Abstract

During recent years, the concept of complex, multi-strata ‘agroforests’ has been emerging as an exciting new model in the range of agroforestry landuse systems and practices. These climax agroforests are biodiverse and economically viable systems established through the enrichment of forest fallows with commercially important perennial cash crops. However, they have really emerged as land use management systems over the last 100 years and are still expanding and evolving. They are not a relic of a former culture, but are a reaction by local people to present day problems, and are thus very dynamic forms of land-use. The best examples of these agroforests are the damar agroforests in Sumatra, and rubber agroforests in Sumatra and Kalimantan. In Sumatra, agroforests produce 80% of the damar resin, up to 90% of the fruits and about 60% of the other non-timber forest products, while in Kalimantan they produce over 70% of the rubber from Indonesia.

The farmers with agroforests have traditionally acquired rights to land through planting and managing the trees, inheritance or land purchase. Recently, insecurity arising from the conflict between indigenous tenure systems and corporate interests on State Forest lands has brought into focus the issue of tenure security for smallholder agroforesters.

The sale of tree products from agroforests provides farm income of approximately $1000 ha per annum, which is well distributed over the year and from relatively low labour and capital inputs. The economic returns from these agroforests are generally in excess of those from other land uses, including plantations, and there are additional and important social benefits. Environmentally, agroforests have been shown to have functions similar to those of secondary forest and tree fallows, while also having substantially greater biodiversity than plantations and being sinks for carbon and methane. In Indonesia, these agroforests, which can be based on several tree species, first appeared some time ago. Consequently, they are considered to provide an attractive alternative to slash-and-burn agriculture, which may be important for the development of sustainable resource management elsewhere in the moist tropics.

INTRODUCTION

There are numerous definitions of agroforestry (Nair, 1989), all of which cite the integration of trees, and other woody perennials, with crops and/or animals, either in spatial mixtures or temporal sequences. This satisfies agronomists and foresters viewing the important socio-economic role of agroforestry from a production standpoint. But it has long been claimed that agroforestry also produces environmental benefits, such as enhanced nutrient and water cycling, through the ecological interactions between trees and crops (e.g. Anderson and Sinclair, 1993). The most recent definition of agroforestry (Leakey, 1996), attempts to embrace these ecological and agronomic benefits within the concept that deliberately mixing plants and animals with different lifespans, foodchains, lifecycles and physical structure, creates a dynamic agroecosystem that can progress through a succession akin to that of natural ecosystems. In this case, it is suggested that the increasing biological and physical diversity of maturing or climax agroecosystems, such as multi-strata agroforests, is associated with an increase in sustainability derived from enhanced ecological processes that influence the functional interactions between the species (Michon and de Foresta 1993; Leakey, 1999). If this can be shown to be true, multi-strata agroforests perhaps represent our best chance of creating sustainable, man-made production systems, which combine local and global environmental benefits (e.g. biodiversity, carbon sequestration and trace gas emissions) with economic production.

The above suggestion, however, does not address the question of where high input monocultures producing the vitally important staple foods (e.g. cereals), which have been bred for growth in full light, fertile, weed-free environments, fit into the agroecosystem. This question is especially important, as we don't have many shade tolerant staple food crops that could be included within mature agroforests. One option is that they are part of the temporal sequence in land use and occupy the land during the first three to five years of the agroecological succession immediately after clearance of the forest or climax agroforest. But another option is that some parts of the landscape are maintained in permanent high input agriculture, and that the advantages of diversifying agroecosystems are achieved by integrating these monocultural systems within a landscape mosaic in which diverse species mixtures are restricted to less fertile and environmentally more vulnerable areas. In other words, high-input systems are segregated from low-input systems, but in a way that provides diversity and integration at the landscape scale. This then raises the question about what are the appropriate scales? Impacts of different land uses can occur at a range of temporal and spatial scales, from individual farm plots (e.g. soil nutrient status), the farm itself (food/income security), to landscape (erosion and pest-parasite dynamics) and watershed scales (biogeochemical and hydrological cycling), through to regional (biodiversity conservation) and global scales (trace gas emissions and carbon sequestration). It is important to realise that an impact, which might be negative at one level, such as a decline in crop production under a tree, may be beneficial environmentally and economically at a larger scale. Research is needed (Frost and Leakey, in press) and some is in progress (van Noordwijk et al., 1995), to develop greater understanding of how the integration or segregation of different land uses is likely to affect food production, environmental degradation, biodiversity and climate change.

There are some excellent examples of climax agroforests in Indonesia, where there has been a long tradition of cultivation of indigenous fruit trees and of trading in spices and other tree products (Michon, 1993). These agroforests, which are frequently juxtaposed with rice paddies or swiddens, make good examples for the study of some of the questions and ideas described above. Elsewhere, multistrata agroforestry systems are mostly localised around the compound; the so-called 'home garden'. These are common throughout the tropics, but are perhaps best developed in south east and south Asia (Soemarwotto, 1987). Agroforests provide a source of scientific insight on how to develop sustainable land uses (Colfer et al., 1988; Michon and de Foresta, 1993), while perhaps removing the incentive for the protection of primary forest that is provided by the existence of extractive resources (Lawrence et al., 1995).

STATUS OF MULTI-STRATA AGROFORESTS

Origin and expansion

The history of the multi-strata agroforests of Indonesia (Michon and de Foresta, 1999) is not well known, but it is clear that they are not a relic of a former culture as some have supposed.

