CIESIN Reproduced, with permission, from: Dregne, H. E., and N-T. Chou. 1992. Global desertification dimensions and costs. In Degradation and restoration of arid lands. Lubbock: Texas Tech. University.

Global Desertification Dimensions and Costs

H. E. Dregne and Nan-Ting Chou


Desertification has been defined by the United Nations Environment Programme (UNEP) as land degradation in arid, semi-arid, and dry subhumid areas resulting mainly from adverse human impact. This 1991 definition is a revision of the definition formulated at the 1977 United Nations Conference on Desertification. The 1977 definition described desertification as the diminution or destruction of the biological potential of the land, which could lead ultimately to the formation of desert-like conditions (UNCOD 1977).

Although the 1977 definition appeared to be a simple, straightforward statement, there have been no end of different ideas of just what constitutes desertification. There has been agreement, only, it seems, that desertification was a land degradation proc ess. Many people believe it to be wind erosion on sandy grazing land or rainfed croplands. At least an equal number contend that desertification is confined to pastoral lands and is due to overgrazing by livestock. To some, desertification is a change of climate toward desiccation, caused either by human or natural factors. To still others, desertification has occurred only if land has become an unproductive wasteland (desert).

Our view is that desertification is a human-induced process of land degradation that can range in severity from slight to very severe and in cause from erosion to salinization to toxic chemical accumulation to vegetation degradation, irrespective of clima te. In this paper, however, desertification will be confined to land degradation in the drylands of the world. More specifically, it will be limited to arid, semiarid, and dry subhumid climatic zones. Hyperarid (extremely arid) climatic zones will be excl uded, except for the case of irrigated land lying in the hyperarid zone, as is the situation in the Nile Valley of Egypt. The rationalization for excluding hyperarid zone land is that such lands, unless irrigated, are incapable of supporting a human occup ancy dependent upon plants, directly or indirectly, for food.


The drylands of the world are those delimited on the map prepared by the United Nations Educational, Scientific, and Cultural Organization (UNESCO) (UNESCO 1977). Subhumid regions shown on the UNESCO map are considered dry subhumid regions, for the purpos e of this study, since they are only the dry part of subhumid regions mapped by other climatologists. Climatic zone areal extent figures, by country, were taken from Hopkins and Jones (1983). The UNESCO map was used to determine the area of drylands in ea ch country of the world.

The principal desertification processes are degradation of the vegetative cover, accelerated water and wind erosion, and salinization and waterlogging. These processes affect the three major land uses in arid regions: irrigation agriculture, rainfed cropp ing (dry farming), and pastoralism on rangelands. Rangeland desertification is primarily a matter of degradation of the vegetative cover through overgracing and cutting of woody vegetation. Rainfed cropland desertification is commonly expressed as increas ed water and wind erosion. Salinization and waterlogging are the principal degradation processes on irrigated land. Wind erosion, like water erosion, affects significant portions of rangelands and can cause damage on sandy irrigated lands.

Other important desertification processes include soil compaction and accumulation of toxic substances such as heavy metals and persistent pesticides. Mining and tourism can cause land degradation at the local level but are of miniscule extent at the glob al scale. So-called dryland salinity is a major small-scale problem in the grazing and rainfed croplands of Australia, Canada, and the United States but the area is limited. Our estimates of land degradation refer to only vegetation degradation, water and wind erosion, and salinization and waterlogging on irrigated, rainfed, and grazing lands.

Land-use figures are derived from the data in the 1986 edition of the FAO Production Yearbook (FAO 1986). Where entire countries lie in the drylands, FAO numbers for irrigated land area were used directly. In countries such as China and India, which have much irrigated land in the humid regions, estimates were made of the fraction located in the drylands. Similar adjustments were made for the rainfed croplands. After irrigated and rainfed farming lands were calculated, the remaining land in the drylands w as considered to be rangeland. The latter is in error to the extent that there are urban, mining, and tourist areas in the drylands. Those numbers are small.

