Restoring Value to the WorId’s Degraded Lands

by Gretchen C. Daily (1995)

The author is with Energy and Resources Group, Building T-4, Room 100, University of California, Berkeley, CA 94720, USA.

Roughly 43 percent of Earth’s terrestrial vegetated surface has diminished capacity to supply benefits to humanity because of recent, direct impacts of land use. This represents an ~10 percent reduction in potential direct instrumental value (PDIV), defined as the potential to yield direct benefits such as agricultural, forestry, industrial, and medicinal products. If present trends continue, the global loss of PDIV could reach ~20 percent by 2020. From a biophysical perspective, recovery of ~5 percent of PDIV is feasible over the next 25 years. Capitalizing on natural recovery mechanisms is urgently needed to prevent further irreversible degradation and to retain the multiple values of productive land.

Rehabilitation of the world’s degraded lands is important for several reasons. First, increasing crop yields is crucial to meeting the needs of the growing human population (1) for food, feed, biomass energy, fiber, and timber (in the absence of a massive increase in the equity of global resource distribution (2). Second, anthropogenic changes in land productivity have deleterious impacts on major biogeochemical cycles that regulate greenhouse gas fluxes and determine Earth’s total energy balance (3). Third, biodiversity preservation depends, in part, on increasing yields on human-dominated land to alleviate pressure to convert remaining natural habitat (4). And fourth, land is frequently a limiting factor of economic output, and its degradation threatens to undermine economic development in poor nations (5, 6) and social stability globally (7).

Here I estimate the rate at which potential direct instrumental value (PDIV) could be restored to degraded lands from a biophysical (as opposed to socioeconomic) perspective. PDIV is the capacity of land to supply humanity with direct benefits only, such as agricultural, forestry, industrial; and medicinal products. It does not incorporate indirect values [for example, ecosystem services (8)], option values, or nonuse values (9) and is thus a conservative measure of value. PDIV is not the same as potential net primary production (NPP), and may even vary inversely with it; for example, average NPP in agricultural systems is typically lower (and DIV higher) than in the natural systems they replace (10). Because PDIV depends on complex and variable factors such as human knowledge and preferences, it is impossible to quantify precisely.

Below I make rough approximations of changes in PDIV on the basis of global surveys of human-induced land degradation. Case histories of recovery from natural or human-induced disturbance are reviewed in order to derive estimates of the time required to restore PDIV to presently degraded lands. Finally, potentially illuminating projections are offered of future changes in PDIV.

Global Extent and Severity of Land Degradation

Land degradation refers to a reduction in the capacity of land to supply benefits to humanity. It results from an intricate nexus of social, economic, cultural, political, and biophysical forces operating across a broad spectrum of time and spatial scales (11). Here I consider only the biophysical agents of degradation that trace directly to human land use since 1945, although other proximate biophysical agents, such as air pollution (12), stratospheric ozone depletion (13), and climate change (14), are also important.

The geographic distribution of degraded land is poorly documented; even less well documented is the severity of degradation, which is typically judged qualitatively (15). The onset of degradation is often masked by intensification of land use that compensates, in the short run, for declines in the natural underpinnings of productivity; however, intensification usually exacerbates degradation, as do natural positive feedbacks (such as the concentration of soil resources by shrubs) (16). Global assessments have been undertaken of degradation of soils (in all biomes), drylands, and tropical forest lands.

Soil degradation. The extent of soil degradation induced by human activity since 1945 was evaluated as ~2 billion ha, or 17% of Earth’s vegetated land, in a recent study sponsored by the United Nations Environment Program (UNEP) (17). Of this, ~750 million ha (38%) are classified as lightly degraded (defined as exhibiting a small decline in agricultural productivity and retaining full potential for recovery); ~910 million ha (46%) are moderately degraded (exhibiting a great reduction in agricultural productivity; amenable to restoration only through considerable financial and technical investment); ~300 million ha (15%) are severely degraded (offering no agricultural utility under local management systems; reclaimable only with major international assistance); and ~9 million ha (0.5%) are extremely degraded (incapable of supporting agriculture and unreclaimable).

The percent of area affected seems regionally to be independent of ecological zone or economic status; for example, it is 20%, 22%, and 23% in Asia, Africa, and Europe, respectively. The direct causes of these forms of degradation (and estimates of the relative importance of each) are overgrazing (35%), deforestation (30%), other agricultural activities (28%), overexploitation for fuel wood (7%), and bioindustrial activities (1%). Global rates of change in soil degradation are unknown. The UNEP study constitutes the first standardized global assessment and is the baseline for planned future monitoring on a decadal basis.