The oldest types of agroforest were based on useful tree species established as home-gardens around temporary settlements. They emerged hundreds of years ago and were associated with the early domestication of forest fruit and nuts in the archipeligo. In areas of shifting cultivation, these fruit-based agroforests remained as fruit tree islands in aging secondary forests, when the settlement moved on. Similar fruit agroforests are presently developed extensively along roads and river banks. Agroforests further developed through the domestication of spices and stimulants, with the advent of long-distance trade in these commodities to the Far East. Thus the earliest examples of commercial agroforest development pre-dated the colonial era. Cloves (Eugenia caryophylla) and nutmeg (Myristica fragrans) were cultivated in the Mollucas around the 5th Century. Then, some 300 years ago, benzoin (Styrax benzoin), the Asiatic incense tree was cultivated in North Sumatra. Then, a little over 150 ago, cinnamon (Cinnamomum birmani), illipe nut (Shoreaspp.) and rattan (mainly Calamus caesius and C. trachycoleus) were brought into cultivation in central Sumatra, Western Borneo and the lowlands of Kalimantan and Sumatra respectively. The concept of agroforests as we know them today, really expanded, however, in both area and geographic distribution, during the late colonial times, around the turn of the 19th-20th Century. This development of great economic, social and environmental importance, was linked to the growing trade in natural products like resin and latex to meet the demand from emerging European industries. This led to the cultivation and domestication of resin (e.g. damar) and latex (e.g. rubber) producing trees. These commercial agroforests are still expanding, principally on the forest margins of Sumatra and Kalimantan. Other commercial agroforests based on fruit and timber species are being developed currently in Kalimantan (de Jong 1994; Salafsky 1994).

There are many forms of these complex agroforests which represent a very dynamic and efficient type of land use, all of which involve the combination of a range of tree species, but they can usually be distinguished by their dominant species (damar, rubber, cinnamon, etc.).

Damar agroforests

Damar resin is used in paint and varnish manufacture and in incense making. The resins of about a dozen tree species are jointly marketed as Damar and, as a consequence, there is considerable variation in quality. High quality damar is classified as ‘mata kucing’ and is produced mainly by two species, of which Shorea javanica is the only one brought into cultivation. At the producer level, there is an important price incentive for harvesting good quality damar. The resin is therefore carefully sorted by the farmers or by local traders and paid according to quality. Then the different qualities are exported. The highest grade goes to Singapore where it is then re-exported to very specific markets, such as the high quality paint factories in Germany. The lower grades form the bulk of what is called on the international market ‘gum damar’. Those who do not have privileged links with high grade damar traders in Singapore therefore receive unsorted damar. The market niche for good quality resin is in fact already occupied, but it could be widened as many users in Europe complain from the lack of quality. This can perhaps be blamed on the organization of the trade chain.

The importance of damar resin for the economy of Sumatra was mentioned as far back as 1783 (Marsden, 1783). The first plantings of Shorea javanica were probably established around 1885, as in 1937 Rappard reported about 70 hectares near Krui, some of which were about 50 years old (Michon et al., 1996). At this time these plantings were considered a promising development (Torquebiau, 1984). In 1936, damar production in the area of Krui was about 200 tons, rising to an estimated 10000 tons at present (Michon et al., 1996). According to recent analyses of satellite imagery undertaken by the Indonesian Department of Forestry, there are now more than 50000 ha of mature damar agroforests in the Krui area, but this excludes immature agroforests that are indistinguishable from other forms of vegetation by these techniques.

It appears that the incentive to plant the first damar agroforests has been the combination of biological, social and economic factors, including:- the high price of the resin; the destruction of the natural forests and consequent social conflicts between villages harvesting from the few remaining natural trees; and the adaptability of the species to cultivation (Michon et al., in press). The first damar seedlings were introduced as wildlings into the swidden fields (ladang) after the rice crops were harvested. Planting trees on swiddens was not a new concept to Krui farmers, as they already had the tradition of replacing the natural fallow with a mixture of coffee, pepper and shade trees. These plantings were productive for 10 - 20 years before being either re-opened for rice cropping or abandoned for a longer natural fallow. Damar introduced amongst these coffee bushes or pepper vines, were found to establish well and to continue growing alongside naturally regenerating vegetation when the coffee and pepper were ageing. Twenty to twenty-five years after planting, the young damar trees were ready for tapping. These cultivated damar trees can to be tapped for a further 25-50 years. These agroforests were subsequently enriched further with other trees planted for their fruits and other tree products. As the agroforests aged, decaying trees were replaced on a tree-to-tree basis, but the whole damar garden was rarely re-opened for a new rice/coffee/damar cycle. Thus a new cash cropping system, has emerged from a traditional rotation system in which subsistence crops were alternated with commercial coffee/pepper gardens and/or natural fallows. A typical damar agroforest is composed of 72% damar (Shorea javanica) and the rest a mixture of Durio zibethinus, (durian) Parkia speciosa (petai), Lansium domesticum (duku), Gnetum gnemon,(tangkill) Syzygium aromaticum(cengkeh), Garcinia parvifolia, Plectonia glabra and other tree species (Torquebiau, 1984). This new landuse system constitutes a permanent land-use unit with trees, a true agroforest.

By 1994, over 80% of the villages scattered along 120km of the Pesisir coast in SW Sumatra had damar agroforests and about half the population were involved in damar production activities (Michon et al., 1996). The damar area is still expanding as more land, formerly devoted to coffee and then abandoned to fallow, is being planted with damar on an increasing scale.

