cal agendas: the (assumed) pastoralism of the ancient tribes of Israel is a force ..... and lead extraction in Roman times (there is a reference to Christian slaves.
A tale of two deserts: contrasting deserti® cation histories on Rome’s desert frontiers Graeme Barker
Abstract Understanding ‘desertication’, the ecological deterioration of arid lands into semi-deserts and deserts, is critical for the future well-being of the fth of the world’s population who inhabit drylands today. Many dryland regions now sparsely occupied by mobile or semi-mobile pastoralists have archaeological remains indicative of much more intensive systems of settlement and land use in the past, so archaeology has frequently been brought into debates about the respective roles of climate and people in causing desertication. While much of this work has been speculative and (in terms of modern ecological theory about drylands) rather simplistic, this paper presents a comparative study of Roman-period settlement and land use on the desert fringes of Roman Africa and Roman Arabia. The two regions provide contrasting histories of dryland farming that differed in terms of: their origins, development, and abandonment; the social and economic contexts of imperialism in which they functioned; farmers’ perceptions of and responses to degradation; the scale of impact of land use on landscape; and the long-term success or failure of the strategies developed to manage processes of desertication. These very different trajectories illustrate how landscape archaeologists can contribute to understanding processes of desertication in the past, and to the desertication debate more generally.
Keywords Desertication; landscape archaeology; oodwater farming; mining; Roman imperialism.
Archaeology and desertication debates Ever since Aubreville coined the term in 1949, ‘desertication’ has been widely used by geographers to signify the ecological deterioration of arid lands into semi-deserts or deserts: McGregor and Nieuwolt (1998), for example, dene desertication as the process by which dryland conditions are brought into areas where such conditions did not previously exist. If hyper-arid zones which cannot be susceptible to further desertication are excluded from the calculation, almost 40 per cent of the world’s drylands today can be World Archaeology Vol. 33(3): 488–507 Ancient Ecodisasters © 2002 Taylor & Francis Ltd ISSN 0043-8243 print/1470-1375 online DOI: 10.1080/00438240120107495
A tale of two deserts
489
described as ‘desertied’, 16 per cent of the overall land area (Le Houerou 1996). However, the term ‘desertication’ has been used in different ways by different scholars in terms of the causal processes its use represents. Some authors have used the term to refer to land degradation caused by climatic change (sustained aridication), but most prefer to use it to refer to the effects of human actions (Fantechi and Margaris 1986). Thomas and Middleton (1994: 9–10), for example, dene desertication as ‘land degradation in arid, semi-arid, and dry sub-humid areas resulting mainly from adverse human impact’. Climate and people can of course work in tandem to degrade dryland environments, as may be the case in the context of global warming today (Le Houerou 1996; Millington and Pye 1994). Drylands support over a fth of the world’s population, and arid and semi-arid lands together support over a third, so understanding the long-term effects of climate and people on arid lands is certainly critical for the future well-being of a large proportion of the world’s population. According to Tolba and el-Kholy (1992), the current rate of desertication is about 60,000 km2 a year, 0.11 per cent per year of the world’s dryland regions, a calculation which would signify that over 70 per cent of the world’s drylands, amounting to a quarter of the world’s land surface, are under threat. The rich archaeology of many of the world’s drylands has commonly been brought into the debates about the causes of desertication, because many dryland regions now sparsely occupied by mobile or semi-mobile pastoralists have archaeological remains suggesting that these same arid and degraded landscapes must once have supported settlement of a very different order of intensity (Barker and Gilbertson 2000). This is nowhere more so than in the case of the archaeology of the southern and eastern frontier regions of the Roman empire, in North Africa and the Levant (Plate 1). Following on from the speculations of early European travellers and explorers (e.g. Barth 1857; Musil 1907; Palmer 1872), archaeologists, historians and historical geographers have frequently debated the implications of the spectacular Roman ruins of these regions for understanding processes of environmental change. Was the ‘greening of the desert’ in Roman times essentially a matter of new know-how, of the introduction of sophisticated technologies and cropping systems? Or was climatic change the key factor, more abundant rainfall allowing crops to be grown by Roman farmers in areas where this is impossible today? And what caused the green elds of antiquity to become the degraded and denuded landscapes of today? Had a return to aridity caused arable farming to be abandoned? Had people sowed the seeds of their own destruction through their greed and stupidity, stripping the landscape for building timber and fuelwood and/or allowing their animals to overgraze the vegetation (Hughes with Thirgood 1982)? Or perhaps had factors such as moral or economic decline caused these societies to abandon maintaining their land in good order? Had unsuitable or over-intensive systems of irrigation ruined soil fertility through salinization? Most of this debate has been characterized more by speculation than evidence-based discussion, usually producing deceptively attractive and straightforward models of change in which agency can be clearly recognized as either climatic change or ‘human impact’, all grist to the mill of popular media treatments of desert archaeology. In many countries, too, debates about past desert settlement have been and are still shaped by present-day political agendas: the (assumed) pastoralism of the ancient tribes of Israel is a force for good in the Old and New Testaments, for example, yet it was ‘destructive Arab pastoralists’ (again,
490 Graeme Barker Plate 1 A Roman-period fortied farm and its ancillary buildings at Ghirza in the Tripolitanian pre-desert, Libya, in an area of less than 25mm of rainfall today. (Kite photograph by G. D. B. Jones).
