Previous PageTable Of ContentsNext Page

ICRAF's strategy for domestication of non-wood tree products

A.J. Simons
International Centre for Research in Agroforestry
P.O. Box 30677, Nairobi, Kenya


Tropical forests are full of tree species that people use every day of their lives for their own needs (food, fodder, medicines, building materials, resins, dyes, flavourings); the so-called non-wood tree products (NWTPs). These species, however, have generally been less studied than their industrial counterparts (Eucalyptus, Pinus spp.). Leakey and Newton (1994) referred collectively to these neglected taxa as the `Cinderella' species since their potential is yet to be unveiled. Yet, if carefully promoted and marketed they raise hopes of alleviating rural poverty and increasing the sustainability of agroecosystems.

There is a growing recognition of the interdependence between forests and agriculture and the role of trees on farms in meeting the tree needs of rural communities and stabilizing land productivity. As such there is no discrete interface between agricultural fields and forests but rather a blurred edge at which poverty commonly abounds. Poverty, in simplest terms, restricts choice and options for development leading to short-term perspectives in the management of natural resources. NWTPs can greatly modify the balance between forests and agriculture, but will only improve natural resource management if they result in reasonable rewards to the people who derive their livelihoods from the forest and surrounding areas (Bennett 1992).

Non-wood tree products

This paper uses the term NWTPs to refer to products arising from individual tree species and excludes the products of the wider forest that are not tree derived (e.g.,, bush meat, mushrooms, etc.). Non-wood is used in preference to non-timber since it excludes not only timber but also fuelwood and poles. In essence, the term NWTPs is restricted to products that are tree derived but for which it is not necessary to fell the tree.

Non-wood tree products have satisfied human needs since before the first tree was felled and long before the Greeks coined the word `economy'. They are currently receiving focused attention from ethnobotanists, conservationists, policy-makers and breeders, as they embody the goals of conservation, development and production activities. With forested areas declining to accommodate agricultural expansion, and with human populations rising, the logical endpoint must be a choice between more people moving into the forest or moving more trees out of it. Many of the tree species that would move out of the forest onto farms would be those that provide NWTPs.

Multipurpose trees, or MPTs, have been variously described and mean different things to different people. One definition (ICRAF 1992) is `Multipurpose trees and shrubs are those that can produce food, fodder, fuelwood, mulch, fruit, timber and other products' . This definition seems to fit well the progenitor species of modern agroforestry, Leucaena leucocephala. The problem is that it also fits many industrial tree species such as those in the genera Pinus and Eucalyptus. In fact, which tree is not multipurpose? Should we then rather talk about trees that primarily provide timber, fruit, gums, fodder or medicines, and collectively refer to them as agroforestry trees.

The needs for research on new food crops have been cited by Arkcoll and Clement (1989) to be danger of reliance on a small number of species, agricultural expansion into marginal lands unsuited to existing crops; avoidance of food imports, emergence of new nutritional recommendations, emergence of new agricultural systems, and the desire for novelty. Interestingly, NWTPs could fulfil all these criteria.


Much research and development expenditure on agroforestry is founded, rightly or wrongly, upon the premise that growing trees on farms will simultaneously restore the environment, conserve tropical forests, generate income for resource-poor farmers and catalyse more sustainable land-use practices. To many, agroforestry is about stand-alone technologies such as alley cropping or improved fallows, although Leakey (1996) provides greater insight with a revised definition of agroforestry. This definition conveys the idea that the various agroforestry practices that have been devised play different roles in the ecological succession towards `climax agroforests' at a landscape scale. The cultivation of trees that provide NWTPs is pivotal in many of these practices. Unlike their crop counterparts, however, they remain in a wild or semi-domesticated form, and are certainly not optimally fit for their roles in the evolving agroforestry practices. Therefore, the selection and management (domestication) of these wild trees providing NWTPs is a key thrust of the International Centre for Research in Agroforestry (ICRAF).

Agroforestry is certainly not a new answer to the problems of the inadequate natural resource base for sustained development. Dubois (1995) in studying forest dwellers in Brazil found three historical levels of intervention in the forest, namely: (1) simple harvesting without intentional management; (2) enrichment of long duration tree-fallow; and (3) systematic manipulation of forest stands or enrichment planting in temporary agricultural fields leading to agroforests. What is new is the concept of applying modern techniques of domestication to the trees of agroforestry systems.


Domestication has recently been accepted as a phrase applicable to agroforestry as evidenced in a number of articles and conference proceedings (e.g., Leakey and Newton 1994). Indeed, as a corollary to the Green Revolution of crop plants, the Woody Plant Revolution is now being heralded (Newton et al. 1994). To domesticate is to naturalize to human conditions and involves human-induced change in the genetics of a plant (Harlan 1975, cited in Leakey and Newton 1994). Put simply, it means to bring into the use of man. It was a topic that so fascinated Charles Darwin that in his great work, The Origin of Species, he devoted the first chapter to `Variation under domestication'.

