Diversity. It is commonly assumed that lands too dry for reliable cropping are all rather uniform. Yet arid lands are more extensive than lands of agricultural or forest potential, and extend across a great range of latitudes and elevations. Even when desert is excluded, the area is large and varied.

Although the cultivation of arid rangeland typically destroys natural vegetation, for no lasting benefit, cropping can be productive where run-off can be captured or irrigation practised. And the scope for other forms of resource use also varies with local conditions; particularly with biodiversity and the use that is made of woody vegetation. Neither customary resource use nor intervention to improve food security can afford to focus exclusively on the grazing resource.

Main rangeland types. Distinction needs to be drawn between three categories of arid rangeland. These are:

1. semi-desert, where vegetation is confined mostly to water courses and depressions, with trees and shrubs contributing more than grass to animal sustenence;
2. areas that are more widely vegetated, but with a ground cover comprising annual plant species and no significant contribution from perennial grasses; and
3. areas of greater biodiversity, with perennial as well as annual grasses, and often with a wide spectrum of woody species also.

While this gradient reflects increasing rainfall, the correlation is with moisture availability rather than with total annual rainfall. And one factor that determines rainfall effectiveness is evapotranspiration, which lapses as latitude and elevation increase. Consequently, "areas of greater biodiversity" arise more commonly and under lower rainfall in temperate and upland zones than in tropical lowlands.

In the lowland tropics, arid rangelands rarely progress beyond an annual grassland type. Such is the case in the Sahel of West Africa, where potential evapotranspiration is consistently high and rainfall is concentrated into a short season of 3-4 months duration. Yet in those parts of East Africa which experience bimodal rainfall (with two seasons per year and hence shorter dry seasons), a mean annual rainfall of 300-400 mm is sufficient to support perennial grasses and greater biodiversity.

Management implications. Such differences are important. Annual grasses can be highly nutritious while they last, but once dry and brittle, they disintegrate more rapidly than perennials under the influence of trampling, termites and wind. Therefore, unless the animals present can turn to browse, they have difficulty surviving to the end of the dry season. In principle, animals such as camels and goats should fare better than cattle and sheep, since the former browse more freely than the latter. However, much depends on the woody species present; and the best browse species may well occur, with perennial grass, in areas of greater biodiversity.

One further aspect of arid zone ecology warrants stress. Not only is rainfall low and erratic but it is also prone to cyclic variation. Historical records over the present century frequently show alternating periods, in which rainfall is consistently above or below the long term average (in as many as 5 years out of 7, or 14 out of 20, depending on the regional cycle). Moreover, years of exceptional rainfall are quite likely to trigger radical changes in vegetation, such that a succession of distinct vegetation types may occupy a site over the course of a century, each with equal claim to characterise the site.

These phenomena affect areas of biodiversity more than annual grasslands. However, semi-desert can show similar shifts, as water courses change alignment and leave established stands of trees and shrubs to die. It is a feature of semi-deserts that the browse plants on which productivity depends are often fed by water flowing into the area from zones of higher rainfall. Indeed, when wells are reliable and browse healthy, semi-desert can be better suited to year-round occupancy than annual grassland.

Pastoralism as a form of occupancy. Pastoralism is the commonest form of occupancy and resource use in arid rangelands; and the only one (apart from large settlements based around permanent water) to raise major concerns of food security. The strategies of resource use employed by pastoralists pivot around drought management and food security; and well-conceived development schemes focus on the same issues.

Past development efforts have not had much success in supporting pastoralism. One reason for this is that pastoralism is commonly treated as a livestock production system; as a source of livestock for consumers in urban and agricultural areas. But this perspective, of pastoralism as an agricultural sector activity, can only be applied usefully where pastoralism is practised on the fringe, locationally and economically, of areas of higher rainfall. It is too crude a caricature to be helpful in drier and more remote areas.

Pastoralism, it should be remembered, was once a dominant geopolitical system on the world stage. Since then, pastoralism has been subjugated to other political and economic systems; but still it operates as a form of territorial occupancy. And like other forms of occupancy, pastoralism is diversified so far as the resource base allows.

