The goal of this case study was to survey and understand the historical and contemporary processes involved in the generation and adaptation of knowledge and technology of trees and tree cultivation by Rwandan farmers. In particular, this study focused on: (1) the dynamic aspects of Rwandan agroforestry systems, i.e. how and why these systems came into being and evolved over time, (2) the active role of farmers in the development of the agroforestry systems, and (3) agroforestry knowledge/technology production processes (i.e. processes of acquisition/adaptation of knowledge/technology from elsewhere, and generation of completely new knowledge/technologies based on farmers' own ideas and efforts).
The primary aim of this study was not to describe the agroforestry systems in the study areas. Agroforestry was used as the medium to understand the socially-constructed nature of the knowledge upon which these systems are based. The visual artefacts (i.e. the things we see in and on farmers' fields) are but the outcome, a reflection of farmer knowledge. Contrary to common assumptions, this knowledge is not static but changes each growing season with the implementation and maintenance of these artefacts. As we saw in the discussions in Chapter VI, putting ideas into practice in order to gain experience with and from them is an important aspect of "gaining knowledge" and of being recognized as "being knowledgeable" about something, be it crop cultivation, livestock rearing, blacksmithing, carpentry or healing practices. Within agroforestry, there are extensive interactions between agroforestry knowledge production and actual tree/crop/animal production activities (the agroforestry systems) on the farm. Often, the two activities are so thoroughly interwoven that it is hard to distinguish experimental practices from everyday practices. This is reinforced by the fact that most (new) knowledge generated through farmers' own efforts does not concern major changes in the systems or in species composition, but consists of serendipitous, seemingly-innocuous innovations involving a better or new use of existing resources. By consequence, to study knowledge production, one must also study the physical, observable artefacts constituting "normal" agroforestry systems and practices. The brief descriptions of particular agroforestry practices given in the main text of this case study provide a context to understand the nature and content of endogenous agroforestry knowledge and the processes involved in producing this knowledge.
This chapter provides a summary of the main findings and recommendations that can be formulated on the basis of these findings. The chapter is divided in four sections: (I )Methodological issues related to the identification of local tree experts; (2) Farmers' perspectives on agroforestry and tree planting practices; (3) Farmer experimentation: A comparison of methods and procedures with scientific research; and (4) Enhancing local communication networks: The key for advancing agroforestry knowledge production and farmer experimental practices.
At this point, there is a note of caution concerning the formulation of recommendations for various outside agents (FAO, CGIAR centres, NGO's, and national research and extension organizations) that aim to strengthen endogenous knowledge production and extension processes. It should be noted that farmer research and extension are, above all, activities of, by and for the farmers. For this reason, any efforts to enhance farmer R&E should be based on their suggestions and be implemented by and in consultation with the farmers. Studies of endogenous knowledge systems and farmer experimentation might not only preserve knowledge itself, but, more importantly, preserve the way in which it has been generated (and continues to be generated) for future generations. The importance of this is all too clear in view of the events of 1994 in Rwanda. But, when outside agents (i.e. people who are not part of an endogenous knowledge system and actively engaged in its reproduction) try to strengthen local knowledge production and diffusion processes, they must be careful not to co-opt and formalize farmer practices and methods, which could harm the system they want to help. The recommendations that are made in this chapter are (with the exception of those related to methodological issues) largely based on ideas and suggestions made by the farmers themselves. The researcher has developed logical extensions and inferences to translate the farmers' words into feasible and practical guidelines for rural development agencies.
Identification and characterization of tree experts: According to the philosophical definition provided by Kagame (1958), knowledge (ubumenyi) implies knowing a plurality of objects or notions. Farmers concurred with Kagame. They equated knowledge of agroforestry with experience in the cultivation and/or use of many different tree species. Thus, the tree experts, as identified through a ranking procedure and community interviews, were all farmers who cultivated many trees of a large species diversity on their land. During the ranking game, informants explained that the larger number of trees and species diversity was possible because experts had more land, more animals and manure, better soils, greater wealth and were more active in tree planting activities.
