Introduction
At the site of Tell Brak, the excavation of the Area TC Oval, a large burnt building dated to the mid-third millennium B.C.E., saw the recovery of large quantities of well-preserved charred macrobotanical remains (Emberling and McDonald Reference Emberling and McDonald2003). The analysis of this material has allowed an in-depth examination of agricultural production and crop management at the site during the Early Bronze Age. Previous discussions of Upper Mesopotamian agriculture during this period have highlighted the importance of water management due to the semi-arid nature of the environment in this region today (e.g., McCorriston and Weisberg Reference McCorriston and Weisberg2002; Riehl et al. Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014). To this end, farmers would have been heavily reliant on winter rains, with even minor climatic fluctuations being potentially devastating for the harvest (Riehl Reference Riehl2009). This paper combines crop stable isotope analysis and functional weed ecology to assess crop growing conditions, as represented by plant remains recovered from the TC Oval building. It also assesses how specific regime choices made by farmers ultimately reduced the overall risk of crop failure at Tell Brak, as well as the role of the city as a centre for mobilisation of staples during the mid-third millennium B.C.E.
Tell Brak
Location and Topography
The site of Tell Brak is situated within northern Mesopotamia, an area that includes modern southeastern Turkey, northeastern Syria and northern Iraq (Fig. 1). The region is separated geographically from southern Mesopotamia by the abuttal of the southern alluvial plains to the Jezira limestone plateau (Lloyd Reference Lloyd1984) and is bounded on the east and west by the Tigris and Euphrates rivers (Oates et al. Reference Oates, Oates and McDonald2001). The site itself is located c. 40 km northeast of the modern city of Hasakah on the Upper Khabur plain (Fig. 2) at an altitude of 357 m asl on a large expanse of flat, rolling landscape ideally suited for rain-fed agriculture (Weiss Reference Weiss and Weiss1986; Wilkinson et al. Reference Wilkinson, Philip, Bradbury, Dunford, Donoghue, Galiatsatos, Lawrence, Ricci and Smith2014). The area immediately around the site is bordered to the south and east by the Wadis Radd and Jaghjagh, respectively. Today these water courses are quite small, and farmers must rely on diesel-powered irrigation systems to provide enough water for crop growth (Charles et al. Reference Charles, Pessin and Hald2010). During the Bronze Age, however, these wadis must have been extremely important, as can be seen from the number of tell sites (e.g., Tell Barri and ancient Nisibis) constructed on their banks, and control of these water sources would have been vital (Oates Reference Oates1990; Wilkinson Reference Wilkinson, French, Matthews, Oates, Oates, Oates and McDonald2001; Ur Reference Ur2010).
Map showing the location of Tell Brak and other northern Mesopotamian sites with archaeobotanical remains mentioned in this paper
Map showing the Upper Khabur Basin (based on Menze and Ur Reference Menze and Ur2012)
Climate
The current climate of this region is typically continental with hot, dry summers and cool winters (BSh semi-arid steppe, Koppen-Geiger classification). Average daily temperatures range from c. 25–35°C in July and August but can increase to highs of around 44°C.Footnote 1 Winter temperatures can drop to below freezing but tend to average around 5–6°C. In terms of rainfall, Tell Brak is located in a marginal area for rain-fed agriculture, lying between the 250–300 mm annual rainfall isohyets (Charles et al. Reference Charles, Pessin and Hald2010). This level of annual rainfall can easily support the cultivation of barley, but it is less suitable for the cultivation of wheat, especially if there are sporadic droughts (Hole and Zaitchik Reference Hole and Zaitchik2007). Modern rainfall levels in general have tended to average 363 mm/year (Hijmans et al. Reference Hijmans, Cameron, Parra, Jones and Jarvis2005), but inter-annual rainfall variability has meant that in some years the precipitation levels do not reach the amount needed to support rain-fed farming. During the last 15 years in particular, this region has become much drier, with recorded rainfall levels in Hasakah barely topping 100 mm/year.
Past climate reconstructions for Tell Brak and northern Mesopotamia have been hampered by the lack of local climate proxy records in the form of pollen cores. Late Chalcolithic and Early Bronze Age local proxy data from archaeobotanical macro-remains (e.g., Miller Reference Miller and Zettler1997; McCorriston and Weisberg Reference McCorriston and Weisberg2002; Deckers and Pessin Reference Deckers and Pessin2010) and stable isotope analysis of crops (e.g., Riehl Reference Riehl2009; Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) have shown that precipitation levels in the region, after dipping during the 5.2kya B.P. event, did recover slightly but that general conditions remained fairly dry. Using speleothem δ18O values from Soreq Cave in Israel (Bar-Matthews and Ayalon Reference Bar-Matthews and Ayalon2011) and the present-day calibration relationship between these values and rainfall has led to the estimation that rainfall was, on average, 300–320 mm/year during the period 2900–2300 B.C.E. (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017). This would place mid-third millennium B.C.E. Tell Brak right on the edge of the ‘zone of uncertainty’ (Wilkinson et al. Reference Wilkinson, Philip, Bradbury, Dunford, Donoghue, Galiatsatos, Lawrence, Ricci and Smith2014), an area of land located between the 200–300 rainfall isohyets where cereal cultivation is thought to be much riskier. This estimate suggests that Tell Brak farmers were heavily reliant on the winter rains for cereal cultivation and that any short periods of drought would have been devastating (Riehl Reference Riehl2009; Lawrence et al. Reference Lawrence, Philip and de Gruchy2021). The location of Tell Brak between the Wadis Jaghjagh and Radd, however, may have mitigated this source of water stress to a certain extent, as they would have allowed Bronze Age farmers to access better-watered soils in certain areas of the urban hinterland (Riehl Reference Riehl2012; Wilkinson et al. Reference Wilkinson, Philip, Bradbury, Dunford, Donoghue, Galiatsatos, Lawrence, Ricci and Smith2014).
Vegetation and Soils
The present-day vegetation of northern Mesopotamia is primarily steppic, bordered by forests to the north in Turkey and desert and alluvial plains to the south in Iraq. Tell Brak itself is located in the ‘Moist Steppe’ vegetation zone (although ‘Dry Steppe’ could also be applicable in the south of the region due to the low average rainfall) and is dominated by grass steppe and land predominantly under cultivation. Trees and woodland are mainly absent, although there are small pockets of Pistacia – Amygdalus forest located near the rivers and wadis (Guest Reference Guest1966). Zohary (Reference Zohary1950) has said that steppic environments consist of open plant communities that are limited primarily by climatic conditions and lack of rainfall. Certainly the local environs of Tell Brak are fairly bare in the dry summer months but are covered with diverse grassland species such as Artemisia herba-alba Asso. and Poa bolbosa L. after the arrival of the winter rains (Zohary Reference Zohary1973). Past vegetation reconstructions of this region have indicated the presence of extensive Quercus woodland (Bottema and Cappers Reference Bottema, Cappers and Jas2000), but during the Late Chalcolithic and Early Bronze Age increasing aridity and human activity in the area led to reduction in woodland and an expansion of grass/shrub steppe (Deckers Reference Deckers, Conard, Drechsler and Morales2011). In terms of soils suitable for arable agriculture, Tell Brak is located in an area of flat land with reasonably fertile calcic xerosols (Wilkinson Reference Wilkinson2003). Furthermore, alluvial soils located on the banks of the nearby wadis would have been enriched with nutrients and potentially provided prime arable land (French Reference French2003). Today, the soils around the site have been affected by erosion and leaching caused by continuous farming activity, leading to a reduction in fertility from that of the Bronze Age (Charles et al. Reference Charles, Pessin and Hald2010).