The information base upon which the estimates in this report were made is poor. Anecdotal accounts, research reports, travellers' descriptions, personal opinions, and local experience provided most of the evidence for the various estimates. Some data were available for Australia and the United States. Both of these countries have conducted comprehensive assessments of land degradation on irrigated, rainfed farming, and range lands. For the country data, it is impossible to estimate the error in the number s of hectares in each degradation class because there are no accepted values against which to make comparisons. To our knowledge, no one except the senior author has ever attempted a global assessment, and very few have published national assessments. An earlier evaluation was published in 1983 (Dregne 1983).

After the area in each of the three land uses had been estimated for the 100 countries, the amount of land in each desertification class (slight, moderate, severe, very severe) in each country. by major land use (irrigated, rainfed, range land) was estima ted. The cost of agricultural production foregone because of desertification was then calculated on a continental and global basis for each land use. Following that calculation, an estimate was made of the cost of rehabilitating desertified land, by major land use, in those developing countries requiring external financial assistance to carry out rehabilitation measures, as determined by the United Nations.

Desertification Status Criteria

The principal criterion for placing land in one of the four desertification classes (slight, moderate, severe, very severe) was the impact degradation has had on economic plant yield, on both croplands and rangelands. Slight, moderate, and severe degradat ion usually are reversible; very severely degraded land is categorized as land which cannot be rehabilitated economically. When money is no concern, virtually any land can be restored to something approximating its original productivity. For croplands, th e slight class represents land that is not desertified at all or shows only little degradation (less than 10 percent loss in potential yield). For rangelands, the slight class includes land that has up to 25 percent loss in potential productivity. The hig her tolerance allowed for slight degradation of rangeland arises from the procedure range scientists employ to evaluate range condition. The 0 to 25 percent productivity loss places rangeland in the best range condition class, variously called good or goo d to excellent by different agencies assessing rangeland condition.

On irrigated land, the principal human-induced degradation problem is salinization, followed by waterlogging. Slightly desertified land has had yields depressed by less than 10 percent, moderately desertified by 10 to 25 percent, severely desertified by 2 5 to 50 percent, and very severely desertified by more than 50 percent. Very severely desertified irrigated lands are badly affected by the accumulation of soluble salts or exchangeable sodium or both and are very slowly permeable to water. Probably close to 98 percent of salt-affected irrigated lands can be reclaimed successfully at economically justifiable costs. Reclamation costs tend to be high but so are the returns to irrigated land.

Desertification of rainfed cropland consists of accelerated water and wind erosion. Water erosion is much more damaging to long-term soil productivity than is wind erosion. Slight, moderate, severe, and very severe soil degradation reflect less than 10 pe rcent loss of potential crop yield, 10 to 25 percent loss, 25-50 percent loss, and greater than 50 percent loss, respectively, the same as for salinity effects. Wind erosion on-site impacts are usually either slight or very severe, with little in between. Very severe water erosion causes deep gullies to form or erodes the soil nearly down to bedrock. Very severe wind erosion is observable as the exposure of unproductive subsoils or substrata or the formation of mobile sand dunes and blowouts. Controlling water and wind erosion with mechanical structures does not commonly restore the soil to its pre-degradation condition. Instead, control stops further degradation and allows soil-improving practices to be applied more effectively.