The differences between tree experts and comparison farmers (the non-experts in this study representing the general farm population) were subsequently confirmed by a socio-economic survey. The survey results showed that tree experts were older, more experience t and further advanced in the family life cycle. They had greater resources (land, labour, animals, manure) and therefore could better absorb risks associated with experimentation. Tree experts were also better informed through travel and more observant of =yew species and practices during travel. In addition, they were more aggressive in seeking help and advice from extension workers, neighbours and/or veterinary assistants, and in pursuing extension and projects to gain access to new technologies, such as improved seeds,
tree seedlings, fertilizer, lime, management practices of trees and crops. Although not all variables included in the t r ,nal survey were mentioned as criteria by informants for the knowledge ranking exercise, the results indicate that informants were able to accurately distinguish groups of farmers within the population on key variables such as resources, wealth and tree species diversity, even if only on a relative scale. The ranking game data and observations made during game implementation provided valuable contextual information about agroforestry practices which proved invaluable in interpreting the overall results of the study. This information could not have been obtained by relying exclusively on the formal socio-economic survey.
It would be wrong to say that comparison farmers did not have any knowledge about trees and tree cultivation, nor understand the need for and importance of trees. They had little land and resources which made their choice of species and species location more critical. The smaller the farm, the more farmers in this group concentrated their efforts on fewer, less-competitive species, resulting in complex systems of low diversity with a high degree of integration of trees and crops. The agroforestry knowledge of tree experts and comparison farmers may, therefore, indeed be different, confirming the initial conclusions of the knowledge ranking exercise. Low diversity/high tree density systems require farmers to have higher levels of management skills and greater knowledge of the various components and their interactions. As agroforestry research in particular concerns the study of the integration and interaction of trees, crops and/or animals, the agroforestry systems of the comparison farmers would make a more appropriate object of study to learn about endogenous agroforestry knowledge and the logic behind these systems.
Beans and bananas are commonly intercropped. Above 1800 metres bananas do not grow well and look scrawny, as in this photograph. Here, the land is being prepared before sowing of beans. The soil is often hoed twice to remove the stoloniferous roots of Eragrostis grass, a major weed. The roots are piled together to dry and are then burned. Photo: JM Olson
This study thus observed a difference in both agroforestry systems and agroforestry knowledge among individuals and groups of individuals. It confirmed observations by Scoones and Thompson (1994) that contrary to scientific knowledge, endogenous knowledge is manifold, discontinuous and dispersed, not singular, cohesive and systematized. As a result of "development" and other political, socio-economic and demographic processes of change, the definition of knowledge defined by a society once largely dependent upon gathering its tree products (i.e. having experience with a variety of tree species particularly concerning their utilitarian aspects) no longer appears applicable to today's situation in which people actively plant and manage trees on their own land. A more extensive kind of "knowledge of trees" is developing which includes knowledge about the planting and management of trees, their integration and interaction with crops and/or animals, as well as knowledge of their utility. It is therefore recommended that:
Future studies of endogenous knowledge of agroforestry combine qualitative and quantitative, participatory and formal data collection methods to provide both complementary and supplementary perspectives on a complex reality (den Biggelaar, 1995).
For this research, it was deemed important, in order to study the various processes involved, to find and interact with the farmers most actively engaged in agroforestry knowledge production. The basic assumptions of this research (that individual knowledge of agroforestry varies because of differences in socio-economic and political position within the community and because of varying biophysical conditions on the farm; and that not all farmers would be equally active in knowledge generation due to different interests, motivations, propensities and abilities towards experimentation) proved to be correct. However, as the nature and content of agroforestry knowledge varies between (groups of) individuals, it was wrong to assume that only some farmers (in this study, those perceived by their fellow farmers to be the most knowledgeable about trees and tree cultivation) would engage in experimentation for knowledge production. In reality, farmers in each of the various "knowledge groups" created by the ranking game informants were engaged in the generation of their own particular kind of knowledge of trees and tree cultivation, although not all farmers were equally active experimenters.
Relying solely on local perceptions of the persons most knowledgeable about trees and tree cultivation to identify experimenting tree farmers (as was done in this research) may, therefore, not provide satisfactory results. As was shown in this study, there are multiple ways of knowing about trees and tree cultivation, and each is based on and evolves from its own forms of knowledge production. It is therefore recorr mended that:
Future in-depth studies of farmers' experimental activities and knowledge production processes should not rely exclusively on the farmers identified by other community members as the most knowledgeable about agroforestry. A sample of farmers stratified into different "knowledge groups" created by ranking game informants may better cover the wide range of agroforestry knowledge and associated knowledge production processes present among (groups of) individuals.