Historical Background
The earliest excavated occupation at Tell Brak is dated to the Ubaid period (c. fifth millennium B.C.E.), although the recovery of Halafian ceramics and Pre-Pottery Neolithic B chipped stone from these levels may indicate settlement in the area as early as the eighth millennium B.C.E. (Oates et al. Reference Oates, Oates and McDonald2001). The site continued to grow into a complex settlement throughout the Late Chalcolithic (4000–3200 B.C.E.), covering an area of c. 100–130 hectares at its peak, with an estimated population of around 20,000 inhabitants (Emberling Reference Emberling, Al- Gailani, Werr, Curtis, Martin, McMahon, Reade and Oates2002; Hald Reference Hald2008; Oates and Oates Reference Oates and Oates1993; Ur et al. Reference Ur, Karsgaard and Oates2011). At the end of the fourth millennium B.C.E. (c. 3200 B.C.E.), a period of aridity (Bar-Matthews et al. Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Riehl et al. Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014) saw the abandonment of many Uruk colony sites in northern Mesopotamia (Lawrence et al. Reference Lawrence, Philip and de Gruchy2021). Several urban sites, including Tell Brak, are thought to have declined in size (Wilkinson Reference Wilkinson2000; Ur et al. Reference Ur, Karsgaard and Oates2011) as city dwellers moved away to more sustainable rural communities (Ur Reference Ur2010). Archaeobotanical evidence from this period at Tell Brak, however, shows a greater reliance on cereals such as einkorn, which have greater water requirements, indicating that the remaining urban community was able to adapt to fluctuating climatic conditions (Charles et al. Reference Charles, Pessin and Hald2010; Lawrence et al. Reference Lawrence, Philip and de Gruchy2021). Certainly, by the first half of the third millennium B.C.E., improving conditions saw a surge in re-urbanization in northern Mesopotamia (Matthews Reference Matthews2004) and by c. 2600 B.C.E. the occupied area at Tell Brak (Fig. 3) had reached 65–70 hectares (Emberling et al. Reference Emberling, Cheng, Larsen, Pittman, Skuldboel, Weber and Wright1999). Moreover, during this period, documentary evidence recovered from the site of Ebla, in northwest Syria, identified Tell Brak as ‘Nagar’, the most important settlement in the area (Oates and Oates Reference Oates, Oates, Oates, Oates and McDonald2001).
CORONA satellite photo showing the ‘hollow ways’ radiating out from the central mound of Tell Brak (Ur Reference Ur2003).
Area TC
Excavation began at Tell Brak in the 1930s under the direction of Max Mallowan, focusing primarily on deposits from the later third millennium B.C.E. Work was resumed in 1976 by David and Joan Oates and continued by a series of field directors from 2003–2011, including Roger Matthews, Geoff Emberling and Helen McDonald, and Augusta McMahon. These studies focused on a range of periods, including the earlier fifth and fourth millennium B.C.E. levels (e.g., Matthews Reference Matthews2004; McMahon and Oates Reference McMahon and Oates2007; Oates et al. Reference Oates, McMahon, Karsgaard, al-Kuntar and Ur2007), the third millennium B.C.E. (e.g., Oates et al. Reference Oates, Oates and McDonald2001) and intensive survey of the urban landscape (Ur Reference Ur2003).
In the 1998 field season, a large Oval building was first identified in Area TC (Fig. 4). This structureFootnote 2, dated to the mid-third millennium B.C.E., was found to cover an area c. 45 x 50 m, organized around two central courtyards (Emberling and McDonald Reference Emberling and McDonald2001). Internally, the building was divided into a number of rooms (Fig. 5), many of which contained objects associated with food storage and processing. Around the outer courtyard were rooms thought to have been used for production of bread. These included Rooms 4, 6, 7 and 8, which were used for grain storage, and the larger Room 2 (c. 2 x 6 m in area) which contained seven small bread ovens along the east and south walls (Emberling et al. Reference Emberling, Cheng, Larsen, Pittman, Skuldboel, Weber and Wright1999). The inner courtyard, by contrast, seems to have been more domestic in character, containing a small kitchen (Room 14) and a corridor (Room 12) with a drainage feature and a large number of broken pottery sherds and stones. This part of the building also had further storage areas thought to be used for agricultural produce, including Room 16, a reception/storage room with benches built along the outer walls and a mudbrick bin, and Room 15, another small store room. Also found in the building were 250 clay sealings from doors and packages, suggesting bureaucratic control over the produce stored and manufactured in the TC Oval. These sealings and features have led to the interpretation that the TC Oval was a public building associated with the administration of grain and bread rations to a segment of the wider population of Tell Brak (Emberling and McDonald Reference Emberling and McDonald2003). The TC Oval building was almost entirely destroyed by fire close to the beginning of Akkadian imperial control in this region (Emberling and McDonald Reference Emberling and McDonald2003). Upon excavation, it was discovered that the building had still been in use at the time of destruction and contained very significant concentrations of intact, well-preserved charred cereal grain. In particular, Room 16 contained piles of pure grain alongside mixed piles of grain and chaff, suggesting that inhabitants were engaged in the final stages of crop processing at the time of the fire. This material was recovered and has been studied and analysed at the School of Archaeology, University of Oxford.
Plan of the main mound at Tell Brak showing Area TC and other excavation areas (Emberling et al. Reference Emberling, Cheng, Larsen, Pittman, Skuldboel, Weber and Wright1999)
Plan of Area TC and the Temple Oval building (Emberling & McDonald, Reference Emberling and McDonald2001)
Previous archaeobotanical work at mid-third millennium B.C.E. Tell Brak
There have been a number of previous archaeobotanical studies carried out on material recovered from Tell Brak (e.g., Colledge Reference Colledge and Matthews2003; Hald Reference Hald2005, Reference Hald2008), and one of these studies (Charles and Bogaard Reference Charles, Bogaard, Oates, Oates and McDonald2001) also focused on the mid-third millennium B.C.E. This previous study was completed on archaeobotanical material recovered from 1978–1984 excavations on both public and domestic areas of the site. The results of this study showed that several cereals, 2-row hulled barley, emmer and einkorn wheat, were commonly found during this period, whilst pulses were also present but much less frequent (Charles and Bogaard Reference Charles, Bogaard, Oates, Oates and McDonald2001). Spatial variation in sample composition indicated that there may have been a contrast between domestic and public production of agricultural goods during this period. For example, in Level 3 of Area FS, sampling of an Akkadian public building (Fig. 4) shows that hulled barley was ubiquitous, but pulses were entirely absent. By comparison, samples taken from domestic contexts in Areas CH and ER included pulses alongside various cereal species. These results led Charles and Bogaard (Reference Charles, Bogaard, Oates, Oates and McDonald2001) to suggest that there were two separate, yet complementary, systems of production in existence during the third millennium B.C.E. at Tell Brak. First, a ‘specialized institutional agriculture’ administered by the temple focused on hulled barley and wheat production as a means of providing bread rations for workers and fodder for palace livestock. Second, a ‘household-scale agriculture’ included the private cultivation of a much wider range of crops including pulses, cereals and other species such as flax. Crop stable isotope analysis has also been carried out on this material as part of a larger investigation by Styring et al. (Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) into farming practices in northern Mesopotamia. These studies will be discussed below along with the new results presented in this paper to form a more comprehensive picture of agricultural production at Tell Brak during the mid-third millennium B.C.E.
