Fiona Marshall & Elisabeth Hildebrand
Cattle Before Crops: The Beginnings of Food Production in Africa
Journal of World Prehistory, Vol. 16, No. 2, June 2002. pp. 99-143
Scheduled consumption and the domestication of cattle
What circumstances might have prompted North Africans to capture and domesticate cattle ? In western Asia, where studies of the domestication of cattle have a long history, scholars have emphasized religious motives (reviewed in Isaac, 1962). These theories are compatible with our emphasis on scheduling, and identification of ceremonies as settings in which domestication is likely. But the scheduled consumption model allows us to identify a broader range of settings in which capturing and domestication of cattle would have been advantageous—contexts in which people need to maintain or increase predictable access to key resources. These are most likely to arise amid ecological perturbations, such as drought or disease, or among sedentary populations clustered around concentrations of resources.We examine the influence of these factors on North African hunter-gatherers immediately prior to domestication, and develop a theory for the domestication of African cattle in the eastern Sahara.
Climate and Society Before Food Production
It is widely recognized that the more arid a region, the greater the variability in the amount, location, and timing of rainfall, both within seasons and between years (Coppock, 1993; Nicholson, 1980). Non-equilibrium systems exist in arid regions of Africa today (less than 300 mm p.a.), where the coefficient of variation in rainfall often exceeds 30%. Unpredictable rainfall causes great variation in the productivity of African savanna ecosystems (Behnke et al., 1993; Ellis and Swift, 1988). Fluctuations have been especially marked in North Africa, where regional feedback mechanisms prolong and amplify climatic perturbations such as droughts (Nicholson, 1989, 1994). The instability caused by repeated cycles of aridity is likely to have augmented concerns about predictability among North African hunter-gatherers during several key periods in prehistory.
Hyperarid conditions prevailed across North Africa during the last glacial maximum. Very dry conditions c. 20,000 BP began to ameliorate c. 12,500 BP, giving way to oscillating wet and dry conditions that resulted from major systemic changes in atmospheric circulation at the end of the last glaciation (Grove, 1993; Hassan, 1997, 2000; Nicholson and Flohn, 1980; Petit-Maire, 1991; Street and Gasse, 1981). The moist climatic regime, punctuated by a cold, dry phase c. 11,350–10,250 BP, peaked c. 9500–8500 BP. Rainfall decreased gradually across North Africa after c. 8000 BP, with regional variation in timing and severity. Wendorf et al. (2001) recognize six humid phases in the eastern Sahara between c. 9500 and 3800 BP. During wetter periods, much of the Sahara was covered with grasslands, the mountain ranges of the Sahara supported mediterranean vegetation, and water levels were high in lakes and rivers (Fig. 1). The eastern Sahara, however, was arid at all times, with severe droughts c. 9500 BP and c. 8700–8600 BP (Hassan, 2000; Wendorf et al., 1984, 2001). The Saharan region saw many short arid phases, and two marked ones c. 8000/7500–7000/6500 BP and 4500– 3000 BP (Grove, 1993; Hassan, 1997). More localized droughts occurred c. 9500–9000 BP at Adrar Bous in northwest Niger, and sporadically between c. 8500 and 7000 BP in the Chad Basin (Barich, 1998). The effects of these fluctuations on subsistence would have been especially pronounced in areas that consistently received scant rainfall, such as the eastern Sahara.
Changes in hunter-gatherer technology and social organization also created contexts that favored domestication. During the last glacial maximum, the Sahara was deserted and population was dense in the NileValley. In some cases, as at Wadi Kubbaniya c. 17,000 BP, settlement remained in one place for several seasons of the year and resource use was intensive. Plants were harvested in ways that may have increased their abundance and diversity, and processed on grindstones. Occupants of such sites also fished, and hunted wild cattle, hartebeest, and gazelle (Wendorf et al., 1989). But these activities did not result in domestication. Later, c. 13,000–12,500 BP, inhabitants of the Nile Valley used plants in similar ways at Tushka, where grindstones are common. At this site, burials with skulls of wild cattle suggest that these animals had symbolic significance prior to domestication (Wendorf, 1968; Wendorf and Schild, 1976).
