Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
3. The development of swidden agriculture
3.1. Past experience: two forms of swidden
''development''
3.2 Technological possibilities for swidden
development
This chapter reviews: (3.1) notable past development experiments attempted with actual local operation of swidden agriculture throughout the world, (3.2) technological possibilities, in terms of various recommendations and findings, and (3.3) development in a broader context, to include types of systems, the socio-cultural sphere, political organization, dual economy, distribution of labour, etc. Although there have been comprehensive reviews of swiddening, usually on a regional basis (e.g., Watters 1971; 1960; Spencer 1966; W. Allan 1965; Miracle 1967), and a few general works on swidden analysis and improvement (e.g., Nye and Greenland 1960; Greenland 1975), the attempt here is to assemble in one place various findings and recommendations dealing with the improvement of swidden cultivation itself, i.e., as swidden agriculture, rather than in terms of its being abandoned in favour of permanent field agriculture.13 As mentioned above, this exploratory approach emphasizes increased yield, system viability, and socio-cultural factors. Chapter 4 will then discuss potential applications for Northern Thailand. It is hoped that the information reviewed in this chapter may also prompt similar development considerations for swidden agriculture elsewhere in the world.
3.1. Past experience: two forms of swidden ''development''
3.1.1
Taungya systems
3.1.2.
Corridor systems
Although Carl Sauer had begun defending, against considerable hostility, the logic and appropriateness of swiddening as early as 1938 (see Peizer 1948. p. 21; Watters 1960, p. 60), what little attention there was to the improvement of swiddening as a viable form of agriculture was rather ambiguous through the following decades. Peizer (1948, pp. 29-32) offered several recommendations on improving swiddening, but in a later work (1957) he treated swiddening as a stage between hunting/ gathering and permanent-field agriculture, implying severe limitations for development, although he did take the position that outlawing swiddening was not "constructive'' (1957, pp. 8 - 9). Spencer (1966, pp. 16 - 17, 73 - 74) talked about "upgrading the economic culture" of swiddening, but he meant by this not the increasing of swidden productivity and viability but, rather, changing the whole of swidden societies in the direction of permanent-field societies. Both Spencer (1966, pp. 12, 27) and Peizer (1948, p.17) recognized that swiddening in Southeast Asia had once been much more prevalent, being practiced in the lowlands, where it has since given way to more intensive forms. But the lesson Spencer drew from this knowledge was that an ongoing trend was occurring whereby swiddening would probably eventually disappear entirely (Spencer 1966, pp. 53, 82-83.) He did, however, come very close to recommending the development of swiddening in marginal areas when he stated (1966, p. 73): "There is ground for believing that shifting cultivation as a cropping system, combined with an economic system that values surpluses for their creative worth in occidental terms, would yield a larger accumulation of wealth, acre for acre, than permanent-field farming of producing land in some parts of the rough country where it now operates." R.F. Watters, perhaps the single researcher who has studied swiddening most widely, recommended several things that could be done to improve swiddening on steep land (1960a, p. 49), but his later work seems to indicate that such recommendations must have been intended as interim measures only, until swiddening could be replaced by other forms of agriculture. In general, he seems to have advocated a change to permanent field farming, arboriculture, or pastoralism, while stressing the need to overcome social and economic obstacles in order to accomplish this (Watters 1971, pp. 233-290). In places, however, Watters too implied that swidden agriculture might be well suited to marginal areas (see Watters 1960, p. 86, discussing upland soils).
I think it fair to say that no researcher until quite recently has ever clearly indicated that he considered swiddening per se to be a potential modern adaptation, capable of agricultural aid and improvement as are other forms of agriculture today. Nevertheless, in actual application, attempts have been made to develop swiddening as such. The major attempts may be grouped under two headings: taungya systems, which combined swiddening with forestry, and corridor systems, which attempted to improve efficiency and organization of swidden regimes.