Rangeland desertification is almost entirely a matter of vegetation degradation, which may be combined with water and wind erosion after vegetation degradation has been initiated. Overgrazing is by all measures the principal cause of rangeland degradation . Cutting woody species for forage, fuel, charcoal production, or construction material is the other major cause of rangeland deterioration. Vegetation degradation includes invasion or increase of undesirable brush species which may actually increase biom ass production on degraded rangelands. Assessment of the desertification status of rangelands was based upon evaluation of the current range condition, then translating range condition into a vegetation productivity estimate. Range condition is a technica l term used expressly by range scientists to evaluate "the current productivity of a range relative to what that range is naturally capable of producing" (Kothmann et al. 1974). For present purposes, a range condition of excellent to good placed the range land in the slight desertification class (less than 25 percent loss in productivity). Fair condition meant moderate desertification (25 to 50 percent loss in productivity), poor condition meant severe desertification (50 to 75 percent loss in productivity ), and very poor condition was equated with very severe desertification (75 to 100 percent loss in productivity). Very severe desertification of rangelands is a consequence of erosion: deep and extensive gullying, exposure of unproductive subsoils by wate r or wind erosion, the formation of mobile sand dunes, or the presence of a repeating sequence of hummocks and blowouts.

Global Status of Land Degradation

Table 1[a, b, c, d] presents data for the area of irrigated land, rainfed cropland, and rangeland used in this report, by c ountry and continent. The total land in those three major uses, plus the hyperarid land area, give the total amount of drylands for each country and continent. Africa has, by far, the greatest amount of hyperarid land, mainly the Sahara Desert, followed b y Asia. Australia and New Zealand, together, are 83 percent dryland, virtually all of it in Australia. There are no Hyperarid lands in Australia or Europe and very little in North America. India, the former Soviet Union, China, Pakistan, and the United St ates are the top five countries in area of irrigated land. India, the U.S.S.R., Australia, Canada, and the United States are tops in rainfed cropland, whereas the ranking for rangeland is Australia, U.S.S.R., China, the United States, and Sudan. Nine Afri can countries (Botswana, Djlbouti, Egypt, Libya, Mauritania, Namibia, Niger, Somalia and Western Sahara) and twelve Asian Middle East countries (Bahrain, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman, Qatar Saudi Arabia, Syria, the United Arab Republic, and Yemen) lie entirely in the arid regions.

Globally, Australia has the greatest amount of drylands, none of it hyperarid. The U.S.S.R. ranked second, again with no hyperarid territory, followed by China, the United States, and Saudi Arabia. Over half of Saudi Arabia is hyperarid.

Estimates of the degradation status of irrigated land are presented in Table 2[a, b, c, d, e]. Salinity is the major problem, usually combined with waterlogging, which caused the problem in the first place. Sodic soils, alone, are of minor extent but saline-sodic soils are fairly widespread, especially in Iraq. For countries having more than 1 million hectares of irrigated land, the worst affected are Iraq (71 percent salinized), Asiatic U.S.S.R. (51 percent), Pakistan (40 percent) and Mexico (36 percent). Globally, the average is 30 percent. Australia has an estimated 15 percent of its irriga ted land salt-affected but there is much uncertainty about the accuracy of that estimate. The correct figure may be considerably higher, judging from some accounts of the severity of the salt and waterlogging problem in the Murray River watershed. Many co untries in Africa and several in Europe have no significant salinization of irrigated land. Asia, with an area of irrigated land four and one-half times that of the next largest continent, has the highest percentage of degraded land (35 percent), followed by North America (28 percent) (Table 3).

Degradation of rainfed cropland (Tables 4[a, b, c, d, e] and 5) i s greater than the degradation of irrigated land. Three countries in Africa have more than 80 percent of their rainfed cropland desertified by water erosion. They are Algeria, Kenya, and Lesotho. Most of the erosion has occurred in the past 50 years in re sponse to high population increases and land use policies. India has the largest amount of rainfed cropland (100 million hectares) and 60 percent degradation. Canada has the lowest percentage of degraded land, with only 8 percent experiencing more than 10 percent loss of potential productivity on its 33 million hectares of dry farming land. The low degradation percentage for Canada and the United States is due to the level topography of the Great Plains, where the vast majority of the rainfed croplands oc cur. Wind erosion is extensive in the Great Plains and destructive to young plants but its effect on long-term soil productivity generally is small over most of the region.