After being deserted during the last glacial maximum, the Sahara was repopulated c. 9500 BP by hunter-gatherers who used ceramics with distinctive wavy-line decorative motifs. This cultural complex, scattered across North Africa, is variously referred to as Khartoum Mesolithic (Arkell, 1949), Epipaleolithic (Close, 1995), or Aqualithic (Sutton, 1977). Sites are concentrated in relatively well-watered massifs and lake basins. Some hunter-gatherers were fairly sedentary and harvested wild plants intensively, especially cereals. At Ti-n-Torha East rockshelter in the Acacus, people built stone structures (Barich, 1987). At this and nearby sites of Uan Afuda and Uan Tabu, they gathered and ground wild grasses, and hunted and possibly managed Barbary sheep (Ammotragus lervia) (Barich, 1987; Cremaschi et al., 1996; Di Lernia, 1999, 2001; Garcea, 2001). Farther south in the Khartoum Nile, sites of the tenth to ninth millennia BP preserve abundant ceramics, as well as wild ungulates, fish, molluscs, grindstones, and wild grasses (Caneva, 1988; Haaland, 1987; Magid and Caneva, 1998; Peters, 1986). With concepts of ownership based on storage facilities and ceramics, some of these groups in the central and southern Sahara probably followed a delayed-return strategy of hunting and gathering (Barich, 1998; Dale et al., in press; Di Lernia, 2001). Farther south in the Khartoum Nile, sites of the tenth to ninth millennia BP preserve abundant ceramics, as well as wild ungulates, fish, molluscs, grindstones, and wild grasses (Caneva, 1988; Haaland, 1987; Magid and Caneva, 1998; Peters, 1986). With concepts of ownership based on storage facilities and ceramics, some of these groups in the central and southern Sahara probably followed a delayed-return strategy of hunting and gathering (Barich, 1998; Dale et al., in press; Di Lernia, 2001).
Fig. 1. Rainfall and physical features of North Africa. Present-day isohyets are shaded; the 200 mm isohyet is widely recognized as the border between the Sahara (desert) and Sahel (grassland). Estimated Sahara–Sahel boundaries for 9000 and 18,000 BP are also shown. Information from Banks (1984), Gautier (1987a), Goudie (1996), and Petit-Maire (1989). These and all other radiocarbon dates are uncalibrated.
Domestication of Cattle in North Africa: Timing and Location
Cattle were the earliest domesticates in Africa. Recent studies suggest that they were probably domesticated from North African populations of wild Bos primigenius by hunter-gatherers of the eastern Sahara 10,000–8000 BP. Their origins are still controversial, but Gautier (1980, 1987a, 2001) and Wendorf (Close andWendorf, 1992;Wendorf et al., 1984, 2001;Wendorf and Schild, 1980) argue for domestic cattle in the eastern Sahara at Bir Kiseiba c. 9500 BP, and Nabta Playa c. 8840 BP (Figs. 1 and 2). Because these sites preserved few cattle bones, evidence for morphological change is difficult to evaluate (Grigson, 2000; Smith, 1986).Wendorf and colleagues buttress the admittedly scarce morphological data with an ecological argument: without human intervention, survival of wild cattle in the arid eastern Sahara would have been unlikely (Close and Wendorf, 1992; Wendorf et al., 1984; Wendorf and Schild, 1998). Cattle are present to the west at Enneri Bardagu´e in the Tibesti by c. 7400 BP and in the Acacus by c. 7400–6700 BP (Garcea, 1995; Gautier, 1987a). There are no early domestic cattle in the Nile Valley.
Recent morphological and genetic research provides some support for Wendorf's hypothesis. Grigson's morphological study (Grigson, 1991, 2000) shows that Egyptian cattle of the fifth millennium BP had long, slender limbs morphologically distinct from those of Eurasian humpless cattle (Bos taurus) and Zebu (Bos indicus). On this basis, she suggests that African cattle may have been domesticated from wild Bos primigenius in Africa. Recent research on genetic variation in breeds of present-day African cattle points to a similar conclusion (Bradley et al., 1996; Bradley and Loftus, 2000; Hanotte et al., 2002).
Genetic distance between African cattle and Asian Bos taurus is sufficient to define the two as discrete genetic populations (Bradley et al., 1996; Bradley and Loftus, 2000). Bradley and colleagues argue that wild cattle in Eurasia, Bos primigenius primigenius, and in Africa, Bos primigenius opisthonomus, diverged by 22,000 years ago, and propose that African populations of wild cattle were domesticated in Africa.
Together, archaeological and recent genetic evidence indicate a single geographic origin for domestic cattle in the eastern Sahara (Gautier, 1987a; Hanotte et al., 2002).