Methods
Soil samples of 30–40 litres were taken systematically from every undisturbed archaeological unit in the TC Oval, resulting in 85 archaeobotanical samples. Areas of particular interest within the building were sampled using a grid system (Emberling and McDonald Reference Emberling and McDonald2001) and on occasion, visible concentrations of charred plant remains were sampled at close intervals by hand to assist in the identification of spatial variation within the plant assemblage. All samples were processed using a flotation machine based on the French design (French Reference French1971). Initial sample scanning was carried out to assess sample richness, with a target of 300 identified cereal grains. From this evaluation, 25 samples were chosen for part of a preliminary study undertaken by Mette Marie Hald and Mike Charles at the University of Sheffield (Emberling and McDonald Reference Emberling and McDonald2001). Full quantification, identification and analysis of these 25 samples and a further selected 33 samples from the TC Oval were carried out at the School of Archaeology, University of Oxford from 2014–2018 (Table 1). Samples for full analysis were chosen due to their apparent archaeobotanical richness, the range of species represented, and their contextual provenance within the TC Oval.
The number of samples studied from the Tell Brak TC Oval building, by room number
All samples were sorted and identified using a Nikon stereomicroscope (x7–80); this included the re-analysis of plant remains identified during earlier studies so that a formal identification criterion could be standardised across the site. Charred plant items were identified using a comparative modern reference collection and relevant published resources such as the Flora of Iraq (Guest Reference Guest1966; Townsend and Guest Reference Townsend and Guest1966–1988) and the Nouvelle Flore du Liban et de la Syrie (Mouterde Reference Mouterde1966). Latin nomenclature for all plant remains follows Zohary et al. (Reference Zohary, Hopf and Weiss2012) and Townsend and Guest (Reference Guest1966–1988). The minimum number of individuals (MNI) was used to quantify all plant remains by counting easily identifiable diagnostic plant item areas (Jones Reference Jones, van Zeist, Wasylikowa and Behre1991). With regard to cereal grains, both embryo and apical ends were recorded separately, but only the most abundant category was used to determine the final grain total. Similarly, for glume wheat chaff, each glume base was scored individually, and spikelet forks were recorded as two glume bases. Seeds from wild or weed taxa were largely scored individually even when fragmented, except for when it was clear that broken fragments belonged to the same seed (cf. van der Veen Reference van der Veen1992).
Correspondence analysis (CA) was used to explore relative compositional variation through the arrangement of samples along a set of axes based on species composition. Associations between species and/or samples are shown by the direction and distance in which they diverge from the central (origin) point of the plot. Samples that cluster have a relatively similar composition, whereas divergent samples are more compositionally distinct. CA was carried out using CANOCO 5 for Windows 8 (Ter Braak and Smilauer Reference Ter Braak and Smilauer2012). In all diagrams, axis 1 (which accounts for the most variation) was plotted horizontally and axis 2 vertically.
Analysis of crop carbon and nitrogen stable isotope values was undertaken to infer water availability and soil nitrogen composition, in order to characterise crop growing conditions and arable land management practices. Within archaeological stable isotopic research, crop water availability is linked to the fractionation of carbon during photosynthesis, and crop manuring status can be inferred from the volatilisation of the lighter nitrogen isotope after soil enrichment. The relative stable isotope ratios of carbon and nitrogen can then be used to infer the extent of human crop management systems and the use of varying husbandry techniques.
Samples representing the four main cereal grain varieties, emmer wheat, hulled barley, ‘small’ barley and ‘small’ wheat, were selected from a range of contexts within the TC Oval as a means of assessing variation in crop growing conditions. In total, 41 subsamples, each containing ten homogenized cereal grains, were selected for analysis (Supplementary Table 1Footnote 3). Cereal grains from representative samples were first screened using Fourier Transform Infrared Spectroscopy (FTIR) to identify possible post-depositional contamination. In each spectrum, peaks characteristic of carbonate contamination (870 and 720 cm-1) were observed. A hydrochloric acid pre-treatment (following Vaiglova et al. Reference Vaiglova, Snoeck, Nitsch, Bogaard and Lee-Thorp2014) was therefore used on all cereal grains to remove carbonate traces. Stable isotope analysis was conducted using a Sercon EA-GSL mass spectrometer at the Research Laboratory for Art History and Archaeology at the University of Oxford. All values were measured with reference to international standards and were calibrated using an internal alanine standard. For δ13C determinations, isotope ratios were normalized to the Vienna Peedee Belemnite scale (VPDB) using IAEA-C6 and IAEA-C7 standards (Supplementary Table 2). Values for δ15N were calculated against the atmospheric composition of N2, using caffeine and IAEA-N2 standards (Supplementary Table 3). All calculations regarding crop stable isotope values were performed using the statistical programming language R (3.2.3). Calculation of Δ13C values (following Farquhar et al. Reference Farquhar, Ehleringer and Hubick1989) was accomplished using the δ13C value of atmospheric CO2 estimated from the AIRCO2_LOESS system (Ferrio et al. Reference Ferrio, Araus, Buxó, Voltas and Bort2005). All results reported are also corrected for the minor effects of charring on δ13C (by subtracting 0.11‰) and δ15N (by subtracting 0.31‰) following Nitsch et al. (Reference Nitsch, Charles and Bogaard2015). Measurement uncertainties for δ13C and δ15N values were calculated using the within-run variability of the raw measurements and the known uncertainty of the two reference standards using the approximation method (Kragten Reference Kragten1994). The average measurement uncertainty for δ13C was 0.094‰ and for δ15N was 0.38‰. On the basis of the difference between the observed and known δ values of an in-house alanine and the long-term standard deviations of the alanine, accuracy or systematic error (u(bias)) was determined to be ±0.125 for δ13C and ±0.2 δ15N (following the Szpak et al. Reference Szpak, Metcalfe and Macdonald2017 protocol).
Functional weed ecology was used to examine the intensity of crop cultivation and the level of labour input related to soil fertility. Discriminant analysis was used to make comparisons between modern weed floras grown under high- or low-input conditions and the archaeobotanical weed data presented in this paper. Modern studies were undertaken in a range of climatically varied regions including semi-arid regions of Morocco and southern Europe (Bogaard et al. Reference Bogaard, Styring, Ater, Hmimsa, Green, Stroud, Whitlam, Diffey, Nitsch, Charles, Jones and Hodgson2018; Jones et al. Reference Jones, Bogaard, Halstead, Charles and Smith1999, Reference Jones, Bogaard, Charles and Hodgson2000). Functional data on specific attributes (i.e., canopy height, canopy diameter, specific leaf area and leaf area per node: thickness) of arable weeds from known agricultural traits were gathered. Discriminant analysis was then performed using a combination of these functional weed traits that successfully separated the modern high- and low-input regimes, to produce a linear equation; this equation was then used to classify the TC Oval archaeobotanical samples based on the functional trait values of the weed species in each sample. Forty-six archaeobotanical samples, each containing ten or more weed seeds identified to species, were included in this analysis. IBM SPSS Statistics 22 was used to perform the discriminant analysis.