Drylands degradation. UNEP has also carried out a series of generally accepted global assessments of desertification (15). In the most recent assessment, desertification refers to land degradation in arid, semiarid, and dry subhumid areas (hereafter called drylands) resulting mainly from adverse human impact (18). Desertification is distinct from natural oscillations of vegetation productivity that occur at desert fringes (19); hyperarid deserts are not considered to be at risk of desertification and are excluded from assessments thereof (18).

The total decertified drylands area amounts to ~3.6 billion ha, or 70% of global drylands area (excluding hyperarid regions). Roughly 2.6 billion ha thereof exhibit no soil degradation, but have reduced crop yields, livestock forage, and woody biomass for fuel and building material (20). Of rangelands, which make up 88% of the drylands area, ~1.223 billion ha (27%) are degraded slightly or not at all; ~1.267 billion ha (28%) are moderately degraded; ~1.984 billion ha (44%) are severely degraded; and ~72 million ha (1.6%) are very severely degraded. Degradation classes are roughly comparable with those defined in the soil survey, and the principal direct causes of degradation are the same ( 18, 21). The rate of abandonment of drylands due to degradation is probably ~9 to 11 million ha year-1 (22). Rates of degradation seem to be accelerating, particularly in developing nations (23).

Tropical moist forest degradation. Land degradation in tropical moist forest afflicts ~427 million ha (24). The present global annual rate of tropical forest clearing (defined as depletion of forest cover to less than 10% in all types of tropical forest) is ~15.4 million ha year-1 (25) and is projected to accelerate (26, 27). In addition, an area of roughly equal size is disrupted, but not cleared outright, through selective logging and shifting cultivation (26). The extent to which clearing and disruption precipitate land degradation is unknown. Rates of abandonment of recently cleared areas, especially in hilly or mountainous regions, of as high as 75 to 100% are indicative of one extreme (26, 28). In general, probably only ~50% of the tropical forest land cleared each year expands the area yielding agricultural benefits, whereas the other half replaces abandoned lands (29).

Total degraded area. As a crude but conservative estimate of the total degraded area, I use the sum of (i) areas affected by soil degradation, (ii) drylands with vegetation degradation but no soil degradation, and (iii) degraded tropical moist forest lands, that is, ~5.0 billion ha (30). This amounts to ~43% of Earth’s vegetated surface.

Time Required to Restore PDIV

There have been few attempts to rehabilitate degraded land on a large scale. Possible rates of recovery can be inferred from studies of succession on land that has experienced volcanic eruption, shifting cultivation, continuous agricultural production followed by abandonment, or reclamation. The time required to restore PDIV varies tremendously with ecosystem type, history and spatial pattern of land use, the degree of alteration of climatic factors, and the types of benefits ultimately desired — those derived from crop cultivation as compared to those derived from extractive exploitation of natural vegetation, for example.

Volcanic eruption. Following volcanic eruption, the regeneration of lost top- and subsoil may be the limiting process with respect to time and difficulty of rehabilitation. At one extreme, the rate of topsoil formation is especially rapid on volcanic ash; a mere 100 years after the 1883 eruption of Krakatau, for example, soil 25 cm deep had formed on a daughter island, Rakata (31). More typical soil formation rates are ~1 cm per 100 to 400 years, however. At such rates it takes ~3000 to 12,000 years to develop sufficient soil to form productive land (32).

Rates of colonization and succession are comparatively swift in the absence of natural impediments (33). Rakata serves as a model for the recovery of a presumably sterilized site 40 km from species source pools (34). Many generalist groups with high dispersal capabilities became reestablished during the first 50 years after eruption. However, important taxa with lower dispersal capabilities, more specialized resource requirements, or higher trophic positions remain poorly represented even today (35). Similarly, 23 vascular plant species were present on Surtsey two decades after its birth (of 450 on the Iceland mainland 35 km away), but only a few had become widely established (36).