The earliest domestic sheep and goat in Africa appear c. 7000–6700 BP in the eastern Sahara and the Red Sea Hills (Close, 1992, 2002; Gautier, 1987a; Vermeersch et al., 1996). They almost certainly come from western Asia (Gautier, 1984a), because there are no wild ancestors for sheep and goat in Africa. Close (2002) argues that sheep and goat came to Africa via the southern Sinai before Near Eastern crop complex, which is thought (Wetterstrom, 1993) to have entered the continent through the Nile Valley. The fact that sheep and goats postdate domestic cattle is further evidence for indigenous domestication of cattle in North Africa. More archaeological data are needed, however, including large, well-dated faunal samples, enclosures with cattle dung, and sites that immediately predate domestic cattle in the eastern Sahara. The origins of African cattle are thus still controversial, but genetic and morphological data together strongly suggest independent domestication in Africa.
Given the sparseness of faunal data from Nabta and Kiseiba and the absence of slightly older sites, specific application of the scheduled consumption model to small-scaled analyses of sites or household economies at the time of domestication is not yet possible. Application of the scheduled consumption model in more general terms is worthwhile, however, because it allows us to explore specific relations between short-term predictability and resource intensification in North Africa.
Domestication of Cattle and the Scheduled Consumption Model
Archaeological evidence suggests that cattle were domesticated in the eastern Sahara during the tenth millennium BP. Nabta, located in the driest part of the Sahara, received too little rainfall at this time (less than 300mm p.a.) to sustain wild cattle. Domestication probably took place slightly farther west, in areas capable of supporting cattle. In marginal areas of the eastern Sahara, human populations concentrated in playa basins (Close, 2001; Close and Wendorf, 1992; Wendorf et al., 1984, 2001; Wendorf and Schild, 1998) or massifs. We argue that hunter-gatherers in these settings domesticated cattle to ensure their predictable availability as a food source. Both ecological perturbations, especially recurring cycles of aridity, and concentration of resources and people in playa basins, could have precipitated an increased need for predictability. Ritual use of cattle may also have provided a specific context in which scheduled consumption would have been especially desirable.
Given that aridity is a recurring theme in North Africa, the question remains: why did domestication occur c. 9000 BP in marginal circumstances, rather than amidst arid conditions c. 17,000 BP, when human populations concentrated in the Nile Valley? We think that subtle variation in rainfall is more important to day-to-day predictability than acute aridity. During the last glacial maximum, environments were so extreme that conditions were quite predictable in most places. The Sahara was predictably dry, and uninhabited. People lived in the Nile Valley, close to water, where other resources were also dependable. In contrast, conditions in parts of the eastern Sahara c. 9000 BP were only marginal. There was sufficient rainfall for arid-adapted resources, but not enough for their distribution to be reliable. Although humans could survive under these circumstances, planning was difficult because the time and place of rainfall and herd movements could not be foreseen.
Another factor that differentiates the tenth millennium from earlier arid periods is the presence of ceramic-using, delayed-return hunter-gatherers in North Africa. Rights to resources and concepts of ownership associated with such groups are important preconditions for herding (Brooks et al., 1984; Meillassoux, 1972; North, 1981). Herding presents more scheduling conflicts for hunter-gatherers than cultivation, because animals, unlike plants, cannot be left for more than a few hours at a time (Marshall, 2000).Without concepts of ownership, individuals would have been unlikely to contribute the extra labor necessary for herding, because any member of the group could have slaughtered an animal at any time.
Like the “perfect storm,” the precise conditions that precipitated domestication occurred rarely in North Africa. They converged during the tenth millennium in areas of the eastern Sahara that were wet enough for wild cattle but dry enough to be risky, and were populated by hunter-gatherers with social organization conducive to resource intensification. Archaeological, genetic, and climatic evidence together suggest that domestic cattle spread from a point origin—perhaps a small playa near the Jebel Marra massif in northwest Sudan, or east of the Tibesti in northeastern Chad— during the tenth–ninth millennium BP.
Hunter-gatherers of this region faced a nonequilibrium rainfall system with generally unpredictable rainfall, as well as short-term climatic fluctuations. We think that these circumstances were challenging, not in terms of absolute abundance of food, but in terms of the predictability with which food could be acquired. In this setting, where plant abundance and prey mobility varied stochastically, the scheduled consumption model predicts manipulation of favored resources during arid episodes.We suggest that local hunter-gatherers intensified their use of wild animal herds rather than their harvesting of wild plants for several reasons. Plant productivity is especially vulnerable to variation in rainfall, because the timing of rainfall relative to plant growth phases is crucial (Le Houérou et al., 1988; Mortimore, 1998). During droughts, ungulates are a more reliable resource than plants because their populations are maintained through movements that exploit local differences in topography, vegetation, and rainfall (Behnke et al., 1993). Following wild ungulates would have been a particularly attractive strategy for hunter-gatherers of the southeast Sahara during the tenth–ninth millennia BP, who faced variability in the amount, location, and timing of rainfall. The alternative, increasing mobility combined with more generalized use of plants, might not have been possible. Generalization would have carried the risk of lowered foraging efficiency (Winterhalder, 1986, p. 378), and the plant component of the diet was already generalized, requiring use of relatively low-rank resources such as grains, which necessitated cumbersome grindstones for processing. Following resilient herds of large wild ungulates, such as cattle or hartebeest, would have reduced risk and constituted generalization by proxy, because the animals processed more diverse plant resources than humans could.
