Community Forestry Note 8
Local technical knowledge and natural resource management in the humid tropics
by Katherine Warner

This forestry note will examine the local technical knowledge (LTK) of the traditional swiddener and how it is utilized for natural resource management in the humid tropics. Starting with a review of the environment of the humid tropics and the problems of natural resource management in the region, the note will go on to an analysis of shifting cultivation as a natural resource management strategy for the tropics. Examples from three major regions of the humid tropics -- the Amazon basin, Southeast Asia and Africa -- will be used to illustrate shifting cultivation practices as adaptations to the local social and physical environment. In the Amazon and Southeast Asia the focus will be on the tribal minorities who have on the whole been very effective in using and maintaining the tropical forest. The focus in Africa will be on the swiddener's response to a less certain environment and the ways in which intensification is occurring.

Tropical deforestation is increasingly a focus of international environmental concern. Current projections of large-scale deforestation of the tropics create a scenario of flooding, drought and wide-scale erosion that would make vast regions unarable. Some recent work on the possible global effect of tropical deforestation has suggested scenarios of a warmer world. Whereas before the tropical forests were seen as a natural resource to be nationally managed, now there is growing sentiment that the tropical forests are a global resource whose management is of international concern. As a result of this new belief, once the grim projections are presented it is asked: What has been done to protect the forests? Who is destroying the forest? Why are not they stopped?

In the past (and even in some instances today) shifting cultivators were the primary recipients of blame for the deforestation of the tropics. Attempts were made to stop them by governments and international organizations, who perceived them as wantonly destroying the natural resources of nations. To blame them and make laws forbidding the cutting and burning of the forests was easy, stopping shifting cultivation was not. Shifting cultivators exist today and will continue to exist well into the future.

Recent studies have shown that much of the blame was misdirected. Rather than wantonly destroying the forest after a clearing has been used for cropping, many shifting cultivators actively reestablish the forest. Shifting cultivation is a complex agricultural system that is well-adapted, under certain conditions, to the environmental limitations of the tropics. It is not primitive nor necessarily destructive. It requires in-depth knowledge of the tropical environment and a high degree of managerial skill to succeed.

This new viewpoint of shifting cultivation has been reinforced by the failure of agricultural development projects in the tropics. As will be shown later, the tropics is a difficult environment in which to intensify production. Projects have failed, in many instances leaving behind grassland where forest had been just a few years before. Yet shifting cultivators in the same region cleared and burnt the forest, planted and harvested their crops, and the forest reestablished itself. Why should the technically sophisticated projects create "green wastelands" and the primitive shifting cultivator forests? Or to ask the question in another way: what do they know, what do they do, and why do they succeed in the tropics when other approaches fail?

As used in this note local technical knowledge (LTK) will refer to practical knowledge of the environment and procurement strategies based on intimate experience accumulated over many generations (Bodley 1976: 48). When studying the local technical knowledge of shifting cultivators, basic data of "environmental resources, plants, animals, land types, soil, water and crops" have to be gathered (Knight 1980: 222). But an ethnobotanical list of plants and classification of soils, etc., although necessary, is not enough. It is not just what a shifting cultivator knows of the environment that is important. It is how that knowledge is utilized. Based on this environmental knowledge and perception, given possible crops, land and labor availability, what does the farmer do? In the study of LTK it is necessary to go beyond categories and attempt to understand how this knowledge is used by the farmer to develop procurement strategies that provide nutritional security.

The swiddener's primary use of environmental knowledge is in making decisions as to what to do and when to do it. This is when that knowledge is put to the test; if it succeeds, it remains in the knowledge pool; if it doesn't work, it may be relegated to the "no longer useful" category and dropped out of the pool. Yet the swiddener's "decision making sequence" depends on more than environmental knowledge; there are also certain constraints or givens that limit the area of choice. These constraints may be social, cultural or environmental (Ellen 1982). Some of these constraints may be of short duration (marital status, young children, illness), others may be constant and relatively unchanging (climatic factors that disallow certain crops). Using LTK and operating within these constraints the swiddener makes decisions and creates a viable food production system.

This perception of the farmer as a decision maker who considers his "biologic and economic resources" and makes decisions "aimed at the achievement of agricultural production and at maintaining soil fertility" supports the current view that the agroecosytem (agricultural system as a component of the larger "natural" ecosystem) is dynamic and responsive, rather than static (Benneh 1972:245). The agroecosystem approach supports the perception of the farmer as an active participant with his culture having coevolved with the environment to create a viable food procurement system (Gliessman 1985:56). As the interactions between man, his culture and the ecosystem create changes, these in turn will encourage other changes as new decisions are made after a reappraisal of the resources. This dynamism, with its complex feedback mechanisms, provides a better understanding of how the swiddener integrates the natural environment and the agricultural system to maintain agricultural production (Gladwin 1983, Olafson 1983, Warner 1981, Benneh 1972).

Although practiced in temperate forest climates in the past, shifting cultivation is an agroecosystem currently found mainly in the humid tropics. The humid tropics is defined as a region with the following characteristics:

1. all months with monthly mean temperatures above 18o C,
2. during the growing period 24-hour mean temperatures above 20oC,
3. more than a 180-day growing period.

This represents an area of almost 2500 million hectares in four regions: Africa, South America, Central America and Southeast Asia (see Table 1). In Africa and tropical America there is a distinct concentration of the tropical humid ecozone within two river basins. In the tropical Americas 75% of the humid tropics is located in the Amazon basin. The Amazon basin is so large that it alone contains over 40% of the total humid tropics (Sanchez 1987). In Southeast Asia the humid tropics includes the mainland and the equatorial islands of Southeast Asia, excluding the upper reaches of the mountains.

Although all the regions share the general conditions of the humid tropics, there is some variation of rainfall between and within the regions. The rains of South America are the most certain, with the least monthly variation, while in almost all of tropical Africa there is a distinct dry season of 1 - 2 months when there is less than 100 millimeters of rain (Richards 1973).

Forest: The natural vegetation of the humid tropics is forest (Richards 1977; Hadly and Lanly 1983). There are two main forest types: the closed forest and open forest (Hadly and Lanly 1983). The closed forest grows where average annual rainfall is above 1600 millimeters. The closed forest has a continuous canopy, is multi-layered, and usually has an abundant undergrowth. Depending on the particular region it can be either broad-leaved, coniferous or bamboo. The floristic make-up may differ but each is adapted to similar conditions: high rainfall and high temperatures (Hadly and Lanly 1983; Richards 1973).

In areas where there is 1200-1600 mm. of rain, the natural cover may be either open or closed forest depending on the length of the dry season, soils, etc. (OTA 1984). Open forests are found where rain is from 900-1200 mm. in regions that are drier than those that support closed forest. The open forest is a mixed forest and grassland vegetation type. The tree canopy is broken but covers more than 10% of the ground.

Closed and open forests are unevenly distributed in the tropical regions. Tropical Africa has only 18% of the closed tropical forests, but contains 66% of the world's open forest. The open forest is characteristic of the drier "edges" of the Congo basin and East Africa. Tropical America has 57% of the world's closed tropical forests, most of that within the Amazon basin. Asia contains 25% of the closed tropical forest, but almost half of it is in Indonesia (Hadly and Lanly 1983: OTA 1984).

It is the closed tropical forest that is biologically the most complex and the richest in species diversity. It is this same forest that is being cleared. Man, especially after the adoption of agriculture as a subsistence pattern, has been responsible for the transformation of an estimated 1000 million hectares of the humid tropics, an area equal to the Amazon basin in size, into semi-desert (Bene et al 1977). The pace of deforestation has quickened during the last 20-30 years, as ranching, plantations and lumbering have expanded and migrants have moved in increasing numbers into the tropical forest (Richards 1977).

Table 1. Extent of warm humid tropics (million ha.)

Source: Ofori, Higgins and Purnell 1986 (citing FAO 1980; 1981; 1982)

When undisturbed, tropical forest ecosystems are stable. The stability of the tropical forest ecosystem is the result of its capacity to "withstand climate and other hazards of the natural environment" (Richards 1977: 230). Several characteristics of the tropical forest create this stability:

1. The humid tropical forest is rich in the number of species of plants and animals. It is the high level of species diversity that provides stability to the forest ecosystem.
2. The tropical forests are highly complex, the most complex of terrestrial ecosystems (Connell 1978). Plants and animals are intimately linked within the tropical forest ecosystem. Animals in the tropical forest fulfill the role played by wind in the temperate forest for seed dispersal and pollination (Hadly and Lanly 1983: 5). Since the tropical forest is far more diverse in species and the animals not far ranging, this reestablishes and maintains local diversity.
3. Since tropical soils are generally poor in nutrients, the tropical forest ecosystem depends on a self-contained, almost closed, nutrient cycle. The nutrients that are cycled in the system are in the biomass, which serves as a form of vegetative storage. The forest itself acts like a giant "sponge" in its recovery and recycling of nutrients, with 65 - 85% of the vegetation's root system found within the topsoil layer (Hadly and Lanly 1983; Uhl 1983; Moran 1981).

The tropical forest ecosystem depends on a self-contained, almost closed, nutrient cycle.