Shifting cultivation. Shifting (swidden) cultivation generally involves slashing and burning of forest patches to create temporary fields that are harvested in a rotation between brief periods of cultivation and longer periods of fallowing. Cultivation typically lasts 1 to 3 years, during which a combination of declining soil fertility, competition from weeds, and pest or pathogen outbreak conspires to diminish yields sharply (37). The plot is then left fallow. Longterm studies of recovery of productive potential in swidden systems are few (38), but fallow periods required to make a system sustainable are ~20 years (ranging between 5 and 40 years) in the humid tropics and may be considerably longer elsewhere (39).

Abandoned cropland and pasture. Rates and paths of natural succession vary widely on abandoned land formerly under continuous agricultural production. The chief commonality is the nonlinear relation between the intensity and duration of land use and the time required for recovery after abandonment. Factors influencing succession on old-fields (land abandoned after some combination of cropping and pasturing) are extremely complex, but the severity of erosion, initial floristic composition, and character of the ex situ seed source are paramount (40). In some areas, initial reestablishment of climax species was observed after 40 years of abandonment; in contrast, highly eroded fields experienced little succession during that period (41).

The conversion to pasture of up to ~43 million ha of Amazon rain forest over the past three decades (42) caused rapid declines in productivity and land abandonment after only 4 to 8 years of use (43). Extrapolation of rates of biomass accumulation and succession over 8 years since abandonment suggests that sites with a history of light use (20% of now-abandoned pasture) could reach forest stature in 100 years, those of moderate use (~70%) in 200 years, and those of heavy use (less than 10%) in 500 years or more (44). These estimates assume no further human impact. In many situations worldwide, recovery of productivity on abandoned land is prevented by burning (45) or episodic human exploitation of regrowth as it occurs.

Even without continued human disruption, however, regrowth of forest may not occur at all (as in the case of fire-climax grasslands (46, 47). For example, an agricultural area of ~3.5 million hectares in eastern Amazonia that was abandoned in the early part of this century had little vegetation aside from scrub and brush 50 years later (48). In India, trees have failed to establish in abandoned, decertified areas adjacent to sacred forest groves despite ample seed sources (49).

Reclamation. Experience in reclamation of degraded areas, although limited, indicates unequivocally that human intervention may be effective (even essential) in ensuring a path and rate of succession that would achieve substantive improvements at time scales relevant to society (50). The potential for accelerating recovery is difficult to assess, as most degraded areas with known histories have not yet recovered. Moreover, recovery is nonlinear (with respect to time), and intervention can only accelerate some phases of the process.

Where land is suited to direct human use and has not been stripped of topsoil, substantial recovery may be achieved in as few as 3 to 5 years with intensive management (51) but more typically may take 20 years (52). However, recovery of self-sustaining, mature ecosystems in areas unsuited for intensive agriculture may take 100 years or more.

Projections of Future Land Productivity

Despite great uncertainties, I venture crude estimates of the present global loss of PDIV and possible future changes therein. The light, moderate, severe, and extreme degradation classes are assumed to correspond to a residual PDIV of 90%, 75%, 50%, and 0%, respectively (Table 1, column 1). These values are conservative in that severely (as well as extremely) degraded land is generally abandoned (17).

It is further assumed that the distribution among classes of the ~5 billion hectares of degraded land is proportional to that of degraded land in the global soil survey (summarized above), for which the data appear most reliable (Table 1, column 3) (53). On the basis of the foregoing evaluation of natural and human-accelerated recovery rates, rough rehabilitation times are proposed for each class of land (Table 1, column 2) (54). These estimates are optimistic in that all assume the higher rate of recovery from ranges of possibilities and that rehabilitation will not be hindered by soil loss, lack of colonists, climate change, further human impact, or other important factors.

Three scenarios of future changes in global PDIV are considered (Table 1) (55). In scenario A, degradation is arrested immediately. In 25 years, complete recovery occurs on 100% of land in the light class and on 50% of land in the moderate class; the other 50% in the moderate class moves into the light category; 0% of the land in the severe and extreme classes recover sufficiently to move up into another class. Scenario B assumes conservative rates of growth of each degradation class (derived in Table 2). Scenario C assumes rates of degradation double those used in B; these accelerated rates approximate what could occur if vigorous measures to prevent and reverse land degradation are not taken.