Rangeland degradation is the most extensive among the three major land uses (Table 6a, b, c, d, e). Few countries have less than 50 percent of their pastoral lands degraded. Overgrazing by livestock is the principal land problem, coupled with cutting of woody species in the many countries where wood is the main fuel source. The high percenta ge of the world's rangeland that suffers from overuse stems from the extensive, low intensity character of pastoral land use, the slow response to land management changes in arid climates, and the social and economic problems associated with reducing live stock numbers on heavily used rangelands. Reducing populations of cattle, sheep, and goats among people living on the margin of survival cannot be done without providing other sources of livelihood and that is rarely available. Australia has a markedly lo wer percentage of degraded rangelands than any other continent (Table 7), primarily because a large part of Australian rangelands are not grazed at all.

Table 8 summarizes the numbers in Tables 2 to 7. Approximately 3 percent of the world's drylands are irrigated, 9 percent are rainfed cropland, and 88 percent are rangelands. The dominance of rangelands, with their high percentage of degraded land, is why the global desertification level is a high 70 percent.

Land Degradation Costs

Two kinds of costs are involved in assessments of the economic impact of land degradation. The first is the income foregone as the result of prior land degradation. The second is the cost of controlling and repairing land damage.

Costs associated with the direct (on-site) effect of desertification on crop and livestock production are the focus of this report. These effects show up in reduced productivity of the rangelands and croplands. Indirect (off-site) costs can be greater tha n direct costs, but they are not discussed here because of the virtual absence of information.

Income Foregone

Productivity losses due to land degradation exact a price that appears as the annual value of agricultural production foregone as a result of the degradation (Table 9). At the global scale, it is difficult to select a single figur e for the cost of degraded irrigated land, for example, because the cash equivalent value of the crop, whether it be wheat or sorghum or corn, varies greatly from country to country. Subsidies, price controls, and foreign exchange rates, among other facto rs, influence price. Despite the variations, one figure was used as the amount of income foregone on irrigated, rainfed, and range land when the degradation was at least moderate in severity. The number used represents, approximately, a 40 percent loss in productivity. A 40 percent loss means that the actual yield was 40 percent less than it would have been in the absence of any degradation. For irrigated land, that represents a $250 (U.S.) per hectare per year reduction in income, $38 on rainfed cropland , and $7 on rangeland (Table 9, footnote). The numbers represent our estimates, based upon a relatively small amount of data, most of it from the United States and Australia (Aveyard 1988; Bureau of Reclamation 1983; Campbell 1990 ; Dixon et al. 1989; Heady and Bartolome 1977; LeHouérou 1989; United Nations 1980).

Table 10 is an excerpt from the publication by Aveyard (1988) on the cost of land degradation on rainfed cropland and irrigated land in the watershed of the Murray-Darling rivers system in southeastern Australia. The watershed li es mostly in the states of New South Wales, Victoria, and South Australia. As far as we have been able to determine, these data represent the only published economic assessment of the damage land degradation has done anywhere in the world over an area of several million hectares.

It is evident that the most extensive and costly land damage done in the Murray-Darling basin is due to soil compaction and an unfavorable change in the soil structure. Shallow watertables, with or without salinization of the soil, are next in importance. Water erosion is much less of a problem in this area, and wind erosion and dryland salinity are of least importance.

On a per hectare basis, the annual income foregone as the result of land degradation amounts to about $132 on land affected by dryland salinization, about $88 for land suffering from water tables close to the surface, nearly $20 for soil structure decline , and about $1.50 for water erosion of rainfed cropland. Wind erosion production losses are approximately $2 per hectare per year. Campbell (1990) estimated that land degradation cost Australia about $750,000,000 annually.

The estimated annual cost of desertification, expressed as income foregone, amounts to nearly $11 thousand million ($11 billion) for irrigated land, $8 thousand million for rainfed cropland, and $23 thousand million for rangeland. The total annual cost is about $42 thousand million in 1990 U.S. dollars (Table 9).