In genetic terms, domestication is accelerated and human-induced evolution. The consequences are either loss or gain of genes, altered gene frequencies or modifications to the way genes are packaged (gene complexes). There is much historical evidence of the human role in shaping present-day vegetation patterns (Briggs and Walters 1984). Even the floral composition of the pristine rainforests of Amazonia has been shown to be affected by the activities of Amerindian cultures (Mabberley 1983). The domestication of trees, then, does not have to occur outside the forest. Other examples of humans changing the species-richness and abundance of desired trees in forests are reported in Asia (Whitmore 1992) and Africa (Shepherd 1992). Domestication of tree species must therefore be seen as a continuum-from unmolested state, to management of trees in forests, to cultivation of semi-domesticates, to monocultural plantation of advanced generation breeding lines. The level of domestication activity will be dictated by biological, policy, market and social factors.

Domestication of any plant species is concerned with selection and management by humans and is not only about breeding per se. Selection can be deliberate or inadvertent. Deliberate choices might be made for fruit characters, growth rate, form, etc., whereas inadvertent selection could be for insect resistance, tolerance of selfing, etc. Of course inadvertent selection could also be carried out for any trait that was genetically linked or correlated to a trait that was being deliberately selected. Traits that confer fitness may either be enhanced or depressed, whichever the form of selection practiced. Management is also linked to the genetics of a plant in that the ability of a plant to be managed in a certain way is often expressed genetically. The direction and speed with which domesticated trees diverge from their wild progenitors will depend upon the size of the population, the heritability of traits under selection, the mating system, the intensity of selection and the inherent variability of the traits.

Potential gain from domestication

Trees are predominantly outcrossing. This results in progeny that segregate with respect to parental traits. This wide genetic variation not only provides buffering against differing environments and management practices but also affords the opportunity for selection. Most quantitative traits (height, biomass, etc.) are under low to moderate additive genetic control, which means that only 10-30% of the variation we see for a trait in a population can be captured in individuals following selection. Using a fairly high intensity of selection (say 1 in 50) we could expect the progeny to outyield the group of parents from which it was derived by approximately 15-25%.

Qualitative traits such as fruit shape, taste, tree form, date of first fruiting are usually much more tightly inherited, from selected individuals. The genetic gain we would expect in one cycle of selection for a qualitative trait would be much higher, such that 60% of the progeny might be similar to their parents.

Cloning is the ultimate means of capturing useful genetic variation in that there is no recombination or segregation of genes. There is a need, however, to screen large numbers of trees to identify clones with superior trait(s). Vegetative propagation, however, is not genetic improvement in itself, and regular testing and introduction of new material is advocated strongly (e.g., Leakey 1991).

Candidate trees for domestication

Notwithstanding further discoveries and taxonomic debates, we can reasonably assume there are 50,000 extant tree species. More than 2500 trees have been described for use in agroforestry systems (most of which provide NWTPs), which is both a fair proportion and a large absolute number of tree species. And yet, ethnobotanical studies continue to reveal a vast array of new tree products of importance to people (food, medicines, fibre, etc. of how best to allocate scarce research resources to such a plethora of species. It is an extremely expensive exercise to carry out the rangewide exploration and collection required for genecological studies, and only after a species has proven itself is this justified. The paradox is that a species may not be able to prove itself until its full intraspecific variation has been tested.

Which species to domesticate will depend on the objectives of domestication and will differ if it is for income generation, germplasm conservation, forest conservation or farm diversification. No studies are available that combine the economic, social and environmental benefits and costs to compare the merits of domesticating various species in the forest or on the farm. Until this lack of knowledge is addressed, species priority-setting will continue to be influenced most by the more readily available information relating to biological and social considerations.


It is clear that extractivists can earn a relatively good living from the forest as evidenced by the study of Allegretti (1994) in Amazonia where annual family income exceeded US$900 from the collection of about 750 kg of rubber and 4500 kg of Brazil nuts. This level of income, however, could not be earned by an infinite number of people in that the amount of Brazil nuts available is finite. Already 40% of the agricultural labour force in the State of Acre is involved in vegetal extractivism (Allegretti 1994) and overharvesting would affect income levels and stability. Overharvesting would also have a negative impact on natural regeneration and the fauna that depend on the Brazil nuts as food.

Clusener-Godt and Sachs (1994) question the extent that extractive reserves are consistent with the five dimensions of sustainable development (social, economic, ecological, geographic and cultural). They present a balanced view of the limitations and possibilities of extractivism without accepting it as the paradigm for agricultural development.

Recognition of the overreliance on extractive reserves led the National Council of Rubber Tappers of Brazil in 1991 to search for agroforestry systems that would diversify production (Allegretti 1994). They concluded that agroforestry systems should use native species and should be preferentially implemented in degraded areas. Not all share this view, with Homma (1994) articulating well the concerns of the agroforestry panacea. He cautions against unbridled enthusiasm for agroforestry, suggesting that it may lead to saturation of markets since a smaller land base is needed. Further, a class structure based on competency would develop since an intensive and well-managed production system is required thus creating social inequities.