Results
Assemblage Overview
The 58 samples analysed from the TC Oval primarily contained a very large quantity of well-preserved cereal grains, cereal chaff, pulses and weed seeds, as summarized in Fig. 6 (also Supplementary Table 4). Cereal grains were, by far, the dominant component of the assemblage, averaging 84%. By contrast cereal chaff (2%) and pulses (0.4%) were much less frequent. Weed/wild seeds (average composition of 13.8% per sample) were slightly higher, with nine samples found to contain over 40% of these taxa. Sample compositions were compatible with the results of crop processing analysisFootnote 4, which will be presented fully as part of a future paper discussing storage context and plant consumption activities within the TC Oval assemblage. Overall, crop processing analysis indicated that cereals recovered from Rooms 4, 6, 8, 9 and 16 had been through threshing, winnowing, coarse- and fine-sieving stages and were likely being stored before final consumption, an interpretation consistent with the proposed function of these rooms (see above). Samples from Rooms 12, 13, 17, 18 and the North Courtyard appear more mixed, potentially indicating their use for multiple activities including crop processing and grain storage.
Percentage stacked bar chart of all samples from Tell Brak showing the proportions of the four major categories of plant remains. Samples have been sorted to emphasize the contrast between those dominated by cereal grains and those dominated by weed/wild taxa
Major Cereal Crops
Four different cereal species were identified from the TC Oval: 2-row hulled barley (Hordeum vulgare L.), the glume wheats emmer (Triticum dicoccum Schübl.) and einkorn (Triticum monococcum L.), and free-threshing wheat (Triticum aestivum L./durum Desf.). Of these, hulled barley was by far the most abundant component of the entire assemblage, totaling 563,629 grains and with 100% ubiquity across the assemblage (Fig. 7). By comparison, the wheat species appear in much smaller proportions; emmer wheat was the most commonly identified, in 55% (32/58) of samples, with free-threshing wheat and einkorn wheat present in 38% (22/58) and 22% (13/58) of samples, respectively. A similar ratio between the glume wheats was also observed in terms of identified glume bases. Emmer wheat chaff was present in 26% (15/58) of samples whilst einkorn wheat chaff was present in 10% (6/58) of samples (Table 2).
Percentage stacked bar chart of all samples showing only major cereal proportions. Samples have been sorted to emphasize the contrast between those dominated by hulled barley grains and those dominated by emmer wheat grains
Summary of the frequency and abundance of major plant taxa
Other Cultivated Crops and Collected Plants
Several pulse crops were also identified within the TC Oval assemblage (Table 2), lentil (Lens culinaris Medik.) and grass pea (Lathyrus sativus L./cicera L.) being the most common, occurring in 21% (12/58) and 12% (7/58) of samples, respectively. The ubiquity and total number of these species were significantly lower than the cereals, however, suggesting that they were not purposely being stored within the building. There was also a small amount of collected fruit/nut material present within the assemblage. These included grape seeds (Vitis sp.), pistachio (Pistacia sp.) and almond (Amygdalus sp.). The number of remains identified (all species had a ubiquity of under 10%) again suggests that these plant items were not being stored in the TC Oval, but were likely to have been gathered and consumed in other areas of the city (Hald Reference Hald2005, Reference Hald2008).
Weed/Wild Taxa
In total, 50 wild/weed taxa were identified from type to species level within the TC Oval, but only 17 were found in more than 10% of samples (see Supplementary Table 4 for full species list). The most frequently identified remains were those of the wild grasses, particularly three species of goat grass (Aegilops crassa Boiss., speltoides Tausch. and tauschii Coss.) and Lolium cf. rigidum Gaud. (see Table 3), as well as Sinapis cf. arvensis L., a species of wild mustard. These taxa are all common arable weeds from cultivated fields and disturbed habitats (Guest Reference Guest1966) and are likely to have been found within the environs of Tell Brak. Additionally, the Aegilops species are known as crop mimics (Anderson Reference Anderson, Singh, Batish and Kohli2006; Barrett Reference Barrett1983) and are very difficult to remove from the crop either in the field or during crop processing due to their appearance and size imitating that of domesticated cereals. There is no evidence that any of the wild/weed taxa were being cultivated as a crop, and certainly the total sum of remains identified is significantly lower when compared with the total sum of cereal grains.
Summary of the frequency and abundance of major weed/wild taxa
Small Cereal Grain Varieties
Throughout the TC assemblage, a number of conspicuously small grains of barley (present in 86%, 50/58 samples) and wheat (present in 34%, 20/58 samples) were identified (Fig. 8). These grains ranged from 0.5–1.5 mm in length, when compared with ‘normal’ cereal grains which ranged from 3–4.5 mm. Morphologically, these grains appear to be small forms of domesticated Hordeum vulgare sp. and Triticum sp. rather than smaller wild species.
Photos of the ‘normal’ and small hulled barley gain varieties (left) and ‘normal’ and small wheat grain varieties (right)
One hypothesis is that these smaller grains are ‘tail’ grains, or the small grains commonly found in the distal florets of the cereal ear (Hillman Reference Hillman and Mercer1981). ‘Tail’ grains develop due to the process of floret growth within the ear. Glume primordia are initiated first, followed by the florets, after which spikelet growth decreases (Kirby and Appleyard Reference Kirby and Appleyard1987). This leads to a gradient of floret development within the ear, with the most mature florets occurring at the base and in the middle of the cereal ear, whilst the top florets tend to be underdeveloped. Glume wheat grains developing in these top or very bottom florets tend to be smaller than grains from the lower and middle florets, and in some cases a grain does not develop at all (Percival Reference Percival1921). The terminal spikelet of emmer wheat contains a single grain, in contrast to the other two-grained spikelets. Tail grains of wheat and barley are approximately two-thirds the size of the larger grain but closely resemble them in shape. Under arid growing conditions or periods of drought, grain, especially at the extremities of the ear, may be under-developed or show signs of shrivelling due to the lack of water (Kirby Reference Kirby, Curtis, Rajaram and Gomez Macpherson2002). Given the semi-arid nature of Tell Brak, the presence of ‘tail’ grains would not be implausible.
Within the TC Oval assemblage, however, the ratio of normal to small grains was lower than would normally be expected if small grains merely represented ‘tail’ grainsFootnote 5. Furthermore, ‘tail’ grains are normally removed, along with small weed seeds and chaff, by sieving during crop processing activities. Clean, stored cereal grain would, therefore, be likely to contain a much higher ratio of ‘normal’ grains to ‘tail’ grains than would be present in the unprocessed ear (Hillman Reference Hillman and Mercer1981; Willcox Reference Willcox2004). As discussed above, the proportions of cereal grains, chaff and weed/wild seeds within the TC Oval assemblage suggest that these stored crops had been through the fine-sieving process. This evidence, combined with the ratios of normal to small grains identified in the assemblage, indicate that the small grains do not primarily represent ‘tail’ grains of the normal hulled barley and wheat crops.