Amazon studies have shown the importance of the root "mat" of the trees in the nutrient cycle. The root mat, made up of the extended roots of trees intermixed with organic matter and mycorrhizal fungi, lies on the top of the soil and covers the forest floor. When leaf litter, twigs, or even fallen trees fall to the forest floor and start to decompose, the root mat absorbs the dissolved nutrients before they can be leached down into the soil (Stark and Jordon 1978). Since 10 -20% of the total biomass dies off and drops to the ground each year, the amount of nutrients recycled through the system is large (Moran 1981).

This system is so efficient that "the concentration of some nutrients in the streams that drain from the forests [is] actually lower than the concentration in the rains falling on them" (Uhl 1983:70). Within the forest not only trees but other plants, as well, have developed a diminished dependence on the soil -- epiphylls, which live on the leaves of trees, are able to absorb nutrients from rainwater and fix nitrogen from the air (Uhl 1983). It is an ecosystem that once established is self-sustaining as long as the rains continue and it is left undisturbed.

Yet the forest, however stable, is not static. Part of the self-sustaining process of the forest is the natural "felling" of the trees. The tropical forest is not an "old" forest, for there is constant change and renewal through the blowing over and falling of trees. The fallen tree creates a gap in the canopy and a patch of sunlight is then able to reach the forest floor. The larger the gap, the larger the microclimate, and the more varied the vegetation in the gap will be from the surrounding closed canopy forest. In an ecosystem where the nutrients are stored in the biomass, a fall of a tree per acre per year provides a substantial nutrient boost (Hadly and Lanly 1983; Uhl 1983; Hartshorn 1978; Whitmore 1978).

The high frequency of tree falls, especially in those areas of the tropics that experience severe storms or cyclones, prevents most trees from ever reaching their full potential in size or age. The successional tree species are dependent on the gaps, since they could not become established without the sunlight and flush of nutrients that a tree fall creates. The particular successional species that becomes established in the gap is determined in turn by the particular plant-herbivore relations in the locality. These factors create a forest mosaic of gaps in the canopy and various stages of growth in the understory that gives the tropical forest its unique diversity of plants and animals. It is a dynamic forest with rapid growth of early successional species and the relatively slow growth of the mature forest species creating a forest of patches in various stages of regrowth within the overall stability of the mature forest (Hadly and Lanly 1983; Hartshorn 1978; Whitmore 1978).

But this stability can exist only within the context of the natural process of renewal. Tropical forests are very vulnerable to man, especially when man enters the forest not with an axe, but with a chainsaw and bulldozer. The very factors of diversity, complexity and closed nutrient cycle that sustain the tropical forest ecosystem in an undisturbed setting cause its fragility when in contact with man. Rainforests, because of the high degree of specialization of the individual species, have a low ability to recover from large-scale disturbances by man (Goudie 1984; Hill 1975). The very complexity of the tropical forest ecosystem that creates stability in a natural state, also makes it vulnerable to man-created disturbance.

This vulnerability is increased by the way in which revegetation of the tropical forest occurs. There are four main pathways for the reestablishment of the forest when a clearing occurs naturally with a tree fall or when the cleared area is small (less than three hectares):

1. the rapid growth of seedlings and saplings present in the shaded understory on the periphery of the opened area, which quickly respond when sunlight becomes available;
2. plant regeneration from the stems or roots of damaged trees;
3. the germination of seeds of fast growing successional species that require sunlight and are lying dormant in the soil;
4. the introduction of seeds from the surrounding area. Forest tree seeds are generally too large to be easily dispersed; they fall onto the forest floor. But the seeds of the pioneer species can be carried in by animals, birds or bats (Janzen 1973, 1975) or by wind. This means that a gap will be initially colonized by pioneer plant species, which may later be replaced in the succession by tree species (Uhl 1983).

While these pathways are effective when the clearings are small, their limitations are apparent when a large clearing is made by logging or the use of a bulldozer. When large areas are cleared using these methods seedlings are left only on the far perimeter with no trees remaining within the clearing to resprout; dormant seeds are scraped up with the forest soil, and reseeding by fauna is impeded since the bare gap is too large to attract birds and bats, or for an animal to feel comfortable to enter (Jordan 1985). Since the reestablishment cycle is adapted to the small gaps that might occur with tree falls, large clearings, especially those made by modern loggers or by the use of bulldozers, make reestablishment of the forest virtually impossible (Jordan 1982; 1985).

Compounding this is the nutrient cycle of the tropical forest. With nutrients stored in the biomass, once the the forest is cleared there is a lack of nutrients available to sustain new plant growth. Without the protection of the forest cover from the heavy rains, the soil washes away, while exposure to the sun hardens the soil. The size of the gap, the removal of the topsoil, and the exposure to rain and sun combine to dramatically slow down the succession to forest. It may take a thousand years for a field of 15 hectares cleared by bulldozer and then weeded, to become forest again (Uhl 1983).

Soils: Although there is great diversity of specific soil types within the humid tropics, the great majority of the soils of the region are nutrient deficient (Jordan 1985). In the humid tropics of Africa, Southeast Asia and the Amazon the problems of phosphorus deficiency, aluminum toxicity, drought stress, and low inherent fertility are common and well recognized (Sanchez 1987; Lal 1989; Moorman and Kang 1978). The amount of rainfall appears to be what creates the poor soils of the region, for if the rainfall of an area exceeds 1000 mm., the soils are usually found to be acidic and leached (Sanchez 1987).

The nutrient deficiencies of the tropical soils are the great limiting factor in tropical productivity. These "old, highly weathered, and excessively leached" soils do support tropical rainforests, but the forests do not depend on the soil for nutrients (Lal 1987:16). Instead, the tropical forest ecosystem bypasses the soil and creates a nutrient cycle based on its own biomass. Unlike the temperate areas where size of the trees in the forest provides a rough measure of soil fertility, the size of the trees in a tropical forest does not indicate the nutrient level of the soils beneath it (Jordan 1982; 1985). Nutrients flow from leaves, fallen trees, etc., through the mycorrhiza and shallow roots of the surface root mat back into the biomass "without ever becoming part of the soil proper" (Beckerman 1987: 64; Went and Stark 1968).

Once deforestation occurs and the forest ecosystem nutrient cycle is broken, the soil loses nutrients and its physical structure is weakened. Although the tropical forest may not have been dependent on the soil for nutrients, the tree roots hold the soil and serve as channels for water infiltration, while the forest litter buffers the soil during the rains (Goudie 1984). When forest cover is removed, the soil is susceptible to compaction, loss of water retention properties, and the loss of important macrofauna (earthworms and termites), which provide nutrients and improve the physical structure of the soil (Lal 1987). When deforestation occurs, the protection provided by the forest for the soil is removed. Deforested sites, especially if more than a few hectares in size, experience accelerated, and possibly severe, erosion when exposed to heavy rains.

However, as with forest regeneration, the size and method of the clearing determines the vulnerability of the soil to erosion. If the clearing is small, no more than 2 or 3 hectares, and surrounded by forest, vegetation will quickly reappear and loss of soil to erosion will be minimal. If the area is large, the soil will quickly decline in nutrients and be vulnerable to erosion. But even a small area can experience severe runoff and erosion if a highly disruptive method of clearing is used.

Table 2. Effects of methods of deforestation on runoff and erosion

Source: Lal 1987

Clearing the forest by traditional and manual means results in less severe soil erosion than occurs on land cleared by mechanized means, especially tree pushers (see Table 2). The method of clearing with the least runoff and erosion is the "traditional" in which machetes and axes are used; the method that has the highest rates is the tree pusher/root rake. The differential rates of erosion are the result of what remains at the site after the forest is cleared. Traditional methods leave tree stumps and untouched root systems with little disturbance of the forest litter -- while the full protection of the forest cover is gone, there are still roots to bind the soil, and litter to buffer the impact of the rain splash. Tree pushers clear a field by pushing the trees over and pulling the roots out of the ground. What is left after clearing is an area of no roots, little litter, and a highly disturbed broken soil surface. On such a site there is severe runoff and erosion with almost 70 times the amount of runoff and a loss of 1700 times the soil as the same area under traditional clearing.

Estimates of the actual number of shifting cultivators vary from 250 million (Myers 1986) to 300 million (Russell 1988). In a world of 5 billion it might appear to be of no great concern how 5% of the population makes its living. But what cannot be ignored is the distribution of shifting cultivators and the large area under these agroforestry systems. Shifting cultivation is the most widespread type of tropical soil management technique. Various types of shifting cultivation are currently practiced on 30% of the world's exploitable soils (Hauck 1974, Sanchez 1976: 346).

There are various definitions of shifting cultivation. The most commonly used defines shifting cultivation as any agricultural system in which the fields are cleared (usually by fire) and cultivated for shorter periods than they are fallowed (Conklin 1957). With the development of the agroecosystem approach and its holistic view of agricultural systems as part of the greater "natural ecosystem," there has been a reconceptualization of shifting cultivation. The agroecosystem approach attempts to integrate "the multiplicity of factors affecting cropping systems" (Gliessman 1985: 18). Whereas many earlier studies described the swidden system as inherently stable and provided a checklist of attributes, more recent work based on an agroecosystem approach has stressed swidden/fallow as part of an overall subsistence strategy, flexibly responding to stress as the social, economic or natural environments change (Gliessman 1985, Altieri et al 1973).