At low densities of human population, the impact of extractivism on the biodiversity of natural forests can be minimal. With increasing extractive pressure, however induced, not only is the diversity of the tree providing the NWTP under genetic erosion, but also the forest itself when we examine recommendations by authors such as Viana and Mello (1995), who call for liberation thinning of Brazil nut trees to enhance productivity.

Homma (1994) prophesies that the extractivist economy is doomed to gradual disappearance. He cites four phases of the evolution of the extraction of plant resources. Expansion is the first phase. The evolution then reaches a stabilization phase in which supply and demand are balanced, which is close to the maximum extractive capacity. This is followed by a decline phase, caused by the reduction of resources and increased costs of extraction. This in turn is followed by the cultivation phase, which had its beginnings in the stabilization phase when technologies and planting materials became available. The model of Homma (1994) is useful to examine where agroforestry can have a meaningful intervention. For instance, conflicts have arisen over the right to collect or market NWTPs whether on state or communal land (e.g., between long-term residents and recent migrants) although with tree cultivation (and tree tenure) this can be proactively minimized even in the absence of land tenure.

Marketing and policy studies

Both agroforestry and NWTPs are multisectoral in nature, and invoke different amounts of government intervention in particular countries. Dewees and Scherr (1996) provide a comprehensive review of the extent and deficiencies of policy and market studies of NWTPs. To increase returns from NWTPs, some authors insist markets must pre-exist, whereas others acknowledge scope for evolving markets. Raintree and Francisco (1994) argue that new markets can easily be promoted through increased supply of existing NWTPs to meet increased demand or through developing demand for new species. The sustainability of NWTP markets (new or existing) preoccupies several studies in the available literature. Of greater importance, according to Dewees and Scherr (1996), is that market information systems are adequate to enable domestication efforts to respond quickly to change.

Another common preoccupation in the literature is the transitional nature of production from small-scale extractivism to monocultural large-scale plantations (Dewees and Scherr 1996, Leakey and Izac this volume). It is clear that the benefits for forest-dwellers and for the agrarian populace may not always be congruent. For example, harvesting of Brazil nuts (Bertholletia excelsa) from natural forest is reportedly threatened from establishment of plantations (Mori 1992). Cross-national perspectives are also evident as typified by the loss of revenue to Latin American countries with the establishment of rubber plantations in Southeast Asia (Browder 1992).

Global benefits (social and economic) of whether NWTPs should be produced on small or large scales are harder to discern when one considers a crop like bananas. Notwithstanding the inequity of three corporations controlling 67% of the world market, there are 300,000 workers in Ecuador who depend on the banana trade and 21% of the labour force in Panama are also dependent on it (ITM 1992). And yet, home consumption and village markets must dwarf the international markets. This illustrates the compatibility of concurrently having multinationally owned monocultures and small-holder agroforests.

Market conservatism will be as important for NWTPs as it is for timber products. It is perplexing that the Amazonian rain forest might contain up to 250 tree species per hectare with 50% having a commercial potential, but only 3-4 species are logged. There are timbers that look like, feel like and work like Swietenia, but because they are not labeled `mahogany' they are very difficult to market. Similarly, many NWTPs may be promoted as substitutes for other products but may be underutilized because of market intransigence, let alone the fact that Arkcoll and Clement (1989) note that many indigenous fruits are remarkably unpleasant or uninteresting to people not accustomed to them.

ICRAF's approach

ICRAF is working in six major tropical ecoregions to help to mitigate deforestation, land depletion and rural poverty through the development of improved agroforestry systems. To achieve these goals ICRAF's research agenda is engaged in a thrust on the domestication of agroforestry trees.

Almost without exception the tree germplasm that is used or envisaged for immediate use in agroforestry is wild and unimproved. `To domesticate or not to domesticate?'. That is the question faced for each of the 2500 species that ICRAF could potentially work with. At a given time some species will warrant domestication while others will not. Consequently, the emphasis of ICRAF's approach to domestication is on developing objective decision-making frameworks that will first ascertain whether domestication should proceed, and then determine in which way and at what level.

In the domestication of agroforestry trees, ICRAF is dealing with an imperfect knowledge base. For instance, in deciding which species have a commercial future problems arise because of lack of information on yield data, absence of selection criteria, inadequate botanic knowledge and consumer acceptance. In addition, ICRAF and its partners will never be able to advance domestication of all agroforestry species. Accordingly, ICRAF uses a number of species in the six ecoregions as case studies to build and optimize the decision-making framework. Many trees that provide NWTPs are represented in each of these ecoregions (e.g., Adansonia digitata in semi-arid lowlands of West Africa, Irvingia spp. in humid lowlands of West Africa, Prunus africana in eastern and central Africa, Uapaca kirkiana in southern Africa, Bactris gasipaes in Latin America). This, together with the strategies and techniques for capturing and using genetic variation (Leakey and Simons, in press) can be used to model approaches to domestication.

Objectives of domestication

The domesticates of tree species that provide NWTPs have the primary purpose of enhancing the productivity and sustainability of agroforestry systems. In turn, it is envisaged that the domesticates may contribute to the conservation of genetic resources of such species and also the preservation of natural habitats (e.g., rainforests, woodlands, savannah, etc.) in which their progenitors reside. It is further desired that the provision of improved germplasm will accelerate adoption and expansion of agroforestry technologies.

