Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

What are the physical effects of deforestation?

Let us proceed from the labyrinth of uncertainty about the causes and effects of the pressures on the mountain forests to a discussion of the hypothetical impacts of deforestation, assuming that the World Bank assessment is correct and the mountains of Nepal will indeed be laid bare by AD 2000. The view that deforestation automatically and most certainly will produce devastating soil erosion, overland flow of water, rill and gully development, rainy-season flooding and dry-season water shortage is so widespread that it is usually taken as one of the fundamental truisms within the concept of 'one small earth'. Where deforestation occurs on mountain slopes, the environmental degradation is assumed to be all the greater than deforestation on the plains, with local damage being proportionate to angle of slope. Linked to this is the assumption that the greater the slope angle the greater will be the impacts on the plains downstream - more destructive floods, siltation, extension (in our case) of the Brahmaputra-Ganges delta, and so on. Hamilton (1987), under the challenging heading 'Myth and Misunderstanding' develops a reasoned attack on these important conventional assumptions.

Hamilton (1987) first argues that the very term deforestation is used so ambiguously that it is virtually meaningless as a description of land-use change: it should be replaced with more precise terms that would be less charged with emotion and better suited to specific areas. It has been used to refer to any or all of the following activities with respect to existing forest: fuelwood cutting; commercial logging; shifting cultivation; forest clearing for continuous annual cropping, for grazing, for food, beverage, or industrial tree-crops, or forest plantations. It has also been used to describe the process of burning, for various reasons, with a degraded forest as the end product. Other activities could be added. For our present purpose - discussion of the hypothetical effects of deforestation - since these various activities produce different hydrological and erosional results, it is emphasized that lack of differentiation only confuses attempts to examine the effects.

Fuelwood cutting, where the products are carried out of the forest manually, will have very different impacts to those resulting, for instance, from commercial logging with heavy equipment and where skidding trails and logging roads may occupy as much as 16-30 percent of the logged area. Similarly, the conversion of forest to sustained annual cropping will have very different impacts depending on the type of crops, and the development (or lack of development) of properly constructed and maintained agricultural terraces, and the application of other conservation practices. 'The aesthetically pleasing, hydrologically and erosionally benign, productive paddy terraces that characterize many parts of the lower slopes of the Himalaya are deforested. Yet generally deforestation is perceived to be always destructive, if not evil' (Hamilton, 1987:257).

From this we must conclude that it is not the cutting of trees per se that causes large increases in on-site erosion, but the methods of cutting and of transporting the products out of the area, and finally the subsequent form of use of the cut-over land (Hamilton, 1983). In most of the humid Himalayan area, harvesting activity should not be described as deforestation unless it is labelled as temporary deforestation. If the affected area is left, and not burned, grazed, or cultivated, a rapid establishment of new reproduction will occur and a spurt in understorey growth will ensure that little bare ground is exposed to high intensity rainfalls. Realization of the confusion caused by semantics led Hamilton (1983) to develop the following headings for chapters of a book on the hydrologic and pedologic aspects of tropical forested watersheds:


1. Harvesting minor forest products
2. Shifting agriculture
3. Harvesting fuelwood and lopping fodder
4. Harvesting commercial wood
5. Grazing on forest land
6. Burning forest land


7. Conversion to forest tree plantations
8. Conversion to grassland or savanna for grazing
9. Conversion to food or extractive food crops
10. Conversion to annual cropping
11. Conversion to agroforestry


12. Reforestation or afforestation

Hamilton argues that even with this degree of specificity, the biophysical consequences of these activities are complex, unknown, or ambiguous. He believes that the emotive term deforestation only compounds the problem and should be eliminated from both technical and popular thinking unless used with qualifications to indicate the actual pattern of forest land-use change.