Reflecting this dynamic view, a more recent definition of shifting cultivation is "a strategy of resource management in which fields are shifted in order to exploit the energy and nutrient capital of the natural vegetation-soil complex of the future site" (McGrath 1987: 223). The emphasis on strategy and agroecosystem dynamics makes shifting cultivation "neither a static nor necessarily stable system of agriculture" but one that is flexible in response to change (McGrath 1987: 223).

Viewing shifting cultivation as a strategy that can be flexible in response to change places shifting cultivation on a continuum with other agricultural systems (which may differ from it in the length of the fallow period, the length of the cropping period, management techniques, etc.) with a movement from one agricultural system to another occurring as a response to changing conditions (Beckerman 1987; Boserup 1965; Raintree and Warner 1986).

As a subsistence strategy, shifting cultivation has not been popular with many governments and international agencies. It is commonly regarded as a waste of land and human resources as well as being a major cause of soil erosion and deterioration. To clear a forest, use the swidden field for a year or two, and then move on to another patch of the forest does indeed seem wasteful if the forest is perceived in terms of timber values alone (Grinnell 1977; Arca 1987). At the heart of the matter is not the cutting of the forest, which foresters do all the time, but the burning of the trees. The concern is not the maintenance (non-disturbance) of the forest so much as who should benefit from its demise. Governments perceive the burning as a misappropriation of resources from the national to the most local (small farmer) level.

In Africa, shifting cultivation is practiced by farmers throughout the humid zone. However, long fallow shifting cultivation has been gradually replaced by intensively used fields close to the home site and long-term rotationally fallowed fields further away (Chidumayo 1987; Getahun et al 1982). Although there is some variation in the actual management practices, crops grown, etc., this intensification of shifting cultivation is occurring throughout the region.

Unlike Sub-Saharan Africa, where everyone belongs to a tribe, in Asia and Latin America the long fallow shifting cultivators have traditionally been ethnic minorities with their own language, religion, values and, in some instances, crops. The government perception of shifting cultivation as a land use system is intricately tied to it being practiced by those who are "outside" the mainstream culture of the country. People who are viewed as being "primitive" since they have a simpler, or merely different, material culture, are also perceived as practicing a "primitive" agriculture, wasteful of resources that could be better utilized by the national "mainstream".

This prejudice has discouraged the emergence of a more objective view of shifting cultivation in many countries. Thus, a land use system becomes judged on the basis of who is practicing it, rather than on its own merits and limitations. In Asia and Latin America the perception of shifting cultivation is further complicated by the fact that it is currently being utilized not just by the "tribos" (tribal minority) or "indos"(local populations), but also by the landless peasant and the frontier migrant. Again, there is indifference, at best, concerning what low status groups are doing, unless it is judged as infringing on the national resources. Both the peasant and the tribos might be perceived as being shifting cultivators, but their respective land use systems are radically different.

The tribos are usually practicing integral swidden, a land use system based on "a more traditional, year-round, community-wide, largely self-contained, and ritually sanctioned way of life." When integral swiddeners enter a new area as pioneers significant portions of climax vegetation may be cleared each year. When the community is well established and little or no climax vegetation is cleared annually they are practicing established integral swidden (Conklin 1957: 2, 3).

The peasants are practicing partial swidden, which, rather than being based on a way of life, reflects "predominantly only the economic interests of its participants" (Conklin 1957: 2). Peasants practicing partial swidden have strong sociocultural ties outside the immediate swidden area and their goals in terms of ownership and productivity differ from the integral swiddener. Rather than being part of a stable community that has historical and cultural ties to the area the partial swiddener may be there only for the purpose of obtaining a crop for a year or two. Such partial swiddeners are primarily permanent field cultivators who make a swidden in addition to cropping permanent fields. In these cases the partial swiddener is practicing supplementary swidden and uses the swidden to supplement the permanent field. A common pattern in Southeast Asia is for the permanent field to be in the valleys and the swidden fields on the hillsides. Another partial swidden system occurs when the cultivator migrates into the forest. Often with little prior knowledge of swidden techniques, this swiddener devotes all his agricultural efforts to making a swidden. This partial swiddener is making an incipient swidden, but in most instances does not have the knowledge to develop a swidden system that can be sustained (Conklin 1957: 3).

These distinctions have been used extensively in the literature, although there is a tendency, especially in South America, to confuse incipient with pioneer swidden. Rather than use the term pioneer as it was originally developed (a tribal integral swidden community becoming established in a new area), the term pioneer swidden is incorrectly used to refer to the swidden practices of peasant migrants who move into the forest, swidden, and later abandon or sell a degraded field and/or establish permanent field cultivation (UNESCO/UNEP 1978: 324; Moran 1987). According to Conklin's original definitions these peasant migrants are not pioneer swiddeners, but incipient swiddeners who degrade because they do not have enough knowledge of the forest ecosystem to do otherwise. Nevertheless, since it has become in recent years the most common usage, for the remainder of this note pioneer swidden will be used to distinguish the practices of migrants from the integral swidden of established, self-contained communities.

With reference to the millions of shifting cultivators mentioned above, it can now be asked how many are pioneer and how many are integral swiddeners? Unfortunately, many governments do not make a distinction between swiddeners as to which are pioneer and which are integral (also referred to as traditional). Since the two swidden systems have very different impacts on the environment, this distinction should be made (Watters 1971). When destruction of the tropical forest occurs, it is the pioneer, not the integral swiddener, who is usually the cause. "Land hungry" migrants, without a background of integral swidden that would give them the knowledge to manage the forest ecosystem, are entering, farming and degrading the forested areas (Olafson 1981: 3; see also Moran 1987: 227; Moran 1983; Watters 1971). A population that resides in an area for one or more generations will have a far more precise knowledge of the local environment than the "dislocated" migrant, who is far more likely to practice a pioneer system, using agricultural methods from the area of origin rather than those suited to the area of resettlement (Moran 1987: 227).

The counter-argument to the position that "swidden is wasteful and causes environmental degradation" is that shifting cultivation appears to be the most effective method for dealing with the ecological realities of the tropical forest (Cox and Atkins 1979). Historically, shifting cultivation has not been limited to the tropics. From the Neolithic on it has been used by agricultural communities throughout the world when confronted by forests. As early agriculturalists moved through Asia, Europe, Africa and the Americas, the forests were cleared and fields appeared. Until very recently swidden was still in use in the spruce pine forest of northern Europe (Cox and Atkins 1979; Russell 1968; Ruddle and Manshard 1981). It continues in the tropics because of the environmental limitations of the region.

Shifting cultivation represents a response to the difficulties of establishing an agroecosystem in the tropical forest. The tropical forest ecosystem is characterized by generally poor but varied soils and extremely diverse flora and fauna, providing few nutrients, but many potential competitor species for food crops. By cutting the forest and burning the felled trees and litter, the swiddener makes use of an "artificial energy pulse" that eliminates competitor species and concentrates nutrients "in order to briefly . . . transfer the energy flow into food crops" (Odum 1971; also Bodley 1976). It is an active manipulation of a patch of the forest and conversion to a more open and useful succession for the cultivator (Rambo 1981: 36; see also Olafson 1983: 153).

With integral swiddeners, however, it is only a temporary intervention in the forest ecosystem. Natural succession begins again, and in many instances swidden practices actively aid in the eventual reestablishment of the forest (Odum 1971; Bodley 1976; Denevan and Padoch 1988a). The form of shifting cultivation practiced by integral swiddeners does not destroy the forest forever; rather, it replaces it with a successional series of regrowth that for the swiddener is more productive than the original forest (FAO 1978).

By having different sites in different areas in different stages of regrowth a variety of ecozones are created (Nations and Nigh 1978). A mixture of crops are harvested and wild plants collected and, since the greatest wildlife potential occurs where there is the greatest diversity of habitats, hunting is improved (UNESCO/UNEP 1978:461). If crop failure occurs, the forest and the created ecozones serve as a famine reserve (Warner 1981; Nations and Nigh 1978).

The strategy of swiddeners makes sense in terms of game theory, for as decision makers they determine how much labour to put into each of the various subsystems so as to receive the best " 'pay-off' under given circumstances" (Smith 1972: 421-22). It is because they utilize more than just the agricultural subsystem that shifting cultivators are sometimes perceived as being "part-time" agriculturalists; in fact they also hunt, fish and gather wild produce for market (FAO 1970). This multi-niche strategy, combining agriculture with hunting, fishing and gathering, with labour being invested as needed, creates an agroecosystem that can be highly productive, stable and sustainable. If one subsystem fails, the utilization of another subsystem can be intensified to provide sufficient food (Warner 1981). In some instances, if the agricultural subsystem loses its reliability because of land shortage or degradation, fishing and gathering may become the central focus of subsistence activities (see Nietschmann 1973).

As more has been learned about tropical soils there has been a growing appreciation of shifting cultivation as representing "ingenious adaptations to unfavourable environments, based on a remarkably complete knowledge of local ecology and soil potential" (Allan 1972a: 217). Acid tropical soils account for one billion hectares of land around the world. Of the one billion, 700 million hectares are in the humid tropics, 300 million hectares are in the savanna, and almost all of this is in the developing world (IBSRAM 1987). The humid tropical environment of the shifting cultivator is one of acid soils.