A second hypothesis is that these small grains instead represent the cultivation of a distinct variety or landrace of small-grained domesticated wheat and hulled barley at Tell Brak. These varieties may have been developed specifically to tolerate arid conditions and could potentially have been grown in less well-watered locations to maximise the use of marginal farming areas. The interpretation of these small grains is explored in greater detail below, with reference to the results of crop stable isotope analysis. Importantly, these small-grained versions of barley and wheat do not resemble poorly developed grains but rather well developed grains of distinctly smaller absolute size (Fig. 8).
Compositional Analysis
Correspondence analysis was carried out on all samples to explore the relationships between cereal crops and associated weed/wild taxa. There are three discrete groups of samples (Fig. 9a–b), those dominated by hulled barley grains around the origin of the plot, those with relatively high proportions of glume wheat grains towards the right (positive) end of axis 1, and samples with a more variable composition of weed/wild taxa and free-threshing wheat toward the top (positive) end of axis 2 (see Supplementary Table 5 for correspondence analysis codes). Each group of samples also has a distinctive set of accompanying weed/wild taxa. Hulled barley-dominated samples contained primarily large-seeded grasses (e.g., Aegilops sp. and Lolium rigidum Gaud.). Glume wheat-dominated samples primarily contained Galium sp. and Silene sp., while the third grouping of samples had the widest range of species with significant quantities of Gypsophila pilosa Huds. and small-seeded legumes. These distinct groupings of cereals and weeds may indicate that each crop was grown under distinct husbandry conditions and that the inhabitants of Tell Brak used a variety of farming strategies.
Correspondence analysis scatterplot of 41 taxa
Correspondence analysis plot of 58 samples with samples represented as pie charts showing the proportions of major plant item categories
Crop Stable Isotope Analysis
Crop stable isotope analysis was carried out on the grains of two-row hulled barley, emmer wheat, small barley, and small wheat (for full results of stable carbon and nitrogen isotope measurements see Supplementary Table 6). Figure 10 shows the Δ13C values from the four cereal taxa. Overall, variability within each taxon is limited (standard deviations are all within ±0.5‰ – cf. Nitsch et al. Reference Nitsch, Charles and Bogaard2015), suggesting that each cereal was cultivated under consistently similar growing conditions. Due to the physiological differences between wheat and barley, such as the earlier ripening of the latter (Araus et al. Reference Araus, Febrero, Buxo, Camalich, Martin, Molina, Rodriguez-Ariza and Romagosa1997; Wallace et al. Reference Wallace, Jones, Charles, Fraser, Halstead, Heaton and Bogaard2013), an offset of positive 1.0‰ was applied to the large and small barley grains. An ANOVA test was used to investigate inter-taxa variability, and the results showed significant differences in Δ13C values between the four cereals (F(3, 37) = 19.38, p = <0.0001). Further tests showed that the differences visible in Figure 10 are statistically significant, in particular the Δ13C values of small barley are considerably different from all other taxa (see Table 4 for the results of all statistical tests). These results denote that whilst emmer wheat, small wheat and some hulled barley were grown under similar medium-low watering conditions, all of the small barley and a few of the hulled barley grains were grown under distinctly drier conditions. Overall, the Δ13C values from all taxa are indicative of rain-fed farming rather than the use of artificial watering, but the values of small barley suggest that farmers exploited a range of growing conditions including the use of marginal arable land.
Δ13C values of the four major cereal taxa. The lower dashed line indicates the beginning of the ‘well-watered’ zone for wheat. The middle and upper dotted lines indicate the beginning of the two potential ‘well-watered’ zones for barley. S. Barley and S. Wheat refer to the small barley and wheat grain varieties
Results of an ANOVA test and post-hoc analysis conducted on the Δ13C values from all cereals with a 1‰ offset applied to the barley
Stable nitrogen isotope values (δ15N) were also measured from all four taxa (Fig. 11). These results showed a wide range of values and were more variable than would be expected from crops grown under consistent growing conditions (cf. Nitsch et al. Reference Nitsch, Charles and Bogaard2015). Moreover, there were also significant differences between the values of hulled barley and the other three cereal taxa (see Table 5 for the results of statistical tests). The manuring bands constructed by Bogaard et al. (Reference Bogaard, Fraser, Heaton, Wallace, Vaiglova, Charles, Jones, Evershed, Styring, Andersen, Arbogast, Bartosiewicz, Gardeisen, Kanstrup, Maier, Marinova, Ninov, Schäfer and Stephan2013) for temperate Europe have been adjusted using past rainfall estimates (cf. Bar-Matthews & Ayalon, Reference Bar-Matthews and Ayalon2011) for Tell Brak to take into account ecosystemic enrichment in 15N caused by aridity (Styring et al. Reference Styring, Ater, Hmimsa, Fraser, Miller, Neef, Pearson and Bogaard2016, Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017). With this in mind, a number of samples still exhibit high δ15N values that would exceed the enrichment caused by aridity alone. The use of manure has been extensively documented from Neolithic communities in south-east and central Europe (Bogaard et al. Reference Bogaard, Fraser, Heaton, Wallace, Vaiglova, Charles, Jones, Evershed, Styring, Andersen, Arbogast, Bartosiewicz, Gardeisen, Kanstrup, Maier, Marinova, Ninov, Schäfer and Stephan2013; Vaiglova et al. Reference Vaiglova, Coleman, Diffey, Tzevelekidi, Fillios, Pappa, Halstead, Valamoti, Cavanagh, Renard, Buckley and Bogaard2021) and Bronze Age societies in the Middle East (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017; Wilkinson Reference Wilkinson1982), where its use improved soil fertility and overall crop yields. The use of manure, however, was usually limited to the immediate settlement environs as it is not easily transportable over long distances (Halstead Reference Halstead2014). At Tell Brak, emmer wheat, small wheat and some of the small barley indicate medium to high levels of artificial enrichment and may reflect preferential manuring. By comparison, hulled barley and the remaining small barley appear to have been grown without the use of manure and/or in areas of the landscape with low levels of enrichment.
δ15N values of the four major cereal taxa, with manuring bands adjusted for aridity (after Styring et al. Reference Styring, Ater, Hmimsa, Fraser, Miller, Neef, Pearson and Bogaard2016, Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017)
Results of an ANOVA test and post-hoc analysis on the δ15N values of the four cereal taxa
The combined results of carbon and nitrogen stable isotope analysis indicate that there was preferential treatment of certain crops in terms of both manuring and better watered soils. Overall, the low Δ13C values of small barley suggest that this crop was grown in much drier conditions than the wheats and hulled barleys. Similarly, the low δ15N values of hulled barley (normal and small-sized) indicate that these crops were cultivated on land without artificial enrichment, whilst wheat crops did receive the application of some manure/midden material.