Rubber agroforests

In Sumatra and Kalimantan and some other areas of Indonesia, agroforests are based on rubber production. These rubber agroforests are commonly found in areas with little agricultural potential as a consequence of acidic, leached soils and grass weeds. Rubber seeds were introduced in the 1910’s from Malaysia and planted in the ladang shortly after the dryland rice. Farmers then abandoned the swidden cultivation after 1-2 years and allowed a rubber-enriched fallow to develop (Gouyon et al., 1993). Jungle rubber agroforests, can be tapped for about 30 years once the trees are about 10 years old. Some tree-to-tree replacement might occur to regenerate the system, but usually, after a maximum of 45 years, the whole stand of rubber is cut down and burned for a new rice/rubber cycle. Rubber agroforests represent a good example of an economically profitable, though still ecologically beneficial, replacement of traditional 15 year natural fallow with a commercial crop (400-500 trees ha-1). Rubber agroforests fulfil many of the environmental benefits of a natural fallow, as the rubber tree effectively replaces pioneer trees in a secondary forest fallow. The rubber agroforest still contains 250-300 other plant species (Gouyon et al, 1993), among which are many common pioneers, but also useful tree species like durian, langsat, mango, rambutan, etc. (Lawrence, 1996).

Rubber agroforests are grown on a large scale; about 1.3 million families farm 2.5 million hectares. In Kalimantan, near Kembera, the proportion of rice fields planted to rubber after the swidden period rose from 25% to 83% between 1990 and 1995 (Lawrence, 1996), again emphasising that agroforests are a very dynamic and expanding form of landuse.

The success of rubber agroforestry is determined by the way it complements the swidden cycle - politically, economically and agronomically (Dove, 1993). Part of this success relates to tenurial issues. As with damar, planting rubber is customarily acknowledged as a regular procedure for securing permanent and inheritable rights on land. The national legal system might acknowledge or ignore these customary regulations, but in practice planted land has a greater chance of being classified as 'utilized' than an empty fallow. Consequently, there is a greater chance to receive compensation in cases of land grabbing by a nationally-supported project. In the political context of the last 32 years, developing agroforests therefore gave farmers a strategy to increase the chances of maintaining their traditional rights on the land.

In Indonesia, rubber agroforests have rapidly become far more important than industrial plantations. As early as 1930, agroforests had overtaken plantations in terms of the planted area, as well as in terms of production. In spite of intensive government programmes for the promotion of monoclonal rubber plantations, agroforests produce about 73% of the nation's rubber. The value of this trade was almost US$1.9 billion in 1995 (International Rubber Study Group, 1997). Barlow et al. (1994) have indicated that these smallholders are the most efficient natural rubber producers in the world. The average yield of rubber from these agroforests is about half the yield of the rubber estates (Penot, 1995), but these low yields (593 kg ha-1) are to some extent compensated for by the other products coming from the agroforests. In the future, clonal smallholder rubber might replace traditional jungle rubber production, if appropriate clones are identified.

Recently, there has been a widening of outlets for rubber wood, which is now used extensively for plywood and furniture. Farmers in Indonesia have made Rps 600000 (US$ 300) per hectare from wood harvested from old rubber plantations (Gouyon et al., 1993).

Cinnamon agroforests

Cinnamon agroforests have been developed along the west coast of Sumatra. For example, in the high Kerinci Valley south of Padang, the farmland can be divided into three areas along a topographic sequence. Paddy rice occurs on the alluvial plain, with mixed tree agroforests on lower slopes, which include cinnamon crops (Cinnamomum burmani) in the understorey as well as larger trees of Durio zibethinus, Artocarpus heterophyllus, Toona sinensis, Aleurites moluccana, etc. in the canopy. More specialized cinnamon stands, sometimes mixed with coffee, are found on the hillsides. Natural forest occurs on the higher parts of the hills (Aumeeruddy and Sansonnens, 1994; Aumeerudy, 1994). The tree crops in this area occupy about 77% of the land. Cinnamon bark can be harvested anytime between 5 and 15 years after planting. Typically these tree crop systems are practised on 20-30 year rotations. However, over this period there are several management options affecting the rotation length, the dominance of different species and the species diversity of the stand (Aumeeruddy, 1994).

In Kerinci, cinnamon cultivation increased from 5950 ha in 1966 to 27534 ha in 1972 and 42577 ha in 1989 (Aumeeruddy, 1994). However, part of the new cinnamon area is under intensive, monocrop cultivation, quite different from an agroforest. Cinnamon cultivation has recently extended beyond its 'traditional' highland limits into the lowlands, and tends to be established, in rubber agroforests.

Other types of agroforests

Tengkawang (illipe nut: oil-producing Shorea species) agroforests are typical of West Kalimantan. Like damar agroforests, they form biodiverse and forest-like permanent agroforests which, once established around settlements, will not be cleared again (Momberg 1993). However, Tengkawang timber is one of the valuable merantis and consequently highly prized by foresters. Because of this and the close structural resemblance of these agroforests to natural forests, they have often been allocated to forest concessionaires by the government and irremediably logged. Consequently, the area of tengkawang agroforests has been drastically reduced in recent years.