Economic diversification. Because arid rangelands rarely possess or attract the resources needed to diversify in other directions, diversification in pastoralism is manifest mainly in livestock; in the several roles which livestock play, as capital, in aesthetics and social relationships, as a source of subsistence and in trade. The value which an individual animal holds in a pastoral economy is determined by the function that it fulfils, relative to the other animals available, and this may be more or less than its market value.

But diversification is not manifest entirely in livestock. Although resources such as oil and precious metals (and sometimes other resources) are acquired by the State, hunting and gathering and localised cropping are commonly practised within the scope of pastoralism. Individuals and families also diversify into artisanal activities and paid employment, either contributing goods and services directly to the pastoral economy or bringing in cash to help maintain their own household and pastoral identity.

Types of pastoral system. It follows that pastoralism takes many forms, with societies varying in the livestock they keep and in the uses they make of livestock and other resources. Usually the livestock are well adapted to the environment concerned, but this is not always the case. For example, a society which has evolved around cattle often retains cattle even when it occupies territory better suited to camels and goats. And the poorer members of society usually keep species or adopt practices different from those of their wealthier neighbours.

The relationship between environment, livestock species and resource use is seldom one of simple cause and effect. These variables lie in a matrix of interacting factors which, taken altogether, serve to differentiate one pastoral system from another. Other factors in the matrix include mobility, diet and economic orientation. Trade-orientated pastoral systems are often associated with a diet based on the daily consumption of grain, and with patterns of transhumance that include annual stopovers in agricultural areas where animals can be traded for supplies of grain. When milk is the dietary staple, peripatetic movement is more likely than a regular pattern of transhumance, since the primary concern then is to seek out whichever areas are best able to keep animals in milk. Agropastoralists who grow their own grain are different yet again, because of the need to be close to their croplands, with draught animals, at set seasons of the year.

 Change in pastoral systems. Furthermore, pastoral systems are subject to change. Not only do existing systems have to adapt to pressures of population and losses of territory but they also respond to changing expectations and opportunities. Thus, motorised camel herding is now practised in some oil-rich countries. Another new category of increasing importance is investment-based herding. This arises when business people or officials invest savings in livestock in their home area. The animals may run with those of the existing system, but their function is different, since they are not used primarily for subsistence or trade but to provide personal satisfaction, capital growth and social security.

The first step, whenever intervention is contemplated in arid rangelands, is to categorise the pastoral system(s) involved. In addition to differentiating them in terms of livestock and economic orientation, and population pressure and mobility, it is necessary to establish how society is organised for purposes of decision-making and controlling access to resources. This is a prerequisite for community participation in development, and helpful too in targetting assistance to the poor through societal mechanisms for welfare support. Some of these aspects are elaborated in the next section, in the context of NRM.

NRM is the principal means by which pastoralists meet domestic needs and pastoral societies secure their future. The essence of pastoral development, likewise, is to ensure sustainable NRM, and to link this with improved social services. Other development objectives may attach, but most World Bank lending in pastoral areas now channels through NRM projects.

Yet the nature of NRM makes it a difficult area in which to intervene. NRM can be readily comprehended as a concept or general objective, but in practice it is the product of three quite distinct entities. The natural system, governed by the laws of nature, determines the resources on offer; the user system determines the utility of those resources; and the larger geopolitical system determines the boundaries and externalities which govern resource utilization. This is not the place to unravel all the complexities of NRM, but some of them emerge in the description that follows of the aspects most affecting food security in arid rangelands. Three key features of NRM are covered, with an additional comment on more comprehensive intervention.

Mobility. Mobility typifies arid zone pastoralism. It is the principal strategy employed for making best use of the resources on offer and for coping with fluctuation and spatial variation in rainfall and feed supply. Short range movements supplement longer range seasonal shifts. Some seasonal shifts are imposed by climatic or other imperatives (such as seasonal snow cover or lack of water), while others are voluntary, guided by the goals and predilections of the pastoralists concerned.