Effective techniques to restore soil fertility are "the pivot of every system of agriculture", and the swiddeners of the tropics have developed a technique that works -- the use and maintenance of the forest to restore soil fertility (Benneh 1972: 235). Recognizing that it is the living vegetation that provides the nutrients to support the crop, the integral swiddener shows a marked preference for field sites with standing mature forest, either "primary" or well established "secondary" (Dove 1983a; Allan 1965; Rambo 1981a; Rambo 1983; Posey 1983). After a burn the nutrients available to the food crops increase, but then quickly start to drop, probably because of leaching and erosion (Andriesse 1977:12-13; Nye and Greenland 1960 and 1964). Nye and Greenland (1964: 102) found the soil within the swidden extremely heterogeneous because of fallen timber, termite mounds and irregular distribution of ash following the burning. These variations will form the microsites that are planted with different crops according to the swiddeners knowledge of which would benefit from rich soils and which would not be affected by poor soils. After the cropping cycle is finished (usually 1 - 4 years) the field is left fallow, although tree crops may continue to be harvested for years. If left long enough the site will recover its fertility; if the site is used too soon, degradation can begin.

It may be difficult to recognize degradation, especially if it is occurring gradually, perhaps over several generations. With swiddeners it is especially difficult since they "appear to be so self-sustaining, so well integrated with their environment" (Street 1969: 106).

In a study that attempted to correlate field usage with soil fertility, frequency of use had a major effect on soil fertility. Arnason et al. (1982) studied two Maya fields, both with the same crop complex (maize as the staple crop planted). One had been under shifting cultivation for 100 years with a fallow period of 5-15 years. The other field had not been used for 50 years. On the field that had been fallowed for 50 years, the yields were twice as high. Phosphorus was suggested as the limiting nutrient. It is interesting to note that the fields are left to fallow after three years by the swiddeners in Arnason's study not because of the recognition of phosphorus loss, but because of the increase in labour needed for weeding.

The implication is that the longer the fallow the better for soil recovery. If long fallows can be maintained, the system should be sustainable. Soil replenishment by fallowing is a response by swiddeners to the need to produce food without recourse to manures, fertilizer or alluvial deposition (Greenland 1974: 5). If long fallow is maintained, the system works; if the fallow period shortens, the soil fertility declines (see Figure 1).

The forest is not only needed and therefore preserved for future fields, but also for gathered food, game, building materials, medicinal plants, etc. -- any or all of which might show degradation or decline before fallow periods grow too short for adequate soil replenishment.

The swiddener's response to a degrading agroecosystem is to move. This is not to suggest that they are indeed the "nomads" of former belief. There is great variation among swiddeners as to their degree of mobility. Some groups cut the forest in the tradition of Conklin's integral pioneers and move on to new village sites often (Kunstadter and Chapman 1978); others may live in permanent villages and make annual treks through the forest at great distances from their villages for hunting (see Posey 1983; 1985). Since the village sizes are usually small (50-250 people) and dispersed, population densities remain low (Harris 1972; 1973). If the population does not increase, most groups can and do stay within a small area for long periods of time, or until their land area is diminished by the fallowing areas being classified as forest reserves or timber concessions.

However it is not unusual for individuals, families and, in some instances, entire villages to move for other than economic reasons. In some societies men move out of their natal area to another hamlet to find a wife and settle there (Warner 1981) or go on journeys that last for years (Dove 1983). Families may move between hamlets or villages to escape interpersonal tensions or engage in extended visits with relatives. Houses, even villages, may be abandoned if there have been deaths. And in the present day, many people may find themselves designated by external agencies (usually the government or a commercial enterprise) for resettlement.

Even within the same regions swidden agroecosystems vary in the emphasis placed on different subsistence subsystems. In some swidden systems fishing is important, in others gathering; homegardens might range from highly productive to virtually non-existent. Although there is subsystem variation in swidden systems, all share the strategy of having potential subsystems that can be intensified as needed. These subsystems may only be utilized when other subsystems fail. Gathering from the forest is a common subsystem, but the intensity of the gathering can vary as needed. If the cultigen (cultivated crops) harvest is good, the food gathered from the forest may be restricted to specially favoured fruits, vegetables or "snacks". But if the cultigen harvest is inadequate, gathering can be intensified to include staples (wild roots, sago, etc.), as well as more fruit and vegetables to support the group until the next cultigen harvest (Warner 1981).

The combination of strategic variability and response to the biological, physical and socio-cultural environment creates a wide array of potential swidden agroecosystems. Swiddeners can plant root crops or seed crops or both; fields may be used for 1 - 4 years and have planted fallow or be left with a few root crops remaining; fields may be left to rest for 5, 10, 25 years or virtually forever; fields may range in size from barely a tenth of a hectare to many hectares and be dispersed or contiguous; swidden fields may be used to supplement hunting and fishing, or for supplementary crop production by farmers whose main concern is their permanent fields. This variety and flexibility is the strength of the swidden agroecosystem (Ruddle and Manshard 1981: 74).

In order to survive, the tropical forest has to make use of the nutrients available in the biotic community. This is the same strategy used by swiddeners. The swidden creates a system of "accelerated decay" that replicates the general sequence of nutrient flow in a tropical forest. Instead of relying on the natural decay of the tropical forest to provide nutrients, the swiddener "accelerates natural decay by the burning of the slashed and felled fields". Because the accelerated decay is less efficient than the natural decay and there is great energy loss, fields quickly decline in fertility (Ruddle and Manshard 1981: 75). To regain their fertility, field sites must be left fallow.

Shifting/fallow cultivation is ecologically sound if forest fallows can be maintained (Moran 1981: 54). Forest fallow, also called "long fallow", is attained when the cleared and planted field is left to regenerate to "high" forest. Traditionally, it was the most common form of swidden in use in the humid tropics by integral swiddeners. If fields are small, the sites, like naturally occurring forest gaps, can "rapidly heal" and regeneration occurs swiftly. The surrounding forest serves as a seed source for the site, as well as protecting it (as it did the swidden field) from winds and erosion (UNESCO/UNEP 1978: 476). Rainforest species are unable to regenerate outside of the forest. By having small fields and retaining "pieces of the original forest" for reseeding the integral swiddener is actively managing the regeneration of the forest (Clarke 1976: 250; Gomez Poma et al. 1972).

The swiddener also uses other techniques of management that favour forest regrowth. While the field is under crops, many swidden groups practice "selective weeding". Herbaceous plants and shrubs that will become part of the desired succession may be cut back, rather than uprooted, and once harvesting of cultigens declines, allowed to regrow. Rather than being cut and burned, trees may just be cut back, so that they will resprout and become part of the succession. Trees that are especially valued may be protected and not cut at all. Having plants and trees already established allows a rapid regeneration of the forest. The swiddener does not have the compulsion to maintain a "clean" field with large patches of exposed soil. Just the contrary, in fact, for it is recognized that uncovered soils are soils that will wash or blow away (Clarke 1976; Ruddle and Manshard 1981). A swidden field is a field not of rows, but of filled spaces.

Ecosystem maintenance creates different stages of regrowth that provide a more diverse array of ecozones for animals. Since secondary forests have a higher carrying capacity for wild animals than primary forests, an anthropogenically created and managed forest improves the subsystem of hunting and strengthens the agroecosystem (Vos 1978: 16, see also Peterson 1981).

Long fallow swidden recreates the diversity, complexity and use of the biomass for nutrients that existed in the forest. The term alternative forest-like structures (AFS) has been used to describe the "resonance" between the forest and the swidden field. Swiddeners actively recreate the forest in their fields so as to "preserve with some stability the analogical relationships between the cultivation cycle and the natural cycle, and to replace the wild species by domesticated ones that fill the same 'functional and structural niches as their wild precedents' " (Olafson 1983: 153 citing Oldeman 1981: 81). In some swidden groups the boundary between forest and fields may blur, as forest species are planted in the swidden and domesticated species in the forest (Olafson 1983: 155 citing Schlegel 1979).

This interpretation of the swidden nicely meshes with agroecosystem analysis, where agriculture is not seen as a system that is separate from the ecosystem of which it is a part. If swidden is a reflection of the forest, it then fulfills the major requirement of being a good agroecosystem since the swidden manager takes into consideration the local biology and attempts to disturb it as little as possible while permitting its periodic reestablishment (Janzen 1975: 54). The integral swiddener changes "selected items of its content" but maintains the "gross pattern" of the forest, and therefore is different from the other users of natural resources who change "generalized biotic communities into more specialized ones" (Ruddle and Manshard 1981: 75). In a difficult environment, the long fallow swiddener has been able to develop an agroecosystem that maintains its natural resource base and achieves sustainability.

Rather than define swidden by listing traits, crops and methods, it is more useful to perceive swidden as a set of strategies for an agroecosystem that evolved in response to environmental conditions. Diversity is highly valued since farmers are aware of the continuing need to match the available varieties to the microsites in their fields. Genetic diversity is maintained by a mixture of natural selection and human preference. Natural selection determines which varieties do well in a damp place, a steep place, a wet year, a dry year, etc. Human preference intervenes through decisions as to which varieties to keep for seed, and which to discontinue.