Rattan agroforests are a common feature in the lowlands of East Kalimantan (Weinstock, 1983, Fried 1995). Like rubber agroforests, they are integrated into an extended shifting cultivation cycle, replacing natural fallows. Rattan vines, established in rice swiddens, usually grow along with pioneer tree species, which they use as support, but mature rattan agroforests also include useful fruit and timber species. The length of the rattan 'fallow' can be reduced to 10 years, but typically it is over 50 years. Their productivity, estimated either in terms of returns to labour or direct income per unit area, is astonishing. Alternated with slash and burn agriculture for subsistence purposes, they represent a very harmonious system of land development in areas of low fertility. Unfortunately, rattan agroforests, as with most other agroforests mentioned above, are established on public forest lands and in areas frequently targeted by industrial plantation development projects. In spite of vigorous protests from the rattan farmers themselves, as well as from various NGOs and scientists, a large part of the rattan area has already been seized and cleared for oil palm or Acacia mangium plantations. The area has also suffered heavily from the drought and fires of the 1989 El Niño season.

Tenurial rights

As for many developed countries, the harmonious development of forest margins in Indonesia is affected by the often confrontational co-existence of two parallel rights systems concerning land appropriation and utilization. On one hand, the numerous, locally-specific, highly flexible and evolving customary systems; and on the other, the universal, often very rigid, national legal system. This problem poses a threat to agroforests, despite their evident proliferation.

Most customary land tenure systems presently acknowledge permanent and transferable rights over land developed through tree planting. Therefore in most examples of agroforest management, the planter or his descendants own, either individually or collectively, not only the trees but also the land. The definition and restriction of agroforest ownership may vary from one region to another, but the concerned rights and duties (in terms of access, use, control or transfer) are acknowledged and usually enforced by all the members of the local community. For example, in Krui, damar agroforests are ‘individually’ owned, but the use and transfer rights, as well as attached obligations, differ between lands acquired by inheritance and those acquired through creation or purchase (Michon et al 1995). Inherited agroforests cannot be sold, nor can trees be cut, without the permission of all brothers and sisters and further transfer to children follows special rules. By contrast, newly created or purchased land can be transferred to anyone, through inheritance or sale, according to the good will of the owner.

The Indonesian Constitution has established inalienable 'State forest lands', which cover 74% of the whole nation and remain under the exclusive authority of the Department of Forestry. On these State forest lands, the Indonesian Constitution theoretically acknowledges pre-existing customary rights. However, the practical translation of this basic law merely tolerates these customary systems, at least as long as the indigenous systems and people they concern do not compete with higher 'national' interests. This 'acknowledgement' is not the same as 'granting permanent rights' as on State forest lands, only temporary use rights can be granted. Even outside the State forest lands, the legal framework does not consider customary property rights as legal ownership of land, because permanent land ownership has to be acknowledged through an official, and costly, titling process. Since more than 95% of the land under customary control is included in these State forest lands, local people cannot expect more than concession rights over resources and land that they have developed for centuries. Practically that means a temporary, often more tacit than official, permission to stay until a large development project (logging, oil palm, pulpwood estate or transmigration) comes and pushes the people out of their ancestral lands.

In Krui, most of the damar agroforests occur in State forest lands. Damar farmers have no legal way of getting official land titles. As long as these State forest lands in Krui remained far from the covetous eyes of large projects, the co-existance between the customary systems and the national legal framework remained more or less in harmony. But everything changed when some timber and oil palm planters realized that there were few projects around Krui and that it was full of highly valuable timber trees, and close to Java. Farmers suddenly discovered that, in the national context, their internally strong tenure system was in fact totally vulnerable.

During the last 15 years, logging, and plantation projects, have destroyed, often in spite of strong local opposition and resistance, large areas of agroforest. This has included all the remaining rattan gardens in South Sumatra and many of those in Kalimantan; many old fruit and tengkawang agroforests in East, Central and West Kalimantan; large areas of rubber agroforests over the lowlands of Sumatra, and some damar agroforests.

This incidence of tenure insecurity holds true for most of the agroforest farmers of Indonesia. Nationally-supported projects are inevitably stronger, in practical legal terms, than indigenous systems and companies easily purchase legal rights that allow them to expel farmers and destroy their systems with a minimum, if any, compensation. This seizing of indigenous lands and rights is facilitated by the fact that forest or land use maps do not reveal the presence of villages and the agroforests. Rattan or fruit gardens, damar or rubber agroforests are all mapped either as 'primary forest', 'secondary forests and bushes', or 'degraded land'.

In recent years, ICRAF and its partners have been involved in studies and dialogue with policy makers aimed at policy changes to give farmers stronger rights over their agroforests. A breakthrough occurred in January 1998, when the Minister of Forestry signed a decree establishing a 'Special Use Zone' inside State forest lands in Krui where farmers could be granted long term and inheritable tree tenure rights (Fay et al., 1998). Another way of helping agroforest farmers to secure their rights is through the scientific promotion, as we are doing in this World Congress, of the agroforest concept itself. This should help to ensure more recognition by government officials, developers and policy makers, of the intrinsic value of agroforests at a local, regional and national level.

Economic and social importance:

Agroforests in Indonesia significantly contribute to the national and international economy, producing about 73% of the rubber latex consumed and exported by the country (this represents about 25% of world’s natural rubber production:- International Rubber Study Group, 1997). They provide 70 to 80% of the damar resin commercially traded inside and outside the country and roughly 95% of the marketed fruits and nuts (Michon and de Foresta, 1995), 63% of the cinnamon, nutmeg, candle nut (Aumeeruddy, 1994), and a significant proportion of rattans and bamboos. Agroforest products play a major role in regional economic development, by supplying regional industry and providing inputs to marketing chains that branch out far beyond regional boundaries.