Developed by nineteenth-century British colonial authorities, the taungya system, or taungya method, was used primarily as an inexpensive means of establishing timber plantations. 15Taungya systems were used extensively in India and Burma and were later adopted elsewhere in Southeast Asia and in parts of Latin America and Africa as well.16 Essentially, the system consisted of paying swiddeners to plant desired species of trees in swiddens, with payment based on the number of surviving trees after one or sometimes several years of swidden cultivation. Lest any misconceptions arise about how this system worked in practice (Eckholm 1976, pp.151 - 152, called it "promising''), it needs to be stressed at the outset that it was a system designed by foresters as a cheap method of establishing timber plantations. As such, it maximized timber, not swidden yields (Kio 1972; FAO 1957, pp. 64-69; Allsop 1949, p. 285). Those who participated in the system were not normal integral swiddeners but, rather, impoverished, failing swiddeners or landless peasants who badly needed the opportunity to make a living (C.W. Allan 1916, p. 534; Allsop 1949, p. 282). It was recognized that the system would not attract labourers unless sufficent population pressure and land shortage existed to motivate the participants to put up with the hardship and/or regimentation which the system often entailed (FAO 1956, pp. 79 - 85; FAO 1957, pp. 64-69; Stebbing 1922, P. 92)
In India and Burma, the taungya method was used principally to plant teak. Allsop (1949, p. 282) says that it worked best in Upper Mixed Deciduous Forest (where teak grew best and swidden yields were sufficient), but it was also practiced on degenerated land (areas of Imperata and bamboo) (C,W. Allan 1916, pp. 533-534). Locations were chosen by the forestry authorities, who laid out the tracts and the spacing of trees and supervised all phases of the operation (C.W. Allan 1916, p. 533; Allsop 1949, p. 283). The system was successful in producing teak plantations. In Burma, between its inception in 1868 and the year 1898, over 52,000 acres were planted to teak by the taungya method, whereas less than 4,000 acres had been planted by the government (Nisbet 1901, p. 58). Once converted to teak, however, taungya areas were useless for swiddening, since the favoured regeneration time was 80 to 100 years (Allsop 1949, p. 284). Thus the system can be seen more as a (frequently exploitative)17 tool of the foresters than as an aid to swiddening, since most of the labourers initially supported in this way not only would be forced to move frequently (Watters 1971, p.19) but would, for the most part, eventually be out of a job. Examples in India and Burma point out that this was further hastened by land scarcity (Manshard 1974, pp.110-111), the very condition that attracted the participants to the system. Kio (1972) points out that usage of the system in the past has been exploitative, but thinks that stressing a mix of timber and farming could avoid this, with cashcropping of tree crops, such as cocoa and fast growing timber, in more populated areas, run on a ''communal ownership" basis. In fairness to the British colonial authorities, experiments were attempted in which forest lands were placed under local village control, but the attempts were abandoned due to "rapid degeneration of the forests" (Allsop 1949, p. 279). Failure here was likely since the lengthy regeneration time would have necessitated a rather unreasonable wait until the swiddeners could reap the fruits of their efforts, and organization needed to distribute benefits was lacking among integral swiddeners.
Sometimes referred to as paysannat systems, corridor methods of improving swidden agriculture were also developed by colonial authorities, this time in Africa. In the Belgian Congo, ''European methods, the practices used on the rich and intensively cultivated fields of Belgium-deep ploughing, clean weeding, thorough cultivation, cover and green manure crops-failed disastrously" (W. Allan 1965, p. 437). Thus, in order to supply food to the growing urban centres and to produce the cash crops which the colonials needed, modified forms of native swidden systems were promoted instead. Modifications were aimed chiefly at efficiency and control, i.e., control over the native cultivators. Allan (1965, p. 444) points out that such systems were not "a striking improvement" and that ''there was more than an element of 'educative' compulsion'' involved. The conditions necessary for adoption of the corridor system were opposite to those of the taungya method: the corridor method was "applicable only under conditions of abundant land, where population densities are well below the critical level for the customary systems of land-use on which the method was based" (W. Allan 1965, p. 441). Whereas taungya techniques could conceivably have some modelling value for swidden development if their more authoritarian and organizational aspects could be overcome, corridor systems, while suffering from this disadvantage too, are even more disadvantageous in that they apparently have little to offer swidden systems already beset by population pressure.