There are many assumptions surrounding this approach, not least that domestication is a beneficial exercise. FAO (1995) lists six advantages and four disadvantages of the domestication of NWTPs that usefully summarize the situation. Advantages include reliable production, relieving pressure on forests, income generation, ease of harvesting, improved growth rates, increased value of the crop. The disadvantages focus on increased susceptibility to pests, loss of ecological function, reliance on new sources of wild seed, added-value benefit to large corporate entities.

ICRAF has a very diverse client group. The resource-poor farmers of the tropics plant trees that provide NWTPs for a variety of reasons (food, income generation, risk aversion through diversification). Fitness of purpose of agroforestry trees is the prime objective of domestication, and this is best ensured through provision of a choice of priority species to farmers.

Priority species

Historically, research priorities among tree species were determined by researchers with varying degrees of research self-interest. Such subjectivity has led to suboptimal use of resources, although this does not mean that no successes have occurred. Rigorous priority setting, however, requires understanding of user needs and preferences, technological opportunities and systematic methods for ranking species. ICRAF, in collaboration with national agricultural research systems (NARs), has developed more objective procedures for priority setting of agroforestry tree species for genetic improvement research (Jaenicke et al. 1995).

The recently published ICRAF/ISNAR (International Service for National Agricultural Research) guidelines on species priority setting represent a great advance in procedural methodology (Franzel et al. 1996). The process described involves seven stages at which the number of species under consideration is consecutively reduced. The steps involved are (1) team building and planning, (2) assessment of client needs, (3) assessment of species used by clients, (4) ranking of products, (5) identification of priority species, (6) valuation and ranking of priority species, and (7) final choice.

These methodologies have now been used in a number of ecoregions, including the humid lowlands of West Africa (HULWA), the semi-arid lowlands of West Africa (SALWA), the lowland forest of the Peruvian Amazon and the Yucatan Peninsula in Mexico. In these regions, and in the miombo ecozone of southern Africa, the lists of the top five species have all contained mostly food-producing trees that provide NWTPs either as the prime purpose or as a secondary purpose.

Genetic Resource Unit

Tree germplasm is unequivocally the primordial input to agroforestry. Great progress has been made in commercial forestry through assembly and testing of rangewide collections of germplasm of timber trees (Barnes and Simons 1994). In contrast, work on agroforestry trees for NWTPs has lagged noticeably behind. Nevertheless, even where comprehensive collections exist of some species there is general under-appreciation of intraspecific variation by agroforesters and farmers alike. For trees providing NWTPs and for agroforestry trees in general, what is planted is usually what is most readily available and does not necessarily relate to germplasm quality (physical or genetic). ICRAF recognizes its role in raising awareness of the importance of quality germplasm.

ICRAF received endorsement to establish its own Genetic Resource Unit (GRU) from the wider germplasm community at an International Consultation held in 1992 (ICRAF 1993). At this meeting, a number of recommendations were taken up by ICRAF, which included the role of ICRAF in establishing field genebanks of material that is difficult to store(recalcitrant), in the need to assist in providing information on quality germplasm sources, and in encouraging ex situ and in situ conservation. ICRAF is respectful of the Convention of Biological Diversity and the need to share benefits and farmers' rights. All collections it carries out are with national partners and follow the voluntary FAO Code of Conduct for Germplasm Collecting, which is in line with ICRAF's policy on intellectual property rights.

In collecting germplasm of agroforestry tree species that provide NWTPs, ICRAF uses different approaches depending upon extant distribution, degree of human disturbance on natural genepools, threat of genetic erosion, reproductive biology (hermaphroditism, monoecy, dioecy) and floral synchrony (Ne/N). One aspect common to all species, however, is whether collections should be carried out using either (1) a random approach, (2) a targeted approach, or (3) a combination of the two methods. In essence, whether phenotypic selection is efficient or not.

Protagonists of the random-collection strategy argue that to sample as much of the variation as possible it is necessary not to narrow down the mother trees from which to collect on the basis of a few traits (e.g., fruit size, fruit quality). On the other hand, advocates of the targeted approach ask why seeds should be collected from trees with bitter fruit since these would never be given to farmers. Genetic theory is used to defend both approaches. Randomists argue that other genes with more latent expression (e.g., disease resistance) might be negatively or positively genetically correlated to fruit traits and therefore targeted collections may inadvertently omit to sample useful variation. Further, randomists question the ability to select accurately trees growing in different microsites, of different age and putatively different previous management (e.g., lopping of branches). Targetists reply that if phenotypic selection is not very efficient then it should not result in genetic variation carrying it out. Although their prime assumption is that since fruit traits are highly heritable then they should be able to capture the desirable phenotypes in progeny of their targeted selections.