assumed, and in this case wrongly so: Ballais 2000; Rosen 2000) that destroyed the Roman urban societies of the Near East and North Africa. Likewise, the farming systems of indigenous peoples in North American drylands have been characterized variously as destructive or benign according to changing colonial and post-colonial agendas (Minnis 2000). The likelihood that many if not most archaeological and historical models of dryland settlement history have been far too simplistic is indicated by the emphasis of modern ecological theory on the complexity of drylands today. Drylands are not homogeneous entities with broadly similar environments, and the processes maintaining or transforming them ‘involve synergies and subleties at a variety of time and spatial scales’ (Spellman 2000: 38). There are major debates about the inevitability or avoidability of humanly induced desertication in drylands today. Beaumont (1993: 474) predicted bleakly that desertication was an inevitable price to be paid by many of the world’s poorest nations:
A tale of two deserts
491
‘land degradation may be a sacrice which has to be paid in order that local populations can survive future drought or famine’. Other geographers, however, have pointed to examples of land degradation in arid or semi-arid regions being reversed by local people making small-scale technological investments such as terrace construction and tree planting, in the context of population growth and market opportunities, for example in Kenya (Mortimore 1998; Thomas and Middleton 1994; Tiffen et al. 1994) and Burkino Faso (Lean 1994). ‘It is clear that difcult problems remain to be resolved before the magnitude and signicance of human–environment–climate interactions today can be fully elucidated, let alone those in the distant past’ (Spellman 2000: 38). One of the problems with much research on desert settlement history has been the lack of inter-disciplinary studies by archaeologists, historians and geographers, working together in the same region (Spooner 1989). Single-discipline research has yielded important information about climate, and/or environment, and/or human settlement, but most models of the structure and agency of landscape change arising from such case studies can be criticized for not being in a position to compare similarly high-quality data sets about all three processes. Modern landscape archaeology, however, integrates the methodologies of the archaeological, social and environmental sciences at the regional scale and frequently over long time scales. Hence it has the potential to address, in Spellman’s terms, the human–environment–climate interactions of the distant past: to identify past societies’ solutions and choices (both good and bad) for living in drylands, to ascertain in what circumstances human settlement was sustainable, when it caused damage from which the landscape was able to recover and when it degraded arid and semi-arid lands to the point of no return. Recent studies of Roman-period archaeology in North Africa and the Levant provide examples of how landscape archaeologists can provide answers to such questions and so contribute fruitfully to understanding processes of desertication in the past, and to the desertication debate more generally.
Roman Africa: the UNESCO Libyan Valleys Survey The UNESCO Libyan Valleys Survey was an inter-disciplinary study of desert-edge settlement and land use in part of Tripolitania in north-west Libya (Barker et al. 1996a; Fig. 1). With annual precipitation today diminishing rapidly from about 100mm in the northern edge of this ‘pre-desert’ zone to less than 25mm, the area has traditionally been inhabited by transhumant pastoralists who camp here in the summer and winter to the south much further into the Sahara. These people are highly mobile, relying on ocks and herds (sheep, goats, camels) and cultivating small patches of cereals on the oors of the dry valleys or wadis after the autumn rains. The ephemeral archaeology created by this lifestyle contrasts strikingly with the hundreds of Roman-period stone farmhouses (gsur or ‘castles’ in arabic) spaced at regular intervals along the edges of the wadis, associated with systems of stone eld-walls that appeared to be prima facie evidence for sedentary farming (Plates 1 and 2). Earlier reconnaissance by Goodchild in particular had indicated that many of these dated between the third and fth centuries AD, and he proposed that they represented an imperially-established system of defence in depth against the desert tribes to the south, with ex-soldiers (limitanei) being settled and allocated land to farm
Figure 1 Tripolitania, north-west Libya, showing the principal landforms and settlements, and the location of the UNESCO Libyan Valleys Survey. The 200mm, 100mm, 50mm and 25mm dashed lines are rainfall isohyets. The contours are in metres.
A tale of two deserts
493
around their fortied settlements (Goodchild 1950a, 1950b, 1952–53a; Goodchild and Ward-Perkins 1949). He argued that the system collapsed when the frontier had to be moved northwards in the context of tribal attacks, but that it was already greatly weakened by the degrading effects of intensive agriculture on the pre-desert landscape (Goodchild 1952–3b). Our project was asked by UNESCO to focus on this archaeology because the Libyan government was keen to understand what had been farmed, and how, to inform its plans for future development. We quickly established that the gsur were only the second phase of an extraordinary orescence in pre-desert settlement in the Roman period. Until the late rst century AD, settlement in the pre-desert was small-scale and mobile, much as in recent centuries, but at that point there was a dramatic transformation in settlement with the appearance of hundreds of courtyard farms, many built in the recognizably Roman style of opus Africanum. During the course of the third century AD these were replaced by fortied versions, the gsur. The gsur in the most arid southern regions were generally abandoned in the fth century AD, but elsewhere we found that most gsur continued in occupation into the sixth or seventh centuries, while in the northernmost part of the study area some of them were abandoned only very recently. Another nding by the project was that inscriptions at several open farms and gsur, or at associated funerary sites, demonstrated
Plate 2 Romano-Libyan oodwater farming in the Wadi Gobbeen in the Libyan pre-desert. A wadiedge diversion wall is visible in the right foreground, and a series of cross-wadi walls down the wadi in the distance, with a fortied farm on the horizon on the right (photograph: G. Barker).
