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
The spread of food production in Africa: why so patchy?
The spread of food production on the African continent was strikingly uneven: hunter-gatherers and food producers coexisted in all regions long after initial contact or colonization by food producers. This was due to different factors in different regions within Africa, most of which related ultimately to the relative predictability of herding versus hunting and gathering. There was variation in the timing as well as the spread of early food production in Africa. Herding spread rapidly but incompletely across the Sahara during the mid-Holocene. As climatic conditions deteriorated during the sixth millenium BP, pastoralists moved south to better-watered regions, such as the Sudanese Nile. Small groups of pastoralists moved into eastern Africa slightly later. In southern Africa, early food production is largely associated with movements of Iron Age mixed farmers into the region c. 2000 years ago. In this section, we focus only on regions that are central to understanding continent-wide variability in pathways to food production. Parts of the western Sahara and the African rainforests that are important to the domestication of plants—and where livestock adoption follows patterns similar to those in other regions—are discussed in a subsequent section on African plant domestication.
Northeast and Northwest Africa
Pastoral occupation at Nabta c. 8000 BP provides an early example of a characteristic African pattern of early food production. The settlement is highly structured but mobile and seasonal, rather than village-based. Site E75–6 has at least two rows of hut floors with associated cooking holes, storage pits, many grindstones, and two large wells that could have been used to water herds of cattle (Banks, 1984; Close andWendorf, 1992;Wendorf and Schild, 1998). A wide range of wild plants were collected. Use of wild sorghum may have been intensive (Wasylikowa et al., 1993, 1997), but E75–6 was only occupied seasonally (Close andWendorf, 1992; Kro´ lik and Schild, 2001).
As rainfall became lower and more variable in North Africa, pastoralism spread unevenly from the eastern Sahara to the Tibesti and Acacus massifs and the west African Sahara between c. 7000 and 5000 BP (Fig. 2). Even after the introduction of stock, the new subsistence economy was generalized, and Saharan pastoralists hunted and sometimes fished (Clark et al., 1973; Gautier, 1987a; Smith, 1980). Early Holocene patterns of plant use also persisted at Uan Muhuggiag and Adrar Bous (Smith, 1980;Wasylikowa, 1993), although smaller quantities of grindstones at some sites suggest less processing of wild plants than previously (Barich, 1987). Localized, unpredictable rainfall necessitated long-distance movements to exploit variable topography and vegetation for water and pasture (Muzzolini, 1993; Smith, 1992a). Herders at Adrar Bous and Acacus sites were more mobile than earlier hunter-gatherers (Barich, 1987; Gautier, 1987b; Smith, 1992a). Upper levels at Torha North and Uan Muhuggiag preserve discontinuous laminae of dung, suggesting seasonal use of the shelters as livestock pens (Cremaschi et al., 1996; Gautier and van Neer, 1977– 82). Although herders seasonally occupied large sites near playas at Nabta and Dakhleh c. 7900–5500 BP (Close, 1992; McDonald, 1998a), ephemeral sites are more common throughout the Sahara (Barich, 1998; Close, 1990; Gabriel, 1987). Pastoral rock art attests to the symbolic importance of domestic cattle (Holl, 1999; Muzzolini, 1993; Smith, 1992), as do cattle burials c. 6500 BP near Adrar Bous and at Nabta (Applegate et al., 2001; Malville et al., 1998; Paris, 2000; Wendorf and Królik, 2001; Wendorf and Schild, 1998). Rituals associated with cattle may have occurred at seasonal meetings of pastoral groups or lineages, and helped to consolidate emerging social and political networks.
Highly mobile pastoral land use minimized competition with huntergatherers. Even after herding was widespread, hunter-gatherers lived near pastoral groups until c. 7000 BP in the eastern Sahara at Dakhleh Bashendi-A (McDonald, 1998a), and until c. 5500 BP in the west-central Sahara at Amekni (Camps, 1968). Spatial variation in conditions made sedentary hunting and gathering unsustainable in different places at different times, and contributed to the patchy spread of herding.
After c. 6500 BP, rainfall decreased. Broader pastoral contacts and exchange systems (McDonald, 1992; Smith, 1980, 1992a)mayhave buffered deteriorating conditions. Green vitric tuff and Amazon stone were traded thousands of kilometers (Clark, 1970; McDonald, 1992). Nevertheless, successful herding was short-term in some places: in the late-middle sixth millennium BP pastoral occupation ceased, at least temporarily, at Dakhleh Bashendi- B, Abu Ballas, and in the Nabta-Kiseiba area (McDonald, 1998a,b). Some herders responded to increasing aridity by emphasizing small stock, increasing mobility (Gautier, 1987a), and moving south to better-watered areas. Much of the eastern Sahara was depopulated by c. 5500 BP, and central Saharan montane sites such asUanMuhuggiag were abandoned by c. 5000 BP (Barich, 1998; Close, 1992). Herders entered the Khartoum Nile by c. 5500 BP (Gautier, 1987a; Peters, 1986), the west African Sahel by c. 4500–3500 BP (Breunig et al., 1996; Holl, 1998; MacDonald and MacDonald, 2000; Smith, 1992a), and the forest margin by c. 3000 BP (Stahl, 1985; van Neer, 2000).