Farmers are experimenters. Different varieties of crops, as well as new crops, are tested and tried in different conditions (Johnson 1972; Manner 1981; Warner 1981). The risk involved is such that experimentation is usually small and only a small component of the agroecosystem is involved, e.g., a small portion of a field is planted in a new crop, or a new variety of a familiar crop is planted in addition to, not in place of, the better known varieties. Forest analogies aside, although a single crop or variety of crop in a field of high diversity might not have as high a yield as it would if planted as a monocrop, the diversity of varieties and crops create a system where even if some crops are attacked by pest or disease, others will survive (Manner 1981).

Diversity exists not only in varieties and crops, but also in the number of fields. It is common to have fields from previous years in production and a new field in preparation. If, as in the Amazon, the system is based on perennials with new fields being made each year, it is possible to have many fields each in a different stage of succession (Denevan et al. 1984). From the perspective of a swidden household there are a wide range of options from which to choose in order to obtain the desired level of diversity. There can be a number of separate fields each with a different cropping pattern -- some fields may be monocropped, others extremely diverse, or there may be a system of monocropped swidden fields with diverse homegardens (Eden 1988).

A household having more than one field in different microenvironments is another way of maximizing diversity and options, as is the practice of having one field cut from secondary forest and another from primary (Warner 1981, Dove 1983). Each field may be small, but by having small fields in different areas a family spreads out subsistence risk in order to minimize "possible crop loss due to flooding, animal pests, and diseases" (Nietschmann 1976: 145). If animals destroy one field, they may not another; if floods wash out one field, another may survive to harvest.

In Africa, rotational bush fallowing is usually a multifield system. There are home fields and "out" or "far" fields. Out fields are the fields that are further from the compound. They are traditionally cropped for a brief phase and then fallowed for many years. Fallow exceeds cropping period. Home fields are closer to the compound and tend to be cropped for longer periods with shorter fallow periods; in some areas they become intensive homegardens. In addition, there is the use of small "wet" areas for dry season fields, and "old house sites, which have a higher than average level of fertility," for more demanding crops (Greenland 1974: 7).

The more diverse and broad-based the swidden agroecosystem, the greater the stability. Through a combination of different crops, different varieties and different fields, the swiddener strives to develop the most stable and sustainable system in order to provide nutritional security.

Integral shifting cultivators in the humid tropics are tribal people. In the Amazon and Southeast Asia this puts the swiddener at a disadvantage since tribal people are minorities in these regions and usually do not have political power nor secure land tenure. They are commonly perceived as being primitive, destructive, and a hindrance to development. In Africa, everyone belongs to a tribe, although particular tribes might be more or less powerful on a national level. To belong to a tribe in Africa is to be part, rather than apart, of the social organization mainstream. Land tenure rights vary depending on previous colonial experience or current land adjudication but, in general, unlike counterparts in Southeast Asia or the Amazon, African farmers in tribal areas will have had in the past, if not in the future, fairly secure usufruct if not ownership of land.

In all three regions shifting cultivators are practicing a traditional farming system. This refers to local systems that "use local products and local techniques," have "roots in the past" and have "evolved to their present state as a result of the interaction of cultural and environmental conditions of a region" (Gleissman 1985: 57). The implication is that a traditional farmer is a member of a community that has resided in a region for many years (at least long enough for an agroecosystem to have developed) and uses local resources rather than imported inputs (Padoch and de Jong 1987:179, Padoch and Vayda 1983, Wilken 1973).

Local adaptation does not make the farmer non-innovative and tied to unchanging methods "derived from individual and social experience" (Wilken 1973). Such an interpretation overlooks, especially with shifting cultivators, the dynamism of a community's adaptation to its environment. Reliance on local materials, energy sources, and the technical knowledge of the community does not imply a lack of willingness to try something new (Padoch and de Jong 1987: 179). Certainly no "traditional agricultural community" is today doing precisely what it was doing a generation ago. A stable community is not a static one, but one that is able to adapt to new conditions. Change need not weaken such a community. In some instances, such as the introduction of new crops, change can improve the procurement systems and increase the stability of the community.

New crops have moved into all the regions of the world. For the humid tropics a period commonly used as a point of reference is 1500 A.D. when contact between the Americas and the Old World began. At this time in South America the primary domesticated staple crops were manioc, maize, sweet potato, potato (in the highlands); in Central America there was maize, usually grown with beans and squash. In this region prior to 1500 there had been movement of maize to the north and south, cassava to the north and into the Caribbean. In Africa there were yams in the humid areas, indigenous rice, millet and sorghum, and in some regions plantains and bananas (originally from S. E. Asia). In Southeast Asia the main domesticate was rice, but there was also millet, sorghum, cocoyam, plantains and bananas. This list represents only the main staples and excludes other crops such as the various pulses, vegetables, spices, etc., that were diffused far from the area of their origin by 1500. It was the farmers who moved these crops around.

A look at what the shifting cultivator of today is planting in the swidden reveals a remarkable willingness to innovate and experiment. Manioc remains the staple in the Amazon area for most groups, but maize, plantains and bananas (which have replaced manioc as the main staple for some groups), cocoyam and rice are grown as well. In Southeast Asia rice continues as the favoured staple, but millet and sorghum have declined, and maize (which has become the main staple in some regions), cassava, yams, and sweet potatoes are grown throughout the region. In Africa maize, manioc, sweet potatoes, cocoyam, and the further diffusion of plantains and bananas have replaced many of the "traditional" crops or lessened their importance.

This diffusion of plants throughout the world has allowed a farmer in an isolated community to become part of the world-wide transformation of cropping systems. It expanded the repertoire of plants and created the potential for a better fit of crops and microsites within the field. It also, in many regions, expanded the amount of potential arable land; land that was too wet, too dry, or too infertile for indigenous plants could now be planted with new crops that would do well in those conditions. In some areas, the higher productivity of introduced crops allowed the restructuring of household labour toward new economic activities or, as in Africa, helped offset the labor shortages that resulted from male outmigration. The addition of new crops to shifting cultivation systems allowed the farmer to become more productive and the agroecosystem more stable and sustainable, as it further adapted to microenvironmental and microsite variation.

Family shredding cassava roots to make flour (Vietnam)

Although the different swidden groups might explain it differently within their own cultural context, the use of natural process is evident throughout the tropics. The shifting cultivator recognizes that the natural processes of the tropics can be utilized as a natural resource. Indigenous resource management is based on maintaining "specific natural processes in order to have specific items" as an outcome of these processes (Alcorn 1989: 64). Rather than expend large amounts of energy to eradicate or override the natural process, the tropical farmer uses the naturally available process for his own ends. Unlike his temperate climate counterpart, the tropical farmer does not have the means to override the natural processes of his environment. Tropical technical knowledge revolves around how to operate with, rather than try to overcome, the natural processes associated with the year-round growing season and rapid succession that result from the high rainfall and high temperatures of the region (Alcorn 1989:69).

Natural processes extend beyond a single agricultural season, and so does the environmental perception of the tropical swiddener. The perception of agricultural succession goes beyond the season and into the next generation as the natural process of regrowth takes place aided and manipulated by the farmer. This manipulation has created anthropogenic forests throughout the tropics (see Balée 1989, also Jorgensen 1978).

This is not to imply that a swiddener could sit down and explain the process of succession or forest ecology and the flow of nutrients in the tropical forest. The individual's knowledge might be encoded in religious belief (e.g., the belief that spirits would get angry if certain things are or are not done), analogy (e.g., the forest is like a parent), or scientifically inaccurate assessments (e.g., seeds will not grow if a certain bird sings). The specific explanation might have no meaning outside the particular culture. But the knowledge system works. Whether it is encoded in religion or myth is not important. What is important is that shifting cultivators understand and use the natural processes of the humid tropics to maintain, not degrade, their resource base.

Although the focus of this paper is on shifting cultivation in the humid tropics it should be recognized that within this broad regional classification there are differences in climate, terrain, population, and historical background that have had a great impact on the existing swidden agroecological systems.

The Amazon basin is one of the wettest regions of the world. About half of the rainfall is generated by the recycling of water within the region, with the remainder having as its source the Atlantic Ocean. The rate of precipitation generally increases from east to west, with the highest rainfall occurring in June north of the equator and January to the south (Hame and Vickers 1983). The Congo Basin is drier; even at its center a "dry season" can occur that lasts up to two months, with rainfall on the periphery of the basin being especially unreliable at the beginning and end of the rainy season (Miracle 1973, Kowal and Kassan 1978).

Unlike the contained basin of the Amazon, Southeast Asia is a sprawling area of ocean, islands, and mainland hills and valleys. About half the land area is continental (Burma, Thailand, Vietnam, Laos, Cambodia, Singapore, and peninsular Malaysia), and the other half is insular (Indonesia, the Philippines, Brunei, Sabah, and Sarawak). The rainfall pattern of Southeast Asia falls into two broad categories: nearly even distribution of rain year round (Malay peninsula, Borneo, Sumatra, West Java, the Moluccas, and the eastern Philippines) and the more common monsoon pattern of a season of heavy rains with a definite dry season (peninsular Thailand, coastal Burma, Kampuchea, Sulawesi and the western Philippines). The driest areas typically receive less than 1500 mm. of rainfall per year (Capistrano and Marten 1989). As is common in island and mountain areas, within a climatic boundary there can be variability from year to year and from site to site. These local climatic deviations from the regional averages create different microenvironments. Microenvironments resulting from the variation of the rain are further differentiated by the localization of soils, forest and riverine/sea resources (Warner 1981).