The analysis suggests that ~ 10% of global PDIV of land has already been lost (Table 3). From a biophysical perspective, recovery of half of this loss may be feasible in 25 years, provided that degradation is halted and strong rehabilitation measures are initiated immediately. In the absence of such measures, a very conservative extrapolation of present rates of degradation suggests that ~16% of global PDIV could be lost in 25 years. At more realistic, accelerated rates of degradation, this loss could reach ~20%. In the latter scenario (C), the land area irreversibly degraded from a socioeconomic perspective (in the severe and extreme classes) would increase by a factor of 1.8 over 1995 levels. These results are most useful for relative, rather than absolute, comparisons.

Costs and Benefits of Rehabilitation

Although a general lack of information on rehabilitation costs constitutes a serious shortcoming (56), the utter dependence of human well-being on productive land makes its continued degradation for short term gain an unwise course. Moreover, the costs of off-site degradation may be substantial (57).

UNEP estimates the direct, on-site cost of failure to prevent desertification during the period 1978 to 1991 at between $300 billion and $600 billion (in U.S. dollars) (58). Currently, the total direct, on-site income foregone as a result of desertification is ~$42.3 billion year- 1 By contrast, UNEP’s estimates of the direct annual cost of all preventive and rehabilitational measures range between $10.0 billion and $22.4 billion.

An enormous potential for recovery is inherent in most land types, but failure to realize this potential can result in rapid, essentially irreversible deterioration. Historically, land degradation has been implicated in the fall of great civilizations (59) and merits serious attention by this one (60).

REFERENCES AND NOTES

1. The United Nations medium-range projection indicates that the world population may reach 10 billion by 2054 ultimately 11.6 billion before halting growth

[United Nations Population Division, Long-Range World Population Projections (United Nations, ST/ SEA/SER.A/125, NY, 1991); United Nations, Population Newsletter, 7 June 1994].

2. J. Perlin, A Forest Journey (Norton, New York, 1989); G. C. Daily and P. R. Ehrlich, BioScience 42, 761 (1992); D. O. Hall, F. Rosillo-Calle, R. H. Williams, in Renewable Energy Sources for Fuels and Electricity, T. B. Johansson et al., Eds. (Island Press, Washington, DC, 1993), pp. 593-652; P. R. Ehrlich, A. H. Ehrlich, G. C. Daily, The Stork and the Plow (Putnam, New York, in press).

3. H. A. Mooney, P. M. Vitousek, P. A. Matson, Science 238, 926 (1987); J. T. Houghton, G. J. Jenkins, J. J. Ephraums, Eds., Climate Change: The IPCC Scientific Assessment (Cambridge Univ. Press, Cambridge,1990); R. A. Houghton, Can. J. For Res. 21, 132 (1991); W. H. Schlesinger, Biogeochemistry: An Analysis of Global Change (Academic Press, San Diego, CA,1991).

4. N. Meyers, A Wealth of Wild Species (Westview, Boulder, CO, 1983); P.R. Ehrlich and A. H. Ehrlich, Ambio 21, 219 (1992); E. Barbier, J. Burgess, C. Folke, Paradise Lost? The Ecological Economics of Biodiversity (Earthscan, London, 1994).

5. C. A. S. Hall, in Ecosystem Rehabilitation, M. K. Wali, Ed. (SPB, The Hague, 1992), vol.1, pp. 101-126; M. Gadgil, Ambio 22, 167 (1993).

6. T. N. Khoshoo, in Ecosystem Rehabilitation, M. K. Wali, Ed. (SPB, The Hague, 1992), vol. 2, pp. 3-17.

7. T. Homer-Dixon, J. Boutwell, G. Rathjens, Sci. Am. 268, 38 (February 1993); N. Myers, Ultimate Security: The Environmental Basis of Potential Stability (Norton, New York, 1993).

8. P. R. Ehrlich and H. A. Mooney, BioScience 33, 248 (1983).

9. For a discussion of economic valuation, see D. W. Pearce and J. J. Warford, World Without End (Oxford Univ. Press, Oxford, 1993).

10. G. L. Atjay, P. Ketner, P. Duvigneaud, in The Global Carbon Cycle, B. Bolin, E. T. Degens, S. Kempe, P. Ketner, Eds. (Wiley, NY, 1979), p. 129; P. M. Vitousek, P. R. Ehrlich, A. H. Ehrlich, P. A. Matson, BloSclence 36, 368 (1986).

11. A. Chisholm and R. Dumsday, Eds., Land Degradatlon (Cambridge Univ. Press, Cambridge, 1987); C. J. Barrow, Land Degradation (Cambridge Univ. Press, Cambridge, 1991); P. Dasgupta, An Inquiry into Well-Belng and Destitution (Clarendon, Oxford, 1993).