The Sudanese Central Nile and Greater Eastern Africa
Chronology and relations between Sudanese and Saharan areas (Paris, 2000; Smith, 1992a) suggest that domestic stock were introduced from the Sahara as it became drier (Haaland, 1992; Hassan, 1997). Cattle, sheep, and goats appear by the sixth millennium BP (Gautier, 1984b,c; Peters, 1986) (Fig. 2). Local assemblages of lithics and ceramics show continuity (Caneva, 1987, 1989; Haaland, 1995; Marks and Mohammed-Ali, 1991), indicating that any movement of Saharans into the region was small-scale, and culture contact was more important than migration to socioeconomic change.
Entry of Saharans may have been eased by prior social links with the Sudan, indicated by trade and common ceramic styles. Compared to the original Saharan herding environments, the Sudanese Nile offered more dependable, productive resources. This area also posed no particular problems for cattle, as it lies within their wild range. Like earlier local hunter-gatherers, pastoralists used large, semipermanent camps near the Nile, as at Esh Shaheinab and Geili (Caneva, 1988; Haaland, 1995; Krzyzaniak, 1991). Domestic animals are the dominant large mammals at many sites, such as Kadero c. 5000–4000 BP, but were added to a wide range of wild animals used by earlier hunter-gatherers (Gautier, 1984c; Haaland, 1992). Unlike Saharan pastoralists, herders in this better-watered landscape are thought to have used plants more intensively than their hunter-gatherer predecessors. Site structure and increased use of grindstones at Kadero 1, UmDireiwa, and Zakiab indicate to Haaland (1981, 1992) that, as early as 5000 BP, pastoral groups were cultivating sorghum that was morphologically wild (Stemler, 1990).
Social differentiation appeared among Sudanese herders by the sixth millennium BP. Clusters of especially rich graves of men, women, and children at Kadero 1 argue for differences in wealth (Krzyzaniak, 1991), but there is no evidence for social stratification. Pastoral intensification and a decrease in wild animal use is also evident at some sites in the Middle Nile after 5300 BP. Despite these developments, the spread of herding was patchy: at Shaqadud, east of the Nile, subsistence focused on wild resources as late as 4000 BP (Marks and Mohammed-Ali, 1991; Peters, 1991). Farther to the east near the Eritrean border, cattle and small stock appear at Atbai sites during the fifth and fourth millennium BP (Fattovich, 1993; Sadr, 1991, pp. 53, 138).
As herders continued to spread east and south of the central Nile, they moved beyond the natural distribution of wild cattle in North Africa and encountered new environmental and epizootic challenges. The earliest domestic cattle in the Horn of Africa date to c. 3500–2500 BP at Lake Besaka and Gobedra (Brandt, 1984; Phillipson, 1977). They spread slowly because of the vertical relief and closed woodlands of highland areas. Dramatic rock paintings of cattle herds in the Horn probably date to this period, and may reflect ceremonies and seasonal gatherings of frontier pastoral groups (Brandt and Carder, 1987; Joussaume, 1981).
Before 4000 BP, small numbers of herders migrated into Kenya from increasingly arid areas of Sudan and Ethiopia (Fig. 2), but herding was not widespread until c. 3000 BP, and extended only to northern Tanzania. Arid conditions c. 6000–3300 BP (Ambrose, 1998) and wild animal diseases (Gifford-Gonzalez, 2000) may have slowed the spread of herding in southern Kenya, and made cattle a less predictable source of food than in more northern areas. Cattle moving into areas with wildebeest and buffalo were exposed for the first time to Bovine Malignant Catarrhal and East Coast Fevers. In this frontier context, the low density of herders would have made seasonal aggregations more important, because it would have constrained other mechanisms for risk reduction, such as intergroup exchange networks, stock loans, and gifts (Gifford-Gonzalez, 1998, 2000), as well as the availability of breeding stock.