Agroforestry: historical aspects: Discussion with individuals and groups of individuals revealed that Rwandan farmers traditionally did not practice agroforestry. Farmers satisfied the household needs for tree products by exploiting the natural vegetation on the hills and along valley bottoms. In addition, they practised "passive agroforestry", an agroforestry where naturally-seeded trees were purposefully maintained on the land. Active tree planting only involved planting of cuttings of Euphorbia tirucalli, Vernonia amygdalina and Ficus sp. as live fences around the compound. With increasing population, more land was needed for the cultivation of food crops, leading to the virtual disappearance of forests and forest galleries from the landscape, and making it less and less possible to depend on natural vegetation to secure the necessary tree products. Since people did recognize the need for tree products and the benefits trees provide, they became more active in planting trees on their own land. Over the years, many indigenous and exotic species have been incorporated into the agricultural systems, facilitated by the overthrow of the kingdom in 1959 and the abolishment of traditional chief-client relationships based on cattle ownership and the necessity of keeping land in pasture free of trees. In addition, research on suitable (exotic) trees and reforestation efforts, started by the Belgians in the 1930s, and by various development programmes since Independence, have helped provide suitable technologies to the farmers to encourage them in their own tree planting activities.
Agroforestry: present situation: This research found that in 1992, almost all farmers in the three study areas practised agroforestry. However, no two farms had a similar agroforestry system because farmers design individual systems that best meet their multiple needs given their resources. This individuality is reinforced by a large species diversity (as was seen in Chapter V, 193 species were found in the study zones), the multiple reasons for planting trees and the multiple uses farmers make of trees. These factors have led to agroforestry systems with very complex arrangements of species over space and/or time.
Tree inventories from previous studies of agroforestry in Rwanda diverge greatly from the results of the present research because they used a different definition of "tree."
Previous studies inventoried woody species (trees and shrubs) defined from a Western conception. Farmers in this research talked about trees using the definition given by Kagame (1958): "Trees are all plants that are not grasses". Thus, the plants found in the indigenous agroforestry systems that correspond to this term for "tree" included not only trees and shrubs (as defined in the Western world), but also annuals and perennials. This contributed to the species diversity found in the study areas and gave an additional layer of complexity to the indigenous systems. However, these "nontrees" (from a Western perspective) should not be ignored as they provide significant benefits to the farmers and often are grown in specific niches not usable for crops or trees. It is therefore recommended that:
Future studies of agroforestry follow the farmers' definition of trees and should not assume that these "non-trees" are weeds of no value..
Farmer-consultants were well aware of, and sensitive to, the biological interactions between trees, crops and soils. These interactions were the main criteria used in decisions about where to plant a certain species within the farm and/or the field, and to evaluate its performance. However, utility and other tangible benefits were cited as the primary reasons behind farmers' choices of certain species. As many species meet the diverse needs of farmers, which species actually are planted depends on gender, farm size, personal interests, motivation, preferences for certain species and the initiative of each person. Women were primarily interested in trees producing food (fruit, spices, seasonings, yeasts, etc.) and medicines, whereas men preferred lumber trees and those producing construction poles. Neither gender expressed much interest in fuelwood trees, and fuelwood did not appear high on the list of farmer priorities to be addressed by agroforestry research.
During many meetings with farmers, it was repeatedly stated that it was imperative to find species and arrangements with the least negative influence on crops and soils in order to further agroforestry practices. In the opinion of farmer-consultants, all trees are competitive and damage the soil, although they stated that all traditional (i.e. indigenous) trees can be used for agroforestry as they do not exhibit these characteristics. This seeming contradiction may be a reflection of their long observation of and experience with planting and managing a variety of indigenous tree species. The passive agroforestry practices of the past may, in fact, have been less passive than assumed or admitted, as farmers may have eliminated trees that did not provide direct benefits, were too competitive or had the wrong form and shape. The "all" in the above statement may indicate that useless and/or competitive indigenous trees have been eliminated over time, leaving only those that are compatible with crops and provide direct, tangible benefits. While farmers consulted in the study were not able to verbalize whether or not they had engaged in experimental practices related to this selection process in the past, it is obvious that these past experiences were and are being used in designing the agroforestry systems found in the study areas at present.