Having quickly reviewed the semantics let us now go on to address some of Hamilton's stipulated myths and misunderstandings. One example of a widespread misunderstanding is the oft-repeated assertion that'tree roots are a sponge' which soak up rainwater during wet seasons and release it slowly during dry seasons - that is, that the forests act as hydrological regulators (Spears, 1982; Myers, 1983). The implications are that cutting down forests causes floods during the summer monsoon season or during major storms, and results in lower water levels, or in rivers totally drying up during the dry seasons (Eckholm, 1976; Sharp and Sharp, 1982; and many others). Watershed research, however, suggests that tree roots behave more like a pump than a sponge. Forest cutting on 94 small catchments world wide, compared with uncut control catchments, resulted in more total water yield from the cut-over areas, with the increase being proportional to the intensity of the cutting (Bosch and Hewlett, 1982). This occurs because tall forests evaporate and transpire more water than other types of vegetation. Furthermore, Hamilton's (1983) review of catchment research showed that in most cases following cutting, increased flows occurred throughout the year, including the dry season. Gilmour (1977), for instance, demonstrated in the humid Australian tropics that a stream that previously had ceased to flow during many dry periods flowed perennially after logging. Hamilton (1987) points out that a notable exception to this effect is the case of cloud forests that intercept precipitation from wind-driven clouds.

The line of reasoning introduced above can be applied to a variety of other myths concerning deforestation. These include: the alleged role of the upper forest canopy in protecting the soil from raindrop impact; lowering of the groundwater table; water yield; and timing and distribution of stream flow. To provide one more example, we will examine the contention that removal of a high-forest canopy by exposing the soil surface to direct impact of raindrops will result in increased soil erosion. It is necessary to take into account the fact that, with an intact high-forest canopy rainwater will accumulate on the leaves and fall to earth as much bigger drops than raindrops. Also, given a height in excess of 20 m the terminal velocity of the larger water drops will not be significantly different from that of raindrops falling on the ground unimpeded. Furthermore, it is the litter and the low vegetation that protects the soil from particle detachment by raindrop impact. Fire or grazing, which remove low vegetation, and litter gathering that exposes bare soil should be reduced, rather than worrying about reduction in the tree canopy if the principal concern is sheet erosion. With a surge in productivity of a shrub and herb layer following forest clearing with more solar radiation reaching ground level, the soil surface may be actually better protected against raindrop impact. Once more, therefore, we are prompted to caution against the dangers of generalization. Hamilton (1987) concludes with the following summary:

1. Cutting of forests, and also their replacement with a different land use (unless the soil surface is sealed), have usually resulted in the free watertable moving closer to the surface, because less water is lost to evapotranspiration.

2. Most forest-cutting catchment research has shown greater streamflow throughout the year, and many studies have shown proportionately greater increases in dry-season flow than in wet-season flow.

3. If cutting of tall forests is followed by conversion to other conservationist land uses then gains in streamflow yield persist.

4. Tree cutting alone does not lead to increases in surface erosion - what is important is the mode of forest harvesting (dragging, skid trails, roads, and especially large-scale mechanization).

5. Much of the sediment in a given river basin is in various temporary storages; thus claims of immediate sediment reduction on a large scale and with fardistant effects following any reforestation project should be critically reexamined.

Some people have blamed 'deforestation' for the catastrophic flooding in June 1985, that killed 237 persons in six Indian states (102 in Kerala alone and 400,000 homes destroyed or damaged) (Reuters, 1985). The real cause was too much rain occurring in too short a time. Some areas received 414 mm in 24 hours, in the monsoon season when soils were already saturated. Even if the entire basin had been forested there would have been disastrous floods. Forests are beautifully functioning ecosystems, capable of producing many products and services on a sustained basis. They must in places be totally protected, and in other places wisely managed. But they will not prevent floods or sedimentation in the lower reaches of major rivers, nor significantly reduce flooding in major storm events. (Hamilton, 1987:262)

The deforestation - changes in hydrologic conditions - soil erosion - sediment-transfer linkages make up yet another instance where assumptions, emotions, and widespread generalizations have been pulled into the Theory of Himalayan Environmental Degradation regardless of the available facts. The difficulty of countering these assumptions, or attempting to expose the myths surrounding them, is exacerbated by the fact that reliable data are available only from a very few sites, most of them located in other parts of the world. A compounding problem is that deforestation, in its many forms, and forest degradation are occurring, and these processes are inextricably linked with the problem of sustaining a viable subsistence agriculture for the great majority of the population that inhabits the lower altitudinal belts of the Himalayan region. Nevertheless, we will conclude this section with the presentation of the results of a detailed and narrowly site-specific study that we believe throw valuable light on several of the critical points discussed above. Gilmour (1986) and Gilmour et al. (1987), within the overall scope of the Nepal-Australia Forestry Project, set out to examine the rather important relationships between rainfall amounts and short-period intensities and soil erosion in the context of type of surface vegetation cover and land use. Gilmour et al. (1987) instrumented five study plots close to the Sindhu Palchok district headquarters of Chautara, about 40 km northeast of Kathmandu. The study area lies at about 1,660 m above sea level and experiences a typical monsoon climate: about 85 percent of the total annual precipitation occurs between June and October.