12. Biological Markers of Air-Pollution Stress and Damage In Forests (National Academy of Sciences, Washington, DC, 1989); O. Loucks, in Changing the Global Environment, D. B. Botkin, M. F. Caswell, J. E. Estes, A. A. Orio, Eds. (Academic Press, London, 1989), p.101; W. L. Chameides, P. S. Kasibhatla, J. Yienger, H. Levy II, Science 264, 74 (1994).

13. R. Worrest and L. Grant, in Ozone Depletion: Health and Environmental Consequences, R. Jones and T. Wigley, Eds. (Wiley, NY, 1989), p.197.

14. G. C. Daily and P.R. Ehrlich, Proc. R. Soc. London Ser. B 241, 232 (1990); M. Parry, Climate Change and World Agriculture, (Earthscan, London, 1990).

15. Some valid criticisms of degradation assessments can be found in B. Forse [New Sci. 1650, 31 (1989)] and in F. Pearce [ibid. 1851, 38 (1992)]; for an overview, see J. L. Dodd, BioScience 44, 28 (1994). Improved, quantitative-methods of assessing land degradation, particularly by remote sensing, are available but remain little used, for example, A. K. Tiwari and J. S. Singh, Environ. Conserv. 14, 233 (1987); T. A. Stone and P. Schlesinger, in Proceedings of the IUFRO S4.02.05 Wacharakitti International Workshop, 13-17 January 1992, H. G. Lund, R. Paivinen, S. Thammincha, Eds. (IUFRO, Thailand, 1992), pp. 85-93; P. Aldhous, Science 259, 1390 (1993); B. V. Vinogradov, Eurasian Soil Sci. 25, 66 (1993).

16. W. H. Schlesinger et al., Science 247,1043 (1990); D. Ponzi, Desertifcation Control Bull. 22, 36 (1993).

17. The UNEP study is by L. R. Oldeman, R. T. A. Hakkeling, W. G. Sombroek, World Map of the Status of Human Induced Soil Degradatlon: An Explanatory Note, rev. (International Soil Reference and Information Center, Wageningen, Netherlands, rev. ed. 2, 1990); see also D. Pimentel, Ed., World Soil Erosion and Conservation (Cambridge Univ. Press, Cambridge, 1993).

18. Status of Desertification and Implementation of the United National Plan of Actlon to Combat Desertification (United Nations Environment Program, Nairobi, Kenya, 1991); H. Dregne, M. Kassas, B. Rozanov, Desertification Control Bull. 20, 6 (1992).

19. C. J. Tucker, H. E. Dregne, W. W. Newcomb, Science 253, 299 (1991).

20. Vegetation degradation occurs in biomes other than drylands. For example, Imperata spp. is an unpalatable weed that occupies an estimated 40 to 100 million ha of potential cropland in Southeast Asia [references in (24); R. A. Houghton, Ambio 19, 204 (1990)]. It forms dense, rhizomatous mats that make cultivation impossible. Rehabilitation has only been achieved with complex, costly, and labor-intensive methods that have not been successfully applied on a wide scale [J. H. H. Eussen and W. De Groot, Wet Gent 39, 451 (1974)].

21. H. E. Dregne, Desertification of Arid Lands (Harwood, Chur, Switzerland, 1983).

22. This estimate is calculated from (18); earlier estimates (53) were ~26 million ha year^-1.

23. A. Grainger, The Threatening Desert (Earthscan, London, 1990); W. Parham, P. Durana, A. Hess, Eds., Improving Degraded Lands: Promising Experiences from South China (Bishop Museum, Honolulu, 1993).

24. This area comprises ~137 million ha of logged tropical moist forest undergoing regeneration; ~203 million ha of rain forest in fallow state under shifting cultivation, much of which is no longer sustainable; and ~87 million ha of cleared montane forest regions [A. Grainger, Int. Tree Crops J. 5, 31 (1988); adapted in part from J. P. Lanly, Ed., Tropical Forest Resource Assessment Project: Tropical Africa, Tropical Asia, Tropical America. (FAO/ UNEP, Rome, 1981)].