Agroforests are economically important for villagers. They usually provide between 50 and 80% of total agricultural income of villagers (Mary et al., 1987; Gouyon et al., 1993; Levang, 1992). The diversity of income sources, as well as the secondary domestic production to meet the household's subsistence needs, are an essential asset in the economic security and welfare of villagers. Agroforests do much to ensure that the quality of villagers lives are reasonable, and help to provide the schooling of children, which is commonly given top priority. Income from regularly harvestable products such as rubber latex, resin, coffee and cinnamon bark, from agroforests cover the cost of every-day expenses; while that from seasonal products such as fresh fruits, clove, nutmeg provide for some of the annual expenses. Other commodities, like timber, provide important sums of money on an occasional basis, to meet exceptional expenses or to serve as savings. This diversity of income is essential as habits of saving money are not well developed and credit is usually expensive or unavailable.

Agroforests also provide food and nutritional security from fruits, vegetables, herbs, spices, etc; as well as the day-to-day needs of the community (construction wood, firewood, thatching and basketry material). In addition, they can be managed as capital assets. For example, a garden or even individual trees, can be "pawned" through special agreements that allow any family to overcome difficult periods without resorting to selling the trees or land (Mary and Dury, 1994; Dury et al., 1996). This economic diversification is favoured by farmers. Security, risk aversion and flexibility of management, are important in a context of potential biological disasters or market failures. Flexibility includes versatility of marketing options, as well as flexible use of family labour, so that more time can be devoted either to intensive rice cultivation, with two crops of hybrid varieties per annum, or to off-farm income generation.

Socially, agroforests are important in terms of the significant contribution that they make to employment, mainly through activities like harvest, transportation of products from the field to the village and to market. These raise significant additional income and other benefits for the community, especially children, and allow landless people to benefit from agroforest activity and from security (Gouyon et al., 1993; Mary and Michon 1987).

Damar agroforests

A socio-economic study of the damar agroforests near Krui, carried-out in 1985, found that the damar resin alone formed 30-50% of the total production value of small rice farms (0.42 ha) with some damar (0.6 ha), that totalled an income of $910 per annum (Mary and Michon, 1987). In a more recent study, mean annual returns from mature damar agroforests were found to be slightly higher: Damar $682, fruits and timber $464, from 127 man days labour per family (de Foresta and Michon, 1997). Importantly, the revenues from the damar resin are spread across the year, in contrast to the seasonal revenues generated from rice (Table 1). Trees are regularly tapped every 2-3 weeks, except in peak periods for labour in rice fields, or in adverse climatic conditions. Revenues from rice come in February ($270) and August ($200). Income from resin sales does show some fluctuation ($22.5-$52.5 dollars per month), that can be due to either climatic or organisational factors. However, this is less marked than that either from coffee, which has irregularities in production, price and marketing, or fruits, that are highly seasonal and unpredictable from to year (Mary and Michon, 1987). In a good fruiting year the revenues from, for example, the fruits of Lansium domesticum (Duku) or Durio zibethinus (Durian) can double the income from the agroforest (Levang, 1992; Bouamrane, 1996). Peasants use the different types of income for different purposes; matching regular income from damar, fish and small animals against regular expenses, seasonal income from tree fruits against annual expenses and irregular income from timber, etc. against exceptional expenses.

Table 1. Cash flow (US$) from rice and damar resin sales from a farm with paddy fields (0.42 ha) and damar agroforest (0.6 ha) in Penengahan, near Krui, Sumatra (Source: Mary and Michon, 1987).

The major difference between rubber agroforests and monoclonal rubber plantations is that, in contrast to plantations, the agroforest represents a low input and economically diverse landuse system, based on a range of products in addition to rubber. Throughout their economic lifetime, rubber agroforests account for up to about 84% of the average income of a household per hectare per year (Gouyon et al., 1993). This compares favourably with smallholder rubber monocultures (Table 2), which through loss of diversity carry a higher economic risk and lower environmental benefits. Farmer income ($2300 from 2ha) has been reported to be slightly in excess of a monoclonal rubber plantation of the same size, and is thus one of the most profitable landuse systems for smallholders in Sumatra (de Foresta and Michon, 1997).

Trees are tapped every 3-5 days, providing cash throughout the year. During the first three years crops cover the bare soil. Later, a rough estimate indicates that the developing bush fallow saves farmers $250 in materials, herbicides and labour against Imperata grass, the main weed of plantation rubber. The young agroforests also provide food self-sufficiency or opportunities for cash generation from rice, bananas, pineapples, vegetables, etc. This cash provides immediate returns against the weeding costs that are needed to protect the young trees and diversifies farmer income. Importantly, the non-rubber tree component in mature agroforests also fulfils the income diversification function (Table 2).

Current research to identify rubber clones adapted to the agroforest environment and hence more productive when mixed with other species is an interesting, but expensive and long term option. A more advantageous strategy would be to focus tree improvement research on product quality, especially if this enhances economic returns (Penot, 1995). The development of better transportation, marketing and processing of the non-rubber products from these agroforests would, additionally, help to sustain farmer income and species diversity (Gouyon et al., 1993).

Table 2. Average farmer’s income (US$) from jungle rubber and rubber plantations, over the economic life of the rubber trees (Source: Gouyon et al., 1993). In 1990: Rp 2000=$1.