In such arid conditions, however, locating herds of ungulates would have been difficult, and access to animal products would have remained unpredictable because of erratic rainfall and the high mobilities and low densities of wild herds. Sporadic access to herds would have impeded attempts to monitor herd size, composition, nutritional status, and the effects of disease and carnivore predation. This would have limited knowledge of their condition, and made scheduled consumption events (from large ceremonies to daily family meals) difficult to plan. Such scheduling would have been especially important to delayed-return hunter-gatherers of the early Holocene in the Sahara, because broad social networks, consolidated by periodic gatherings, would have helped to spread risks in an uncertain environment. A ceremonial role for cattle at such gatherings would have provided social, as well as dietary, motivation for humans to achieve or maintain predictable access to cattle through control of herds.
Cattle would have been the logical focus of intensification for many reasons. Wild cattle would have been the main meat source for Saharan hunter-gatherers (Hassan, 2000). They were grassland-adapted herd animals (Gautier and van Neer, 1989), and would have been the easiest large North African ungulate to tame. Barbary sheep are territorial and found in small groups, and gazelle and other antelope are notoriously difficult to domesticate (Clutton-Brock, 1981; Diamond, 1997; Haltenorth and Diller, 1988, p. 105; Lewis, 1977), whereas cattle have proven amenable to domestication multiple times in different parts of the world (Bradley et al., 1996; Grigson, 1989; Meadow, 1996). This is probably due to their size (energetic efficiency), rapid growth, and behavioral characteristics (ease of breeding in captivity, lack of territoriality, and well-developed dominance hierarchy) (Clutton-Brock, 1981; Diamond, 1997). Tame animals could have been controlled by corralling them overnight in brush enclosures. Provisioning (wells or salt), taming, bleeding, milking, and selective breeding would have followed. Taming cattle and protecting them from predators at night would have required substantial commitment of labor, however (Marshall, 2000), and the sustained effort required is likely to have occurred only among delayedreturn hunter-gatherers with storage technology (ceramics) and concepts of ownership.
We have stressed the importance of predictability in the domestication of cattle. During droughts, yield and predictability both would have declined, however. Might people have corralled cattle to increase yield rather than predictability? Precise figures for the number of large animals that hunter-gatherers would have been able to kill under these circumstances are difficult to obtain, but a !Kung hunter in a similarly semiarid environment kills up to five large animals per year (Lee, 1979, p. 231). Hadza hunters kill substantially more animals: six to nine large animals per hunter per year (O'Connell et al., 1992). In contrast, African pastoralists today harvest only 4–8% of the cattle herd a year; in areas where drought and disease are common, this represents the surplus to the growth needs of the herd (Dahl and Hjort, 1976). To kill as many animals a year as does a single !Kung hunter would require keeping a herd of at least 125 cattle, an improbable scenario in the early stages of domestication. Given the slow rate of growth of cattle herds, and the fact that early domesticates are smaller, not larger, than their wild ancestors, it is unlikely that cattle were domesticated to increase yield.
We hypothesize that delayed-return Saharan hunter-gatherers of the tenth–ninth millennium followed herds, and subsequently domesticated cattle in order to increase day-to-day predictability and reduce risk by manipulating a resource that could move to exploit localized favorable conditions. Controlling movements diminishes the risk of not finding animals, allows evaluation of condition and predictable access, and creates a dense, movable concentration of resources. Day-to-day control of herds and subsequent domestication of cattle also fit with well-known strategies for increasing longterm predictability: storage, mobility, and sharing (Halstead and O'Shea, 1989). Keeping cattle is a form of storage on the hoof (Close and Wendorf, 1992; Legge, 1989), and mobility and sharing of resources such as water and grazing are probable features of early pastoralism in the Sahara. Nascent cooperative social and political links among far-flung early herding groups would have passed on information and established safety networks (Legge, 1989; Robinson, 1989; Ryan et al., 2000). These could have developed through resource sharing, stock loans, ceremonies, and other social bonds. Cattle domestication in North Africa established mobile herding, and pastoralism rather than settled agriculture, as the earliest form of food production. In the following sections, we discuss the wide-ranging influence of herding on the subsequent spread and development of food producing economies in Africa.