494 Graeme Barker that their builders were in fact indigenous Libyans, ‘Romano-Libyans’ as we termed them, rather than colonists or limitanei (Mattingly 1983, 1987). Detailed eld and laboratory studies in geomorphology and palaeoecology by the project team established that the climate of the Tripolitanian pre-desert over the past 3000 years has in fact been essentially the same as today, apart from a wetter episode that may equate with the Little Ice Age in Europe (Gilbertson and Hunt, with Smithson 1996). In this part of Roman Africa at least, climatic change clearly cannot be cited as a primary reason for the greening of the desert or for its piecemeal abandonment (Gilbertson 1996). Instead, we were able to show from investigations of the eld systems associated with a representative series of open farms and gsur that Romano-Libyan farmers practised rather sophisticated oodwater farming. They built walls to divert surface runoff after seasonal storms, so that they could channel the oodwaters down from the plateau edges into the elds they laid out on the wadi oors (Gilbertson and Hunt 1996; see also Plate 2 and Fig. 2). By collecting together the surface runoff from a large area and channelling it into a small area of protected elds, these farmers were able to grow a wide range of crops that needed much more water than was available in terms of annual rainfall levels. These crops included three cereals (six-row hulled barley, durum or hard wheat and bread wheat), three pulse crops (lentil, pea and grass pea), four oil plants (olive, safower, linseed or ax and castor), ve Mediterranean fruits (grape, g, pomegranate, almond, peach), ve African fruits (date palm, water melon, wild pistachio, the Christ-thorn and sumach) and three herbs (purslane, dill and celery) (Van der Veen et al. 1996). Many farms have stone pressing structures for making oil and wine, and vat capacities indicate that most farms were producing a surplus well beyond the needs of the inhabitants. Pastoralism, by contrast, appears to have been small scale: the principal animals kept were sheep and goats, their dung indicating that they were stalled part of the year on the farms. Their water and pasture needs would have been competing with those of the people and crops, and the abundance of wild animals such as gazelle in the farm middens suggests that they were hunted as much for their meat as to protect the vulnerable crops on the wadi oors. The conclusion of the project was that indigenous Libyan élites in the pre-desert switched to cash-crop farming in response to the opportunities of Romanization in the rst century AD (Mattingly 1996). The indicators of surplus production in the pre-desert farms chime with epigraphic evidence from the major Roman fort of Bu Ngem (see Fig. 1 for its location) that the local population was supplying the military with agricultural commodities such as olive oil. Oil was probably also sent to the urban markets on the coast. The shift to fortied farms in the third century AD was primarily an indicator of internal social tensions and competitive behaviours among these élites. The decline of the system was also a gradual process: people in the most marginal locations soon returned to traditional lifestyles, whereas gsur farming continued apace further north, mostly in response to Roman markets; and in certain wadis local warlords appear to have retained control over their peasant farmers and maintained their oodwater-farming systems until the Arab conquest and beyond. The study of sediments associated with several farms, for example in midden deposits, building inlls and cistern conduits, found indicators for local landscape erosion at a few sites that might reect the pressures of overgrazing and fuel-wood collection. However,
Figure 2 Plan of the Romano-Libyan oodwater farming systems in the Wadi Gobbeen in the Libyan pre-desert.