The corridor system was used mostly in the Belgian Congo, but even there the system was not continued afterwards in independent Zaire, indicating its general lack of popularity among the cultivators. As Eckholm (1976, p. 141) puts it, "its passage into the history books has not been widely Lamented.'' The system apparently was begun by the cotton companies as a means toward gradual intensification, not as an end in itself (W. Allan 1965, pp. 441, 442). De Coene (1956) says that it was an adaptation of the local Bantu swidden system, with three goals: (1 ) conservation of soil fertility, (2) increase in productivity, and (3) "cessation of nomadic agriculture." Basic elements in the system included the demarcation of parallel, continuous belts, or corridors, with an optimal width of 100 metres each, running east-west to get maximum sunlight. Usually the strips were farmed one year each, alternately, i.e., moving through the oddnumbered strips then beginning again, moving through the even-numbered strips. Thus a series of twenty strips would allow a regeneration time of twenty years. The 100-metre width was important, since beyond that distance regeneration was found to suffer due to lowered re-colonization from adjacent (forested) strips. The length of strips was much less important (de Coene 1956; W. Allan 1965, p. 438).
Designation of swidden plots along the corridors was of two types (de Coene 1956). In one type, all plots were assigned to individual swidden families in the beginning; in the other, swidden plots were reallocated each year. In both cases, overall ownership remained "collective." The first type had the advantage of stimulating better care for the land, since each family had its own perpendicular swath across the corridors, and thus it was advantageous to care for the plot actually being worked as well as adjacent forest fallow, since both would eventually be farmed again by the same family. in the second type, the advantage was that land could be reallocated according to the changing size of the families and thus matched with the labour available (de Coene 1956). In both types, the rigid geometrical pattern would make "corridors" unsuitable for the varying terrain and forest mixtures found in marginal mountain environments, unless considerable modification were attempted. Even on the flatter lands where it was practiced in Africa, it meant that poorer land could not be avoided and fallow time could not be matched individually to appropriate regeneration (see Dumont, cited in W. Allan 1965, p. 442). But the principal disadvantage of the corridor systems was the use of imported organization and outside (European) supervision, antithetical to the values of integral swidden society. In supervising these systems, however, the Europeans arrived at some useful technical findings concerning the improvement of swiddening, some of which will be referred to in later sections of this chapter.
For both the taungya and corridor systems, the imposed organization and regimented lifestyle were unsuited to integral swidden society. It has been seen that the taungya system generally did not attract integral swiddeners, while the dissolution of the corridor system with the termination of colonialism suggests that this system too was unsuited to integral swiddeners, for the same reasons. In both systems, then, the problem with swidden "development" was not so much the technical adaptations tried but rather the interface between technology and society, both systems paying little heed to what was most suitable in this regard. This latter viewpoint will be examined in the final section of this chapter. From a purely technical standpoint, however, the experiences with the taungya system suggest that there may be ways successfully to integrate agriculture and forestry. The corridor experiences, on the other hand, suggest that swiddening in itself need not necessarily be either destructive or non-productive, and may be amenable to improvements.
3.2 Technological possibilities for swidden development
3.2.1. Viability through forest fallow
3.2.2.
Productivity
3.2.3. Cash-cropping and other income sources
The possibilities for improving swiddening must be considered in terms of what is meant by "improvement.'' Two major aspects are the focus here: increased yield and long-term viability.19 The two are interrelated, since to increase yield without viability might "mine the soil," sacrificing sustained yield for immediate gains (Janzen 1973, p. 1216; Datoo 1976, pp. 10-11). This has happened among swiddeners-particularly non-integral swiddeners, but the pioneer swiddeners of Northern Thailand are also an example. At lower population densities their system was indeed viable, but under present conditions it is not. Some researchers feel that loss of viability through environmental degradation is likely when returns from cash crops are particularly attractive (Dasmann et al. 1973, p. 63; Janzen 1973, p. 1216). Thus, cropping practices must be carefully monitored so that viability is not sacrificed for increased returns.