Functional Ecological Analysis of Weeds and Intensity of Cultivation
The intensity of cultivation was assessed relative to known modern farming regimes, using the functional ecological analysis of weed data from the TC Oval. A discriminant function extracted to distinguish high- and low-intensity agricultural systems in southern Europe and Morocco (Fig. 12a) was used to classify the archaeological samples, based on the functional ecological attributes of identified weed taxa (Bogaard et al. Reference Bogaard, Hodgson, Nitsch, Jones, Styring, Diffey, Pouncett, Herbig, Charles, Ertuğ, Tugay, Filipovic and Fraser2016, Reference Bogaard, Styring, Ater, Hmimsa, Green, Stroud, Whitlam, Diffey, Nitsch, Charles, Jones and Hodgson2018). Of the functional attributes measured (see Methods, above), specific leaf area was found to be the most important attribute in terms of discriminating between both the ethnographic datasets and the archaeological data. Figure 12b shows the results of this discriminant analysis and clearly demonstrates that the majority of samples from Tell Brak are located towards the low-intensity end of the spectrum, with only one sample classified as high-intensity. This sample was taken from Room 12 and had a mixed composition of hulled barley and glume wheat, but also the lowest probability classification in the assemblage (76%) suggesting that the categorization of this sample as high-intensity is not reliable. All other samples were classified with very high probabilities (over 90%), and 40 samples were classified with 100% probability. When compared with modern weed datasets, the Tell Brak samples are significantly lower than all other results, but some points do correlate with samples from modern low-intensity agricultural regimes which are characterized by limited use of manuring, no large-scale irrigation, and low rates of disturbance (weeding and tillage), such as fields in Haute Provence, France and rain-fed terraces in Morocco. The results from Tell Brak suggest therefore that crops stored in the TC Oval were managed in a similar fashion. They also reinforce previous functional ecological analysis carried out on other mid-third millennium B.C.E. samples from Tell Brak (Bogaard et al. Reference Bogaard, Styring, Ater, Hmimsa, Green, Stroud, Whitlam, Diffey, Nitsch, Charles, Jones and Hodgson2018) and are indicative of an extensive, low-input agricultural production system in use during this period.
A combined plot of modern field survey samples in relation to the discriminant function extracted to distinguish between high (black) and low (white) intensity crop husbandry regimes, b – Plot of Tell Brak samples in relation to the discriminant function extracted to distinguish between high and low intensity crop husbandry regimes
Discussion
The Role of Cereals at Tell Brak
At Tell Brak, the cultivation and use of cereals is well attested from multiple periods (Colledge, Reference Colledge and Matthews2003; Hald, Reference Hald2008), and this evidence has been further strengthened by the results from the TC Oval building. Cereals appear to have been the primary source of food for human consumption, with by-products used for animal fodder, and cereals may also have been an important trading commodity between other settlements to the north and the major cities of southern Mesopotamia via the Euphrates River (Forrest et al. Reference Forrest, Mori, Guilderson and Weiss2004). Of the cereals recovered from the TC Oval, 2-row hulled barley was by far the most commonly identified and was the dominant component of the entire assemblage. It is likely that the 2-row variety was favoured over the more productive 6-row due to its lower water requirements (Townsend et al. Reference Townsend and Guest1966: 85), whereas 6-row hulled barley has been recorded in southern Mesopotamia, presumably under irrigation (Renfrew Reference Renfrew1984). All recovered hulled barley from the TC Oval was in a well processed state with very little barley rachis or weed/wild remains identified. This level of cereal processing would have been very labour intensive and time-consuming, suggesting that this grain was intended for human consumption rather than for use as animal fodder.
The glume wheats, emmer and einkorn were also commonly recovered, albeit in much lower quantities than hulled barley grain. Of these species, emmer was the most frequently identified, whereas einkorn was only occasionally recognised, indicating that it may have been a crop contaminant of emmer rather than a cultivated crop in its own right. In a similar manner, free-threshing wheat was also found in only a small number of samples and often separate from the clean hulled barley grain. This would suggest that both glume wheat and free-threshing wheat were still important crops for the inhabitants of Tell Brak but that they were not the primary focus of production within the TC Oval building. Furthermore, if as suggested by Emberling and McDonald (Reference Emberling and McDonald2003) the TC Oval was a public building for the production and administration of grain rations to workmen, it is possible that the emmer and free-threshing wheat were intended for another form of domestic human consumption within the structure. Further discussions around crop consumption at Tell Brak will be part of a future paper on the TC Oval assemblage. This will combine crop processing data, storage context, and material culture with ongoing bioarchaeological research at the site to expand current interpretations of agricultural production and consumption at Tell Brak during the Bronze Age.
The Question of Trade and Crop Importation
A source of debate with regard to the agricultural economy of Tell Brak is to what extent the archaeobotanical cereal grains recovered from the TC Oval are representative of crops grown in the immediate environs of the city, or whether staple foods could have been imported from other urban settlements in northern Mesopotamia (McCorriston Reference McCorriston1995; Wilkinson Reference Wilkinson2000). Varying lines of evidence including the identification of early stage crop processing activities from the mid-third millennium B.C.E. at Tell Brak (Charles and Bogaard Reference Charles, Bogaard, Oates, Oates and McDonald2001) and presence of sherd scatters and ‘hollow ways’ (Fig. 3; Ur Reference Ur2003; Wilkinson Reference Wilkinson1994) indicate that some form of arable agriculture was taking place at the site. However, during this period, semi-arid conditions meant that Tell Brak was located in a marginal area for rain-fed agriculture and even small changes to annual rainfall levels could have had a catastrophic effect on crop production (Riehl Reference Riehl2009; Wilkinson Reference Wilkinson2003). This has led to suggestions that Tell Brak would not have been able to survive periods of even minor drought without the assistance of regional trade networks (Wilkinson Reference Wilkinson2000) In particular, it has been proposed that crops could have been imported from other sites further to the north, such as Tell Leilan, that were situated in wetter areas more favourable for rainfed agriculture (Smith, Reference Smith and Weiss2012; Weiss et al. Reference Weiss, deLillis, deMoulins, Eidem, Guilderson, Kasten, Larsen, Mori, Ristvet, Rova and Wetterstrom2002).
The limited textual sources from Tell Brak during the Early Bronze Age period have, however, precluded the identification of these trade networks and made it difficult to ascertain exactly where traded crops might have originated. Instead, the assessment of crop stable carbon isotope values provides direct evidence of the growing conditions relevant to crops from the site. With this in mind, it is possible to compare the Δ13C values from crops stored in the TC Oval with values from mid-third millennium B.C.E. crops recovered from Tell Leilan (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017), where higher rainfall levels would certainly have allowed for cultivation around the site, as well as other southerly sites in the middle Khabur valley (Riehl et al. Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014). Crops from the more northerly sites are expected to have been grown under wetter conditions due to the higher annual rainfall in this area. If the crops from the TC Oval originally came from these sites, the stable carbon isotope values should be similar to these northerly values. Comparison of the TC Oval results with values from Tell Leilan (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017), however, indicates that the Tell Brak material was grown under significantly drier conditions overall (see Table 6). Furthermore, even the values from the TC Oval emmer wheat, thought to have been grown in relatively well-watered soils, still appear drier than glume wheats from Tell Leilan.
Results of Welch's two sample t-test between Δ13C values from Tell Brak and Tell Leilan (based on Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017).