Farmer-consultants indicated that they value knowledge about the uses and benefits of trees, especially for medicinal purposes, more than knowledge of tree planting and cultivation. This is not surprising as people have been using trees for various purposes for centuries, but the intentional planting and management (outside the home compound) is a relatively recent phenomenon. As women were better in identifying trees and more skilled in treating everyday ailments using a wide variety of trees, they were considered more knowledgeable about trees and tree cultivation practices. As utility and tangible benefits were the primary reasons behind farmers' choices of certain species; and as women were deemed more knowledgeable about tree uses and benefits; and as men were said to be more knowledgeable about, and had the responsibility for, tree planting and management. It is recommended that:
_ Extension workers initially focus on women farmers to convince them of the benefits of new tree speciess and other agroforestry technologies, and subsequently explain to the men how to go about putting these improved technologies into practice on their own farm. As many species (especially medicinals and fruit trees) were planted by the men on the suggestion of their wives, this would not involve a major change in practices for the farmers. It would, however, involve a change in the habits of extensionworkers who primarily talk to men. Involving both men and women in the decision-making process about new species, where and how to plant and manage them would ensure that new agroforestry technologies will he more acceptable to, and adoptable by, a wider range of farmers.
Agroforestry: future perspectives: Although farms in the three study areas were already very small and continue to diminish in size every generation, farmers considered agroforestry the only solution to obtaining tree products in the future. With the increasing competition between trees and crops for a limited land base, farmers recognized that decisions concerning species selection and arrangements were becoming more and more difficult. The present differentiation of agroforestry systems will, therefore, become even more pronounced in the future with increasing fragmentation of farms. It is therefore recommended that:
Researchers and extensionists take the multiple goals and needs of farmers, and differences in availability of and access to resources, into account in the search for new species and intercropping arrangements of trees and crops.
Agroforestry development efforts in Rwanda in the late 1980s and early 1990s were often narrowly focused to address a limited number of problems defined by outsiders (for most projects, these problems were fuelwood, green manure and mulch to combat declining soil fertility, and erosion control'). Most projects came up with one (standard) solution, ignoring differences in resource availability, goals and needs of farmers. There are two problems related to such narrow focus. First, new technologies proposed may not be perceived as a solution by the farmers. For example, in the case of green manure for soil fertility improvement, farmers believed that plants by themselves could not add anything more to the soil than what they took up to grow, so how could they improve the soil? Second, some farmers' needs may not be identified as major problems in a diagnostic exercise but researchers and extensionists should not assume that these other needs (for medicines, spices, fruit, twine, etc.) are already being met or will automatically be met if farmers adopt new agroforestry technologies. In this regard, it is important to keep differences in agroforestry practices and tree planting objectives in mind, particularly as they relate to gender and farm size. For example, the results of this study indicate that women and micro-holders were primarily interested in planting trees that contribute to the family food supply, whereas men and larger land owners preferred to plant trees for timber and construction poles. Women on both small and large farms also expressed great interests in planting medicinal trees.
As these differences in objectives and context indicate, collaboration between farmer and researcher to identify problems and opportunities, and to look for solutions, helps develop a range of technologies that reflect these differences. In all cases, existing agroforestry knowledge and practices should form the starting point from which to move towards solutions. To help bring this about, it is recommended that:
With the help of participatory approaches, local people and communities identify. implement and evaluate their own priorities for tree growing, as this will generate more reliable research and extension agendas than top-down approaches (adapted from Scherr, 1992).
The farmer-consultants themselves made research and extension recommendations, which they felt would best support their efforts in advancing agroforestry practices. They recommended that:
_ There should be, more research on suitable species to increase the number of options available to farmers. Farmers do not want to be constrained in their choices and put into a straightjacket by research and extension which offers them standardized solutions to the very diverse conditions of their farms. They demand a cafeteria system of new species' and technologies suitable for a range of biophysical and socio-economic conditions from which they can choose according to ' their own needs, goals and resources. They stressed that the availability of a variety of tree species (both indigenous and exotic) and tree management options was important to enhance their traditional strategy of diversification to overcome economic and ecologic uncertainties.