Another cause for careful consideration in the course of swidden development concerns carrying capacity arguments. In chapter 1 cataclysm arguments concerning the collapse of swiddening were reviewed, and it was suggested that integral swiddeners were probably not the causal agents in such scenarios. But it is conceivable that substantial technological improvements in swiddening might have the effect of substantially raising the number of people supported, leaving a larger number of swiddeners living at equally impoverished levels (or worse) and creating a situation wherein any remedial action would be even more difficult than it is now. This is one of the concerns that has prompted researchers to advocate the abandonment of swiddening rather than its improvement. It is also clear, however, that this particular problem is not confined to swiddening alone-it is a common problem of development efforts in general.
Connected to these issues are the question of controlling mechanisms within swidden systems, the nature of swidden society, and the relationship between technological changes and social changes. It was pointed out in chapter 2, for example, that a principal controlling factor among secondary forest swiddeners in Northern Thailand today is the amount of labour necessary to operate the system, specifically the heavy weeding requirements. If these swiddeners are to have time to devote to increasing their income, an increased yield per labour unit would be a desirable improvement. But if this could be accomplished, other mechanisms would need to be effected to prevent environmental degradation, now kept in check by the weeding requirements. Similarly, increased productivity per unit of land would allow for an increased population density without increasing standards of living, unless mechanisms existed to limit access by others. Such mechanisms have generally not developed in swidden society. Indeed, private, long-term ownership of specific swiddens is rather antithetical to the values of the society and to the integrated socio-ecological methods by which swiddeners make a living (and further exacerbated by governmental prohibitions against land rights for swiddeners). It can be seen, then, that technological improvements in swidden agriculture cannot be divorced (and developers must not try to divorce them) from the broader implications that include the socio-cultural realm. In this section of the study (3.2), technological possibilities will be discussed, but, in light of the above, the reader is cautioned that this discussion cannot stand alone and it is not intended that techniques described here be advocated without a very searching consideration of their broader implications. Some of these implications will be dealt with in the following section (3.3).
That swidden productivity can be improved, while attending to viability, was recently argued persuasively by D.J. Greenland (1975). Greenland feels that the intelligent application of modern but low-cost, low-energy techniques could be expected to more than double swidden yields. Prior to Greenland's landmark paper, various other separate findings and recommendations have occasionally been published. Some of the more convincing of these will be reviewed here. The discussion is confined to (Evergreen) forest swidden systems. Swidden systems in savanna grasslands or drier bush areas would require considerably different treatment from that discussed here. Discussion will cover improvements in viability in swidden systems (particularly in promoting regeneration in forest fallow) (3.2.1), improvements in productivity in the swiddens (3.2.2), and cash-cropping possibilities and other sources of cash income (3.2.3).
3.2.1. Viability through forest fallow
A distinguishing characteristic of swidden agriculture is that it uses a relatively long fallow period per unit of land, relative to the cropping period, and that the fallow period itself is primarily distinguished by the regrowth of natural vegetation. This, of course, requires a larger land area than permanent-field systems, and it is this extensive nature of swidden cultivation that has been most objected to. Eckholm (1976, p. 141), for example, feels that the corridor system was an improvement on native swidden agriculture but was still inappropriate because it was ''still an extensive use of land that required large empty areas to sustain a relatively small annual product" (Eckholm's emphasis). Gourou (1956, p. 337) and many others make the same point. It has been argued here, however, that, while such an argument was valid for fertile lowlands where population density was great, there is as yet no suitable substitute for swiddening in many marginal areas of the world. Thus the objection that swidden agriculture is too extensive must be countered with the question, in comparison to what? If the comparison is with temperate zones, the objection is not valid, since tropical agroecosystems are of a very different nature, as the colonials in Africa found out the hard way. In areas of poor soil but with abundant tropical vegetation, shifting agriculture works but permanent field agriculture does not because of the ability of tropical vegetation in the fallow quickly to reclaim cleared ground in preparation for the next cycle of cultivation; thus it is important to retain the fallow (Dasmann et al.1973, p. 48). In comparison to other forms of tropical agriculture, Eckholm himself points out (1976, p.139): "For the unpleasant truth is that, for many tropical areas of Africa, Latin America, and Southeast Asia, no alternative food production system to shifting cultivation has yet proven both biologically and economically workable.''20
There are other reasons as well for the preservation of forest fallow which swidden agriculture entails. Gomez-Pompa (1972, p. 763), for example, has credited swidden agriculture with preventing the mass extinction of tropical forest species. The net decrease in natural biotic production is seen by many as a problem of modern resource system processes (see Woodwell 1974). The use of forest fallow in swidden systems can be seen as an enlightened and valuable exception to this general trend .21
Some researchers feel that forest fallows should not be manipulated, that there is little value and a fair degree of danger in trying to "improve" them. A principal value of the system, after all, is the degree to which it is integrated with the natural environment, species composition and diversity in the natural forest being preserved by unmanipulated natural regrowth during fallow. Nye and Greenland (1960, p.136), although they also offer recommendations for fallow improvement, point out several obstacles: (1) that it is the fastest-growing species that are the most important and these are usually the ones already growing wild in the area; (2) that leguminous fallows are of no advantage, since the nitrogen status of the forest fallow in most areas is already good; and (3) that the comparative effect of green manuring is in question (it was promoted in Nigeria but no one adopted it).