ICRAF is using several tree species (e.g., Inga edulis, Sclerocarya birrea, Uapaca kirkiana) that provide NWTPs to provide hard evidence to determine which approach is more appropriate. Since most fruit tree species that ICRAF is working with have recalcitrant seeds, live-genebanks are being established which have a conservation as well as a utilization purpose. These are typically set up as progeny trials to allow determination of genetic parameters (heritabilities and genetic correlations). In this way, the reduction in additive genetic variance (in juvenile traits-e.g., height, form) from targeted selections can be calculated. In parallel with this study, molecular characterization (Dawson et al. 1995) is being carried out to identify the genetic bottleneck from targeted selections. As species come into fruit it is possible to compare parental fruit phenotypes with those of their progenies.

Quantifying and forecasting demand

To justify investment in large-scale seed or clonal production or in a formal breeding programme there needs to exist a strong demand for the species. Yet without improved material of new species little demand may develop. Most of the trees that provide NWTPs are found in natural forest or communal lands and are largely uncultivated. ICRAF distinguishes two groups of clients, namely: (1) the rarer group with a tradition of tree planting or nurturing natural regenerants; and (2) the common group with no tree cultivation experience.

Those in the first group are easier to target with improved materials since the leap of faith required by them is only that the new germplasm is superior to their wild or semi-domesticated germplasm. But superior in what way? ICRAF recognizes that it is crucially important for farmers to be involved in the identification and selection of traits for improvement. A key question that remains is whether farmers would buy improved material when they have the option of collecting their own seed or cuttings from fields or the forest. Street et al. (1992) carried out an economic analysis of growing agroforestry trees in Haiti and concluded that when farmers had to pay for seedlings the net present values were negative, irrespective of the economic product.

To stimulate demand for improved replacements (free or otherwise) the new germplasm would have to be markedly and demonstrably superior. Assuming farmers were interested in, for instance, greater fruit production; how closely can they discern differences between germplasm and thus express demand for particular sources. There would likely need to be an actual difference in quantity of greater than 30% for farmers to appreciate the difference. Length of fruiting season, for example, it would likely have to be two weeks longer or 25% of the fruiting period, whichever is the lower. It is important, therefore, to weigh up the biological threshold for improvement (genetic gain) versus the farmers' threshold for improvement. Using clones will usually push the biological threshold much higher than using sexual propagules, and this may mean the difference between farmer acceptance or not. It is important to clarify, however, that just because the biological threshold is above the farmer's threshold does not guarantee that there will be adoption.

The second, and larger group of clients are the farmers for whom the tree planting philosophy has not taken root. Yet, if farmers are to be convinced to adopt tree cultivation it will most generally be first with trees providing income or food security. In this regard, NWTP agroforestry trees are the ideal candidates. ICRAF considers that if better germplasm were available then adoption would proceed at a faster rate. This explains the great importance ICRAF attaches to assembling rangewide collections of germplasm (both targeted and random).

For either of these two groups of farmers, domestication can be (1) done for them, (2) done in conjunction with them, or (3) done by them alone. ICRAF conceives that the second method is generally the most appropriate for trees providing NWTPs. Participatory breeding by farmers is encouraged and in such cases ICRAF is careful to provide them and national research institutions with sufficient diversity from which to make selections.

Demand for germplasm of a particular species in agroforestry development projects has often been stimulated by the provision of incentives (cash, livestock, food, fertilizer, etc.). In these instances, it is difficult to dissociate the real demand from that created by the incentives. To fill a void in policy aspects of tree germplasm, ICRAF is to hold a workshop on this subject in November 1996.

Understanding that germplasm availability (improved or not) is often cited as a constraint to increased tree planting, ICRAF is using a number of case studies to examine how much germplasm will be needed, by whom, when, and for how long. It is crucial that this demand is real and not just perceived. For instance, demand might be overestimated if NGOs or other organizations ask for large quantities of planting materials in advance of well-meaning extension projects that subsequently underachieve in relation to their planting targets.

Germplasm delivery pathways

If improved germplasm is to be provided to farmers, what are the delivery pathways currently available or in need of creation to ensure availability of improved germplasm? Very little information is available for agroforestry trees on the supply-side parameters. Indeed, even the seed regulatory frameworks (variety development, variety release, seed certification, distribution and sale) for many agricultural crops in developing countries are poorly formed and are considered unresponsive to the needs of resource-poor farmers (Tripp 1995). We would be in a far better position to have impact with domestication efforts if we better understood the pathways of germplasm delivery, namely:

The link between the demand and the supply of germplasm of a species determines the efficiency with which improved material is used. The demand side of germplasm delivery is being assessed by ICRAF at three levels, namely that of (1) the farmer, (2) the disseminator (e.g., NGOs, community programmes), and (3) the distributor (NARS, international projects, etc.). Demand from farmers is being determined using farmer surveys. These examine aspects that are both qualitative in terms of the useful traits they desire and quantitative in terms of the amount of germplasm they require. The demand from disseminators is being assessed through questionnaires and direct contact with various groups. The distributors are similarly polled using both questionnaires and direct contact.