The five study sites were chosen to represent various stages in the process of reforestation, ranging from very heavily grazed and trampled grassed land (typical of much of the degraded grassland that is available for reforestation) to a relatively undisturbed religious' forest which is as close to a natural forest site as could be found. Intermediate sites included a formerly heavily grazed and trampled grassland that had been planted with Pinus patula five years earlier and from which grazing animals are excluded, a site similar to the previous one but which had been planted with Pinus roxburghii twelve years earlier, and a site, also planted with P. roxturghii twelve years previously but which carried a cover of low shrubs and herbs about 40 cm high plus a few scattered remnants of the original forest. Here too grazing animals are excluded. The field programme consisted of determining values of nearsaturated hydraulic conductivity for various layers of the soil profile of each of the five sites at increments of 0-0.1, 0.1-0.2, 0.2-0.5, and 0.5-1.0 m depth from the soil surface; in other words, an experiment was set up to determine whether or not soils under different surface conditions, including different degrees of compaction, could absorb the heaviest rainstorms to be expected in their vicinity. It was considered that such measurements would facilitate characterization of the nature of water movement through the soil profile.

At each site twenty-five measurements were taken for each of the four soil profile levels. For all sites the hydraulic conductivity values showed little variation within the deepest layer (0.5-1.0 m). Variability increased with closer proximity to the surface, with maximum differences occurring in the 0 0.1 m layer which showed a range of 39-524 mm/hour. The site with the lowest value represented the totally deforested area that had been heavily grazed and trampled for many decades. Site 2, reforested five years earlier showed a slight improvement, while Site 3, reforested twelve years earlier showed further improvement with a value of 183 mm/ hour. As near as can be determined from examination of old photographs and discussions with the local villagers, these three sites had been used in a similar manner for several decades prior to the reforestation of Sites 2 and 3. Gilmour et al. (1987) concluded that the measured differences are due to changes in soil surface conditions brought about by the change in land use (reforestation). Of particular importance was the reduction in compaction by the exclusion of grazing animals and an improvement in soil aeration through greater activity of soil micro-flora and fauna associated with the higher levels of soil organic matter.

From these results, however, it still could not be determined how much time is required to return heavily grazed and trampled sites to near natural conditions, but it is assumed that it would be many decades. However, the data obtained were adequate for comparison with the actual characteristics of rainfall events. Thus the authors postulate that if, for instance, the hydraulic conductivity of the 0-0.1 m soil layer at Site I is only 39 mm/hour yet the rainfall intensity never exceeds this value, then all the incident rain would infiltrate the soil profile and no surface runoff would occur. This would apply provided the soil was not already saturated before the onset of the specific rainstorm. The fact that the surface layer value of Site 4 is thirteen times higher than that of Site I would have no practical significance. Thus the next step was to consider rainfall characteristics.

The annual precipitation for the experiment site is 2,008 mm, with 85 percent occurring between June and October inclusive. It is necessary, therefore, to isolate the summer high-intensity events. Since rainfall short-termintensity data are not available for the actual field area, data from Kathmandu were used. This data set was considered by the authors to give an adequately close approximation to actual field conditions. This is based upon the many years of project personnel experience in the two areas.

Examination of the Kathmandu climatological record (1971-79) allowed the compilation of 5-minute rainfall intensities expressed as equivalent hourly intensities in mm/hour to enable a direct comparison with the measured soil hydraulic conductivity values (also in mm/hour). The highest 5-minute intensity value during the period 1971-79 was 120 mm/hour and was recorded on four occasions. Maximum 24-hour rainfall totals during the period 1972-82 varied from 59 mm to 117 mm. A summary of all the data on 5-minute rainfall intensity peaks shows that incident rainfall intensities exceed the soil absorption capacities for Site 1 on an average of 6.7 days per year and for Site 2 on an average of 3.7 days per year. The remaining three sites, according to these data, have hydraulic conductivities that exceed rainfall intensity peaks on all occasions. Thus, in the worst possible situation (that is, heavily grazed grassland as represented by Site 1), overland flow is likely to occur on 6.7 occasions for an average summer monsoon period.