Nevertheless, measures can be taken throughout the swidden cycle that will help protect forest fallow and speed forest regeneration, thereby contributing to both productivity and viability in swidden systems. These include (1) site selection, (2) cutting and clearing practices, (3) burning and fire-guarding practices, (4) length of plot use, crop types, assemblages, and sequences, (5) weeding practices, and (6) fallow species selection. Before discussing each of these in turn, it should be pointed out at the beginning that all the basic requirements of a stable system of agriculture are met by swiddening if the fallow period is long enough (Greenland 1975, pp. 841 - 842). Thus real efforts should be made to limit population in swidden communities (if it threatens fallow time), to provide for outmigration if population cannot be internally limited, to guard against further reduction in fallow time, and to promote longer fallow periods where forest is already being taxed. This will be discussed further in section 3.3.
1. Site Selection
If sites are not too large and are separated by fairly mature forest belts, regeneration will be promoted both through nearby sources of reclamation (e.g., seed infiltration) and through the lesser likelihood of fires escaping into old swiddens. The surrounding belts also serve as windbreaks to reduce evaporation and protect the soil (Watters 1960, p.19). The corridor system experiences suggest that a systematic use of 100metre belts on an east-west alignment for both cropping and fallowing may best accomplish these purposes if terrain is suitable, but this system is disadvantageous for other reasons (regimentation, lack of flexibility, unworkability in most mountain areas, inability to avoid areas of poorer soil, etc.; see discussion in 3.1. above). British experiences in Burma suggested the use of small, narrow clearings, uncut strips between clearings, avoidance of areas near ridges and streams, and maintenance of virgin forest pockets (Allsop 1949, p. 281). Peizer (1948, pp. 29-32) recommended a 60-foot belt between fields, narrow plots parallel to contour, and avoidance of headwater areas. In general, the way in which a group is organized and its preferred labour distribution arrangements will affect the size and location of swidden plots. Large villages whose members prefer to site swiddens contiguously will have a problem with fallow regeneration. The problem is further complicated by various trade-offs in swiddening goals. Small separated plots, for example, may help regeneration but they will mean more labour required for fencing and make pest and animal invasion more difficult to control. Another factor is the methods by which a group decides whether or not an area is ready to be swiddened again. This is almost always not based on counting the number of fallow years, but rather on observing such factors as the state of vegetation (see, e.g., de Schlippe 1955, p. 37). Any fixed fallow time applicable to all areas might be inadvisable, since regeneration time will vary widely due to microenvironmental conditions, but vegetative indicators are also dangerous, since there is a lag time between changes in soil and changes in vegetation (Watters 1960, p. 83). Appropriate measures here will be discussed further below.
2. Cutting and Clearing Practices
Fallow regeneration can be significantly affected by cutting and clearing practices. The pollarding or trimming of the larger trees is much more advantageous than clear felling (cf. Peizer 1948, pp. 29-32), but felling is to be preferred to ringing and/or firing around the base of trees (Seavoy 1973, p. 528). In Northern Thailand, enlightened Karen and Lawa practices with regard to forest strips above and beside swiddens and leaving (lopped) larger trees scattered within the swiddens promote very rapid regeneration.