- New tree species to be introduced should be compatible and noncompetitive with crops, non-shading, have tap roots and the ability to be used as live stakes. Ideally, and this was suggested in particular by many of the very small farmers, research should introduce trees that produce fruit and (fuel )wood simultaneously.
- Research and extension should ensure that new technologies (species) are made more accessible to farmers.
Identification of specific shortcomings, instead of the wholesale dismissal of local technologies as inadequate, can be an important mechanism to generate knowledge: it can be one of the means with which local knowledge can be developed in a fertile interaction with other knowledge. In this way, specific areas of local technology can be improved so that existing shortcomings are eliminated and a solution will be found with greater local relevance than whichever `external' technology (van der Ploeg, 1991).
Many farmers intuitively followed this advice. It was their strategy to solve particular problems and shortcomings by borrowing from any sources available (other endogenous knowledge systems, research and extension, etc.) to arrive at workable and profitable solutions for their conditions. It was, however, difficult to distinguish new from existing practices or to differentiate experiment from normal practice. For farmers, each season is an "experiment" in which new knowledge is obtained and new ideas are generated. The tree experts consulted in this study, therefore, described knowledge production through experimentation (igerageza) as an activity interwoven with everyday agricultural activities, not separated from them as is the case in the scientific knowledge system. These findings confirm the results of Box (1988) with cassava experimenters in the Dominican Republic, in which one of his research subjects explained that experimentation can only be understood if it is grounded within everyday life experiences and activities.
Although Plumería alba is known to botanists as an ornamental shrub with sweet smelling flowers, farmers in Maraba regard it as a medicinal plant and call it umudwedwe. Women use its sap to increase lactation and it is given to cows for the same reason. Interestingly, neither its local name, nor its medicinal uses were known to botanists and foresters. The tree was not found in Karama and farmers in that area knew nothing about the tree or its medicinal properties.
However, in spite of the interweaving of experimentation and normal production practices, experimentation was a conscious effort on the part of the farmers to build upon the body of endogenous agroforestry knowledge. The diversification of species and its resulting increase in complexity of land use systems resulting from farmers' experimental efforts were a deliberate strategy used by the farmers to overcome ecologic and economic uncertainties, and approach a better and more secure livelihood. Their experimentation with new agroforestry technologies was considered necessary to discover benefits not available from existing species and practices, or to determine whether they could get similar benefits more quickly and efficiently. Thus, although farmers were reluctant to try new technologies without having seen them in practice or use somewhere else, the majority of tree experts chose to try species with which they were unfamiliar in the species trial set up as part of this study (see Chapter VI). Farmer-consultants justified their choice by stressing the importance of experimenting with new ideas and technologies for development and progress. Within the group of tree experts, this was expressed as a greater sense of curiosity, a desire for knowledge, an internal drive for discovery and recognition for leaving descendants a better world. There was an implicit understanding that not experimenting would lead to stagnation and compromise an already precarious existence.
A fundamental problem facing experimenting farmers was the supply of new technologies to test on their farms. They identified the availability and accessibility of a range of options as necessary for their experimentation and system development. However, the options to be offered to farmers do not have to originate from scientific research. They can also come from other endogenous knowledge systems around the world. Farmers are able to make qualified assessments of what can (potentially) work in their own situation, but they are obviously limited to their immediate surroundings for information and ideas. An important role can be played by FAO, ICRAF, and other CGIAR centres to fill the gap. It would be ideal to take farmers on guided farm tours in different regions and countries to see new technologies in practice and to talk with their colleagues directly, but this is often impossible due to costs and restrictions on travel (especially across borders). Modern technology (video, cdrom, 3-D computer programmes) could provide an answer by taking farmers on "virtual" farm tours to expose them to new species, practices and systems from around to world. These media could show how farmers in other countries or regions have dealt with (similar) problems and what solutions they have devised to solve these problems. These media should be visually oriented, as farmers stressed the importance of seeing technologies in action in order to determine their suitability for their own situation. It would not be necessary to provide specific solutions, particularly not for the experimenting farmers. A glimpse of what other farmers have developed could provide ideas to push experimental activities in new directions. As FAO, ICRAF and other CGIAR centres have a global presence; and as they have collected and documented extensive information on agricultural and agroforestry systems from many different countries and/or indigenous peoples. It is recommended that:
- The multinational organizations and international research centres develop and distribute visual documentation (video, film, cd-rom, etc.) of indigenous agroforest y systems and practices, and farmers experimental methods and procedures related to the development of same. They also should assure their availability and accessibility to NGO's, farmer organizations, extension services and other persons and institutions interested in furthering farmer research and extension practices.