496 Graeme Barker there was no evidence to suggest that environmental degradation was on such a scale as to have been a signicant factor in the decline of the system and the return to traditional pastoral lifestyles (Gilbertson 1996). In fact, there are several indicators that many Romano-Libyan desert farmers in Tripolitania were well aware of the need for conservation strategies: some sluices had been blocked, presumably so that smaller quantities of oodwater could enter the elds; stalling was presumably in part for the production of manure to fertilize the elds; and the concurrence of evidence for stalling and for active vegetation growth in some sediment sequences indicates awareness of the relationship between animals and vegetation, and of steps being taken to maintain it to the farmers’ advantage. The cultivation of tree and ground crops, the widespread harvesting of water, the improvement of forage, the maintenance of controlled grazing systems, the construction of walls and terraces, the concentrated inputs of nutrients from animals – all these components of the land use system must have tipped the balance at many locations in favour of the surface vegetation cover and soil stability, and against uvially-accelerated soil erosion. (Gilbertson 1996: 277)
Roman Arabia: the Wadi Faynan Landscape Survey The Wadi Faynan Landscape Survey is investigating the landscape history of the Wadi Faynan in southern Jordan, with a primary focus on the 10,000 years of the Holocene from the Neolithic to the present day (Barker 2000). The Wadi Faynan is one of a series of major wadi systems that dissect the western escarpment of the Jordanian plateau and ow westwards down into the Wadi Arabah rift valley (Fig. 3). The catchment of the wadi is about 15 kilometres west/east and 5 kilometres north/south, and we have concentrated on the central sector, surveying in detail an area measuring just over 30 km2 (Fig. 4). The climate on the escarpment is semi-arid, with an annual rainfall of some 200mm, so the villages there practise Mediterranean-style farming, growing cereals, olives, vines, and so on, whereas the Wadi Faynan, only a few hours’ walk down the escarpment, is rainless for most of the year and sparsely vegetated. Traditionally it has been used for seasonal grazing by Bedouin from plateau settlements, though today it is used mainly by landless nomadic pastoralists who are based in the Wadi Arabah. At the centre of the wadi, where a series of tributaries join to form the main channel, there is a major settlement of Nabataean (pre-Roman Late Iron Age), Roman and Byzantine (late Roman) date called Khirbet Faynan (Plate 3), the ‘Ruin of Faynan’. The hillslopes around the settlement are black with slag, the residues of ancient smelting, and the surrounding hills are honeycombed with ancient mineshafts, for the area is rich in copper and lead ores. The history of mining and metal-working in the Faynan region has been the subject of intensive study by a team from Bochum Mining Museum. They have demonstrated that industrial activity began in the Chalcolithic c. 5000 BC and expanded in scale in the following millennia to climax in Roman and Byzantine times (Hauptmann 1989, 1992, 2000; Hauptmann et al. 1992). Khirbet Faynan is commonly identied as the
A tale of two deserts
497
Figure 3 The location of Wadi Faynan within its region.
settlement or town of Phaino mentioned in classical sources as the regional control centre of copper and lead extraction in Roman times (there is a reference to Christian slaves from Palestine and Egypt being consigned to work its mines). Extending for some 5 kilometres to the west of Khirbet Faynan is a substantial eld system demarcated by hundreds of drystone walls (Fig. 4, Plates 3 and 4), its surface pottery indicating primary use contemporary with the settlement. Clearly, the abundant archaeological remains of the Wadi Faynan indicate patterns of settlement and land use in classical antiquity very different from modern Bedouin pastoralism. Through the ve seasons of eldwork the eld system, some 1000 elds, has been dissected by painstaking study of every eld surface and wall, the archaeology of the surrounding landscape has been mapped in considerable detail, and the resulting archaeological sequence contextualized within an environmental sequence established by geomorphology, palynology and geochemistry (Barker et al. 1997, 1998, 1999, 2000). The rst major contrast with the Tripolitanian study is the fact that the eld system represents a
Figure 4 The survey area of the Wadi Faynan Landscape Survey, showing the ancient eld systems at the centre and archaeological sites mapped around it. The topography shown is from a photogrammetric map, the boundaries of which do not extend as far as the boundaries of the survey area.
A tale of two deserts
499
Plate 3 Looking north-east across the ancient eld system in Wadi Faynan to Khirbet Faynan (the prominent hill in the right middle ground); the elds visible are mainly Nabataean in date (see Fig. 5) (photograph: G. Barker).
palimpsest of oodwater farming activity that began c. 4000 BC. At that time, in the context of signicant aridication, Early Bronze Age farmers started to build simple check dams across the oors of low-gradient and minor wadi channels to trap storm waters and the sediments being moved by them. By the early rst millennium BC, the Edomite Iron Age, more sophisticated methodologies were developed, whereby the oodwaters were gathered by massive diversion walls laid out along hillslopes and channelled to wadi-oor elds. Nabataean farmers in the centuries immediately before the Roman annexation of the region (AD 106) used the same principles of diversion-wall oodwater farming, but they applied them much more effectively by concentrating their efforts on the several small wadis dissecting the steeper slopes on the southern side of Faynan (Plate 3 and Fig. 5). They built check dams across the wadi channels high up on the hillslopes, linked to short diversion walls that carried the oodwaters along the contour of the slopes. The oodwaters were then managed through simple sluices (gaps) and spillways (step structures) onto a succession of terraced elds below. In the Roman/Byzantine period the eld system was considerably enlarged across the almost at oor of the main wadi (Plate 4 and Fig. 5). An ingenious system of long channels was constructed to tap the water in the tributary wadis and divert it hundreds of metres to a series of discrete ‘estates’ laid out on the oor of the main wadi. Our GIS modelling of water ow indicates that water resources were shared down the length of the eld system. Wadi Faynan had become a highly organized imperial estate organized by the Khirbet Faynan garrison, with extensive eld systems carefully
500 Graeme Barker
Plate 4 Looking west from the top of Khirbet Faynan across the main channel of the Wadi Faynan to the ancient eld system; the elds visible are mainly late Roman/Byzantine in date (see Fig. 5) (photograph: G. Barker).