Not all desertified land is in a position to be reclaimed economically. Our estimate is that nearly all irrigated land will pay back the cost of rehabilitation in greater income generated as the result of the rehabilitation. Reclamation of irrigated land is costly, primarily because of the need for an effective drainage system, but crop yields will generally be high since irrigation removes the water constraint on plant growth. Rainfed cropland, however, does face a water restriction on crop growth. We es timate that the rainfall restriction is large enough in the drier parts of the rainfed cropping zone to reduce the area of land capable of producing a favorable cost-benefit ratio to 70 percent of the total rainfed cropland. For rangeland, the moisture co nstraint is even greater, which leaves only about 50 percent of the rangeland capable of a favorable cost-benefit ratio when rehabilitated. Out of the total global desertified land (3,592 million hectares), approximately 52 percent (1,860 million hectares ) can pay back the cost of rehabilitation. The income foregone by not rehabilitating that 52 percent is about $563 thousand million over a 20-year period (Table 11).

Cost of Rehabilitation

Land rehabilitation costs are those incurred for stopping further degradation and to restore the land to something approaching its original undegraded condition (Bishop and Allen 1989; Bojo 1990; FAO 1971; Girt 1990; Holmberg 1990; IFAD 1987; United Natio ns 1980). Rehabilitation requires three to five years for irrigated land after an acceptable drainage system has been installed, perhaps five to 10 years to improve eroded rainfed cropland, and as much as 50 years to bring rangeland in the drier areas to a good range condition.

There are reasonably good data in the files of the Food and Agriculture Organization (FAO) of the United Nations, the World Bank, and government agencies in several countries that give the cost of land rehabilitation. There is a very wide range in costs, depending upon the availability of trained technical people, large equipment, spare parts, fuel, transportation, and other items. We selected a cost of $2,000 per hectare to improve irrigated land, $400 for rainfed cropland, and $40 for rangeland. FAO fil e data provided the principal data on cost of improving irrigated land and rainfed cropland. Australia and the United States were the major source of costs of rangeland. These figures are low for some cases and high for others.

Rehabilitating all of the desertified land in the world is not economically profitable, for reasons noted earlier. For the desertified land for which we estimate rehabilitation would pay, the cost of doing so would be about $213 thousand million (Table 12). Reclamation costs per hectare are 50 times greater for irrigated land than for rangeland. Returns are about 35 times greater per hectare per year for irrigated land.

Cost-Benefit Analysis

Over a 20-year period, in 1990 dollars, the annual cost of income foregone on the desertified lands capable of producing a favorable cost-benefit ratio when rehabilitated is $28 thousand million. The annual rehabilitation cost is approximately $11 thousan d million. According to our estimates, the benefits are 2.5 times higher than the costs. This number is independent of discount rate, inflation, and interest rates as long as it is denominated in 1990 dollars. It could be affected by government subsidies, costs of human displacement, etc. The gain in productivity from rehabilitation extends beyond the 20-year period but is not considered in this benefit calculation. If it had been, the benefit would have been significantly higher, assuming good land manag ement which is essential if rehabilitation is to be successful. The economic returns from land improvement should continue indefinitely.


Irrigated land occupies about 3 percent of the drylands, rainfed cropland about 9 percent, and rangeland about 88 percent. Approximately one-quarter of the irrigated land is degraded (desertified), half of the rainfed cropland, and three-quarters of the r angeland. Asia has the highest percentage of its irrigated land desertified, Africa carries that distinction for rainfed cropland, and North America has the poorest evaluation for rangeland. Australia has the lowest percentage of its drylands in a degrade d condition, largely because a considerable amount of its rangeland is practically unused.