3. Burning and Fire-guarding Practices
Most researchers agree that stringent fire control measures to avoid the escape of fires into other areas and to prevent accidental fires (as well as those set intentionally by others) are instrumental in preventing the spread of Imperata and promoting rapid regeneration. Burning destroys the young shoots, and they will after a time cease to sprout (see de Coene 1956).
4. Length of Plot Use, Crop Types, Assemblages, and Sequences
A general rule of forest regeneration after swiddening is that the fewer the years in which the plot is cropped, the faster and fuller the regeneration. In most swidden environments, cropping for only one year at a time is most appropriate in this regard. Very little erosion occurs in the first year, nutrient supply is not exhausted, necessary wild species survive, and regrowth is then rapid (Nakano 1978; Peizer 1948, p. 32; Kellman 1969; Spencer 1966, p. 34; Farnworth and Golley 1973, p.152). The maintenance of habitat under these conditions has been scientifically confirmed by testing in Thailand and in the Philippines (Nakano 1978; Zinke et al.1970; Sabhasri 1970; Kellman 1969).
From the standpoint of fallow regeneration, crop types, assemblages, and sequences can also be quite important. The corridor experiments (in humid equatorial zones) showed that annual weeded crops (such as groundnuts, maize, soybeans, and cotton) did the most harm, whereas perennials (such as banana and cassava) did the least harm and required the least care (de Coene 1956). Taungya experiences in non-equatorial monsoon areas suggested that a number of crops were not deleterious to forest regrowth: beans, soybeans, groundnuts, cucumbers, pumpkin, cabbage. Cotton and cassava were found acceptable, but rice was not recommended. Sugar cane, maize, and bananas were acceptable if the plantation trees planted were quick-growing species; otherwise they became shaded out by these crops (FAO 1957, pp. 64, 69). Taungya goals, of course, were to maximize timber, whereas corridor goals were in crop yields and viability for re-use for cropping purposes. Harris (1971) suggested that vegetative crops are probably better suited to the humid equatorial tropics, and seed crops to tropical areas with less moisture and more pronounced dry seasons. Comparative regrowth characteristics under varying macro-climatic conditions have not been adequately studied, but they are undoubtedly extremely important (cf. Watters 1971, p. 46 ff.). In general, regrowth is much faster in moister equatorial regions (Nye and Greenland 1960, pp. 76-78).
Crop assemblages and sequences also affect regeneration. In general, the more diversified the swidden, the less weeding is necessary, the more niches can be used for production, and the less likelihood there is of exhausting any particular nutrient, which would inhibit regeneration. Staggered planting and harvesting within the growing period allow for the most thorough and convenient use of swidden capabilities. If crops are to be planted in a second consecutive year, it is advantageous from the regrowth standpoint that the second series of crops be of different species than the first. Semi-perennials and perennials that are given a ''head start" on fallow regrowth require little weeding. Little by little the forest reclaims the plot, with harvest becoming less and less in succeeding years. Non-demanding crops which can ''blend into'' the fallow period without excessive weeding allow fallow regeneration while still providing productivity22 (see point 6 below on fallow species selection) .
5. Weeding Practices
In general, the less the soil is disturbed (minimum or zero tillage) and the less weeding that takes place, the faster and fuller is the forest regeneration (Kellman 1969; Nakano 1978; Greenland 1975; Lal 1975). Nevertheless, as pointed out in chapter 2, Karen secondary forest systems are quite conservational, with rapid regrowth, despite heavy, continuous weeding. Several variables are important here. Nye and Greenland (1960, pp. 76 - 78) point out that weed growth (and thus forest regrowth) is much faster in the humid equatorial zones that lack any pronounced dry season. But these weeds are woody species that can be slashed off so that crops can keep ahead of them, and thus weeding is usually not a severe problem, particularly if vegetative or semiperennial crops are planted. In the monsoon areas where annuals are planted, however, weeds tend to be more of a problem. This is probably particularly true of rice. Cowgill (1961) showed that weeds were not a principal problem in Meso-American corn swiddens. Perhaps this was because the corn was quickly able to get above and shade out the weeds. in rice swiddens, however, using the Northern Thailand examples, weeds would seem to be much more of a problem. For regeneration, the type of weeding done is also important: if weeding is done by hand or with a sickle, a fair percentage of underground suckers survive to resprout during fallow; hoeing, however, is apt to be much more destructive (Nakano 1978). Greenland (1975, pp. 843-844) envisions a number of methods that could be combined to reduce weeding requirements and raise productivity: mixed and relay cropping techniques, breeding of crops to do well in mixed cropping situations, zero tillage and mulching of weeds and plant residues, etc. Eckholm (1976, p. 150) recommends a similar array of techniques.