The supply side of germplasm is being examined by compiling a list of suppliers. This is being put into a database that links species information with individual suppliers. Easy-to-fill-out questionnaires have been sent out to several hundred suppliers. Surveys of large- and small-scale tree nurseries are also being carried out in several countries. Further, the possibilities of private sector involvement and contract multiplication and delivery are being investigated by ICRAF and its partners.

Tree improvement

Breeding of most commercial forest tree species has been in progress for less than 40 years (Barnes and Simons 1994). Using a strategy of recurrent selection for general combining ability and seed production from progeny tested clonal orchards, average gains of around 20% were attained in the first generation of selection (Barnes and Simons 1994). Is this a strategy that should be followed for agroforestry trees providing NWTPs. Irrespective of the strategy followed, a key question is `whose strategy is it'?-that of governments, of research projects, of international organizations? The most important question for genetic improvement is whether farmers are setting the breeding objectives. It is fair to say that most literature on improvement of non-industrial trees, equates superiority with greatest growth, yet it is recognized that farmers may want more stable production than enhanced but variable production (Simons 1992). In essence, is biggest (and probably most competitive in agroforestry systems) the best? Kanowski (1993) concluded that there has been a `northern temperate bias' to tropical tree improvement and there exists a need to re-evaluate methods.

In the absence of comprehensive survey information of what farmers want from improved germplasm it seems prudent to pass on as much genetic variation to them as possible. In some situations, ICRAF is also promoting efforts to broaden the genetic base of genetically depauperate landraces (e.g., Alnus, Grevillea). Wei and Lundgren (1991) call for development of breeding methods that combine the goals of gain and diversity. Gain is associated with economic return while diversity is a more nebulous concept. The loss of diversity is dependent on the selection method used and the selection intensity (Wei & Lundgren 1991). To date, the improvement of agroforestry trees has largely followed the methodology used for commercial tree breeding, which in turn borrowed exactly the methods used for agricultural crops (Barnes and Simons 1994). Certainly, agroforestry trees and agricultural crops share many biological attributes, and classical genetic theory holds true across both groups. However, a number of attributes relevant to improvement strategies distinguish them. Most notable are their size and their perenniality.

There is a need to carry out many of the sequential steps of tree improvement in parallel to reduce the time to delivery of improved materials. Investigations into the basic biology (e.g., reproductive biology) of species are underway (see Ladipo et al. this volume, Leakey and Jaenicke 1995), which are especially important if it is the reproductive produce (fruit, seeds) that are the NWTPs. Germplasm multiplication is being carried out in advance of identification of the most superior material. This approach, therefore, means that while some multiplied materials may never be used, nucleus amounts of superior germplasm are sure to be available when demand is incipient.

Agroforestry trees are biologically a very heterogeneous group, as are their users, the resource-poor farmers of the developing tropics. A single strategy for improvement will not suffice for all agroforestry species, given these prerequisites and an awareness that improvement activities would differ for trees if they were (1) exotic or indigenous, (2) a new or an existing introduction, (3) of low value or high value, (4) for home use or for sale, (5) subject to recurrent improvement or not, (6) sexually or vegetatively propagated. What is clearly needed is a decision-making framework, first to determine whether improvement is worthwhile and second to identify a specific strategy for improvement. This will lead to strategies that are not only species specific but also ones that are location or market specific.

Decision-making framework

Agroforestry offers a flexible environment in which to domesticate trees providing NWTPs at an intensity and pace to suit the local populace. Decisions on improvement of agroforestry trees are being made throughout the tropics, although these may be flawed without consideration of the full spectrum of determinants. Various categories of such determinants are given in figure 1. The challenge ICRAF faces is to integrate these parameters and factors into a usable decision-making model. As an example, Maghembe et al. (1994) usefully describe the relevant research needed for fruit trees indigenous to the miombo woodlands of southern Africa. It is by using the case study species from the different ecoregions that progress will be made in the generation of a usable framework for domestication.

Genetic conservation

Conservation of tropical trees and forests is a subject that has received considerable attention post-UNCED, yet the term often remains undefined. From a species-centric perspective, one can seek to (1) preserve the habitats where the species exists, (2) prevent the species from going extinct, (3) preserve all possible genes or genotypes of that species, or (4) preserve all genetic variation of a high utility value. It is axiomatic that these four aspects are considered simply as conservation, and confusion can arise as to specific objectives of conservation efforts.

w3735e7a.gif (13845 bytes)

Figure 1. Determinants of strategies for genetic improvement of agroforestry trees

Without doubt, tree species are best conserved in situ in pristine forests. Recognition of the difficulties of such an approach for species of important human use has led to calls for conservation through utilization. A laudable, yet subjective, concept. From experience with agricultural crops it can be seen that greater utilization of a species can lead to a reduction in the amount of genetic variation used and, hence, conserved. Indeed, the use of improved material is based on selections that alter gene frequencies for favourable traits.

The challenge of linking the conservation of high-value trees providing NWTPs to their use lies in expanding the cultivation of the species. This can be done most easily by having germplasm available that is perceived by farmers to be superior to alternative species or to material available in the wild or semi-wild state (i.e., of high utility value).