managed and farmed to feed the industrial workers operating the mines and smelting works. In striking contrast with the gradual (and indeed partial) demise of oodwater farming in the Tripolitanian pre-desert, this remarkable agricultural enterprise went out of use in its entirety, relatively quickly, in the later centuries of the Roman empire. Alongside palynological studies to measure landscape impacts in terms of activities such as wood-cutting and grazing, we have used geochemical techniques (EDMA – Energy Dispersive X-Ray Micro-Analysis) to measure the nature and scale of environmental pollution caused by the copper and lead mining and smelting – the mining waste produced pollutants that were washed out into the landscape by water ow, and the smelting threw out pollutants into the atmosphere that in due course were also encorporated into sediments. There is overwhelming evidence from these studies that the collapse of intensive settlement and land use in the Wadi Faynan, and the development of the present-day degraded landscape, were directly related to the intensity of Roman industrial activity (Gilbertson et al. 1999; Hunt and Mohammed 1998; Pyatt et al. 1999). For the Early Bronze Age, alongside indications of small-scale soil erosion in the geomorphological record, we have pollen evidence for vegetation degradation on a small scale that probably reects a combination of grazing impact and fuel-wood cutting for smelting, together with geochemical signatures of small-scale localized environmental pollution from smelting. By contrast, the classical landscape was progressively and
A tale of two deserts
501
Figure 5 The distribution of (above) Nabataean and (below) Late Roman and Byzantine pottery in the ancient eld system in Wadi Faynan. Nabataean farmers concentrated their oodwater farming activities on steeper slopes on the southern margins of the system (the part of the eld system shown in Plate 3), whereas Roman/Byzantine farming expanded over the almost at terrain of the northern sector (the part of the eld system shown in Plate 4).
massively eroded, degraded, and polluted. Pollen of Nabataean and early Roman date indicates a steppic landscape around Khirbet Faynan in which olives and vines were being cultivated, but the landscape then rapidly developed into the extremely desertic environment that we nd today. Charcoal residues collected from the mining sites indicate that fuel-wood had to be brought down from the plateau by Roman miners, whereas local sources were still being used in Nabataean times (Engel 1993). Erosion made the oodwater farming systems increasingly less effective as gullying lowered the depth of water ow below diversion walls. Most strikingly, geochemical studies of a long sediment sequence in ponded deposits immediately below Khirbet Faynan indicate extraordinarily high levels of metallurgical pollution in Roman times, many times higher than modern safety limits – Figure 6 shows levels of copper, but these are also mirrored in terms of lead, barium and many other toxins. Furthermore, we have established that the modern biomass and cover values of barley plants growing in the wadi increase signicantly with distance from the contamination ‘hot spots’ such as Khirbet Faynan (Pyatt et al. 1999). The implication is that metal pollution was probably also affecting the productivity of the crops being grown in the eld system
502 Graeme Barker
Figure 6 The distribution of copper (in parts per million – ppm) through sediments by Khirbet Faynan, the focus of the Roman mining and smelting industry in Wadi Faynan. The sequence extends from c. 2500 years ago (far right) to the present day (left).
– perhaps the very high levels of late Roman/Byzantine pottery and other settlement debris strewn over the elds indicate increasingly desperate attempts to maintain soil fertility by intensive manuring, though in fact this simply served to carry metallurgical slags and associated pollution further out into the elds. The inescapable conclusion is that Roman and Byzantine industrial activity and its associated agricultural support system created the degraded landscapes of the modern Wadi Faynan desert. There are in fact still high levels of pollution in the modern fauna and ora living on and around the ancient settlements, resulting in high levels of copper traces, for example, in the dung, urine and milk of the Bedouin ocks (Gilbertson et al. 1999; Pyatt et al. 1999). With this evidence that Roman-period pollutants are still getting into the food chain, work is in progress to assess the levels of heavy metal pollutants in human skeletal remains from a Roman/Byzantine cemetery near Khirbet Faynan, the expectation being that metallurgical pollution is likely to have had major effects on the health of the Roman workforce from inhalation, skin contamination and bioaccumulation from animal and plant foods.
Conclusion The Tripolitanian pre-desert in Libya and the Wadi Faynan in Jordan were both marginal landscapes on the fringe of the Roman empire, the former in Roman Africa and the latter in Roman Arabia. Both were transformed by systems of oodwater farming through the centuries of Roman imperialism, in climatic contexts fundamentally the same as today.
