Caboclos, Amerindians, extractivists, and most small-holder colonists are also swidden agriculturalists, although Schwartzman (1989) comments that the Acre rubbertappers are poor practitioners of this agrotechnology. Each family therefore has varying amounts of previously deforested land and will deforest a little more each year for their subsistence requirements. Agroforestry offers numerous alternative practices that have potential to productively enrich these tracts. These can be roughly divided into low- versus high-diversity systems. Generally, each family also has a tract of forest, in various stages of degradation, that would be suitable for silvicultural or forest management interventions. Most of these practices can be considered as high-diversity systems also, because a large portion of the resident biodiversity will be maintained.
Low Diversity Agroforestry - Numerous low-diversity agroforestry systems are widely used in the humid tropics. A lot is known about their establishment and yields. This knowledge base is obviously attractive, but may have limited value for the income-forest initiative because of the a priori requirement for diversity.
As Ewel (1986) and Myers (1986) point out, however, a mosaic of different agroecosystems and natural ecosystems is probably more viable than any single type of system for several reasons. 1) The producer can concentrate energies on the agroecosystems and allow the natural ecosystems to continue undisturbed. 2) Each agroecosystem has different labor requirements that can be distributed throughout the year, thus utilizing labor more efficiently. 3) Each agroecosystem will yield a different combination of products that can go to market at different times. Other advantages of diversity can probably be exploited through mosaics as well. Thus, several types of low-diversity systems can be combined within the productive unit to form a high-diversity mosaic acceptable to the producer.
1) The Cacao Under Coconut Model - Nair (1979) and numerous other authors have written on the value of this model in various parts of the humid tropics. Numerous species could substitute either the coconut or the cacao in this model. Johnson (1983) has shown that the larger palms, especially the single-stemmed one", make good upper-story components in agroforestry systems. Cacao or coffee under leguminous trees is a similar model.
In this model, coconut or a similarly statured palm or tree with a canopy that transmits enough light for an understory crop to grow is planted at or near its ideal monoculture spacing. Cacao or coffee is then planted in the spaces between the upper-canopy species. Both cacao and coffee are moderately shade tolerant and yield well in this light shade. This model generally yields more than either species in monoculture (Nair 1979).
2) The Tamshiyacu Model - Padoch et al. (1988) report on the sequence used by the farmers of Tamshiyacu, Peru, to form near monoculture plantations of umari (Poraquieba serecia) from moderately high-diversity traditional swidden systems. A sequence of short-lived trees, shrubs and herbaceous plants are harvested early in the sequence and removed or die out before they compete seriously with the umari. The model yields a very low-diversity mature system but is interesting because the species tolerates pests and diseases and has a strong local market. The Brazil nut is occasionally used in Tamshiyacu instead of umari and might be used similarly elsewhere. The number of species that can be planted at high density in Amazônia is limited, however. Peters et al. (1989) list several that naturally occur in nearly monospecific stands and therefore have potential for this type of model.
The question is: are these low-diversity models appropriate starting points for income-generating forests? In terms of producer acceptance, ease of installation and management, facility for marketing greater volumes of product, etc., these models are certainly appropriate, especially in already degraded areas. If biodiversity conservation is also a primary objective, however, then these models are less appropriate on a large scale. Consequently, they would be suitable as components of an income-forest, agroforest, subsistence swidden mosaic.
High-Diversity Aqroforestry - Numerous high-diversity agroforestry systems are found throughout the humid tropics. Many of these are of ancient origin, others appear to be due to increased population pressures (Michon et al. 1986). They are apparently stable and economically viable, although their economic value is seldom easy to quantify (Alcorn 1989).
1) The Home Garden Model - MacDicken (1990), Soemaroto (1987) and many others have reported on the highly diverse home gardens found throughout the humid tropics. These may have 50-70 or more useful species in an area of less than a hectare. They supply a wide range of subsistence needs and frequently surpluses for market, thus supplying family needs that would otherwise cost money as well as generating a small income flow.
These gardens could be expanded away from the home, much as is done in Pohnpei, Micronesia (Raynor 1989), or among the Huastec Maya of southern Mexico (Alcorn 1989). Non-economic plants are tolerated and may even be encouraged to enhance "ecosystem services", such as shade, erosion control, microenvironment control, etc. (Alcorn 1989). This type of expanded home garden with some fallow characteristics appears to provide a solution acceptable to both producers and the conservation movements that are currently supporting sustainable development programs.
