Thomas K. Rudel, Kevin Flesher, Diane Bates
and Sandra Baptista are in the Department
of Human Ecology, Rutgers University, New Brunswick,
New Jersey, United States.
Peter Holmgren is Project Director of the Forest
Resources Assessment Programme, FAO Forestry
Department, Rome.
Findings of an inventory of the literature on tropical deforestation, and analysis of the patterns of causation reflected in the literature, carried out in the context of the Global Forest Resources Assessment 2000.
The literature on tropical deforestation has grown at such a rapid rate over the past decade that it has become difficult for interested persons to keep up with it. Approximately 50 studies of tropical deforestation appear annually in the literature. This mushrooming of information has important implications. The uncoordinated way in which studies are conducted and published challenges efforts to integrate the data for closer monitoring of the tropical deforestation problem, and also challenges efforts to use the literature in the development of policy initiatives to combat deforestation.
This article summarizes the findings of a recent eight-month effort to inventory the literature on tropical deforestation, carried out in the context of FAO's Global Forest Resources Assessment 2000, which has as one of its objectives the estimation of forest cover change in all countries (see Box below). The article highlights current patterns of information about tropical deforestation, and summarizes patterns of causation represented in the literature.
The first task in describing the literature on tropical deforestation was to build up an inclusive bibliography on the subject. The analysts began with a list of 120 citations collected during the early 1990s, and then updated and expanded it through mid-2000 by:
The creation of the bibliography was a snowballing process: newly discovered articles led to the discovery of other unread articles. Over six months more than 1 250 references on tropical deforestation were accumulated. The references were catalogued using ENDNOTE, a computerized bibliographic software system.
The bibliography was then culled to remove the following:
More than 300 such articles were thus removed from the original bibliography. An additional 100 articles proved difficult to obtain but appear to be substantively useful contributions to understanding tropical deforestation and are still being sought.
Global Forest Resources FAO regularly reports on the world's forest resources through its Forest Resources Assessment Programme. Its next report, the Global Forest Resources Assessment 2000 (FRA 2000), will review the status of the world's forests at the end of the twentieth century. This effort is being carried out with the assistance of donors, partners and member countries, as well as in partnership with the Economic Commission for Europe for compilation of country data for industrialized countries. The objective of FRA 2000 is to provide a source of information and knowledge on the world's forest resources. FRA 2000 is expected to stimulate discussion at all levels as well as decision-making on the management and protection of forests on a global scale. By actively sharing information, FRA 2000 will create a forum for the continuous exchange of information and knowledge. It will promote a greater understanding of the issues surrounding the world's forests and their importance to human survival. FRA 2000 results are mainly organized by country and are being released throughout 2000 on the FAO Forestry Web site (www.fao.org/forestry/fo/country/nav_world.jsp). |
It is important to be aware of the limitations inherent in this effort. Many reports on land-use trends or deforestation in particular locations are authored by personnel in local non-governmental organizations or government agencies and are not published in internationally recognized journals, but are rather circulated among a largely local audience. The search strategy has undoubtedly missed most of these "grey area" reports, which are sometimes cited in other publications but are difficult to obtain through libraries. In addition, there is a language bias: although the study covered articles in four languages (English, French, Spanish and Portuguese), most bibliographies are dominated by publications in English. This fact partly reflects the prevalence of English as the international language of science, but it may also reflect a lack of access to the relevant literature, particularly in French-speaking Africa.
New global forest cover map developed by FRA 2000
Each of the 825 articles in the final list was read and coded as follows. The continent, subcontinent and country described in the reference were identified using FAO's standard codes. A second entry recorded the geographical scale of the deforestation processes described in the reference: did it concern a single community, a region within a country, a country, a cluster of countries, a continent or all of the tropical places in the world? Another code represented the historical dimension: in which decade or series of decades did the events described in the reference occur?
Items were also coded for the type of information used to draw conclusions about deforestation:
The coding scheme also recorded whether a study offered quantitative estimates of forest cover change (7) or loss (8). If an article used several different types of information, all of them were included on the coding sheets.
Items that did not involve the collection of any primary data in or from a rain-forest region were coded as relying on secondary sources (category 6). Thus an article presenting a new model of tropical deforestation but using already collected census and forest cover data to test the model would be coded as relying on secondary sources, as would an article drawing conclusions based on deforestation rates calculated in another report.
Finally, each reference was categorized by the cause or causes of deforestation cited, using a typology of 20 different causes (see Box). (The greatest number cited in a single reference was 14.) To ensure that the substance of the authors' arguments was conveyed accurately, a cause would be listed only if it was described and subject to analysis in the body of the article (not only in an abstract or introduction).
