬁned by Peasnall (2000) and were occupied by hunter-gatherers.
Two of these sites oﬀer convincing bioarchaeological evidence for sedentism: plant and bone remains and, especially, the growth bands on clam shells (Unido tigridus) suggest that Hallan Cemi was occupied year-round (Rosenberg et al. 1998: 34). The M’lefaat bone assemblage ¸ includes the remains of birds that are present in Mesopotamia in winter only (Dobrowolski 1998: 225–7). Except perhaps at Hallan Cemi, where circular clay platforms were interpreted ¸ as possible silo bases (Rosenberg and Redding 2000: 47), no storage facilities were found.
The four sites are located in a region for which few archaeobotanical analyses have been undertaken. Most of the other Epipalaeolithic/PPNA sites for which plant remains have been published are located within the present-day potential vegetation zone of the steppe.
In contrast, Hallan Cemi, Demirkoy, Qermez Dere and M’lefaat have the advantage of ¸ ¨ being located along an ecological transect running from the potential vegetation zone of the steppe to that of an open oak forest.
Hallan Cemi ¸ Hallan Cemi is located in south-east Turkey, 40km north of the city of Batman, on the ¸ west bank of the Sason Cayi, a tributary of the Batman river, itself a tributary of the ¸ Tigris. It lies within the present-day potential vegetation zone of an open oak forest. It was discovered in 1990, during a survey that aimed at identifying sites threatened by the construction of dams (Rosenberg and Togul 1991). A rescue excavation was carried out from 1991 to 1994 by Rosenberg and his team.
Hallan Cemi is considered to be the oldest fully settled village site known so far in ¸ eastern Anatolia (Rosenberg and Redding 2000). At least four building levels were identiﬁed. The lower part of the semi-subterranean round building structures were made of stones, but a large quantity of burnt clay fragments with impressions of wood sticks or reed bundles suggests that the higher parts were made of wattle and daub (Rosenberg and Davis 1992: 3). The building structures were organized around an open central area over 184 Manon Savard et al.
15m in diameter, containing ﬁre-cracked stones and a high density of animal bones.
Though rubbish dumping is not excluded, the zooarchaeological evidence and the presence of dark lenses suggest that the central area might be associated with primary deposition remains of meat preparation (Rosenberg pers. comm.). A series of nineteen new AMS dates on carbonized seeds suggest a relatively short occupation during the ﬁrst half of the tenth millennium BP (uncal.) (ORAU in press).
The chipped stone industry is made from obsidian, ﬂint and chert and has strong typological links with that of Zawi Chemi Shanidar and other late Zarzian sites (Rosenberg 1999: 29). Projectile points were rare, but most could be classiﬁed as variants of the Nemrik point (Rosenberg 1994: 129).
An impressive quantity of querns, mortars, handstones and pestles was found at Hallan Cemi. Sandstone was most commonly used for both mobile and stationary ground tools, ¸ along with limestone and metamorphic rocks (Rosenberg et al. 1995: 56). The querns belong to both trough and basin types, and some reach up to 50cm in length (Rosenberg et al. 1998: 7). The ground stone assemblage also includes stone bowls made from limestone and from a grey-green chloritic stone (Rosenberg and Redding 2000: 50). Those made from chloritic stone are often elaborately decorated with incised naturalistic or geometric designs. Elaborately decorated pestles made from the same chloritic stone or, occasionally, sandstone, were also found. They were extensively conserved and refashioned until they were too small to be used (Rosenberg and Redding 2000: 50–2).
The bone assemblage is dominated by wild ovicaprids, followed by red deer, and boar.