Restricting customary mobility usually reduces food security. It is particularly important to keep open transhumance routes and drought retreats that allow access to flood plains and areas of higher rainfall. Converting such areas to cropland does not improve food security if, through the loss of key resource areas, the whole of the regional pastoral system is thrown into disarray. Cattle-based systems, especially, can be decimated by the loss of just a few hundred hectares of flood plain.

That is not to say that customary mobility ensures optimal NRM. Voluntary movements can have their origins in factors unrelated to NRM, and there may be better alternatives. However, intervention still requires sensitivity, since to disrupt socially-driven patterns of movement (such as joining clan gatherings at a set time and place, or moving according to phases of the moon) may imply undermining customary authority that is essential to maintaining effective NRM overall.

Drought Management. Much of customary NRM is an exercise in drought management. Mobility is used both to avoid drought, by moving elsewhere when drought strikes, and to absorb drought, by seeking out unused grass or browse (whether left by design or because of relative unpalatability). But mobility is not the only coping mechanism. People, as well as livestock, can step up their use of bush foods. They can also ensure, with the help of natural selection, that their livestock are drought-tolerant; and societal welfare support mechanisms can be put to use to help those hardest hit by drought.

One feature of the drought management strategies of pastoralists is that they tend to be reactive rather than pre-emptive. Where external assistance can help is in developing early warning systems and drought preparedness. The strategic offtake of livestock has a place here, to allow animals to be removed before they become valueless and die. However, the design of offtake systems has to be adapted to meet the different circumstances of subsistence and trade-orientated systems.

External agencies also have a role in supporting restocking programmes post-drought. The safest approach is to assist in the distribution of local animals through the societal welfare support system. Importing animals and distributing them as government or NGO largesse runs the risk of aggravating grazing pressure with ill-adapted animals allocated to undeserving cases outside of societal control.

Administration. Another critical feature of customary NRM is that it is segmented, with different layers of pastoral society holding responsibility for different aspects of management. Historically, there are examples of strong leadership exercising tight control across the whole spectrum of NRM, but typically each resource (water, grazing, trees of value, etc) is managed separately, under the control of family or neighbourhood groups. Higher levels of societal structure are concerned primarily with maintaining territorial boundaries and societal ethics, and intervene locally only when ethical issues arise.

State governments may assume responsibilities in NRM, but seldom have the staff, knowledge or budget to administer NRM effectively in remote pastoral areas. They are best advised, therefore, to concentrate on providing an enabling framework and to encourage herders' organisations to manage the rest. Where pastoralists represent a linguistic and cultural minority, local government might usefully be groomed to provide necessary administrative support.

Further Improvement. Serious intervention in NRM calls for detailed knowledge of the pastoral systems concerned. Mostly this implies field study, because most texts treat pastoralism as homogenous or not much affected by the variations outlined earlier. Hence, when customary resource tenure is extolled, it is without differentiating tenure from access, or one resource from another. And when mobility is extolled, it is without analysis of the rationale for movement, or of how development should proceed when mobility is already curtailed. Moreover, pastoralism is often presented as "opportunistic range management". In practice, opportunism works well in wet-season grazing areas, but tends to be anti-social in dry-season grazing areas and anarchic when applied to the resources of others. It also brings much suffering to families, in the course of enabling societies to survive drought and other perturbations.

Sweeping assertions about pastoralism arise partly as a by-product of advocacy; of hammering home the point that customary practice is better suited to arid rangelands than ranch-style management. However, they arise also from the dearth of in-depth studies of the workings of pastoral systems and ecological processes. Participatory rural appraisal (PRA) may suffice to get a process project started, letting monitoring data guide subsequent inputs, but embarking on development in a state of relative ignorance is always a high-risk strategy.

Improving understanding of ecological processes has particular priority in arid rangelands. The arguments that favour opportunistic range management focus too much on grass, relative to browse, and show little regard for the effect of periodicity in rainfall and for the successional processes that can be observed at work in some areas. Economic and social perspectives also have to be drawn into the interpretation of what constitutes overgrazing. A depleted range ecosystem may suit the present expectations of pastoralists, but if these expectations were to change, then management would need to shift rangelands closer to their natural state.