A tale of two deserts
503
But here, as case studies in desert settlement and their implications for understanding desertication processes, the comparisons cease. In Tripolitania, systems of sedentary settlement and intensive land use involving cash crop farming developed rapidly, probably within a century, as indigenous élites took advantage of new opportunities offered by urban settlement growth on the coastal littoral and the supply needs of local Roman garrisons. In the Wadi Faynan, equally remarkable systems of sedentary settlement and land use were at their zenith in the centuries of Roman power, but in this case they represented the climax of a sequence of increasingly sophisticated systems of oodwater farming that began in the Early Bronze Age. From the third or fourth centuries AD onwards the Tripolitanian pre-desert farming systems gradually reverted to extensive pastoral-dominated systems of land use, and in a piecemeal way, with more favoured localities in the northern sector continuing to sustain sedentary settlement supported by irrigation farming to recent times; whereas in the Wadi Faynan the elaborate systems of water control, and the substantial settlements sustained by them, seem to have been almost wholly abandoned in the Byzantine period. In the Tripolitanian pre-desert we found no signicant evidence for humanly induced environmental degradation, and indeed several indicators of attempts by the Romano-Libyan farmers to practise sustainable systems of land management, whereas in the Wadi Faynan there is abundant evidence of catastrophic degradation resulting from the contemporary activities of miners and farmers. These two frontier regions display remarkably contrasting trajectories of landscape transformation within the same context of Roman imperialism, characterized by both wise and foolish decisions, and by both activities which had minor impacts from which the landscape recovered and activities with devastating impacts both on the people who were responsible for them but also on generations to come. It must also be emphasized that these differing trajectories cannot be assumed to be region-wide – other arid regions in Roman Africa and Roman Arabia undoubtedly have other desertication histories, and there are certainly no simple correlations between climatic change and human responses in terms of dryland settlement. Thus there is clear evidence in coastal North Africa, for example in Tunisia and Libya, for massive alluviation during the centuries of the Roman empire, with dams and aqueducts being overwhelmed by alluvial sediments (Ballais 2000; Vita-Finzi 1960, 1969). This was caused mainly by agricultural exploitation, which was certainly very intensive in many regions – olive farming in the hills behind Lepcis Magna was on an industrial scale, for example (Mattingly 1996) – though the deleterious effects of this land use were probably exacerbated by a climatic shift to wetter weather. In Israel, coastal molluscan sequences (Geyh 1994), terrace formation in the Negev desert (Goldberg 1994) and higher Dead Sea water levels (Frumkin et al. 1994) have all been cited as evidence for a wetter climate in the Levant (or at least the regions more exposed to Mediterranean weather patterns) in the rst two centuries AD, yet in the Negev oodwater farming systems seem to have been on a very small scale at this time compared with, for example, those of the Tripolitanian predesert (Rosen 2000). The peak of desert farming in the Negev was in fact in the Byzantine period, which was certainly arid, and signicant agricultural activity continued well beyond the abandonment of the Byzantine cities at least until the ninth or tenth centuries, and the nal abandonment of the central Negev took place at a time of climatic amelioration, not aridication (Rosen 2000).
504 Graeme Barker Modern desertication theory can learn a great deal from the archaeological record of the desert margins of the Roman empire in North Africa and the Near East. There was no simple evolutionary development from simple to complex or extensive to intensive systems of oodwater farming, and there were no straightforward correlations between particular human activities and particular environmental impacts. The two case studies emphasize above all the complexity of ancient societies’ perceptions of these marginal and precarious environments, of their responses to the constraints and opportunities presented to them, and of the manner in which their activities impacted on the landscape. Here as elsewhere, dryland agricultural and pastoral histories were ‘formed and changed within specic place-bound, social, historical, and ecological contexts’ (Widgren 2000: 262). In terms of the different degradation histories summarized in this paper, though, perhaps two general, though no doubt simplistic, observations can be made: the ecological sustainability of the oodwater farming in the Tripolitanian pre-desert was rooted in bottom-up decision making, with the wall systems demonstrating detailed local knowledge of landscape and water ow, whereas in the Wadi Faynan the extraordinary but short-lived and erosive agricultural system created by Roman farmers was established largely within the context of imperial power, imposed control and top-down decision making. These ndings certainly resonate strongly with the experience of development agencies working with dryland farmers today (Lean 1994; Mortimore 1998).
Acknowledgements I should like to acknowledge the support of the many funding bodies that nanced the eldwork and laboratory studies of the two main case studies cited in this paper: for the UNESCO Libyan Valleys Survey, notably the British Academy, the Leverhulme Trust, the Society for Libyan Studies, UNESCO and the Universities of Leicester, Manchester and Shefeld; and for the Wadi Faynan Landscape Survey notably the Arts and Humanities Research Board, the British Academy, the British Institute at Amman for Archaeology and History (now part of the Council for British Research in Levant), the Humanities Research Board, the Society of Antiquaries and the Universities of Aberystwyth and Leicester. The paper draws on the research of the many scholars in the two projects, but in particular on the work of Russ Adams, John Dore, Crispin Flower, David Gilbertson, Annie Grant, John Grattan, Chris Hunt, the late Barri Jones, David Mattingly, Hwedi Mohammed, Paul Newson, Brian Pyatt, Tim Reynolds, Roberta Tomber and Marijke van der Veen. I am particularly grateful to Paul Newson for supplying a GIS map of pottery distributions in the Wadi Faynan eld system (Fig. 5). School of Archaeology and Ancient History, University of Leicester, Leicester LE1 7RH, UK
References Aubreville, A. 1949. Climats, Forêts et Désertication de l’Afrique Tropicale. Paris: Société d’Éditions Géographiques Maritimes et Coloniales.