Unlike the Amazon basin and Africa, the terrain of the swiddener throughout Southeast Asia is one of hills and valleys. Heavy rainfall combined with this terrain makes hillsides difficult for intensive agriculture, with erosion easily occurring at the cost of the hills but to the benefit of the lowlands, where fertile alluvial soils form the basis for wet-rice culture in the region (Capistrano and Marten 1986).

Population densities in the Amazon basin are low. The indigenous populations throughout the Americas were decimated by Old World diseases at the time of contact. In the Amazon the initial epidemics were followed by the persecution and disenfranchisement of many of the indigenous groups. In response to these pressures there was a movement by some survivors away from contact into the inaccessible areas within the forest. "Detribalization" of areas also occurred, where the residents were ancestrally tribal but were no longer practicing their indigenous customs or part of an identifiable group. Scattered populations were brought together by the Christian missions and resettled (Roosevelt 1989).

The low population densities currently found in the tribal areas are more reflective of the effect of these pandemics and persecutions of the past than of the carrying capacity of the Amazonian indigenous agroecosystems. What knowledge was lost with the pandemics of the past and the persecution that has continued to the present? This is difficult to assess. In small societies, although there may be people who are recognized as knowing more than others about plants, animals, medicines, ritual, etc., everyone knows enough to do all of the basic tasks of a man or woman in the society. The more authoritative knowledge might be lost, but the everyday "know how" remains. In studies of Amazonian peoples it appears that their indigenous knowledge is certainly complete enough to allow them to develop and maintain a diversity of procurement activities.

As in the Amazon basin, areas of Africa in the past experienced depopulation as a result of contact with the West. The slave trade played a similar role in Africa as did the Old World diseases introduced to the New World. Currently, however, Africa has the highest intrinsic growth rate in the humid tropics (2.6%). Indigenous beliefs and marital patterns that favoured large families in the past are still strong enough today to encourage a large number of offspring. The continuation of high fertility, with a cessation of deaths due to inter-tribal warfare and raiding and the growing availability of modern medical services, has led to the increase in population growth rates. The high growth rate exerts pressure on the traditional field rotation systems (Pieri 1987). It is a problem not so much of numbers of people, but of how quickly the numbers are increasing. If a village doubles its population within a generation, there may not be enough land to continue the existing rotation system, nor can the traditional means (such as open aggression against another tribe) be utilized to acquire more land.

In Southeast Asia population densities vary greatly in the region depending on urbanization and land use systems. Current swidden population densities range from a low of 12 persons per km² (northern Laos) to 35 per km² (northern Thailand) (Boklin 1989, Kunstadter 1978b). As with their Amazonian counterparts, integral swidden is being practiced by the tribos, the tribal people, of Southeast Asia. Culturally, linguistically, and religiously different from peasant "lowland" society, they have little political power and are regarded as being inferior. Usually swiddeners are perceived as "squatters" rather than "owners" and disputes between logging operations, migrants, and swiddeners are increasing. The response to in-migrating population pressure on resources has been out-migration, wage labour and, when feasible, agricultural intensification.

Although there are exceptions, indigenous Amazonians and Southeast Asians are predominantly village people. They live in small settlements, rather than in individual homesteads. Although a family may spend a period of the agricultural cycle in a temporary house on the swidden field, their primary residence will be in a settlement. In Africa individual homesteads can assume the characteristics of a village. A polygynous household with several wives, married sons and their wives may become a village in size and function.

The settlement site of the village itself may be chosen by criteria other than the quality of nearby agricultural land. Throughout the tropics, in an area where there are several ecological zones (mountains, forest, grasslands, flooded areas) a village may be sited in a transitional zone that provides access to each ecological zone and its resources (Posey 1983).

In areas of the Amazon where there is one dominant ecological zone, criteria used in making a decision for a site for a village are concerned with community well-being: raw materials for rituals, plentiful game and/or fish, good visibility to avoid surprise raids, and availability of water. These criteria may take precedence over the inherent fertility of the soils near the proposed village site, not because of ignorance of soils, but rather because of the utilization of manioc, the staple crop of many groups in the Amazon (Moran 1989). Manioc is well adapted to tropical soils and will grow in soils that are nutrient deficient, acidic, and contain high levels of aluminum toxins. The tolerance of manioc for poor soils allows other criteria to be used for village sites.

Both in mainland and island Southeast Asia, swiddeners are predominately hill people, making use of the slopes for good drainage for their fields. As the Amazon and Africa demonstrate, swidden is not tied to a hilly terrain. The dichotomy of hill and valley, swiddener and padi farmer, that exists in Southeast Asia is the result of historical factors rather than agronomic principles. Swiddeners have been pushed into the hills away from the valleys by later arrivals to their areas. They have adapted to the hillside and have identified the hills as their agroecological site. On the mainland, integral swiddeners favour small river valleys for residence. Although the inner islands of Indonesia are currently farmed by permanent field farmers, integral swiddeners live on many of the other islands of Southeast Asia and are the predominant populations in parts of Sumatra, Sabah and Sarawak.

Throughout the humid tropics the general pattern is for each family to be responsible for its own field. Whether living in longhouses, individual houses, or villages in which a shaman or elder selects the block of forest that the village will use in a particular year for swidden, each household has the autonomy to make decisions concerning crops, labour and microsite utilization. Even if, as in Southeast Asia, there are communal regulations concerning irrigated terraces, swidden fields are regarded as being individually owned and managed (Prill-Britt 1986). However, while swiddeners are usually more loosely organized than their peasant counterparts, highly structured communities do occur. For example, the agricultural schedule of the Lua' and Karen of northern Thailand is tightly regulated by the shaman-elders, who decide which areas of the managed forest reserve will be cut for swidden, when it will be cut, and when it will be burned (Kunstadter 1978c, Keen n.d.). However what appears to be more common is for the village or hamlet leader(s) to have authority to settle interpersonal disputes, while agricultural activities, unless they infringe on the rights of others, are the concern of the individual household (Weinstock 1986).

The swidden household, therefore, has to make a series of decisions concerning the management of the agricultural component of the agroecosystem. These decisions are guided by the resources available, the individual's knowledge of how to make use of these resources, the rules and preferences pertaining to residence, the religious beliefs and sanctions of the society, and the labour resources available within the household.

There are six stages in the swidden cycle at which the swiddener is required to make crucial decisions concerning location, scheduling, crops, and labour inputs: site selection and clearing, burning, planting, weeding and protecting, harvesting, and succession. A poor decision at any of these stages might well mean smaller harvests, or perhaps no harvest at all.

Given the goal of diversity, how do swiddeners choose their fields? An integral swiddener usually has the right to make the field anywhere in the forest. Rights to returns from labour are recognized, so a family "owns" the harvest of its fields. In Southeast Asia and the Amazon, sharing of food occurs within the settlement and is encouraged, but the harvest "belongs" to those who clear and maintain the field. Since the potential field can be, theoretically, anywhere in the forest, site selection operates within minimal constraints on availability of potential sites. From the swiddener's viewpoint s/he is surrounded by thousands of hectares of forest, all of which at the initial stage of decision making are potential fields.

A swiddener in the humid zones of Southeast Asia and the Amazon basin will usually have a choice between primary forest and secondary forest, whereas in Africa it is increasingly rare for there to be a primary forest available for fields (Okigbo 1982). Since in many swidden societies a field will be planted more than once, the choice will have to fulfill present and projected needs. The site selection depends not only on soil fertility requirements, but also on distance from the house or village, year-round accessibility of the site (whether on a river, over a steep mountain, etc.), potential crops and labour availability, as well as supernatural constraints (sacred groves, presence of spirits, etc.) (Dove 1983; Warner 1981; Brokensha and Riley 1980; Debasi-Scheng 1974; Nietschmann 1973) (see Figure 2).

Soil fertility is recognized by swiddeners as being related to forest growth. A mature forest is usually considered as having soils that are good for the crops (Dove 1983; Warner 1981). This is confirmed by soil research that links nutrients to biomass in the tropical rain forest ecosystem; the greater the biomass, the more nutrients available to the crops (Richards 1952; Jordon 1982; Poulsen 1978). While there is a preference among swiddeners for mature forest, different groups have different preferences as to whether the forest should be primary or mature secondary (Conklin 1957; Nietschmann 1973; Rambo 1983; Beckerman 1987).

Many swiddeners simply express a preference for primary forest, and then go on to the next stage of the decision-making for the site. Other groups, however, do distinguish between the soils or topography in their area and classify sites according to these distinctions. In the Philippines the hillside residence of swiddeners makes terrain of prime importance (see Figure 3). The preferred swidden site is on a hillside with a regular slope, for a broken terrain increases the difficulty of clearing, weeding, guarding, etc. (see Conklin 1957).