Such virtual farm tours need to be carefully conceived to guarantee that farmers receive the right message. Anthropological studies of farmer perceptions of messages from televised media show that there are fundamental differences in how people from different cultures perceive the images they see (Crowley, personal communication). The following suggestions are made to take such differences in perceptions into account:
Traditional healing practices are still widespread. This woman, one of the tree experts consulted in the case study, specializes in treating ringworm in children. The oval balls drying on banana leaves are made of clay mixed with leaves from various trees and plants. The clay preserves the healing powers of the leaves and masks their identity. It also facilitates application.
Experimental methods and procedures: The processes by which knowledge is added, used and lost to the stock of knowledge of a social group are similar in both the endogenous knowledge system and the scientific knowledge system. In both systems, we find a certain specialization and differentiation between a small group of knowledge producers (generaters, adapters, acquirers) and a large group of knowledge users. But there are also fundamental differences in the way knowledge and technology are produced and validated.
The tree experts did not use specific research methods and procedures for tree and tree cultivation experiments. Trees take several years to mature or to yield usable parts, so most farmers (even the large ones) could not afford to tie up land for several years to experiment with new tree species or arrangements. Thus, tests of new tree species or tree cultivation methods took place within existing fields and crops, which explains the interweaving of experimental and everyday agricultural activities talked about earlier in this section. The topic (most often a new species) and objectives of experimenting and the methods to be used depend on:
While there is no uniform method for testing new agroforestry technologies, farmer tree experts did adhere to more or less standard variables to evaluate the experiment:
Risks associated with experimental methods: The tree experts were well aware of the risks in trying new, untested technologies as can be seen from the above list of evaluation criteria. However, they stated that such risks were also associated with normal production practices; weather, pest and disease outbreaks, and market conditions cannot be predicted. The emphasis on observing (new) species and practices on other farms (to study their stature, growth pattern and possible negative effects on soil, crops and other trees) aimed at minimizing risks, since knowing potential risks helps in dealing with them. However, risks would not automatically preclude them from trying new technologies, especially if the utility and benefits were perceived to be greater than the risks.
Validation of experiment results: The major difference in knowledge production between experimenting farmers and scientists is not found in experimental procedures or trial evaluations, but in the way new knowledge and technologies are validated. In the scientific knowledge system, the primary aim of experimentation is the advancement of knowledge. There is an active communication of experimental results (in journals, conferences, seminars, etc.) among scientists and researchers. By contrast, knowledge production in the endogenous agroforestry system is primarily use- and user-oriented. Consultants consider communication of experimental results an important aspect of being recognized as "being knowledgeable", but there was not much apparent effort to share new knowledge. Validation of new ideas/knowledge comes when efforts are imitated by other farmers (neighbours and friends), in other words by the final technology users, not by fellow experts engaged in knowledge/technology production. It is therefore recommended that:
· Knowledge sharing and extension of the results of farmer experiments (i.e., experimental outputs) were beyond the scope of this study. This research was concerned with the sharing of ideas as inputs for farmers' experimental activities. However, farmer-to-farmer communication of experimental results is important to enhance indigenous agroforestry practices and will make a fruitful area of research. Some topics that should be addressed in studies of informal extension networks have been provided in the section "Research questions requiring future investigation" below.
Knowledge sharing and extension of the results of farmer experiments (i.e., experimental outputs) were beyond the scope of this study. This research was concerned with the sharing of ideas as inputs for farmers' experimental activities. However, farmer-to-farmer communication of experimental results is important to enhance indigenous agroforestry practices and will make a fruitful area of research. Some topics that should be addressed in studies of informal extension networks have been provided in the section "Research questions requiring future investigation" below.