Farmers at the forest edge can reduce pressure on natural stocks and thus conserve genetic resources by cultivation of high-value trees. In this way, the raw germplasm can be protected in the natural ecosystem while the germplasm of greatest utility value can be conserved and promoted through the cultivation and domestication of trees on farms. This germplasm is the priority material for conservation by ICRAF. Dubbed circa situ conservation (near the site of origin), this approach runs the risk of contaminating the natural genepools with new genes or changing the frequencies of existing genes. However, this accelerated evolution is far preferable to gene loss or species extinction, especially if through agroforestry it is simultanously benefiting farmer welfare and providing other valuable environmental services.


The development of new tree crops on a sufficient scale in sustainable agroforestry systems is a challenge to link genetic improvement research with marketing to ensure income generation and food security. Domestication is both about product development and product creation. ICRAF is aware, however, that greatest focus should be on existing markets while keeping watch for emerging ones. Nevertheless, wherever focus is placed, research on input markets (e.g., planting material ) is as important as research on output markets in understanding small-holder investment in and output of NWTPs.

The use of agroforestry may avoid many of the problems with monoculture plantations reducing the risk to farmers of crop loss and the need for agrochemical application. Agroforestry systems where NWTPs trees could play a role are numerous and include homegardens, scattered trees in fields, multistrata systems and boundary plantings. The undomesticated nature of agroforestry trees and the marked intraspecific diversity within them augurs well for genetic improvement through selection and breeding. The real challenge then is not just the mechanics of improvement but rather the reconciling of the potential for genetic advance with the practical realities of farmer needs, farmer perceptions and the imperfect framework for germplasm delivery.

Domestication is human-induced change, or more aptly farmer-induced change, in the genetics of a plant, and it will differ in magnitude and direction from farmer to farmer, species to species and location to location. The formulation of objective decision-making frameworks is essential to ensure optimal use of resources and provide opportunities for maximal adoption. In several cases it may be that we do not need enhanced domestication (i.e., semi-domesticates may suffice), while in many other situations accelerated domestication may offer distinct advantages. Farmers may plant trees but in many cases are reluctant to thin them or cull out poor performers, and in the absence of silvicultural improvement, genetic improvement becomes increasingly important. What is required is a cohort of barefoot tree breeders!

Conservation is an integral part of ICRAF's strategy although not solely for curatorial purposes. ICRAF seeks to promote conservation of genetic resources through use, and also through contributing to policy formation and regional and international collaboration and coordination. In essence, we are interested in genes for sustainable development. In this regard, we believe there is scope for coevolution of the cultivated and `wild' populations.

From surveys of both buyers and suppliers of agroforestry germplasm it is clear that we are dealing with an imperfect market. There is no premium paid for quality (physical or genetic) and there is little appreciation of intraspecific diversity. Germplasm, almost without exception, is marketed under the species name only. Greater awareness of differences among trees of a given species and attachment of names to such differences are needed.

IUFRO, in their summing up of the recent quinquennial conference noted that `IUFRO should look at further domestication of tree species. In addition, greater attention should be given to agroforestry systems'. ICRAF is taking up the challenge of domesticating trees that provide NWTPs in agroforestry systems, to ensure that much of the unwritten agenda of the 20th century is completed before we tackle UNCED's Agenda 21. It is envisioned that we may soon hear a modification to the old adage `you can't see the wood for the trees' which would be `you can't see the trees for the products'.


The domestication work at ICRAF is financed by numerous donors. Most significantly the governments of Britain, Canada, Denmark, Germany and Japan are providing timely support as domestication activities at ICRAF take on enhanced importance.


Allegretti M.H. 1994. Policies for the use of renewable natural resources: the Amazonian region and extractive activities. p 14-33. In: Extractivism in the Brazilian Amazon: perspectives on regional development, ed. M. Clusener-Godt & I. Sachs. MAB Digest 18. UNESCO, Paris.

Arkcoll D.B. & Clement C.R. 1989. Potential new food crops from the Amazon. p 150-165. In: New crops for food and industry, ed. G.E. Wickens, N. Haq & P. Day. Chapman and Hall, London.

Barnes R.D. & Simons A.J. 1994. Selection and breeding to conserve and utilize tropical tree germplasm. p 84-90. In: Tropical trees: the potential for domestication and the rebuilding of forest resources, ed. R.R.B. Leakey & A.C. Newton. HMSO, London.

Bennett A.J. 1992. Dynamics of forestry and agriculture at the forest edge. p 139-148. In: Wise management of tropical forests 1992, Proceedings of the Oxford Conference on Tropical Forests, ed. F.R. Miller and K.L. Adam. OFI, Oxford.

Briggs D. and Walters S.M. 1984. Plant variation and evolution. 2nd ed. Cambridge University Press, Cambridge. 412 p.

Browder J.O. 1992. The limits of extractivism: tropical forest strategies beyond extractive reserves. Bio Science 42:174-182.

Clusener-Godt M. and Sachs I. 1994. Extractivism in the Brazilian Amazon: perspectives on regional development, MAB Digest 18. UNESCO, Paris. 88 p.