Gilmour et al. (1987) next use Caine and Mool's (1982) data to show the effects of a I -hour duration storm of 40 mm/ hour. A storm of this magnitude would have a recurrence interval of only once in twenty-five years and is the type of event likely to cause a degree of local flooding. At the most degraded site (Site 1) such a severe storm would generate only I mm/hour of surface runoff. The inference is that deforestation under these conditions would have little effect on the generation of any major flooding. Conversely, the improvement in soil surface conditions caused by reforestation would result in the absorption of an additional I mm/ hour of rainfall - hardly the kind of result likely to have any noticeable beneficial effect on local downstream flooding and certainly not on distant downstream flooding, such as on the Ganges Plain. The authors conclude that their findings from a select research site in the Nepal Middle Mountains support the results of the majority of researchers (see reviews by Boughton, 1970; Hewlett, 1982; Hamilton, 1983; Gilmour, 1986) that deforestation and reforestation would not, by themselves, influence major flooding. 'Major floods are primarily influenced by precipitation factors in association with the geomorphology of the catchment' (Gilmour et al., 1987:247).

The other factor to be considered is the extent of poor-quality catchment cover and the degree to which this can be improved. Data collected by Nield (1985) indicate that about 7 percent of Nepal's Middle Mountains had a cover of degraded grassland. Even if all of this could be reforested (a mammoth task in itself) the likely effect on flood reduction, partly on account of the broken and scattered nature of the grassland, would be insignificant.

The remaining point to be considered is the degree to which a wellvegetated and protected soil surface would decrease the incidence of overland flow and thereby reduce local soil erosion. Planting trees on mountain sides can decrease the number of shallow landslides, or conversely, tree removal can increase the likelihood of shallow landslides (Hamilton, 1987). Surface soil erosion is widespread both in the local field area of Gilmour et al. and more widely throughout Sindhu Palchok and Kabhre Palanchok districts, especially on the over-grazed and trampled grassland areas characterized by Site I of the study. Reduction in this process of local environmental degradation would undoubtedly be achieved as reforestation is extended. Moreover, establishment of trees can impart a greater margin of safety against shallow landslides (Hamilton, 1987). This would be of considerable importance in terms of improved maintenance and site productivity and is probably one of the major benefits to accrue from regional and local reforestation programmes. A more general consideration of reforestation, in terms of species selection and methodologies is beyond the scope of the present discussion.

A number of general conclusions can be drawn from this discussion on the hypothetical and perceived impacts of deforestation and reforestation in the context of the wider Theory of Himalayan Environmental Degradation. First, as Gilmour et al. (1987) infer, rainfall total amounts are lower and the incidence of storm periods with high-intensity downpours is much less frequent than has been widely assumed for a monsoon climate. While the data set upon which this conclusion is based, involving extrapolation from Kathmandu to Chautara, is not completely satisfactory, it does contrast with the earlier assumptions that deforestation must be disastrous in terms of increased soil erosion due to highintensity rainfall on an unprotected or compacted soil surface. This is perhaps another assumption that has contributed to much misunderstanding, although certainly much more data on rainfall amounts and short-period intensities and a much longer period of observation are needed. The short period of observation and the single study site should be carefully borne in mind; extrapolation to other areas is not yet warranted. We should also exercise extreme caution in this regard because, as indicated in Chapter 2, as more precipitation-recording stations yield new

The pressures V&l Himalayan forests and their role data, examples of extreme annual totals (that is, in excess of 5,000 mm) increase in number. Nevertheless, the combined results and reasoning of Gilmour et al. and Hamilton must be considered as cautionary tales leading to the recommendation that even on a local scale the widely cherished myths concerning the evils of deforestation, and conversely, the beneficial effects of reforestation are likely to lead government and development agencies into policy decisions with unattainable objectives. There are enough real benefits from maintaining natural forest, and from reforestation of degraded lands, without having to rely upon questionable values that simply sound attractive specifically, the dubious promise of reducing the annual floods and siltation far downstream on the floodplains of the Ganges and Brahmaputra.

Contents - Previous - Next