While the two knowledge traditions vary in the nature and the content of their knowledge, the primary difference is in the distribution of that knowledge. Communication networks for knowledge sharing and distribution among tree experts or between tree experts and other farmers, were neither very extensive nor very well organized. Farmerconsultants identified this virtual absence of local communication and information exchange networks as a major barrier to agricultural and agroforestry development. To enhance communication among farmers in general. It is therefore recommended that:
Better communication is needed among farmer-experimenters to enhance endogenous agroforestry knowledge production through the sharing of methods, procedures and results. Opportunities exist for research, extension and NGO's to identify (potential) farmer-experimenters and organize them around particular problems to be solved or technologies (species, practices or arrangements) to be tested. A fixed group of socalled "pilot farmers" involved in pretesting and/or demonstrating all new technologies offered by a project or research (as was used by PAP-Nyabisindu in Kibingo) is not recommended, as, for various reasons, few technologies spread beyond the group of pilot farmers to other members of the community.2 Additionally, it limits farmers' own experimental practices and "technology development from within." As interests and resources vary among farmers, different farmers or groups of farmers may be involved in experimenting with different technologies (for example, Chamaecytisus palmensis for stake production or Euphorbia tirucalli for erosion control), and multiple networks may be in operation simultaneously. It is therefore recommended that:
These farmer study clubs should be organized, run and (if possible) financed by the farmers themselves in order to ensure their independence and sustainability. In addition, independence will ensure that problems addressed and solutions devised will be truly based on farmers' own beliefs, values, norms and experimental activities, even though they may decide to borrow ideas and technologies from elsewhere. As it will be difficult, if not impossible, to organize and run such clubs with farmers scattered in a wide geographic area, the clubs will of necessity be of a local nature. By consequence, the problems they will address and solutions to these problems will be of a local nature. Research, extension and NGO's may, however, help facilitate information exchanges between or periodic meetings of study clubs in different villages, regions, countries or even continents, especially for clubs struggling with similar problems.
NGO's, for example, could provide information exchange services through newsletters that include lists of problems the study clubs are addressing and stories about specific solutions devised. Information exchanges between clubs in different countries and continents could be made possible through, for example, the Ecovolunteer Programme of the Environment Liaison Centre - International (ELCI) in Nairobi, Kenya. This programme at present links more than one hundred NGO's across the globe through the Internet precisely for this purpose.
More in-depth research is needed to study pathways of agroforestry development and agroforestry knowledge generation. Research should be directed to investigate intensification-extensification processes of agroforestry systems across agro-ecological zones, land use systems and/or countries resulting from farm fragmentation and declining farm sizes.
The basic assumption of this study was that not every farmer would have the same knowledge of agroforestry nor be equally engaged in agroforestry knowledge production. Thus, a large part of the study was devoted to identifying the farmers most knowledgeable about trees and tree cultivation and most active in creating new knowledge about the subject. These farmers were then surveyed in detail about their knowledge and experimental practices. However, due to differing pathways of agroforestry intensification (high diversity/low density vs. low diversity/high density) both agroforestry systems and agroforestry knowledge differ among (groups of) individual farmers. A logical extension of this finding is that the kind of knowledge produced, and, by consequence, the processes by which this knowledge is produced may vary among (groups of) individuals as well. As this study focused primarily on the group of locally-identified tree experts, a future study aimed at the group of comparison farmers (or, even better, aimed at both groups) would be a welcome addition to the topic of farmer research and extension. Such a study should look into the question of differences in topics, objectives, methods, and evaluation criteria used by experimenting farmers who follow different strategies in agroforestry development.
An interview being conducted with a tree expert about the history and management of his woodlot.