A tale of two deserts
505
Ballais, J.-L. 2000. Conquests and land degradation in the eastern Maghreb during classical antiquity and the Middle Ages. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 125–36. Barker, G. 2000. Farmers, herders, and miners in the Wadi Faynan, southern Jordan: a 10,000-year landscape archaeology. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 63–85. Barker, G. and Gilbertson, D. 2000. Living at the margin: themes in the archaeology of drylands. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 3–18. Barker, G., Gilbertson, D., Jones, B. and Mattingly, D. 1996a. Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis; Vol. 2, Gazetteer and Pottery (ed. G. Barker). London: Society for Libyan Studies; Paris: UNESCO. Barker, G. and Gilbertson, D., with Hunt, C. O. and Mattingly, D. 1996b. Romano-Libyan agriculture: integrated models. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 265–90. Barker, G., Creighton, O. H., Gilbertson, D. D., Hunt, C. O., Mattingly, D. J., McLaren, S. J. and Thomas, D. C. 1997. The Wadi Faynan Project, southern Jordan: a preliminary report on geomorphology and landscape archaeology. Levant, 29: 19–40. Barker, G., Adams, R., Creighton, O. H., Gilbertson, D. D., Grattan, J. P., Hunt, C. O., Mattingly, D. J., McLaren, S. J., Mohammed, H. A., Newson, P., Reynolds, T. E.G. and Thomas, D. C. 1998. Environment and land use in the Wadi Faynan, southern Jordan: the second season of geoarchaeology and landscape archaeology (1997). Levant, 30: 5–26. Barker, G., Adams, R., Creighton, O. H., Crook, D., Gilbertson, D. D., Grattan, J. P., Hunt, C. O., Mattingly, D. J., McLaren, S. J., Mohammed, H. A., Newson, P., Palmer, C., Pyatt, F. B., Reynolds, T. E. G. and Tomber, R. 1999. Environment and land use in the Wadi Faynan, southern Jordan: the third season of geoarchaeology and landscape archaeology (1997). Levant, 31: 255–92. Barker, G., Adams, R., Creighton, O. H., Daly, P., Gilbertson, D. D., Grattan, J. P., Hunt, C. O., Mattingly, D. J., McLaren, S. J., Newson, P., Palmer, C., Pyatt, F. B., Reynolds, T. E. G., Smith, H., Tomber, R. and Truscott, A. J. 2000. Archaeology and desertication in the Wadi Faynan: the fourth (1999) season of the Wadi Faynan Landscape Survey. Levant, 32: 27–52. Barth, H. 1857. Travels and Discoveries in North and Central Africa. London: Longman. Beaumont, P. 1993. Drylands: Environmental Management and Development. London: Routledge. Engel, T. 1993. Charcoal remains from an iron age copper smelting slag heap at Feinan, Wadi Arabah (Jordan). Vegetation History and Archaeobotany, 2: 205–11. Fantechi, R. and Margaris, N. S. (eds) (1986) Desertication in Europe: Proceedings of the Information Symposium in the EEC Programme on Climatology Held in Mytilene, Greece, 15–18 April 1984. Dordrecht: Reidel. Frumkin, A., Carmi, I., Zak, I. and Magaritz, M. 1994. Middle Holocene environmental change determined from the salt caves of Mount Sedom, Israel. In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean (eds O. Bar-Yosef and R. Kra). Tucson, AZ: University of Arizona, Radiocarbon Press, pp. 315–22. Geyh, M. A. 1994. The paleohydrology of the eastern Mediterranean. In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean (eds O. Bar-Yosef and R. Kra). Tucson, AZ: University of Arizona, Radiocarbon Press, pp. 131–45. Gilbertson, D. D. 1996. Explanations: environment as agency. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 291–317.
506 Graeme Barker Gilbertson, D. and Hunt, C. O. 1996. Romano-Libyan agriculture: walls and oodwater farming. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 191–225. Gilbertson, D. and Hunt, C. O., with Smithson, P. A. 1996. Quaternary geomorphology and palaeoecology. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 49–82. Gilbertson, D., Grattan, J. and Pyatt, F. B. 1999. Environmental impacts of ancient mining and smelting activities: initial EDMA studies. Pp. 262–9 in G. Barker, R. Adams, O. H. Creighton, D. Crook, D. D. Gilbertson, J. P. Grattan, C. O. Hunt, D. J. Mattingly, S. J. McLaren, H. A. Mohammed, P. Newson, C. Palmer, F. B. Pyatt, T. E. G. Reynolds and R. Tomber. Environment and land use in the Wadi Faynan, southern Jordan: the third season of geoarchaeology and landscape archaeology (1997). Levant, 31: 255–92. Goldberg, P. 1994. Interpreting late Quaternary continental sequences in Israel. In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean (eds O. Bar-Yosef and R. Kra). Tucson, AZ: University of Arizona, Radiocarbon Press, pp. 89–102. Goodchild, R. G. 1950a. Roman Tripolitania: reconnaissance in the desert frontier zone. The Geographical Journal, 115: 161–78. Goodchild, R. G. 1950b. The Limes Tripolitanus II. Journal of Roman Studies, 90: 30–8. Goodchild, R. G. 1952–3a. Farming in Roman Libya. The Geographical Magazine, 25: 