The authors' choice of terminology undoubtedly influenced the ways in which some items were coded. One analyst might cite a colonization programme as the primary cause of deforestation in a location and say little about how the construction of roads, as part of the programme, accelerated land clearing in the region. A second author, looking at the same situation, might feature road construction as a primary cause of deforestation in this region. Under these circumstances the two items describing the same situation might be coded differently.
Clearing for agriculture has been a driving force of deforestation in all regions; here, trees have been destroyed for rice cultivation in Thailand
- FAO/10458/A. WOLSTAD
In addition, the use of multiple coders (six individuals, all graduate or undergraduate students with an interest in human ecology) increased the potential for discrepancies in coding. Coders were given special training to improve the reliability of the coding. To measure the magnitude of the remaining error, a reliability exercise was carried out in which two coders read the same items independently and the degree of agreement was calculated. Intercoder reliability was quite high for information sources (90 percent) but was lower for causes of deforestation (64 percent). This reliability score makes the use of the data in small samples somewhat problematic but does not prohibit their use in large samples.
Another potential source of error in the analysis stems from the use of the same term to describe different phenomena. This problem is most evident with definitions of deforestation. Some analysts have referred to a selectively logged area as deforested even though 40 to 50 percent of the canopy survived the cut. For other authors, depleting the canopy to less than 10 percent of its original size constituted deforestation (the FAO definition). The term deforestation is also sometimes interpreted as logging of primary forest - not taking into account eventual regeneration into secondary forest.
A similar problem occurs when analysts discuss population increase. Studies in South Asia and Africa frequently refer to population increase as a cause of deforestation, usually pointing to high rates of natural increase among already densely settled populations of smallholders. Analysts in Latin America also sometimes cite population increase as a source of deforestation, but here the term refers to the in-migration of relatively small populations of settlers into sparsely settled frontier regions. Such differences in meaning seem to be an almost inevitable analytic cost in highly aggregated analyses. The offsetting benefits from this scale of analysis come in the form of patterns of information use and argument that only become apparent when the literature on tropical deforestation is analysed on a global scale.
Cutting trees for fuelwood has led to deforestation on this hillside in Oaxaca, Mexico
- FAO/20302/J. SPAULL
The data set created from the coded ENDNOTE information was analysed using SPSS (Statistical Package for the Social Sciences) software for cross-tabulations. These analyses revealed the patterns of information use and conclusions described below.
Coding scheme used for attributed causes of deforestation 1. Timber or logging 2. Ranching - cattle and other livestock 3. Plantation agriculture 4. Smallholder agriculture 5. Fuelwood (household use) 6. Road construction (including railways) 7. Mineral extraction 8. Hydroelectric development 9. Industrial processing 10. War 11. International market forces 12. Government policies and subsidies 13. Population increase 14. Poverty or income level 15. Education level 16. Legal status of land tenure 17. National debt 18. Colonization 19. Urbanization 20. Fire |
The trajectory of accumulation of information about tropical deforestation exhibits a recognizable pattern. Throughout the 1980s, as concern about the problem grew, the rate of publication also increased, from eight publications in 1980 to 41 publications in 1989. Since 1990 the rate of publication has remained relatively constant, between 45 and 60 publications per year. Half of the publications on tropical deforestation have come out since 1992.
Slightly more than two thirds of the publications on tropical deforestation have a clear geographical referent. These are distributed unevenly across countries. The remaining almost one third of the publications discuss a particular aspect of the problem in an abstract way with no geographical referent, or have a global focus.
Several general patterns are apparent in the geographical distribution of studies across the different regions containing tropical rain forests (Table 1). First, not surprisingly, the numbers of studies are distributed roughly in proportion to the extent of rain forests across nations; those countries with large rain forests are studied frequently, while countries with small forests are studied rarely. The correlation coefficient between the size of rain forests and the number of deforestation studies in a country is 0.842 (p < 0.001).
TABLE 1. Where have the studies been done? | |||
Region |
Percentage of studies |
Percentage of tropical forests, 1990a |
Mean annual deforestation rate, 1981-1990a (%) |
Central America |
18.3 |
4.30 |
2.19 |
South America |
36.0 |
48.19 |
0.97 |
West Africa |
4.4 |
5.95 |
0.85 |
Central Africa |
3.7 |
14.08 |
0.57 |
East Africa |
8.4 |
8.61 |
0.73 |
South Asia |
4.7 |
6.07 |
1.43 |
Southeast Asia |
22.5 |
12.71 |
1.40 |
Oceania |
1.9 |
0.09 |
0.20 |
a Data from FAO, 1993. | |||
Analysts also tend to study places with the highest deforestation rates (correlation coefficient = 0.402, p < 0.01). This tendency probably accounts in part for the disproportionate - relative to forest area - number of deforestation studies carried out in Central America and Southeast Asia (Table 1). The concentrations of studies in Costa Rica (0.08 percent of tropical forests, 3.66 percent of studies), Ecuador (0.70 percent of forests, 5.50 percent of studies) and the Philippines (0.46 percent of forests, 5.19 percent of studies) may be partly explicable in these terms. From an information management point of view, this pattern has the happy consequence of providing the most information about deforestation processes in those places that have had the highest rates of deforestation, although there are some exceptions, notably in Central Africa.