Cromwell E. & Wiggins S. 1993. Sowing beyond the state. ODI, London. 94 p.

Dawson I.K., Simons A.J., Waugh R. & Powell W. 1995. Diversity and genetic differentiation among subpopulations of Gliricidia sepium revealed by PCR-based assays. Heredity 74:10-18.

Dewees P.A. & Scherr S.J. 1996. Policies and markets for non-timber tree products. Environment and Production Technology Division Discussion Paper No. 16. IFPRI, Washington. 76 p.

Dubois J.C.L. 1995. Silvicultural and agrosilvicultural practices in Amazonian Native and Traditional Communities. p 67. In: Invited Paper Abstracts, IUFRO XX World Congress, Tampere, Finland.

FAO. 1995. Non-wood forest products for rural income and sustainable forestry. FAO Technical Report No. 7. Non-wood Forest Products Division, FAO, Rome. 127 p.

Franzel S., Jaenicke H. & Janssen W. 1996. Choosing the right trees: setting priorities for multipurpose tree improvement. ISNAR Research Report No. 8. ISNAR, The Hague. 87 p.

Harlan J.R. 1975. Crops and man. American Society of Agronomy/Crop Science Society of America. Madison, Wisconsin.

Homma A. 1994. Plant extractivism in the Amazon: limitations and possibilities. p 34-57. In: Extractivism in the Brazilian Amazon: perspectives on regional development, eds. M. Clusener-Godt & I. Sachs. MAB Digest 18. UNESCO, Paris.

ICRAF. 1993. Annual report 1992. International Centre for Research in Agroforestry, Nairobi, Kenya.

ICRAF. 1993. Proceedings of the International Consultation on the Development of the ICRAF MPT-Germplasm Resource Centre. ICRAF, Nairobi, Kenya. 228 p.

ITM. 1992. Third world guide. Instituto del Tercer Mundo, Montevideo, Uruguay. 632 p.

Jaenicke H., Franzel S. & Boland D. 1995. Towards a method to set priorities amongst multipurpose trees for improvement activities: a case study from West Africa. Journal of Tropical Forest Science 7:490-506.

Kanoswki P.J. 1993. Mini-review: forest genetics and tree breeding. Plant Breeding Abstracts 63:717-726.

Leakey R.R.B. 1991. Towards a strategy for clonal forestry: some guidelines based on experience with tropical trees. p 27-42. In: Tree improvement and breeding, ed. J.E. Jackson. Royal Forestry Society of England, Wales and Northern Ireland, Tring, England.

Leakey R.R.B. 1996. Definition of agroforestry revisited. Agroforestry Today 8(1):5-7.

Leakey R.R.B. & Jaenicke H. 1995. The domestication of indigenous fruit trees: opportunities and challenges. p 15-26. In: Proceedings 4th International BIO-REFOR Workshop, Tampere, Finland, ed. K. Suzuki, S. Sakurai, K. Ishii & M. Norisada. BIO-REFOR, Tokyo.

Leakey R.R.B. & Newton A. 1994. Domestication of tropical trees for timber and non-timber forest products. MAB Digest No. 17. UNESCO, Paris. 94 p.

Leakey R.R.B. & Simons A.J. In press. When does vegetative propagation provide a viable alternative to propagation by seed in forestry and agroforestry in the tropics and sub-tropics? In: Problem of forestry in tropical and sub-tropical countries: the procurement of forestry seed-the example of Kenya.

Mabberley. 1983. Tropical rain forest ecology. Blackie and Son Ltd, London. 156 p.

Mori S.A. 1992. The Brazil nut industry-past, present and future. p 241-251. In: Sustainable harvest and marketing of rain forest products, ed. M. Plotkin & L. Famolare. Island Press, Washington.

Newton A., Moss R. & Leakey R.R.B. 1994. The hidden harvest of tropical forests: domestication of non-timber products. Ecodecision (July):48-52.

Raintree J.B. & Francisco H.A. 1994. Marketing of multipurpose tree products in Asia. In: Proceedings of an International Workshop, Baguio City, Philippines, December 1993. Winrock International, Bangkok.

Shepherd G. 1992. Managing Africa's tropical dry forests. Overseas Development Institute, London. 117 p.

Simons A.J. 1992. Genetic improvement of non-industrial trees. Agroforestry Systems 18:197-212.

Tripp R. 1995. Seed regulatory frameworks and resource-poor farmers: a literature review. Agricultural Administration Network. Network Paper 51. Overseas Development Institute, London. 53 p.

Viana V.M. & Mello R.A. 1995. Ecology and management of Brazil nut populations in extractive reserves in Xapuri, Acre, Brazil. Poster 61, p 37-38. Poster Abstracts, IUFRO XX World Congress, August 1995, Tampere, Finland.

Wei R-P. & Lindgren D. 1991. Selection effects on diversity and genetic gain. Silva Fennica 25:229-234.

Whitmore T.C. 1992. An introduction to tropical rainforests. Oxford University Press, Oxford. 226 p.

Previous PageTop Of PageNext Page