These results indicate that the TC Oval crops are unlikely to have been grown in the higher rainfall zone of Tell Leilan and, by extension, not at other northern sites located in higher rainfall conditions. Instead, the carbon stable isotope results from Tell Brak correlate roughly with carbon isotope results from other mid-third millennium B.C.E. sites in the middle Khabur valley (Riehl et al. Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014) located within the same rainfall isohyet (see Table 7). Generally, the results from the TC Oval are a fraction higher than those from the middle Khabur sites, and this may reflect Tell Brak's slightly higher annual rainfall due to latitude, but overall comparison of these values indicates that the TC Oval crops were most likely to have been grown within the environs of Tell Brak. This does not lessen the role of Tell Brak with regard to the trade of goods and staple products with other sites in both northern and southern Mesopotamia, however, and it is likely that the site was mobilising as well as growing crops during this period (Powell Reference Powell, Aerts and Klengel1990).
Mean stable carbon isotope values measured from Tell Brak, Tell Leilan (based on Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) and other mid-3rd millennium BCE sites within the middle Khabur valley (based on Riehl et al. Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014).
Evidence for agricultural practices from the TC Oval
The results of crop stable isotope analysis and functional ecology, from the crops stored in the TC Oval, combined with other archaeological and topographical evidence, provide insights into how the landscape around Tell Brak was farmed. Overall, agricultural production appears to have been large-scale and ‘extensive’ with relatively low inputs of labour, water and manure. Within this system, however, certain aspects of the farming regime would have been tailored to overcome the environmental challenges faced by the inhabitants of Tell Brak, specifically the preferential treatment of some cereal species.
The results of stable carbon isotope analysis of cereal grains suggest how the natural hydrology of the landscape around Tell Brak was exploited. In general, none of the cereal species appear to have been grown in markedly wet conditions, but a portion of the hulled barley and the small-grained barley variety do appear to have been grown in substantially drier conditions than the emmer or small-grained wheat. This could indicate that both varieties of hulled barley were grown in marginal soils for agriculture, possibly because barley is more tolerant of drought than wheat (Hillman Reference Hillman1985; Riehl Reference Riehl2009), while wheat was grown in better watered soils near the wadis. Certainly, the isotopic values from the small-grained barley represent the driest conditions found in the TC Oval assemblage, and it is likely that the production of this variety would have required less water than their larger grained counterparts (Alghabari and Zahid Ihsan Reference Alghabari and Zahid Ihsan2018). To date, these smaller grain varieties have not been reported from other sites in Mesopotamia, suggesting that they may have been developed and cultivated specifically at Tell Brak to take advantage of the range of farming conditions present around the main mound.
The use of manure or any other form of fertilization for agriculture is barely mentioned in Bronze Age texts; as Postgate (Reference Postgate1992: 172) suggests, it was perhaps of no concern to the official scribes. Use of manuring/middening in the fields at Tell Brak has, however, been documented archaeologically from scatters of third millennium B.C.E. abraded pottery sherds, supporting the identification of specific cultivation areas around the mound (Ur and Colantoni Reference Ur, Colantoni and Klarich2010; Wilkinson Reference Wilkinson1994). Similarly, manure use has been identified at Tell Brak through nitrogen stable crop isotope analysis (Bogaard et al. Reference Bogaard, Styring, Ater, Hmimsa, Green, Stroud, Whitlam, Diffey, Nitsch, Charles, Jones and Hodgson2018; Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) and suggests that individual households had access to cereals grown under a range of conditions. The results of nitrogen (δ15N) stable isotope analysis from the TC Oval seemingly mirror the carbon results discussed above, with a clear separation between the values for wheat and barley. Results from hulled barley and the majority of the small-grained barley fell within the lowest manuring band, indicating that manure/middening material had not been applied to the field in which these crops grew within the last 3+ years (Bogaard et al. Reference Bogaard, Heaton, Poulton and Merbach2007; Fraser et al. Reference Fraser, Bogaard, Heaton, Charles, Jones, Christensen, Halstead, Merbach, Poulton, Sparkes and Styring2011). By contrast, the higher isotope values for emmer and small-grained wheat show that these crops were grown in soils artificially enriched through the application of medium-high inputs of manure. Furthermore, if wheat crops were grown in the better watered soils near the wadis, these alluvium soils would also be naturally enriched with nutrients (French Reference French2003).
As discussed above, the use of manure has two limiting factors: availability and transportation. These limitations suggest that wheat crops were grown in immediate ‘infield’ areas of Tell Brak, where access to manure or household midden material would have been less constrained. By contrast, barley crops may have been cultivated in the ‘outfield’ region further from the urban centre. This model of landscape use fits with the ‘halos’ of sherd scatters located near the city (Ur Reference Ur2003; Wilkinson Reference Wilkinson1994) and supports the existence of a manuring/middening spectrum which fades in intensity with distance from the main mound (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017).
The cereal economy and crop production in north Mesopotamia
Evidence from the TC Oval fits with wider evidence for the cereal economy in Mesopotamia. Archaeobotanical evidence from sites in both northern and southern Mesopotamia, alongside documentary evidence from Tell Beydar (Van Lerberghe Reference Van Lerberghe, Ismail, Sallaberger, Talon and Van Lerberghe1996) and palace archives in the south (Postgate Reference Postgate1992: 170), indicate that cereal cultivation was the backbone of the agricultural economy in this region during the Bronze Age (see Table 8 for summary of northern Mesopotamian sites in Syria with identified cereal grain remains). The staple cereal economy is thought to have supported the expansion of Late Chalcolithic populations during the early fourth millennium B.C.E., as well as sustaining these communities through a period of urban ‘devolution’ at the beginning of the Early Bronze Age (Wilkinson Reference Wilkinson2000) and the subsequent phase of re-urbanization that followed in the third millennium B.C.E. (Ur Reference Ur2010).
- Summary of Bronze Age northern Mesopotamian sites in Syria with identified cereal remains
The predominance of hulled barley seen at Tell Brak is consistent with data from a number of other sites in northern Mesopotamia, including Tell Leilan (Miller Reference Miller, Van Zeist, Wasylikowa and Behre1991) and Tell Bderi (van Zeist Reference van Zeist1999). Archaeobotanical evidence from the Khabur Basin Project (McCorriston and Weisberg Reference McCorriston and Weisberg2002), however, suggests that hulled barley was not always the primary cereal crop cultivated in this region. Instead they suggest that during the fifth and fourth millennia B.C.E., glume wheat was the dominant crop and that the cultivation and consumption of hulled barley did not expand until the early third millennium B.C.E. Evidence for these changes to cereal production systems have been identified from a number of sites, such as Tell es-Sweyhat in eastern SyriaFootnote 6 (Miller Reference Miller and Zettler1997), Tell Atij in western Syria (McCorriston Reference McCorriston1995), and sites within the middle Khabur Basin (Hole Reference Hole1991; Zeder Reference Zeder1994), and from cuneiform evidence in southern Mesopotamia (Jacobsen and Adams Reference Jacobsen and Adams1981). Reasons for this change from wheat to barley are often attributed to the period of aridity at the end of the fourth millennium B.C.E. and the greater tolerance of hulled barley to drought and saline conditions (Hillman Reference Hillman1985; Nesbitt Reference Nesbitt and Duru1996; Riehl Reference Riehl2009).