2) The Swidden-Fallow Model - Denevan & Treacy (1988), Flores Paitán (1988), and others have described the swidden-fallow system in Amazônia and elsewhere. Most extractivists already practice some form of this system, although it is frequently much less elaborate than is the case among the Bora and other Amazonian Indian groups. The more elaborate techniques used by the Amerindians enrich the economic-plant diversity in the fallow and tolerate volunteer non-economic plants as well. Balée (1989) has suggested that these systems are probably the origin of most high density stands of Brazil nut (called castanhais), many palms (babaçuzais, tucumãzais (Astrocaryum tucuma), etc.), and other anthropogenic forest areas in Amazônia. This model, then, is definitely appropriate for the planting of income forests.
3) S. Flores Paitán's Sequential Experiments - Based upon the swidden-fallow model, these experiments are designed to yield subsistence and marketable products during several decades, until the timber species reach marketable size (S. Flores Paitán, Univ. Amazônia, pers. com. 1990). Plantation geometry is based on a 2 × 2 m or 3 × 3 m grid, with species having distinct growth habits, sizes, competition and pest tolerances occupying the 2 × 2 m, 4 × 4 m, 8 × 8 m, 12 × 12 m, 16 × 16 m or greater nodes.
The design is simple, flexible and easily understood and adapted by producers. Although the current experimentation is relatively new (< 15 years), it has been well accepted by several colonist and native communities around Iquitos, Peru. While current practice discourages non-economic volunteer plants and aims for a "climax" of timber species, this could easily be modified to attain the objectives desired for an income forest.
The agroforestry models mentioned, both high and low diversity, require the elimination of the forest as a starting point, and so are suitable for the swidden areas that each caboclo, Amerindian, small-holder colonist or extractivist prepares each year. The practices presented below start from the intact forest or from forests in various stages of degradation, but which still have some forest structure remaining.
Silviculture and Forest Management - Forestry has many practices that can be adapted from a timber production model to an income forest model. Also, timber may, in fact, be one of the numerous alternatives extracted from an income forest, as long as it is managed as a sustainable product, rather than being "mined". Poore (1989), however, point out that sustained-yield forest management is not currently practiced on more than an experimental scale in Amazônia, although many private and state businesses will argue violently that they are practicing it. Nonetheless, the technologies exist and have been used in various countries during the last century. As Palmer (1989) points out, the technologies themselves are biologically viable, although the socio-economic milieu in which they are used may not be conducive to an economically successful intervention. Of the numerous practices available, only two will be discussed here.
1) Enrichment planting - Numerous variations of this practice exist and have been experimented with in Amazônia and elsewhere. At the National Research Institute for Amazônia (INPA, Manaus), the Tropical Silviculture Department has used narrow trails cut in the forest or second growth areas to plant timber species at quite high densities. As an example, a trail is 4 m wide and separated from the next by 10-15 m of relatively undisturbed forest. Economic species are planted along the trail at 2-4 m intervals. In order to assure adequate growth of the desired species, the trails must be maintained open, both at ground level and at canopy level, until the economic species reach the canopy level (N.P. Fernandes, INPA, pers. com. 1990). On poor ultisols near Manaus, INPA has observed moderate growth rates of andiroba (Carapa guianensis, Meliaceae, an income-forest candidate with oil-rich seeds), with low incidence of tip-borer (Hipsiphylla grandella), in this system (P.T.B. Sampaio, INPA, pers. com. 1990). Other timber species are also giving good results.
An ecologically attractive alternative is to use natural or created gaps in the forest to plant a selection of economic species. This has the attraction of using natural forest dynamics, i.e. gap creation and forest regeneration in them, to enrich the forest. This was tried by a company working in the Madeira River watershed south of Manaus, but did not succeed because they did not adequately manage the gaps until the economic species could compete successfully with the natural regeneration, which is always very vigorous in the native forest. Nonetheless, this alternative form of enrichment planting deserves further evaluation because it involves lower levels of intervention than other forms.
In a forest that has been partially degraded by selective logging, numerous gaps and extraction roads are left in the forest (de Graaf & Poels 1990). Enrichment schemes in these areas would significantly improve the regeneration of the forest with a higher level of economic species. Again, any interventions in this type of forest would require continued management of the area, especially to control vines and less desirable competitors.
Although timber has been the objective of all Amazonian experimentation with enrichment plantings to date, many of the species used have been multipurpose species, like the andiroba mentioned above. Enrichment plantings have the additional advantage of introducing high-quality germplasm into the forest, which is not generally practiced with the next alternative.