Hunter-gatherers of the fifth millenium BP near Lake Turkana and in central Kenya are thought to have added herding to local hunting or fishing strategies, because lithics show continuity with earlier East African traditions (Ambrose, 1984a; Bartheleme, 1985). Use of the Pastoral Neolithic funerary complex at the northern Kenyan site of Jarigole (Nelson, personal communication, 1998; Gifford-Gonzalez, 2000) may have reinforced extensive social networks among dispersed early pastoral groups. Farther south in the central Rift Valley, the earliest domestic stock are found at very low densities in a hunter-gatherer occupation, RBL2.1, c. 4000 BP at Enkapune Ya Muto rockshelter (Marean, 1992).
A mosaic of pastoral and hunter-gatherer groups coexisted in southern Kenya and parts of northern Tanzania from >4000 BP onwards. After 3500 BP, two distinct specialized pastoral cultures emerged: the Elmenteitan at sites like Ngamuriak, and the Savanna Pastoral Neolithic at Narosera and Crescent Island Main (Bower, 1991; Robertshaw, 1990). Both cultures relied on intensive use of livestock, and made little use of abundant wild ungulates (Gifford-Gonzalez, 2000; Marshall, 2000). At the site of Enkapune Ya Muto, contemporary Eburran 5 hunter-gatherers had lithic technology and microlith styles similar to those of earlier hunter-gatherers, and consumed large quantities of wild fauna and limited stock (Ambrose, 1984b). The few domestic animals are attributed to gifts from pastoral neighbors, raiding, or limited herding (Marean, 1992). Nderit ceramics similar to those found on pastoral sites also attest to interaction between Eburran hunter-gatherers and nearby herders (Ambrose, 1998).
Pastoral use of the landscape was mobile and extensive, did not destroy hunter-gatherer habitat, and allowed local hunter-gatherer subsistence and social organization to continue (Gifford-Gonzalez, 2000; Marshall, 1986, pp. 248–249). Gifford-Gonzalez (1998) argues that a likely pastoral strategy for reducing the risk of moving into new areas would have been to integrate with local hunter-gatherer groups, perhaps through marriage alliances. In this way, some hunter-gatherers would have assimilated into herding groups, but herders could fall back among hunter-gatherer groups in case of stock loss to drought or disease. Social and economic systems continued to be fluid in the central Rift Valley until recent times: hunter-gatherer and pastoral groups interacted regularly, some hunter-gatherers adopted food production, and pastoralists periodically suffered stock losses (Marshall, 1994; Mutundu, 1999; Spear and Waller, 1993).
Adoption of food production in southern Africa followed a different trajectory from that in the northern half of the continent. Although small groups of pastoralists are thought to have entered parts of southern Africa early, disease was a barrier to domestic stock. The introduction of domestic animals and plants is closely associated with the rapid spread of Early Iron Age farmers c. 1600 BP. Mixed agriculture became common in the extreme east, and herding in the west of South Africa. South of the Orange River, hunting and gathering continued across the interior of the subcontinent.
The earliest pastoralists may have been so mobile and patchily distributed as to be archaeologically invisible in some places. It has long been thought that early herders spread from Zimbabwe and Zambia south, and that stone-using Khoisan groups may have brought sheep and pottery from the Zambezi through Namibia to the Cape by 2000 BP (Klein, 1984; Smith, 2000). Some scholars suggest, however, that ceramics and small stock both spread south through exchange networks between hunter-gatherers and Iron Age farmers, rather than via Khoi migration (Mitchell, 1996; Sadr, 1998). There are few sites dating to this period, but sheep are directly dated at Spoegrivier in Namibia c. 2105 BP, and Blombos Cave in the southern Cape c. 2000 BP (Henshilwood, 1996; Sealy and Yates, 1994, 1996) (Fig. 2). Faunal evidence from other sites, previously thought to indicate early herding, may in fact be more recent. Pastoral sites occur at low densities, and show a generalized subsistence based on sheep herding and variable use of wild animals and plants.
Early Iron Age groups in southern Africa after 2000 BP lived in fairly permanent settlements and relied on livestock, especially sheep and goats, and African grains and pulses (Maggs, 1984). Cattle became more numerous after c. 1500 BP, but fish, molluscs, tortoises, and wild mammals continued to be important (Plug and Voigt, 1985; Voigt, 1987). Mixed farming was not continuously distributed across far southern Africa as it was just south of the equator. Because of livestock diseases, winter rainfall, and desert areas, farming was confined to the eastern half of South Africa.Away from the continental margins, immediate-return hunter-gatherers continued to exploit flexible and predictable resources such as tubers and mongongo nuts, which were more evenly distributed in space and time than were cereals in North Africa. In areas where hunter-gatherers and food producers coexisted, they had variable relations, ranging from trade (Bousman, 1998; Kinahan, 1996; Smith et al., 1991, 1996;Wadley, 1996) to clientship (Denbow andWilmsen, 1986; Schrire, 1992). Over much of southernmost Africa, agriculture was not adopted until recent times (Deacon, 1984a,b; Smith, 1992a).