Figure 3. Southeast Asia: local topographic classification

Use of soil colour categorization of the soil is common throughout the region. In the Amazon, for example, black or dark soils are regarded as the best, a bit of ethnoagronomic wisdom that laboratory analysis supports (Balée 1989, Johnson 1983). Also of importance is texture; manioc as a root crop requires a soil that is loose in texture so that the tubers can develop (see Figure 4). Among the Machiguenga the forest cover is not perceived as being indicative of good soils since "trees always grow in the forest," regardless of whether or not the soil is good for crops (Johnson 1983). The Kuikuru distinguish between forest on black or red soils, and clear the forest on the black soils for the more nutrient demanding crop of maize. The taste of the soil can also be used, with "sweetness" being an indicator of a better soil (Hill and Moran 1983).

Figure 4. Amazon: local soil classification

In the Philippines (Figure 5) a similar attempt at correlating colour of soil and texture to specific crop needs is present. According to these categorization systems, soil is distinguished as to whether a specific crop grows well if planted there. Attempts are made to match specific soils to specific crops for the best combination. These categorizations should not be interpreted as broad "fertility" classifications, they are more concerned with matching crop to soil type.

Figure 5. Southeast Asia: local soil classification

In Africa farmers recognize that crops requiring fertile soil do well if planted on termite mounds.

Termite mounds are often favoured sites for swidden fields. In Africa farmers recognize that crops requiring fertile soil, such as okra and pumpkin, do well if planted on the mounds. Recent studies on the properties of termiteria have shown that the mounds do indeed have higher levels of bases, soil water, organic matter, silt and clay than the adjacent soils (Nyamapfene 1986, Arshad 1982, Mielke 1978). The development of cash crops in Africa as a component of the agroecosystem has been successful because of the knowledge of farmers concerning the relationship of soil colour and vegetation to soil fertility. In Ghana, for example, farmers when choosing a site for cocoa trees prefer the reddish brown upland soil rather than grey sandy soil, and look for the presence of certain trees on the potential site. Occurrence of trees such as Cylicodiscus gabunensis and Ricino dendron hendolotii is perceived as indicating soils good for cocoa, "while poor cocoa soils are associated with Mallotus opposilifolius and Aracia pennata (Bennah 1972: 252).

While it is recognized that fields from primary forest require less weeding and may give higher yields, primary forest requires more work in cutting and takes longer to dry for burning (see Dove 1983; Freeman 1970). The future uses of the proposed site are also considered; if, as with the Iban and Tagbanwa, the fields will be cropped for a second year, then the extra labour investment in clearing mature forest may be considered worthwhile (Dove 1983; Warner 1981).

To find a site with primary forest (and, if it is considered, with a particular soil or terrain) is just one step in the decision-making process in site selection. Specific location of a site requires judgements that take into consideration the utilization of other resources of the agroecosystem as well as residence customs and labour availability. Since travelling is by foot (or in some regions by boat), the field cannot be so far from the household residence that too much time is spent going and coming. "Too much time" spent travelling to the field and back is culturally defined, and depends on the perceived opportunity costs of pursuing agricultural rather than other activities (Vickers 1983). If other activities (hunting, fishing, gathering) will be carried out, the field site must not be so far away that it will curtail them. Since agriculture is only one component of the agroecosystem, the time spent on swidden activities must be limited, agricultural activities cannot absorb the time that other economic activities require.

Residence moves by individuals or by a village or hamlet occur when some components of the agroecosytem demand time and energy that should be spent on others. The multi-economic niche strategy requires that the various components be in harmony with one another so that the agroecosytem can retain its stability. If fields are too far to allow hunting or fishing, or when, because of game depletion, hunting requires expending time that should be devoted to agriculture, a change of residence or field site occurs. In Africa, where swidden is usually supplementary to permanent fields and, in many areas, cash tree crops, the site is tied to a fairly close radius around the village or homestead. The farmer must find the best swidden site within the area, but in areas of land shortage the head of the extended family may be the final arbiter, since the individual farmer will have to obtain his permission before final site selection (Engle et al 1984).

The specific resources of a potential site in the forest are also considered. The Chacibo of Amazonian Bolivia favour sites near the vicinity of Brazil nut trees so the women can collect the nuts when tending to the fields (Boom 1989). At the time of site selection, thought must also be given to the harvest, for a distant site will mean hours of drudgery carrying heavy loads of the harvested crops back to the homesite. To ease the burden of travelling between home and field, field houses may be built in which family members will stay off-and-on for the season while maintaining a house in the village (Salick and Lundberg 1989). Among some groups, families will build watch houses in which to stay during the day to scare animals and birds, but the family members will return to their home every evening (Warner 1981).

All of these variables of soil, distance, crops, other economic activities, etc., require that the swiddener have not only the environmental knowledge to judge the agricultural quality of the field site for agriculture but also the managerial ability to judge whether its location will allow other important economic activities to be maintained.

Once a potential site is found that satisfies soil and distance requirements, supernatural factors may also have to be considered. In some societies rituals are performed to test whether the field is a good one, e.g., free of bad spirits. If the portents are bad, it will be abandoned as a potential field site and another chosen (Warner 1981). If the supernatural gives approval, the swiddener then has to consider his array of potential crops/varieties and their suitability for the proposed field. Are the slopes a bit too steep for rice? too wet for maize? Will a favoured rice variety do well in poorly drained soils? If the site meets these criteria, the clearing will begin.

Then comes the labour consideration of how big to make the field (see Figure 6). Although communal task groups may occur, the main swidden work group for most tasks is the independent household family (Weinstock 1986). Usually, if more labour is needed than the household can supply, exchange labour arrangements will be made. The resulting labour force may be communal in appearance, but individuals within the group will be accruing or repaying labour obligations. Amazonian societies that are engaged in sporadic warfare, such as the Yanoama, may engage in group clearing of primary or mature forest and then divide and manage the field individually (Smole 1989).

It is universal in swidden societies for men to clear the high forest, yet the size of the finished field is determined by more than a man's ability to clear. Factors such as how much time a man can spend clearing without sacrificing other economic activities, and how large an area the family labour will be able to keep weeded and protected also have to be considered (Debasi-Schweng 1974, Engel et al 1984). The ambitions of a family will also play a part. To make a large field is a necessity in many societies when a family wants to acquire status, since generosity with food, entertaining with feasts, etc., are prerequisites for prestige. Although there is no limitation on how small an area a family can farm, there is a limit on how large a field a family can manage. Few swiddeners attempt fields larger than two to three hectares, although there might be swidden fallows of the same or greater size that are visited, occasionally weeded, and sporadically harvested.

The decision of how large a field to clear usually hinges on another decision as well -- when to clear the field. There is a relationship between the pattern of rainfall and the attention given to the scheduling of clearing. The heavens and earth are scanned for signs that the time has come (see Figure 7). In areas of more or less constant rainfall, fields are cleared throughout the year. Swiddeners in these ever-wet regions clear fields as needed, usually when there is slack time in their pursuit of other resource activities. Where there is more variation in rainfall or a marked dry season, there is an attempt to utilize the dry period to get a "good burn." These periods of no rain may be quite sporadic and of short duration, a few days here and there of dry amidst periods of greater or lesser rain. The goal in such areas is to be able to "catch" these dry days (see Figure 8).

A much better burn is possible where there is a dry season than in the constantly humid areas. In areas of a marked dry season of two to three months, swiddeners cut the forest during the waning days of the rain and leave it to dry. Swiddeners attempt to time the clearing to optimize the potential burn: if the trees are cut too soon and heavy rains continue, the vegetation will rot rather than dry and not burn well, but if the field is cut too late, it might not dry in time for burning and planting (Carneiro 1983, Johnson 1983).

Figure 7. Southeast Asia: indicators of when to start clearing the swidden field

Methods of clearing are consistent throughout the tropics. There are two stages: underbrush is first cleared followed by the trees. The clearing of large trees requires time and skill. Since pioneer swiddeners initially moving into forested areas often have little experience with felling large buttressed trees, they often hire integral swiddeners to clear the trees. Among integral swiddeners, themselves, felling trees is regarded as a dangerous task that requires experience, so young men may ask, or even hire, more able men to cut the larger trees (Warner 1981).

The well documented central African chitemene swidden system is based on a farmer cutting or lopping trees from an extensive area, carrying the cuttings to a central area, which when burned will become the swidden field site (Fosbrooke 1974, Schlippe 1956, Richards 1939, Peters 1950, Trapnell 1953, Manshard 1974). These fields are usually circular and may include a termite mound (Schultz 1976, Schlippe 1956, Mielke 1978). Although labour intensive, the chitemene system is unique in utilizing the nutrients stored in the biomass of a large area (the "out-field" where trees are cut/lopped may be 8, 12 or even 20 times greater than the "in-field" area burned and cultivated) to enrich, once burned, a relatively small field site (Ruddle and Manshard 1981, Chidumayo 1987, see also Haug 1983, Vedeld 1983).

Selective cutting is a common management technique for maintaining forest succession. Species that are valued are spared during clearing, although some may be coppiced or cut at waist height (Fosbrooke 1974, Denevan et al 1984). Trees good for timber, nuts, oil, and fruit are routinely protected if either on the forest edge or within the field itself. These trees may be protected throughout the period of cultivation, and when the field is left to fallow they will form the basis for the first stage of forest succession (Denevan et al 1984, Engle et al 1984, Yandji 1982).

In summary, the decisions of where and what size to make a field, and how and when to clear it, require a swiddener to have an intimate knowledge of the physical environment, labour availability for the swidden component of the agroecosystem, crop requirements, and the future agricultural, raw material, etc., needs of the family. These decisions are linked to similar decisions made in the past and decisions that will be made in the future. The goal of having previous fields in various stages of succession depends on consistently making the right decision concerning the right place for the field site.