Other explanations suggest that increased hulled barley production was connected to an expansion in animal herding and equivalent need for large quantities of fodder (Miller Reference Miller and Zettler1997). This is supported by texts from Tell Beydar that list hulled barley as a fodder crop (Van Lerberghe Reference Van Lerberghe, Ismail, Sallaberger, Talon and Van Lerberghe1996), as well as a regional expansion in cattle herding during this period (Zeder Reference Zeder and Smith2003). At Tell Brak, glume wheat does appear to have been more prevalent than hulled barley during the Late Chalcolithic (Hald Reference Hald2005). Hulled barley when identified, however, was found in a number of storage contexts and, contrary to the theory outlined above, was thought to have been destined for human consumption, in the form of bread or possibly beer (McCorriston and Weisberg Reference McCorriston and Weisberg2002; Postgate Reference Postgate1992: 170), due to its highly processed condition (Hald Reference Hald2005: 122). As discussed above, hulled barley from the TC Oval was recovered in a similarly well-processed state, suggesting that it was intended for human consumption.
In direct contrast to the increase in cultivation of hulled barley, the cultivation of glume wheats seems to have decreased within northern Mesopotamia during the Bronze Age. At Tell Brak, the occurrence of glume wheats decreased from the Early Uruk period (Colledge Reference Colledge and Matthews2003), and in the north Syrian Euphrates region the cultivation of emmer wheat is radically reduced during the Early Bronze Age (van Zeist and Bakker-Heeres Reference van Zeist and Bakker-Heeres1988). Similarly, when all the archaeobotanical results from Tell Brak are examined chronologically, a decrease in the cultivation and use of glume wheats is apparent (c.f. Hald Reference Hald2008). This trend can also be seen in the TC Oval data. Glume wheat grain still forms an important portion of the assemblage but was recovered in significantly lower quantities when compared with hulled barley. There have been a number of theories regarding this decrease in glume wheat production, including social choice and the worsening climatic conditions present in the Late Uruk period favouring the production of hulled barley. McCorriston and Weisberg (Reference McCorriston and Weisberg2002) also suggest that the reduced visibility of glume wheat within the archaeobotanical record during the third millennium B.C.E. is because these cereals require different methods of processing and so were stored separately from free-threshing hulled barley crops. Certainly, extra crop processing stages are needed to de-husk glume wheat spikelets (Hillman Reference Hillman1985; Jones Reference Jones1990) leading to the suggestion that they were grown primarily within the domestic sphere and stored at the level of the individual household (McCorriston and Weisberg Reference McCorriston and Weisberg2002) rather than in large public buildings such as the TC Oval.
The results of crop stable isotope analysis from the TC Oval and the theory of ‘extensive’ crop production at mid-third millennium B.C.E. Tell Brak, as suggested by functional ecological analysis, also correlate with the findings of other stable isotope studies from sites such as Tell Leilan, Hamoukar, Tell Sabi Abyad and Tell Zeidan in northern Mesopotamia. Previous work by Riehl et al. (Reference Riehl, Pustovoytov, Weippert, Klett and Hole2014) and Styring et al. (Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) has found that there appears to be no significant changes to crop Δ13C values from the Ubaid period to the Bronze Age (5300–2500 cal. B.C.E.) at all sites in this region, a factor that would be expected if crops were reliant solely on variable annual rainfall. Instead, this lack of isotopic variation indicates that there was some type of water resource management, even if this was only the strategic sowing of certain crops in better-watered soils around the site. Work by Styring et al. (Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) has also shown that wheats and pulses, throughout sites in northern Mesopotamia, tend to be better watered than hulled barley crops. This again corresponds directly with the stable isotope results from the TC Oval and indicates that the preferential treatment of some crops was a strategy to maximise yields in poor arable conditions at Late Chalcolithic and Bronze Age settlements in this region, rather than a practice unique to one site.
In terms of nitrogen crop stable isotope analysis, results from Araus et al. (Reference Araus, Ferrio, Voltas, Aguilera and Buxó2014) and Styring et al. (Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) indicate that there was a general decrease in cereal grain δ15N values through time at various sites in the Near East and northern Mesopotamia. This has been linked with an overall decrease in soil fertility due to the increased exploitation of arable areas (Araus et al. Reference Araus, Ferrio, Voltas, Aguilera and Buxó2014) and the cultivation of more marginal agricultural zones. Additionally, Styring et al. (Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) have linked the decrease in δ15N values with an equivalent increase in site size at Tells Leilan, Hamoukar, Sabi Abyad, Zeidan and Brak as populations expanded during the Bronze Age. Increased population sizes would have required the production of greater quantities of grain and the expansion of agricultural areas beyond the immediate environs of the main settlement. These ‘outfield’ plots would necessarily receive lower inputs of manure due to the difficulty associated with the transportation of this resource over long distances. The combined stable isotope and functional ecology results from the TC Oval seem to fit within this model (Styring et al. Reference Styring, Charles, Fantone, Hald, McMahon, Meadow, Nicholls, Patel, Pitre, Smith, Soltysiak, Stein, Weber, Weiss and Bogaard2017) and, as was discussed above, may indicate that wheat crops were grown within the ‘infield’ areas close to the Tell Brak main mound whilst barley was primarily cultivated within the ‘outfield’. The combination of these crops within the same storage area then would suggest that agricultural land ownership was spread over both of these disparate areas and that this spatially variable method of production may have been a risk-buffering mechanism.
Conclusion
The multi-stranded analysis of archaeobotanical material recovered from the TC Oval has provided a wealth of new information about agricultural production at Tell Brak in the mid-third millennium B.C.E. The use of traditional archaeobotanical techniques, combined with crop stable isotope analysis and functional weed ecology, has shown not only what was being grown, but also potentially how it was being grown and where. The TC Oval itself seems to have played an important administrative role at the site as a storage depot for fully processed cereal grains, as well as potentially a centre for the distribution of worker's rations. Within the building, the predominance of clean hulled barley suggests that, contrary to previous theories, hulled barley was prepared for human consumption, although this would not preclude its use as a fodder crop as well. Additionally, the identification of small-grained varieties of hulled barley and glume wheat indicates that the farmers of Tell Brak were growing a wider range of cereal varieties, including small-grained versions suited to poor growing conditions, than has been identified from other sites in the region. The use of crop stable isotope analysis has revealed that these different cereal crops were likely grown in different areas of the Tell Brak landscape to take advantage of the uneven environmental conditions. These results suggest that glume wheat crops were grown in wetter soils associated with the nearby wadis, while hulled barley, and the small barley variety in particular, were grown in drier areas with little artificial enrichment.
This use of both ‘infield’ and ‘outfield’ areas around the main city would have allowed the cultivation of large areas of land with limited labour resources while also maximising the potential arable yield by exploiting the stress-tolerant nature of barley in marginal environmental conditions. Finally, comparisons with carbon stable isotope studies from other nearby sites within the Upper Khabur plain suggest that the crops stored in the TC Oval were not imported from areas with higher precipitation levels to the north. Instead, the stored crops appear to have been grown locally, in the arable catchment surrounding the city. Overall, these results indicate that, even though mid-third millennium B.C.E. Tell Brak was located in a semi-arid region on the edge of the ‘zone of uncertainty’ (Wilkinson et al. Reference Wilkinson, Philip, Bradbury, Dunford, Donoghue, Galiatsatos, Lawrence, Ricci and Smith2014), careful agricultural management and regime choice allowed farmers to overcome environmental challenges and maximise arable production.
Open access