Pathways to food production in northern and southern Africa offer an interesting contrast. Diamond (1997, p. 132) suggests that southern regions lacked a critical mass of potential domesticates, and that domestic plants spread south slowly because of Africa’s north–south axis. Although the ranges of wild sorghum and rice extend into southern Africa (Harlan, 1992a), Diamond notes that cattle and most African cereal crops are not found wild south of the equator. Factors affecting continental patterns of rainfall also differ between Sahel and southern Africa in ways that may have affected pathways to food production. Despite significant arid areas within southern Africa, the subcontinent as a whole is generally cooler and wetter, and lacks the local feedback mechanisms that prolong droughts in the Sahel (Nicholson, 1989, 1994). Increasingly unpredictable rainfall and the resultant stresses on Saharan hunter-gatherer groups during the Holocene may not have had parallels in the Southern Hemisphere.
Another interesting contrast with northern Africa is that southern regions have no evidence for early ceramic use, and little for delayed-return subsistence strategies (but see Sadr, 1998). Digging sticks and digging stick weights for harvesting underground storage organs are found in the archaeological record from early periods (Deacon 1984a,b). Holocene huntergatherers in southern Africa were able to exploit high-ranked tubers and nuts, resources that are consumed immediately. Thus, factors that promote use and storage of relatively low-ranked resources such as wild grasses may not have operated in southern Africa. Finally, the southward spread of a fully developed Early Iron Age agricultural complex made the domestication of local plants less likely than in other parts of Africa, where early herders without domestic plants were the first food producers in many regions.
Why So Patchy?
As during the initial domestication of cattle, concerns about predictable access to animal products would have shaped decisions to adopt livestock, or to move them to new locales. The relative predictability of herding versus hunting and gathering varied from place to place, however, depending on local food resources, terrain, diseases, and social settings.
African cattle were well-adapted to the grasslands of north Africa that constituted the original wild range of Bos primigenius: wetter parts of the Sahara and northern Sudan.There, pastoralism spread quickly because cattle allowed hunter-gatherers to use a broad range of vegetation types, intensifying subsistence while exploiting localized environmental instability. Herding was more sustainable than sedentary hunting and gathering in an arid landscape with unpredictable resources, and social relations among huntergatherer groups were well established.Trading networks between the Sahara and the Sudanese Nile had long existed, and in the Sudan cattle had access to a dependable water source. In contrast, herders moving into eastern Africa had no previous relations with local hunter-gatherers, and limited access to other herders, breeding stock, and social safety networks. Eastern Africa also lies outside the wild range of cattle, and wildlife diseases made livestock a less predictable source of food. In part because of these diseases, pastoralism was not widespread farther south until it was integrated with crop cultivation.
Pastoralists tend to cope with climatic variability by exploiting spatial heterogeneity, rather than by modifying the landscape. Early herders thus affected hunter-gatherer resources less than settled agriculturalists would have done, and extensive, shifting land use allowed pastoralists and huntergatherers to coexist, contributing to the continuation of hunting and gathering. Furthermore, in many areas, herding offered no particular advantages over existing strategies of hunting and gathering, but would have required much more labor. Despite environmental deterioration in the Sahara, hunting and gathering persisted for millennia after stock became widely distributed. The Sudanese Nile offered reliable access to plants and fish without investing in herding. The abundance of game in East Africa would have offered higher returns from hunting than herding. In eastern and southern Africa, the sustained commitment of time and labor required by herding would not have fit well with preexisting immediate-return hunting and gathering strategies supported by the region’s comparatively predictable nuts, fruits, tubers, and game. In Africa, hunting and gathering continued both as an independent strategy and as a component of generalized pastoralism. This pattern contrasts with the spread of farming in many other regions of the world.
Despite the success of herding as the earliest form of food production in Africa, several characteristics of herding (as compared with agriculture) contributed to the uneven spread of food production, and ensured the continuation of hunting and gathering groups. Pastoralists are mobile, with relatively low population densities. They are quite specialized, depending mainly on three domestic species. In addition, the continous labor of herding presents more scheduling conflicts than farming does for hunter-gatherers adopting food production: domestic plants can be more easily left than animals (Marshall, 2000). The form of the earliest food production in Africa, pastoralism, and the spatial variation in relative predictability of hunting and gathering versus food production, both contributed to the patchy spread of food production so distinctive of Africa.