Figure 8. Desanâ agricultural calendar

Burning is essential for a good crop with a minimum of labour. There are six beneficial effects of burning (Rambo 1981: 5 - 9):

1. Clearance of unwanted vegetation from the field;
2. Alteration of soil structure, making planting easier;
   The heat of the fire changes the texture of the earth and makes it more friable. Walking on a burned field is like walking on tiny ball bearings that roll underfoot. This loose texture is easy to plant with a dibble stick and provides a good seed-bed (see also Conklin 1957; Tivy 1987).
3. Enhancement of soil fertility by ashes;
   When the vegetation is burned, large quantities of nutrient rich ashes are deposited on the soil surface providing the newly planted crops with the benefits of the biomass that has grown on the site (Sanchez 1976: 363 - 365; see also Dove 1983; Tivy 1987).
4. Decrease in soil acidity;
   Since plant ashes are generally alkaline, with burning there is an increase in soil pH. This helps with one of the more serious problems of tropical soils, aluminum toxicity, since an increase in soil pH reduces the exchangeable aluminum (Moran 1981: 116 - 117, Popenoe 1960: 100).
5. Increase in availability of soil nutrients;
   The heating of the soil makes the stock of stored nutrients available to plants (Nye and Greenland 1960: 71 - 72).
6. Sterilization of soil and reduction of the microbial, insect and weed populations.
   The heating of the soil controls weeds and reduces insect, nematodes, and various pathogen populations (Glass and Thurston 1978: 110). The elimination of weed seeds means less weeding, which is why swiddeners associate high forest and good "hot" burns with little weeding and high yields.

It is recognized by swiddeners that a good burn improves the yields of the fields and reduces the time spent in weeding. The problem is how to get a good burn? Whereas site selection and clearing are activities over which the swiddener has control, the results of burning depend to a large degree also on luck. A swiddener can do an exemplary job of site selection and clearing, only to obtain low yields because the rains came too soon for the field to burn well. The decision as to when to burn is usually one that is made by the individual, although, for example, among some of the hill tribes of Thailand the decision is made by the elders and the entire village burns its fields on the same day (Keen n.d., Kunstadter 1987).

Choosing the time to burn is difficult since for a "good burn" it must be done after the wood is dry, but before the onset of the rains. In the perhumid zone around the equator the dry season may be so short as to be effectively non-existent, and burning is difficult (Harris 1973: 252). Rather than praying to the gods or spirits for rain, in the equatorial region the prayers are for the rains to stop so that the vegetation will burn (Vickers and Plowman 1984). Since it is such an important decision, which will have ramifications throughout the rest of the swidden cycle in both labour and productivity, the decision of when to burn is fraught with anxiety.

In many swidden societies this anxiety is allayed by rituals or, perhaps more effectively, by reliance on environmental indicators (leaves sprouting, sighting of birds, etc.) that "tell" that it is the proper time to burn (see Figure 9) ( Richards 1985). With or without rituals, anxiety exists.

Ideally a field will be burned just prior to the coming (or increasing) of the rains. If it is burned too soon after clearing, the vegetation will not be dry enough and weeds might start establishing themselves in the burnt field. This would mean the field would have to be weeded prior to planting (Warner 1981: 20). A poor burn will require a secondary burn. Vegetation that has been partially burned will be put in piles, sometimes mounded around unburnt logs, and then burned again. In some of the wetter areas this will have to be done repeatedly until the field is judged to be adequately burned. In a community there are always individuals (it is more a matter of how many than how few) who have fields that have not burned well, and it would be a rare swiddener who sometime during his lifetime did not experience a poor burn (see Box 1).

If the society is one in which a family may have several fields, for example, when one field is cut from primary forest and another from the previous year's field, then one field may be burned earlier than the other so as to increase the odds of having at least one field mesh with the rains (Warner 1981). Again, it is an attempt to minimize risk through a strategy of diversity and variation.

Figure 9. Local indicators of the coming of the rains and the optimal time to burn

In most swidden societies burning is a male task. If the field is on a hillside surrounded by forest, a common burning technique is to start at the bottom of the hill and work upwards. Using a torch, fires are started throughout the field and special care is given to large felled trees. If a field shares a border with a cultivated field the fire is commonly started on the shared border and directed toward the slashed field.

Escape fires can occur. There is a nonchalance regarding escape fires and the potential destruction of forest in most of the perhumid tropics. This can partly be explained by the wet conditions of the forest in this zone -- a fire will usually not escape far and little damage will occur. In the Amazon, for example, the forested areas are so large that the areas burned by escape fires are only a small part of the total forest. The accidentally burned forest is perceived as being able to recover rapidly, especially in the perhumid areas. In the drier areas, however, the fires will escape further and substantial damage can occur, with large trees being burned and falling. This may still not be regarded as a problem. Since the hunting in these areas will be good, the burnt forest becomes an enriched resource. Gardens may even be planted in the areas burned by the escape fires and regarded as a low labour windfall with potential yields (Ruddle 1974).

Within the fields, the vegetation that was selected and spared during the cutting will be protected from the fires. The area around a favoured tree, for example, may be cleared so that the fire will not come close enough to permanently harm it. The protected vegetation will remain in the field throughout the cropping period and will become part of the natural succession to forest.

Once the swidden is burned, the decision must now be made as to when and what to plant. The decision to start planting is a crucial one. After burning there is a layer of nutrients on the fields that will be rapidly washed away by rain. In perhumid areas the swiddener will quickly plant a burned field. In areas with a dry season, there is a need to get the field planted quickly once the rains begin so that the plants can utilize the nutrients before they are lost to the system. In Africa it was estimated that a week's delay in planting could result in a 1/3 reduction in yields (Porter 1970). The decline is a result of the leaching of nutrients by the rains, and to a lesser degree the water shortages that occur as the season continues. The seeds planted when the ground is dry will "put out extensive root systems, taking advantage of the ephemeral presence of the large quantities of phosphorus and other minerals. Late planted crops developing in moist or saturated soil build less extensive root systems and are more vulnerable to drought, should it occur later in the season" (Porter 1970: 193).

The decision to plant is still further complicated by the uncertainty as to whether the rains have indeed started, or whether it is simply a short period of rain that will be followed by another period of drought. How to tell that the rains have started? It is common for swiddeners in regions that have a dry season to have environmental "cues" that foretell the coming of the rains. The climatic shifts reflected in winds, cloud movements, and color of the sky (red at sunset or sunrise, blackness in the afternoon, etc.) are studied and discussed (see Figure 10).

In West Africa climatic cues are supplemented by what Richards (1985: 47) refers to as "leaf indicators" (the leafing of specific plants), as well as the songs of certain birds. Throughout Africa and Southeast Asia when termites swarm it is interpreted as a sign that the "true" rains have begun, rather than the "false" rains that are followed by the return of the dry season. Do these "cues" accurately foretell? Further study is needed on these cues to determine their accuracy, especially the objective rather than interpretive ones, such as leafing (Richards 1985). In any case, by watching for these cues the swiddener becomes sensitive to his environment, which probably gives as good a basis for the decision as is available to him, as well as relieving some of the anxiety surrounding his decision.

Since swiddeners have such a variety of crops, they can stagger the plantings in relation to the conditions under which the specific crop will do best. Crops, or specific varieties of crops, that can do well in relatively dry conditions are planted first, to be followed by crops or varieties that demand moist conditions. As with burning, if there is more than one field, there is a tendency to diversify even further, so that one field may be planted earlier than another, perhaps with different crops, in the hope that at least some of the crops in one of the fields will be planted under what turns out to have been optimum conditions.

Unlike the Western farmer who sits on the tractor and "works large and regular areas . . . and must, to some extent, take the bad with the good", the swidden farmer is down on the ground, can examine at first hand every inch of the field, and can be selective, matching crops to soil, drainage, shade, etc. (Allan 1965: 87). It would probably be more accurate to state that what is perceived by the swiddener is not one field, but many microsites, each with its own characteristics. These characteristics are noted and used when planting is done (Wilken 1973; Denevan et al 1984; Conklin 1957; Warner 1981, Salick and Lundberg 1989 ). When a swidden field is planted the visual result, as viewed by the outsider, is a mixture of plants that defies his idea of order. But to the swiddener, the field is a reflection of the soil variation in the fields and the plants that will do best in each microsite.

Figure 10. Southeast Asia: local indicators of the time to plant

When a field site is chosen, trees and plants already growing there may be protected because of their edibility, medicinal uses, fiber content, or other economic values. In addition to these advantages, there is also the benefit of leaving bits of the existing vegetation in the field as providers of shade, mulch, wind protection, climbing poles for vines, etc. This form of microsite management alters crop climates by forming larger areas of desirable characteristics (usually protection from heat and sun) and preserving these characteristics within the crop zone ( Padoch and de Jong 1987; Wilken 1973: 545).

This intensive microsite management would be impossible in huge fields. It is the size of the swidden that enables it to occur. The small swidden field that appears so chaotic is the end result of the application of the best traditional knowledge concerning old crops, new crops, preserved vegetation, soils, and microclimate manipulation. (Stigter