There have been few comparative studies in SE Asia of the economic returns from complex agroforests and other land uses, but some are in progress. However, outside Indonesia, agroforests have been shown to be economically attractive. A study carried-out at an experimental level in Yurimaguas, Peru suggests that multistrata systems based on Cedrelinga catenaeformis, Colubrina glandulosa, Bactris gasipaes underplanted with Coffea arabica, Eugenia stipitata, Inga edulis and Centrosema macrocarpum arranged in a geometric pattern (ICRAF, 1997) are economically a very attractive landuse option (Table 3). Similarly, in Brazil, açai agroforests have 4-5 fold greater net returns to labour than sugar cane swiddens (Hiraoka, 1995).

Table 3. Comparison of net present values and internal rates of returns of production systems in 1985 and 1991 prices using a 15-year time horizon, Yurimaguas, Peru (Source: ICRAF, 1995)

The social importance of landuse systems can be critical for their adoption. The emergence and the present success of these tree-based landuse systems seems to be a good example of farmers responding to the four situations which Arnold and Dewees (1997) have reported as requirements to make tree-based options attractive to farmers. These are the need to:

• maintain supplies of tree products as natural stocks decline due to deforestation or loss of access,
• meet growing demands for tree products as populations grow, new uses emerge or external markets develop,
• maintain agricultural production in the face of declining soil quality and increasing land degradation
• contribute to risk reduction and risk management in the face of tenure insecurity, fluctuations in income, seasonal demands on labour or to provide reserves to buffer against times of stress or emergency.

Limiting factors for expansion

One of the limiting factors to agroforestry with trees with marketable products, especially for perishable products such as fruits, has been the lack of adequate transportation and marketing (Gouyon et al., 1993). This is alleviated by good road access allowing traders to come in from other areas to meet urban consumption demands. The development of commercial tree crops in the Kerinci valley of Sumatra expanded in the 1920’s by the creation of a road link to Padang. More recently, the improvement of the road from Krui to Jakarta has boosted the commercialization of fruits, and consistently increased the income from the damar.

Another limiting factor in the future of agroforests concerns the potential reduction in average agroforest land holdings per household, due to decreasing land availability. In the frontier zones of Jambi in Sumatra, surveys by Gouyon et al. (1993) found that households own 5 ha of rubber, while those in south Sumatra where most land is already developed have 2.5 - 3 ha, the minimum that is economically viable. With increasing land scarcity, labour opportunity costs and standards of living, will jungle rubber remain profitable? It seems likely that ways of making agroforests more profitable will have to be found. There are two options: i) further diversification, for example into quality hardwoods, and ii) further selection and improvement of tree species to enhance the yields and quality of their non-timber products to market specifications. More systematic and scientifically supported selection of the best damar producers could help the damar farmers meet the increasing demand for high grade resins in the international market. In the same way, selection of rubber clones adapted to the levels of competition, or shade, in an agroforest environment may be better than either wild rubber or the clones selected for performance in monocultural plantations (Gouyon et al., 1993; Penot, 1995).

Lastly, one of the most threatening factors for further expansion of agroforests concerns land availability. Large areas of forest are still unused in Sumatra and Kalimantan, but it is unlikely that this land will be available for smallholder agroforests. Most forest frontier areas of Indonesia, lie in State forest lands, and are targeted by government-sponsored, large-scale plantation projects for oil palm or timber production. However, one promising area for further agroforest expansion could be the rehabilitation of degraded Imperata ('Alang-alang') grasslands that are not coveted by corporate agriculture (de Foresta and Michon, 1997). New policies conferring tree tenure to farmers in the 'Alang-alang' grasslands on State-owned land would consequently be very important for both the well-being of landless families and for land rehabilitation.

Environmental importance

Research on the benefits of agroforestry to soil fertility have primarily focussed on short-rotation agroforestry practices with leguminous tree species, managed as improved fallows or as intercropping systems (Cooper et al., 1996). While most of the hypotheses relating to benefits of agroforestry to nutrient cycling and fertility management are not conclusively proven (Sanchez, 1995), there are many reports of positive effects. Experiments at Yurimaguas in Peru have compared different agroforestry practices, including the already described multistrata agroforest (ICRAF, 1997). Data from this experiment comparing soil nutrient stocks under the different land uses, have shown that the changes over 10 years under multistrata agroforests are similar to those in secondary forest and in 8 years old fallow after shifting cultivation (Table 4). The conclusion was that complex multistrata agroforests were able to maintain a level of soil fertility superior to that of 20 year-old forest, despite supporting the production and nutrient off-take arising from ten years cropping. The same experiments also demonstrated that trees in complex agroforests have well developed symbiotic mycorrhizal associations and support large populations of soil macrofauna, especially earthworms (ICRAF, 1997).

Table 4. Percentage change after 10 years in the nutrient stocks under different landuse management systems at Yurimaguas in the humid lowlands of Peru (Source: Alegre and Bandy, in press) * = 8 years after 2 years of shifting cultivation

Multistrata agroforestSecondary forest (20 years)Fallow* after shifting agricultureLow input agricultureHigh input agriculture Peach palm plantation with Centrosema Carbon

In a study in Kalimantan, the physical and chemical properties of soils in rubber and fruit agroforests and fallows were superior to those of primary forest (Lawrence et al., in press), leading to the conclusion that these man-made systems may be sustainable from the point of view of soil fertility. In this context, it is interesting that the farmers in Krui often state that, since the last 50 years of intensive colonization by damar agroforests of former shifting cultivation areas in the hills, the water supply for ricefields in the valley bottoms has considerably improved in quantity and regularity.