70–80. Goodchild, R. G. 1952–3b. The decline of Libyan agriculture. The Geographical Magazine, 25: 147–56. Goodchild, R. G. and Ward-Perkins, J. B. 1949. The Limes Tripolitanus in the light of recent discoveries. Journal of Roman Studies, 39: 81–95. Hauptmann, A. 1989. The earliest periods of copper metallurgy in Feinan, Jordan. In Archaemetallurgie det Alten Welt/Old World Archaeometallurgy (eds A. Hauptmann, E. Pernicka and G. A. Wagner). Bochum: Deutsches Bergbau-Museum, Der Anschnitt, Beiheft 7, pp. 119–36. Hauptmann, A. 1992. Feinan/Wadi Feinan. American Journal of Archaeology, 96: 510–12. Hauptmann, A. 2000. Zur frühen Metallurgie des Kupfers in Fenan/Jordanien. Bochum: Deutsches Bergbau-Museum, Der Anschnitt, Beiheft 11. Hauptmann, A., Begemann, F., Heitkemper, E., Pernicka, E. and Schmitt-Strecker, S. 1992. Early copper produced at Feinan, Wadi Araba, Jordan: the composition of ores and copper. Archaeomaterials, 6: 1–33. Hughes, J. D., with Thirgood, J. V. 1982. Deforestation, erosion and forest management in ancient Greece and Rome. Journal of Forest History, 26: 60–75. Hunt, C. O. and Mohammed, H. A. 1998. Palynology of the Khirbat barrage reservoir sediments: preliminary results. Pp. 21–3 in G. Barker, R. Adams, O. H. Creighton, D. D. Gilbertson, J. P. Grattan, C. O. Hunt, D. J. Mattingly, S. J. McLaren, H. A. Mohammed, P. Newson, T. E. G. Reynolds and D. C. Thomas. Environment and land use in the Wadi Faynan, southern Jordan: the second season of geoarchaeology and landscape archaeology (1997). Levant, 30: 5–26. Lean, G. 1994. How stones can hold back the Sahara. The Independent on Sunday, 16 October: 16. Le Houerou, H. N. 1996. Climate change, drought and desertication. Journal of Arid Environments, 34: 133–85. Mattingly, D. J. 1983. The Laguatan: a Libyan tribal confederation in the late Roman empire. Libyan Studies, 14: 96–108.
A tale of two deserts
507
Mattingly, D. J. 1987. Libyans and the ‘limes’: culture and society in Roman Tripolitania. Antiquités Africaines, 23: 71–94. Mattingly, D. J. 1988. Oil for export: a comparative study of Roman olive oil production in Libya, Spain, and Tunisia. Journal of Roman Archaeology, 1: 33–55. Mattingly, D. J. 1996. Explanations: people as agency. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 318–42. McGregor, G. R. and Nieuwolt, S. 1998. Tropical Climatology, 2nd edn. Chichester: Wiley. Millington, A. C. and Pye, K. (eds) 1994. Environmental Change in Drylands: Biogeographical and Geomorphological Perspectives. Chichester: Wiley. Minnis, P. E. 2000. Prehistoric agriculture and anthropogenic ecology of the North American Southwest. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 271–87. Mortimore, M. 1998. Roots in the African Dust: Sustaining the Drylands. Cambridge: Cambridge University Press. Musil, A. 1907. Arabia Petraea II: Edom. Vienna: Alfred Holder. Palmer, E. H. 1872. The Desert of the Exodus. New York: Harper & Bros. Pyatt, F. B., Grattan, J. P., Barker, G. and Mattingly, D. 1999. King Solomon’s miners – starvation and bioaccumulation? An environmental archaeological investigation in southern Jordan. Ecotoxicology and Environmental Safety, 43(3): 305–8. Rosen, S. A. 2000. The decline of desert agriculture: a view from the classical period Negev. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 45–62. Spellman, G. 2000. The dynamic climatology of drylands. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 19–41. Spooner, B. 1989. Desertication: the historical signicance. In African Food Systems in Crisis, Part 1, Micro-Perspectives (eds R. Huss-Ashmore and S. H. Katz). New York: Gordon & Breach, pp. 111–62. Thomas, D. S. G. and Middleton, N. J. 1994. Desertication: Exploding the Myth. Chichester: Wiley. Tiffen, M., Mortimore, M. and Gichuki, F. 1994. More People, Less Erosion: Environmental Recovery in Kenya. Chichester: Wiley. Tolba, M. K. and el-Kholy, O. A. (eds) 1992. The World Environment 1972–1992. London: Chapman Hall. van der Veen, M., Grant, A. and Barker, G. 1996. Romano-Libyan agriculture: crops and animals. In Farming the Desert: The UNESCO Libyan Valleys Archaeological Survey, Vol. 1, Synthesis (ed. G. Barker, with D. Gilbertson, B. Jones and D. Mattingly). London: Society for Libyan Studies; Paris: UNESCO, pp. 227–63. Vita-Finzi, C. 1960. Post-Roman changes in Wadi Lebda. In Field Studies in Libya (eds S. G. Willimott and J. I. Clarke). Durham: University of Durham, Research Papers Series 4, pp. 46–51. Vita-Finzi, C. 1969. The Mediterranean Valleys: Geological Changes in Historical Times. Cambridge: Cambridge University Press. Widgren, M. 2000. Islands of intensive agriculture in African drylands: towards an explanatory framework. In The Archaeology of Drylands: Living at the Margin (eds G. Barker and D. Gilbertson). London: Routledge, One World Archaeology 39, pp. 252–67.