2) Natural regeneration - An example of this practice is provided by de Graaf & Poels (1990). The CELOS Management System combines careful harvesting technology with a significant silvicultural intervention to create a sustained-yield forest management system based upon natural regeneration of economic species. Natural regeneration is defined as the use of naturally occurring seeds, seedlings and saplings of the desired species, without introduction of outside germplasm, to repopulate the disturbed area. This is precisely the way a forest regenerates from a disturbance in the absence of humans. In the CELOS system, humans direct natural regeneration, however, so as to increase the abundance and the growth rate of desirable species. On poor oxisols in Suriname, de Graaf & Poels (1990) report that a 20- to 25-year timber harvest cycle is possible. Although all experimentation to date with this and similar techniques has been with timber species, there is only one important limitation to adopting this practice directly for the creation of income forests.
The limitation is the genetic and market quality of the naturally regenerated population of the desired species. In forestry, the best quality trees are removed during harvest, leaving second-quality germplasm and juvenile material who's quality has not yet been examined. Over time, i.e. over several regeneration cycles, this results, inevitably and unintentionally, in the degeneration of the genetic base of the desired species. In forestry terms, the population will present fewer trees with straight stems and fast growth. This limitation wild be directly felt with any of the multipurpose forest species used in income forests (eg with andiroba).
With species that are not also potential timber crops, such as palms or copaíba, for example, natural regeneration may not provide a sufficient number of plants within the population that yield well or have sufficiently high-quality products for market. If this is the case, poor-quality individuals can be selectively eliminated from the area, allowing for the slow increase in abundance of high-quality individuals. Over time, this will result in the genetic improvement of the desired species, which will almost always be necessary to supply growing markets.
Given the limitations in genetic and market quality that may be present in the initial populations in any given area, it would make sense to combine enrichment plantings with regeneration management. This will increase the abundance and market quality of the managed populations, facilitating harvest and guaranteeing reasonable prices because of good quality.
The high-diversity models discussed above are directly relevant to the income forest precisely because they are highly diverse and can contain significant biodiversity. Flores Paitán's experimental work, for example, shows that high-diversity agroforestry systems can be readily adopted by modern producers with a market orientation. The use of enrichment plantings and natural or managed regeneration should also be readily adopted by producers, because they are already generally practiced to a certain extent by most caboclos, Amerindians, and small-holder colonists and only require orientation to point them in the direction of the income-generating forest.
As Gregersen et al. (1989) point out, any set of plants and plantation designs offered to the public will have to be site specific, both in ecological and cultural terms. Thus, the whole gamut of models discussed in the above paragraphs, plus numerous others that exist elsewhere in the humid tropics, have a potential role to play in developing income-generating forests. Bwel's (1986) emphasis on mosaics can guide the adoption of these different models into a coherent whole.
The models selected must be attractive to the users, principally in terms of the economic returns that the user can expect, but also for esthetic or cultural reasons important to them (Alcorn 1989). As researchers and planners, we visualize the income forests as being compatible with the conservation of tropical biodiversity in general. But the users are interested more in economic returns. Consequently, we can recommend most intensively those models and management practices that are flexible enough to allow for the accumulation of non-economic species in the matrix or at least within a mosaic that leaves space for biodiversity.
Once a list of candidate species and agroforestry or forestry models has been selected by an extractivist, colonist, Amerindian or other community, a research and development effort by state, national or international agencies and private conservation and entrepreneurial groups must be organized in collaboration with the community. This effort will include a genetic/agroforestry component, a marketing/entrepreneurial component, and an extension component. Some of the activities necessary are:
1) Product information availability - The importance of the market has been mentioned previously. Even before a product goes to market, however, information on its characteristics must be available to the commercial segments of the chain that will take it to market. The R&D institutes-and agencies that support this program must prepare documentation that can be distributed to all segments of the producer/merchant/consumer chain. This information must include not only health and safety data, product quality characteristics (chemical/nutritive), but expected yields, harvest periods, ease of processing, storage and transport, etc., obtained either from the literature or from directed research.
2) Selling the idea to the producers - The forest peoples are well aware of what they need (Allegretti 1990), but they have already been deceived many times by governments and entrepreneurs. Therefore, the extension of the income-forest idea and its new crop and agroforestry/forestry systems R&D must be done openly and honestly. This is best done in detailed discussion with forest producer groups, who will be expected to suggest and approve candidate species, species combinations, plantation layout, etc., based upon the best information that researchers, planners and the producers themselves can provide.