In addition to these local benefits, climax agroforests may have important global benefits. Preliminary results from Jambi in Sumatra suggest that like secondary forest, soils in rubber agroforests act as sinks for methane and might off-set a substantial fraction of the methane emissions from paddy fields (van Noordwijk et al., 1995). Maturing rubber agroforests also appear to be a net carbon sink (van Noordwijk et al., 1995.

Preserving biodiversity would be another important global benefit. In most farming systems non-crop species are in conflict with the income/food needs of the household. In contrast to this all too common situation, complex agroforests are an example of where biological diversity provides economic returns. These returns come in the form of supplementing farmer income, reducing dependancy on a single crop while, at the same time, allowing expansion of planted area with minimal capital and labour input, and reducing the need for shortened fallows. It is this unique quality that makes complex agroforests of great importance to those interested in sustainable natural resources management (de Foresta and Michon, 1993; Michon and de Foresta, 1999 ).

Recent data from Indonesia (Michon and de Foresta 1990, 1995) has shown that agroforests contain a large proportion of the species found in natural forests and very substantially more than in monocultures (Table 5). Thus, with their capacity for production and economic returns to farmers, agroforests are a valuable compromise between the conservation of tropical forest biodiversity and profitable use of natural resources (Michon and de Foresta, 1990; 1995; Leakey, 1999). There is, however, a need for research within agroforestry to determine the relationships between intensification, biodiversity and ecological functioning at different spatial and temporal scales (van Noordvijk et al., 1995). At higher scales (from watershed upwards) a landuse mosaic with corridors between forest/agroforest patches may be the most appropriate means of acquiring the scale to achieve some level of ecological equilibrium. Interestingly, this is often the pattern in Indonesia, where paddy rice is grown in the valley bottoms and agroforests on the valley sides.

Table 5. Biodiversity in Indonesia agroforests: observed numbers of species (Source: Michon and de Foresta, 1995; Leakey, 1999).

The damar agroforests are an example of a landuse system that has been developed by farmers without the involvement of scientists. Within the context of the domestication of agroforestry trees for the production of non-timber forest products (see Leakey et al., 1996), they have been described as 'domesticated ecosystems' (Michon and de Foresta, 1996), and thus as one of the pathways for bringing new species into cultivation (Leakey and Simons, 1998). In modern agriculture and forestry, crop domestication has usually led to intensively-managed monocultures and to landuse practices that are only sustainable with large inputs of fertilizer and pesticides. The complex agroforests of Indonesia run counter to this twentieth century model of agricultural production and thus give hope that new and more environmentally desirable landuse systems can be developed.

Potential for rehabilitation of 'alang alang' grassland

Concerns about the limited land available for agroforest establishment in the forest areas could perhaps be resolved by the promotion of agroforests to rehabilitate the expanding areas of Imperata cylindrica (‘alang alang’) grassland due to the spread of slash-and-burn agriculture. Optimism that the establishment of agroforests could be adapted to this situation stem from the evident need for an approach based on a natural plant succession and the striking structural and functional resemblance between agroforests and natural forest succession (de Foresta and Michon, 1997). This is enhanced by the economic flows which minimise unproductive periods of landuse. Conversion of ‘alang-alang’ grasslands to rubber agroforestry has been calculated to be profitable up to a real discount rate of almost 15%, although the real opportunity cost of capital probably falls somewhere between 15-20% in Indonesia (Tomich et al., 1997).

MODEL FOR OTHER TROPICAL AREAS

The Indonesian experience with complex multistrata agroforests stands out from those in other parts of the world, however it is not totally unique. There are other examples, albeit usually more localized and smaller in both individual and overall scale. Where the scale is as big or bigger, as in the use of shade trees over coffee and other perennial crops in Latin America (Beer, 1987), the degree of complexity is not as great as the Indonesian agroforests. Multistrata home gardens are found in West Africa (Okafor and Fernandes, 1987; Watson, 1990), Tanzania (Fernandes et al., 1984), Micronesia (Raynor and Fownes, 1991), Philippines (Fujisaka and Wollenberg, 1991), Thailand (Gajaseni et al., In press), China (Saint Pierre, 1989), Peruvian Amazon (Padoch and de Jong, 1987), Brazil (Hiraoka, 1995), Sri Lanka (McConnel and Dharmajala, 1973; Luu, 1989), Guatamala (Gillespie et al., 1993), Mexico (Gliessman et al., 1981; Alcorn 1984) and many other places (Fernandes and Nair, 1986).

In West Africa subsistence farmers do already enrich the forest fallow away from the home compound with cocoa, indigenous and exotic fruits, vegetables and other crops, so the jump to an agroforest, like those found in Sumatra, is not enormous (de Foresta and Michon, 1993; Gockowski and Dury, 1999).

In their study of deforestation and farming in the Philippines, Fujisaka and Wollenberg (1991) indicated how agroforests emerged as a sustainable agroecosystem after a period of logging and slash and burn agriculture (Table 6). This provides hope for the spread of sustainable alternatives to slash and burn agriculture that could perhaps be transferred to other areas of the humid tropics. The CGIAR's 'Alternatives to Slash and Burn' Programme, led by ICRAF, is aimed at this objective through the development of 'best-bet' agroforestry alternatives that are adapted to meet the needs of local people in different continents, as well as the global community (van Noordwijk et al., 1996; Tomich et al., 1998). Similar objectives are being pursued by REBRAF in Brazil.