Of course, it can be assumed that the number of studies in a given geographic area is influenced by other factors besides forest cover and deforestation rates - for example, ease of travel and work in a country and proximity of the region to centres of research. For rather obvious reasons, researchers have not studied rain-forest regions plagued by civil unrest. For example, the search did not find any studies of deforestation trends or processes in Angola, Guinea-Bissau or Mozambique, all nations that have experienced extensive civil unrest since the 1970s. In the Americas, civil unrest probably explains at least in part the paucity of deforestation studies on Colombia relative to its high rates of deforestation, as compared with studies on nearby Ecuador, Panama and Costa Rica. Given the constraints faced by field researchers in conflict areas, remote sensing studies of deforestation processes may have a particularly important role in these settings.
The research methods employed by analysts of deforestation processes have changed during the past two decades (Table 2). General studies drawing on secondary sources and first-hand accounts by field researchers predominated in the early publications about the problem. From the 1980s to the 1990s, well-funded remote sensing and survey-based studies increased in frequency, while first-hand accounts of deforestation processes declined in number. The number of studies based exclusively on secondary sources (and thus strictly speaking not generating new information) has also declined somewhat, although such studies remain numerous (about 46 percent of published research in the 1990s).
TABLE 2. Tropical deforestation studies categorized by primary information source: trends over time (number of studies) | |||||
Decade |
Remote sensing |
Survey |
Field observation |
Secondary source |
Total |
Pre-1980 |
7 |
7 |
34 |
40 |
88 |
1980s |
27 |
15 |
101 |
189 |
332 |
1990s |
47 |
42 |
55 |
132 |
276 |
Total |
81 |
64 |
190 |
361 |
696 |
Given the greater measurement precision attainable through remote sensing, household surveys and combinations of the two, the increased use of these methods is a positive development. The geographical distribution of studies based on remote sensing is relatively broad: only 27 percent of them concern Brazil, while significant numbers have been carried out in Ecuador, Madagascar and the Philippines.
The survey of the tropical deforestation literature had two purposes: to describe the knowledge base that has accumulated over the years and to describe the patterns of causation that prevail in the literature. Tables 3 and 4 display the data on the causes of deforestation. In effect the study represents a survey of experts, rather than a systematic comparison of directly observed deforestation processes, to ascertain whether or not there are large regional differences in the attributed causes of deforestation and whether or not the patterns of attributed causation have changed over time. Therefore the results need to be interpreted with caution. They are, for example, affected by the rate at which experts publish their work. The judgements of researchers who publish more will have more weight in this type of exercise than the judgements of researchers who publish less. Similarly, conditions in a frequently studied country such as Ecuador will count more in characterizing regional tendencies than will conditions in a less studied country such as Colombia. However, given the difficulty of performing the explicit comparative work that would be needed to produce more comparable cross-national and cross-continental results, the method used here may have some value.
TABLE 3. Regional variations in attributed causes of deforestation (number of studies citing cause) | ||||||||
Causes |
Southeast Asia |
South Asia |
East Africa |
Central Africa |
West Africa |
Central America |
South America |
Total |
Logging |
80 |
12 |
7 |
6 |
12 |
32 |
58 |
212 |
Ranching |
4 |
7 |
7 |
1 |
4 |
57 |
88 |
168 |
Plantation |
33 |
6 |
8 |
3 |
9 |
25 |
46 |
132 |
Smallholder agriculture |
77 |
14 |
27 |
7 |
14 |
59 |
102 |
300 |
Fuelwood |
18 |
11 |
23 |
4 |
4 |
9 |
12 |
81 |
Roads |
19 |
2 |
3 |
1 |
4 |
29 |
84 |
142 |
Population increase |
41 |
9 |
21 |
7 |
10 |
22 |
34 |
144 |
Colonization programmes |
33 |
2 |
3 |
1 |
3 |
43 |
100 |
185 |
Market expansion |
28 |
4 |
11 |
3 |
4 |
27 |
44 |
121 |
Public policy |
54 |
8 |
16 |
3 |
5 |
41 |
90 |
217 |
Total |
128 |
27 |
48 |
22 |
25 |
106 |
207 |
563a |
a The total includes only those studies that have a clear geographical referent. | ||||||||
Some analysts have observed that the causes of tropical deforestation vary dramatically by region (Rudel and Roper, 1996). Table 3 explores this possibility by cross-tabulating the attributed causes of deforestation by the region in which the studies took place. When the percentage row for a particular cause deviates significantly from the percentage row at the base of the table (the distribution of studies by region), the results imply that that particular cause is perceived as more important in some regions than in others.