6. Fallow Species Selection
As pointed out above, attempts to improve upon actual fallow regeneration itself should first be carefully considered. The natural ecosystem has adapted to the local conditions through millions of years and is probably the most suitable to the area. Attempts to improve upon it are not apt to be well enough informed in regard to local ecosystem complexity that costly and irreversible errors can be avoided. General recommendations and findings have been made, however. Kellman (1969, pp. 45-46) found that regeneration time during fallow was dependent on the faster-growing species present: the faster growing the species, the sooner fertility was restored. On the basis of these findings, Kellman recommended the use of softwood species in order to speed regeneration and allow for more frequent usage of swidden plots without habitat deterioration. Planting quick-growing species in fallowed swiddens has also been recommended by Peizer (1948, p. 32) and Nye and Greenland (1960, p.137). Nye and Greenland (1960, p.137) also suggested that species might be selected for more rapid phosphorus and potassium accumulation. Tests could be performed to select species that could best do this. The importance of microenvironmental factors, however, makes it advisable that fallow regrowth be tested under local conditions, in relation to soil type, forest type, swidden crops, etc., to determine the limiting factors in the restoration of fertility. Species composition could then conceivably be manipulated to promote more rapid regeneration. Experiments in planted fallows in Surinam showed that fallow time could be lowered (with improved yields), but the method was costly and labour-consuming (Kroon, cited in Ruthenberg 1971, p. 48).
Overall, then, there are a number of fruitful directions to be explored for swidden usage techniques that could promote and protect fallow regeneration and thus restoration and protection of soil fertility. Manipulation of species within the fallow itself may be possible. Such manipulation should be done very cautiously, however, in order to protect ecosystemic viability. Improvements during fallow should also aim to preserve the essential complexity of the tropical forest ecosystem. The use of single species fallows, as often advocated for one reason or another, is inadvisable unless environment is already so degraded that such a measure is needed to check Imperata invasion (see Nye and Greenland 1960, p. 137). For those who argue that tropical forests should be maintained in climax states or that tropical forest ecosystems should not be manipulated at all, it needs to be pointed out that, under present conditions, such an option is no longer possible. Dasmann et al. (1973, p. 70), for example, point out that secondary forest in the tropics is now an unavoidable reality that must be faced. Spencer (1966, p. 127) believes there is no such thing as a virgin forest, unmanipulated by man, remaining in Southeast Asia. Eckholm (1976, p. 152) points out that unmanipulated forests are no longer possible in most of the world, and that the real choice now lies between enlightened forest usage and its wholesale destruction. For the forest fallows of the swiddeners, then, changes in the natural ecosystem are already occurring. Intended changes, directed toward increased viability, are preferable to unintended changes resulting from the many competing interests that now threaten the swiddener's custodianship of the remaining tropical forests (see Clarke 1976).
The essential problem with productivity in the tropics is that while shallow trophic webs are the most productive, they are also the least stable (Watt 1968, p. 50). In temperate climates, the lower rainfall, milder temperatures, and presence of dry and cold seasons have allowed for the evolution of simpler, "seasonal restart" ecosystems. When such ecosystems are simplified even further to produce for man, there is relatively much less danger of loss of stability (viability) than when tropical ecosystems are simplified to a similar degree. Swidden agriculture is particularly suited to the tropics because it preserves this essential diversity and complexity, both within the swidden field and by the use of forest fallows. The alternative to retaining a sufficient degree of diversity and complexity is that wide oscillations will occur that will cause massive famines (Watt 1968, p. 50). For example, yearround production of monocultures (such as doublecropped paddy rice) allows the hyperbolic expansion of pests and diseases. Thus far, these have usually been checked by chemical pesticides, but the evolution of pests that are immune to these treatments is outrunning our ability to come up with new pesticides, while at the same time dosages have been increased to such an extent that the dangers of using these control measures outweigh their benefits. It should be clear by now that tropical agriculture is essentially different from temperate agriculture, and that diversity and complexity must be an essential feature of agriculture in the tropics, whether or not it will mean a "loss'' in productivity.23 Consequently, it is likely that development of agriculture in tropical zones will not be able to proceed in a manner similar to that in temperate zones.