The development of basic understanding of the biophysical and economic factors affecting the success of multistrata agroforests is in progress in Peru (ICRAF, 1997), and being planned in Cameroon (Leakey, 1998). Similarly, the canopy development in species mixtures is also being studied in mixed species plantations in Costa Rica (Menalled et al., 1998), where enriching forest fallows with timber species has been shown to have beneficial impacts on soils and the economics of subsistence farming (Montagnini and Mendelsohn, 1997). The inclusion of valuable timber species at low density in agroforests is certainly an option (de Foresta and Michon, 1990; Gouyon et al., 1993) that requires much greater consideration worldwide, as many of these timber trees are suitable species for agroforestry. Indeed agroforestry is seen as a form of integrated pest management to overcome the problems of mahogany shoot borer, Hypsipyla grandella(Matos et al., in press).

One of the concepts of the new definition of agroforestry (Leakey, 1996) is that over a period of time a number of agroforestry practices can be integrated within a given area. This would result in a patchwork of trees and crops, in which trees for the production of high value products (fruits, resins, timber, fodder, medicinal products, etc.) become an upperstorey. Within any particular climatic zone, the resultant climax agroecosystem would resemble a mature natural ecosystem. This concept differs from, but builds on, many of the approaches to agroforestry implemented over the last few decades (Cooper et al., 1996), as for example the linear upperstorey tree system being developed in Uganda (Peden et al., 1997). In this regard it is interesting to note that, in highland East Africa, formerly deforested areas of intensive farmland are now increasingly being planted with trees (Holmgren et al., 1994; Place, 1995; Place and Otsuka, 1997); being less subjected to soil erosion (Tiffen et al., 1994). Together these optimistic developments have led to the suggestion that 'the future of trees is on farm' (Sanchez et al., 1998). This on-farm planting of trees is thought to represent the fifth stage of forest status and follows after deforestation (Shepherd and Brown, 1998).

Table 6. Comparison of productivity, stability and sustainability of different land uses in the Philippines following deforestation and the transition to agroforest. (Source: Fujisaka and Wollenberg, 1991).

SystemsProductivityStabilityEquitabilitySustainability Primary tropical forest Disturbed (logged) forest Forest-crop-fallow mosaic Secondary forest and tree plantations

Commercial / illegal logging Settlers resource extraction Settlers home gardens Annual cereal cropping Annual tomato cropping Perennial crop agroforestry

In recent years, a programme has been initiated to domesticate the indigenous trees producing timber and non-timber forest products for growth in agroforestry practices (Leakey and Jaenicke, 1995; Leakey and Simons, 1998). This is based on farmer-derived species priorities (Franzel et al., 1996). As with the case of rubber already mentioned, the development and inclusion of improved cultivars of the dominant trees in these complex agroforests is one way of increasing the economic returns for the farmers. In this approach, the genetic improvement and domestication of a tree is not associated with a loss of species diversity, as it is deliberately not targeted at the development of a monoculture. There could be a risk attached to this; if the market opportunities rise high enough to attract big business it is likely that they would revert to a policy of plantation monoculture (Leakey and Izac, 1996). It is interesting, however, that a few companies (e.g. Daimler-Benz) are breaking away from this mould (Panik, 1999) and promoting small-scale production of natural products by local communities. This departure from monocultural practices follows the early history of rubber in Indonesia, which started as an estate crop introduced by large-scale planters, but was later replaced by smallholder jungle rubber agroforests. Thus domestication need not destroy the beneficial relationships between environment and small-scale production and the intensification of smallholder farming can still be associated with biodiversity and sustainable ecological functioning.

The example represented by the Indonesian agroforests runs contrary to the view of Crook and Clapp (1998) that market-oriented forest conservation is unlikely to be a successful strategy. The Indonesian agroforests are clearly an example of farmers developing a successful strategy towards forest conservation that is based on market forces. As such, it represents a good example of farmers using their own intuition and skills about sustainable management. It is interesting that this strategy is in tune with a 'Woody Plant Revolution' (Leakey and Newton, 1994), and that it perhaps points the way by which international development agencies should seek to implement the 'Doubly Green Revolution' (Conway et al., 1998).

CONCLUSION

Complex multistrata agroforests, as epitomised by the damar and rubber agroforests of Indonesia, represent a sustainable, farmer-derived landuse that is a response to dwindling natural forest resources and declining income from extractive resources. Though not entirely new in conception, these agroforests are an emerging and constantly evolving landuse system which, in its modern form, is both profitable for smallholder farmers and biodiverse. It is achieved by enriching and extending the normal fallow of swidden agriculture. This enriched fallow can either become a permanent perennial crop system, or have a lengthened duration, in which commercially important tree species provide cash income, as well as other economic, social and environmental benefits. It occurs were traditional land tenure rights are secure enough to allow for investment in long-lived trees, and has the potential to be expanded to other areas by provision of appropriate tenurial rights (de Foresta and Michon 1993).

The sequence of events (forest exploitation, extraction, settlement, annual crop farming and perennial cropping) presented by Fujisaka and Wollenberg (1991) represented the passage from a sustainable natural forest through less sustainable landuse practices to a situation culminating in a sustainable agroecosystem based on agroforestry. This is in accord with the new definition of agroforestry that sees agroforestry practices as a means to achieving a climax agroecosystem (Leakey, 1996). The agroforests of Indonesia are an excellent example of how this can lead to socially, economically and environmentally desirable resource management system and provide indications of the socio-economic and physical conditions required for the development of sustainable resource use in the forest zone of the humid tropics. They therefore provide insights that may be crucially important for the development of similar landuse systems in other parts of the tropics in Africa and Latin America.