Once the idea is accepted, all stages of planning and execution of the research and development must be done in collaboration with the community. Rocheleau (1987) discusses the whys and hows of working directly with the community to be benefitted by any development project.
3) Provenance evaluation and germplasm selection - Wood (1990) outlines the process for agroforestry species and slight modifications of this process can easily be introduced to arrive at a full-fledged income forest. Arkcoll & Clement (1989) emphasize the importance of a wide genetic base, as most perennial species are outcrossing and thus highly variable. One outstanding provenance can make a species, while a dozen inferior ones are not worth pursuing. Once a wide genetic base has been obtained, selection for quality and yield is essential to develop high-quality plants and products for the consumers.
Germplasm collections, however, are expensive to develop, maintain, characterize and evaluate correctly. Since most forest communities will not have the resources to do this in the classical way, their individual home gardens and swidden plots can be used. Most Amazonians, and probably most other tropical peoples, use their home gardens for just this type of germplasm evaluation (Saragoussi et al. 1990). The R&D agents who work with the extractivists on developing new crops must maintain good records of germplasm location and assist the producers in evaluating those species unfamiliar to them. In this way, large amounts of germplasm can be evaluated and selections made with relatively low costs.
4) On-farm experimentation with selected models - Because an income forest may take 15-30 years to attain full stature, depending upon the route taken to establish it, the agroforestry/forestry experimentation must be done on farm. This is also necessary because large areas of Amazônia (> 100,000 km2) are already degraded and require urgent action to become productive again and because this time frame is similar to that for worst-case projections for the elimination of the Amazonian biome from most of its current area.
Flores Paitán's sequential and geometric designs have proven to be acceptable to producers in Peru and can be easily understood and implemented by producers elsewhere. They yield abundantly and continually throughout their development, which is a major reason for their attractiveness. They are also inherently attractive to university-trained R&D staff for the same reasons, and because they are perceived to be more "scientific" than the apparently random location of plants in Amerindian swidden plots. Where Amerindian knowledge and experience is available, however, this should be exploited. All on-farm experimentation should be done within a mosaic pattern, since this is a way to increase diversity at the farm level and thus reduce the risk of failure for the producer.
5) Entrepreneurial development - Most new crops or products require an active entrepreneurial participation to develop lucrative markets. Entrepreneurial capability is often lacking in Amazônia and elsewhere in the humid tropics, however, not only among the forest peoples but among the researchers and conservationists who are attempting to assist these peoples. Cultural Survival Enterprise's Rainforest Marketing Project has run into this barrier at the local level (J.W. Clay, Cultural Survival, pers. com.).
The development of local and regional entrepreneurs must be fostered in order to guarantee the success of this type of venture. In Brazil there are several agencies that support the small and medium entrepreneur, as well as cooperative ventures. Some international foundations specialize in training Third World groups through consultancies and courses organized with retired entrepreneurial and managerial experts from the developed world. A combination of these ideas with current efforts to market forest products can help develop the required know-how.
6) Getting long-term financial support for R&D - The international research and development agencies should be receptive to this idea since it includes both new crop development and agroforestry, both of which are "hot" areas for funding. This financial support for research and development should not be confused with the development of the commercial chain that will take the products from the forest to the consumer. This chain must remain free from subsidies that can distort the true economic value of a product or species. Most forest peoples desire a helping hand to become independent, not a hand-out to remain dependent.
The income-generating forest idea is extremely attractive, both for forest conservation and for sustainable development in the tropics. As outlined above, its implementation is feasible in agroecological terms. Agroforestry systems and forest management practices can play important roles in developing income forests, as they are known and will be easily accepted by the forest peoples. The agroforestry systems, especially, will provide an economic return during the years that they take to mature, and they are flexible enough to accommodate a great number of species.
If the species are chosen with care and if marketing channels and local entrepreneurs can be fostered, they will be economically viable. Nonetheless, once the necessary research starts it will be never-ending, since successful new extractivist products will be taken into plantation in other tropical areas and newer ones will be necessary to replace them.
Given the idea's potential for tropical-forest conservation and sustainably improving the quality of life of the forest peoples, the research and development necessary to design, plant and manage income-generating forests is well worth the investment.
I would like to thank Drs. Johannes van Leeuwen, INPA, and Jason W. Clay, Cultural Survival, for numerous discussions that helped form the ideas presented here; James H. Fownes, Univ. Hawaii at Manoa, for critical review of early drafts of this chapter; John R. Palmer, Tropical Forestry and Computing Ltd; and Ghillean T. Prance, Royal Botanical Gardens, Kew, for critical review of later drafts of this chapter.