Thus the literature indicates a disproportionate influence of logging on the deforestation processes in Southeast Asia, while deforestation in the arid and populous regions of East Africa and South Asia seems to have been driven particularly by demand for fuelwood. Population increase seems to drive deforestation to a greater extent in Africa and Asia than it does in Latin America. Colonization programmes, associated road building and an expansion in cattle ranching have apparently induced people to clear tropical forests in Latin America. Smallholder agriculture, plantations, market expansion and public policy seem to operate with equal intensity as driving factors in all the regions.
Table 4 explores the degree to which the citing of factors driving tropical deforestation has changed over time. A number of factors appear to remain essentially unchanged. Logging, plantation expansion, smallholder agriculture, road building, population increase and demand for fuelwood were cited as frequently in studies from the 1990s as in studies from before 1980. Several factors that are particularly operative in Latin America, for example ranching, appear to have become less important over time. This change may reflect a slowing of the expansion of cattle ranching into the forest when Brazil cut subsidies for enterprises in the Amazon basin in the late 1980s (Browder, 1994) - or at least a waning of researchers' interest in studying the subject.
Table 4 suggests that two factors associated with globalization have grown in importance. More researchers in the 1990s cited the spatial expansion of markets (through growth in urban populations, improvements in transportation and the search for raw materials in more remote settings) as a cause of deforestation. In addition, the increased foreign debt which has resulted in structural adjustment agreements has led to a renewed emphasis on the expansion of export crops at the expense of the forest (Kaimowitz, Thiele and Pacheco, 1999). It is of course difficult to know whether these changes in patterns of attribution represent real changes in the relative significance of causes or just changes in the salience of these factors in the researchers' minds. More detailed work on region-specific changes in the relative importance of the different causal factors may help answer this question.
TABLE 4. Trends over time in attributed causes of deforestation (number of studies citing cause) | ||||
Cause |
Pre-1980 |
1980s |
1990s |
Total |
Smallholder agriculture |
56 |
193 |
150 |
399 |
Logging |
41 |
149 |
123 |
313 |
Ranching |
36 |
123 |
72 |
231 |
Colonization |
37 |
122 |
67 |
226 |
Population |
27 |
105 |
83 |
215 |
Plantations |
28 |
83 |
68 |
179 |
Market expansion |
17 |
79 |
81 |
177 |
Roads |
27 |
80 |
66 |
173 |
Fuelwood |
11 |
77 |
52 |
140 |
Debt |
0 |
28 |
18 |
46 |
Total |
88 |
335 |
268 |
691 |
The research described in this article shows that many scientific studies on tropical deforestation are available, but they are unevenly distributed geographically. Central America and Southeast Asia receive disproportionate amounts of attention, whereas Central Africa receives little attention. Places with high deforestation rates are studied more frequently, but places with significant amounts of political unrest are studied less frequently.
Construction of roads and bridges was identified as an important factor in deforestation in Latin America, as pictured here in the Ecuadorian Amazon
- T. RUDEL
Trends in the scientific literature on tropical deforestation seem both problematic and promising. The uneven geographical distribution of studies across regions persists and hampers efforts to monitor deforestation. At the same time, the spread of reliable remote sensing and household survey methods for measuring deforestation promises increased understanding of the phenomenon. Taken together, these points highlight an information management opportunity. By publicizing the unevenness in information about tropical deforestation across nations, FAO hopes to induce researchers to carry out studies that might close the analytic gaps and in so doing improve efforts to monitor changes in tropical forest cover.
Bibliography
Browder, J. 1994. Surviving in Rondonia: the dynamics of colonist farming strategies in Brazil's northwest frontier. Studies in Comparative International Development, 29(3): 45-60.
FAO. 1993. Forest Resources Assessment - 1990: tropical countries. FAO Forestry Paper No. 112. Rome.
Kaimowitz, D., Thiele, G. & Pacheco, P. 1999. The effects of structural adjustment on deforestation and forest degradation in lowland Bolivia. World Development, 27(3): 505-520.
Rudel, T. & Roper, J. 1996. Regional patterns and historical trends in tropical deforestation: a qualitative comparative analysis. Ambio, 25: 160-66.
1 This article has been adapted from Forest Resources Assessment Programme Working Paper No. 27. FAO, 2000.