Increases in productivity, then, will have to be of an essentially different nature. The outlines of a viable system of tropical agriculture which is capable of development would seem to be that the system will have to be not only more diverse at the local level, but also more closed than temperate zone systems. Units of the system (farms, villages, regions) will have to be much more capable of meeting their own needs than are temperate zone units (Holdridge 1959, p. 278). Careful, precise inputs will be needed to replace those exports necessary for people to participate in the modern economy (Dickinson 1972, p. 222). Units of production, adapted to local conditions, will be smaller than in temperate regions and, because of smaller input/output flow, it seems inevitable that human labour, ''carefully and patiently applied,'' will continue to be an essential component (Ophuls 1977, p. 60).
Probably the most specific technological assistance needed for the improvement of swidden productivity lies in the area of increasing yields in mixed cropping situations, an essential characteristic of the swidden field. Mixed cropping has been advocated as the best overall agricultural strategy for the tropics (Igbozurike 1971), and research is now being undertaken in this direction (e.g., the Ford Foundation sponsored the Multiple Cropping Management Project at Chiang Mai University). Such research efforts could be broadened to cover the particular requirements of swidden agriculture.24 Greenland (1975) has recently taken the lead in addressing the question. He notes that "the full potential of mixed and relay cropping has yet to be realized." Particular mixes can be selected for maximum yield, and plants can be bred to do well in mixed cropping situations (Greenland 1975, p. 843). Trenbath (1974) comprehensively reviewed available findings on the productivity of mixtures and concluded that most mixtures would yield a total biomass productivity in between the expected yield of either component separately (i.e., no productivity disadvantage to mixing), but some mixtures would probably yield more than either component separately. In swidden agriculture, however, where mixing is advantageous for a number of reasons other than total productivity alone (diet variety, for example), mixtures could be selected for maximum yield in crops desired in greater quantity, and for lower yield in others (nitrogenfixing legumes, for example). In marginal mountain areas, the use of mulching and zero tillage techniques on sloping ground can also be expected to increase swidden yield as well as to ensure long-term viability (see Greenland 1975, p. 843, table 2; Lal 1975).
When fertilizers are needed to increase productivity, Eckholm (1976, p.150) makes the important point that chemical fertilizers are too expensive for most tropical cultivators; that what is needed is cropping sequences (with legumes), other forms of green manure, mulches, animal waste, and acid-tolerant crops for areas of high soil acidity. Dickinson (1972) has underscored this point, emphasizing the value of an essentially ''closed system'' for tropical cultivators. Even if chemical fertilizers were not becoming increasingly expensive. the swidden cultivator would be better off if he did not have to rely heavily on such products for subsistence crops. Since these crops will, in the main, be consumed at home and not sold, cash outlays for their production would have to be subsidized from other income sources. Increased production in these cashproducing areas could conceivably allow the swiddener enough income to participate rewardingly in the modern economy. but only if his food production were locally supported and required little or no cash outlay. Consequently, locally obtained, nocash-cost sources for fertilization are much to be preferred. There is much room for innovation in this area. Farnworth and Golley (1973, pp. 153, 154) conceive of plant byproducts being fed to cattle and the manure being used for fertilizer, or, even better, of manure being used in fish ponds to raise fish locally and the fish pond residue then being used for fertilizing fields. Mulching techniques have already been mentioned. Greenland (1975, p. 844) points out that, when fertilizers are needed that cannot be locally produced, technology of fertilizer application can be adapted in order to reduce its cost. Small "knapsack" sprayers, for example, could be used (Greenland 1975, p. 844).25