25. Food and Agriculture Organization of the United Nations, Forest Resources Assessment 1990; Tropical Countries (FAO, Rome, 1993); this estimate is subject to debate [N. Myers, in The Causes of Tropical Deforestation, K. Brown and D. W. Pearce, eds. (University College Press, London, 1994), pp. 27-40; D. Skole and C. Tucker, Science 260, 1905 (1993)]

26. N. Myers, Deforestation Rates in Tropical Forests and Their Climatic Implications (Friends of the Earth, London, 1989).

27, _________, Environ. Conserv. 20, 9 (1993).

28. R. Buschbacher, C. Uhl, E. Serrao, in (6), pp. 257-274.

29. R. A. Houghton, BloSclence 44, 305 (1994); increasingly, abandoned lands do not revert to forest (27).

30. This ignores all other land types.

31. S. Hardjowigeno, Geojournal 28, 131 (1992).

32. J. Skoupy, Desertification Control Bull. 22, 5 (1993); F. R. Troeh and L. M. Thompson, Soils and Soil Fertility (Oxford Univ. Press, New York, ed. 5,1993).

33. Impediments include (in Amazonia, for example) the aerial extent of pastures, which frequently cover hundreds or even thousands of hectares and make reestablishment of forest problematic because ~90% of the tree species have animal seed dispersers, of which very few venture into open pasture (43). Seed and seedling predators, particularly some ant species, are abundant in pasture and will remove experimentally placed seeds within minutes of placement [C. Uhl, in Biodiversity, E. O. Wilson, Ed. (National Academy Press, Washington, DC, 1988), pp. 326-332]. Microclimatic conditions (air and soil temperatures and humidity) are harsh and further limit seedling survival [C. Uhl, J. Eco. 75, 377 (1987)].

34. M. B. Bush and R. J. Whitaker, J. Biogeogr. 18,341 (1991); I. W. B. Thornton and T. R. New, Geojournal 28, 219 (1992); but see R. W. Suatmadji, A. Coomans, F. Rashid, E. Geraert, D. A. McLaren, Philos. Trans. R Soc. London Ser. B 322, 369 (1988).

35. I. W. B. Thornton, T. R. New, R. A. Zann, P. A. Rawlinson, Philos. Trans. B. Soc. London Ser. B 328, 131 (1990).

36. S. Fridriksson, Arct. Alp. Res. 19, 425 (1987); Environ. Conserv. 16,157 (1989).

37. D. R. Harris, Geogr. Rev. 61, 475 (1971); H. Ruthenberg, Farming Systems in the Tropics (Clarendon, Oxford,1971).

38. Most experimental studies monitor the recovery of sites that were slashed and burned but not actually cultivated [but see O. P. Toky and P. S. Ramakrishnan, Agro-Ecosystems 7, 11 (1981)].

39. W. J. Peters and L. F. Neuenschwander, Slash and Burn: Farming in the Third World Forest (Univ. of Idaho Press, Moscow, ID, 1988).

40. F. E. Egler, Vegetatio 4, 412 (1954); E. D’Angela, J. M. Facelli, E. Jacobo, ibid. 74, 39 (1988); D. D. Steven, Ecology 72, 1076 (1991); R. W. Myster and S. T. A. Pickett, ibid. 75, 387 (1994).

41. C. Keever, Ecol. Monogr. 20, 229 (1950); F. A. Bazzaz, Ecology 49, 924 (1968).

42. P. M. Fearnside, Ambio 22, 537 (1993)

43. C. Uhl et al., J. Ecol. 76, 663 (1988).

44. R. Buschbacher, C. Uhl, E. Serrao, in (6), pp. 257-274.

45. C. Uhl and R. Buschbacher, Biotropica 17, 265 (1985).

46. Grasses often invade areas cleared of other types of vegetation, thereby establishing ecological conditions favoring their persistence [C. M. D’Antonio and P. M. Vitousek. Annul Rev. Ecol. Syst. 23, 63 (1992)]. A state of arrested succession and lowered soil fertility often results, which sometimes strongly hinders efforts to establish crops or other vegetation (20).