Iraq has the highest percentage of degraded irrigated land, at 71 percent. Salinity has been a severe problem in that country for at least two thousand years. The Central Asia republics of the former Soviet Union also have large amounts of salinized irrig ated land, at 51 percent or, perhaps, even more. Three African countries (Algeria, Kenya, and Lesotho) have more than 80 percent of their rainfed cropland damaged by water erosion. Lesotho is said to have particularly severe erosion problems. The rangelan ds in most countries have more than 50 percent of the land degraded. Some range scientists claim that a few countries have 100 percent of their rangelands degraded.

The cost-benefit ratio for rehabilitating the 1,860 million hectares of desertified land that can be repaired economically is about 1:2.5. Approximately 48 percent of the desertified lands are estimated to have a favorable cost-benefit ratio.


Aveyard, Jim, 1988. Land degradation: Changing attitudes-why? Journal of Soil Conservation, New South Wales 44:46-51.

Bishop, Joshua, and Jennifer Allen, 1989. The on-site costs of soil erosion in Mali. World Bank Environment Working Paper No. 21, Washington, D.C. 71 p.

Bojo, J., 1990. Benefit-cost analysis of the farm improvement with soil conservation project in Maphutseng, Mohale's Hoek District, Lesotho. In John A. Dixon, Paul B. Sherman, and David E. James (editors), Dryland Management: Economic Case Studies, Earthscan Publications, Ltd., London, 98-112.

Bureau of Reclamation, 1983. Colorado River Basin Water Quality Improvement Program, Status Report. Bureau of Reclamation, U.S. Department of the Interior, Denver, Colorado. 274 p.

Campbell, Andrew, 1990. The greening of Australia. Our Planet, Vol. 2, No. 1, p. 7-8

Dixon, John A., Paul B. Sherman, and David E. James, 1989. The Economics of Dryland Management, Earthscan Publications Ltd., London. 302 p.

Dregne, H. E., 1983. Desertification of Arid Lands. Harwood Academic Publishers, New York. 242 p.

FAO, 1971. Land and water use and management in the Chambal Irrigated Area, Rajasthan, India. Agricultural Engineering. FAO AGL:SF/IND 60, Technical Report 4. 34 p.

1986. FAO Production Yearbook. Food and Agriculture Organization of the United Nations, Volume 40, Rome. 306 p.

Girt, J., 1990. Land degradation costs in Canada: A recent assessment. In John A. Dixon, Paul B. Sherman, and David E. James (editors), Dryland Management: Economic Case Studies, Earthscan Publications, Ltd., London, p. 294-303.

Heady, Harold F., and James Bartolome, 1977. The Vale Rangeland Rehabilitation Program: The Desert Repaired in Southeastern Oregon. USDA Forest Service Resource Bulletin PNW-70. 139 p.

Holmberg, G., 1990. An economic evaluation of soil conservation in Kitui District, Kenya. In John A. Dixon, Paul B. Sherman, and David E. James (editors), Dryland Management: Economic Case Studies, Earthscan Publications, Ltd., London, p. 56-71.

Hopkins, Stephen T., and Douglas E. Jones, 1983. Research Guide to the Arid Lands of the World. Oryx Press, Phoenix, Arizona. 391 p.

IFAD, 1987. Programme special de conservation des eaux et des sols et agroforesterie dans le plateau central. Fond International de Developpement Agricole Rapport No. 0172-BF, Rome, Italy. 147 p.

Kothmann, M. M., et al. (editors), 1974. A glossary of terms used in range management. Second edition. Society of Range Management, Denver, Colorado. 36 p.

LeHouérou, H. N., 1989. The Grazing Land Ecosystems of the African Sahel. Springer-Verlag, Heidelberg. 282 p.

UNCOD, 1977. Round-up, Plan of Action and Resolutions. United Nations Conference on Desertification, Nairobi, Kenya. 43 p.

UNESCO, 1979. Map of the World Distribution of Arid Regions. Explanatory Note, UNESCO, Paris. 54 p. Map scale 1/ 25,000,000.

United Nations, 1980. Study on financing the United Nations Plan of Action to Combat Desertification. United Nations General Assembly, A/35/396, New York. 128 p.