Most erosion in deserts is beyond feasible control. Some of the huge areas of bare rock, steep bouldery slopes, sand dunes and pebble-covered plains barely change from year to year; others erode, sometimes quickly, but the processes are seldom accelerated by human agency. In some agricultural systems, like run-off agriculture, erosion is actually welcomed (Evenari and others 1971). Rain-fed agriculture is only possible on the desert margins, and here too there is an equivocal message about erosion. Farmers in the northern Kordofan province of Sudan can produce crops from sandy soils in arid areas, but only in wet years (Figure 5.8). The sandy soils are easy to cultivate with hoes (usually the only ground-preparing tool available) and have good water-holding and wateryielding characteristics. The sandy fields may suffer wind erosion in dry periods, but many of the sands are deep enough to withstand millennia of erosion before they become unproductive. There are many such soils in arid areas. Few farmers can be convinced that erosion of these soils should be controlled, and anyway have inadequate labour to invest in it (Warren and others 2001).

Erosion, of course, is what produces the sediment that silts reservoirs. Some of it comes from high, tectonically active mountains, like the Himalayas, as in the case of the Tarbela and Mangla dams. Here, the natural rate of erosion on slopes and in river channels vastly exceeds the erosion caused by people; conserving agricultural soils would make little impact (Ives and Messerli 1989). In the U.S. Southwest, desert areas feed some of the sediment that is filling reservoirs like Lake Powell (behind the Glen Canyon Dam), and here too it appears that soil conservation would have little impact. Fifty years ago the consensus of scientific opinion was that overgrazing had exacerbated erosion in the Southwest, and if this were true, it might have been controllable. Some 30 years ago the consensus shifted to the belief that most erosion occurred in years with intense summer rains. These wet years have now been linked to El Niño cycles (Hunt and Wu 2004). Moreover, it appears that erosion rates in parts of Arizona are unaffected by vegetation cover in the short-term (Ritchie and others 2005). A more effective way of managing sediment in reservoirs is to build small sediment-holding dams upstream of the main dams (Catella and others 2005). Small dams beneath the water could hold back siltation from the outlets of the main dams (as proposed for the Tarbela and Glen Canyon dams). Although effective in the short-term, neither of these systems will give more than temporary relief.

Dust is the most significant natural output from the deserts (see Chapter 3 and Uno and others 2005). Should/could it be controlled? Control would be barely feasible for the quite natural processes that create about 90 per cent of dust in areas like northern Chad (Figure 5.9) or western China (Zhang and others 2003). Moreover, control, in the unlikely event that it succeeded, would interfere with the role of dust as a global fertilizer, as in the agro-ecosystems and forests of West Africa, the forests of the northeastern Amazon basin, the forests of Hawaii, and, most critically, the oceans, where the supply of iron-rich desert dust regulates biological productivity and may thus help to control global CO2 (Dutkiewicz and others 2006).

Dust from agricultural operations can be a severe and expensive nuisance, but seldom in deserts. In deserts, dust is a real pollutant downwind of desiccated lakes, such as the Aral Sea and Owens Lake in California, and here control is feasible and desirable. It is a taller order to ask for the control of military manoeuvres in the desert, like those that created large clouds of dust in the desert war in North Africa in the 1940s and the two Gulf Wars.