47. J. Ewel, Biotropica 12, 2 (1980); M. Kellman, ibid., p. 34,

48. E. G. Egler, Rev. Bras. Geogr. 23, 527 (1961).

49. P. S. Ramakrishnan, in (6), pp. 19-35.

50. J. Walls, Ed, Combating Desertification in China (United Nations Environment Program, Nairobi, Kenya, 1982); J. Ewel, Annu. Rev. Ecol. Syst. 17, 245 (1986); E. B. Allen, Ed., The Reconstruction of Disturbed Arid Lands (Westview, Boulder, CO, 1988); J. Cairns, Ed., Rehabilitating Damaged Ecosystems (CRC, Boca Raton, FL, 1988); R. Lal and B. A. Stewart, Eds. Advances in Sod Science: Soil Restoration (Springer-Verlag, NY, 1992); D. Saunders, R. Hobbs, P. Ehrlich, Eds., Reconstruction of Fragmented Ecosystems (Surrey Beatty, Chipping Norton, 1993); B. H. Walker, Ambio 22, 80 (1993).

51. V. M. Kline and E. A. Howell in Restoration Ecology, W. R. Jordan III, M. E. Gilpin, J. D. Aber, Eds. (Cambridge Univ. Press, Cambridge, 1987), pp. 75-83; P. Singh, in (6), pp. 51-61.

52. P. M. Blaschke, N. A. Trustnum, R. C. DeRose, Agric. Ecosyst Environ. 41, 153 (1992).

53. This value is conservative relative to other estimates that classify only 57% (as opposed to my assumption of 84%) of arid lands as retaining at least 75% of their productivity [H. E. Dregne, Desertification of Arid Lands (Harwood, Chur, Switzerland, 1983); J. A. Mabbutt, Environ. Conser. 11, 103 (1984), Desertification Control Bull. 12, 1 (1985).

54. These are comparable to estimates given in N.-T. Chou and H. E. Dregne, Desertification Control Bull, 22, 20 (1993).

55. These projections assume that the average PDIV of degraded land is equivalent to that of nondegraded land. This may be an exaggeration, but the issue is complex and geographically varied; moreover, the assumed rates of recovery are sufficiently optimistic as to make the overall calculations conservative.

56. But see J. A. Dixon, D. E. James, P. B. Sherman, Eds. Dryland Management: Economic Case Studies (Earthscan, London, 1990); M. M. Mattos and C. Uhl, World Dev. 22, 145 (1994).

57. G. Upstill and T. Yapp in Land Degradation: Problems and Policies, A. Chisholm and R. Dumsday Eds. (University of Cambridge, Melbourne, Australia, 1987), chap. 5; E. H. Clark, J. A. Haverkamp, W. Chapman, Eroding Soils: The Off-Farm Impacts (The Conservation Foundation, Washington, DC, 1985).

58. This excludes other costs such as human suffering (18).

59. R. McC. Adams, Heartland of Cities (Univ. of Chicago Press, Chicago, 1981).

60. I gratefully acknowledge comments by R. Adams, A. Ehrlich, P. Ehrlich, R. Hanson, J. Holdren, K. Holl, P. Matson, S. Schneider, P. Vitousek, and two anonymous reviewers. Supported by the Winslow Foundation, the Heinz Foundation, and W. Alton Jones Foundation, an anonymous grant, and by P. Bing and H. Bing.

*These rates are derived from the mean rate of land degradation from 1945 to 1990, assuming that all currently degraded land was so rendered during that period (5.0 x 10⁹ ha per 45 years = 111 x 10⁶ ha year^-1). From this conservative estimate (given that rates of degradation have accelerated) is subtracted the rate of degradation to the severe and extreme classes, yielding 98.5 x 10⁶ ha year^-1. Equal rates of growth of the light and moderate classes are assumed and degradation in tropical moist forest (TMF) subtracted, yielding 0.5 (98.5 x 10⁶) – (14.2 x 10⁶) = 35.0 x 10⁶ ha year^-1.

†These rates assume that 100% of disturbed (but not clear-cut) TMF become lightly or moderately degraded, along with 84% of the clear-cut TMF (as 15% and 1% of the latter become severely and extremely degraded, assuming the same proportions as found in the soil degradation survey (17)]. Assuming equal partitioning between the light and moderate classes, the rate is 0.5 [(15.4 x 10⁶) + 0.84 (15.4 x 10⁶)] ha year^-1 for each class.

‡Assuming 10 x 10⁶ ha year^-1 become severely or extremely degraded (see above) in the same relative proportions as reported for the soil degradation survey (17), namely, 15/16 and 1/16, respectively.

§Fifteen percent of clear-cut forest [0.15 (15.4 x 10⁶) ha year^-1].

||One percent of clear-cut TMF [0.01(15.4 x 10⁶) ha year^-1].