A topic that has not been dealt with in great detail in this study is the informal extension of ideas, technologies and practices developed through farmer experimentation. This research only touched on the subject as far as it concerned the input of ideas for farmer experimental practices, but did not look at how results (the output) of these experiments diffuse through the community. Farmers did consider communication of their experimental results an important (if not the most important) aspect of their being considered "knowledgeable about trees and tree cultivation," but they did not appear to make an active effort to spread the results of their research efforts. This seeming contradiction calls up many questions and would make an interesting topic for study:
Endogenous agroforestry knowledge is neither stagnant nor inherently "worse" than scientific agroforestry knowledge. One can study and compare the two knowledge traditions, but one cannot, and should not, judge them by each other's standards. Science-derived agroforestry knowledge and technologies now dominate (or appear to do so) because their adherents (researchers and extensionists) are more efficient in distributing information on a large scale, not necessarily because their products are inherently better. In reality, both the scientific and endogenous knowledge system are hybrid systems that have incorporated each others' technologies and ideas, even though researchers and extensionists do or may not want to recognize this. On the other hand, the farmer-consultants in this study did recognize the value of agricultural research, since without research they would never have been able to obtain tree species such as Grevillea, Eucalyptus, avocado, coffee, cypress and pine, and crops such as cassava, potatoes, wheat and climbing beans that now dominate the rural landscape in the study areas. Both farmer-derived and researcher-derived agroforestry knowledge and technologies have their strengths and weaknesses and both have a role in developing and furthering agroforestry technologies and practices. The major strategy to enhance farmer research will, therefore, be a synthesis of the two knowledge systems (den Biggelaar, 1991) taking the strengths and weaknesses of each into account.
The greatest strength of formal (Western scientific) research is that it can access information, ideas, and technologies from a much greater number of sources and a wider geographical area than farmers can. Formal research, therefore, has an important role in agroforestry development, not so much to design specific agroforestry systems but to increase the number of options (species, management practices) available to the farmers. The design of specific agroforestry systems is better left to the farmers, as they are much better skilled in incorporating technologies generated by each of the knowledge traditions in ways that are locally applicable and beneficial.
For this to happen, an understanding of the processes of knowledge production is of major importance. Future studies of endogenous knowledge systems should, therefore, put greater emphasis on process-oriented research. However, most research on endogenous knowledge to date has been aimed at collecting specific facts (soil classifications, tree uses, indigenous medical or veterinary practice, etc.) that sit on dusty bookshelves because no one knows what to do with the collected facts. They become dead, static representations of ever-evolving, dynamic knowledge systems. This study has been interested primarily in knowledge generation processes. For this reason, it has been short on specific details about particular agroforestry systems, tree arrangements or species uses. The latter are only facts to help understand the logic of what farmers do and the reason behind it. By themselves, facts are meaningless; it is people who give meaning to them. This is perhaps best stated in the following quote from Jahn (1990): ".... Kuntu [modality, determination of a `thing'] is of higher value than Kintu [being without intelligence or thing; in Rwandan philosophy, these include minerals, plants, animals, objects, material goods, etc.] and so they gladly alter things and their organization yet keep their own style.
Two research assistants conduct a ranking game with a farmer and son, who were herding their cattle in a pasture in Karama.
They [the Africans] are convinced that it is not the thing which determines the style and the person, but that man through his style can and must give the thing a meaning, that the dignity and force of man lie in his capacity to give meaning to things, even when the things themselves were made for a explicit purpose" (text in `[ ]' added by this author). For example, a diagnostic study of agroforestry systems in Burundi contains a number of diagrams of farms showing tree locations, but this does not contribute to an understanding of how the system functions, why it was put together, how it evolves over time and which forces influence these changes. The drawings are but Bintu (things), they have no meaning. This case study has moved beyond this, not to show the Bintu (specific agroforestry systems and practices found in the study areas) but Kuntu (the modality and determination of the systems and practices) as indicated in the subtitle: "A case study of the dynamics of agroforestry systems and agroforestry knowledge".
The development of a synthesis among knowledge systems will increase the effectiveness of ongoing scientific agroforestry research and development as well as empower, legitimize and enhance the existing endogenous capacities for identifying problems and developing solutions (den Biggelaar, 1991). Farmers seem to have an intuitive sense of the importance of such a synthesis between the endogenous and exogenous knowledge systems. The major challenge will, therefore, be to convince researchers, extensionists and policymakers of the necessity and benefits of such a synthesis. This synthesis should not lead to a formalization of farmer experimental methods nor to a relaxation of the rigor of scientific research. The goal of the synthesis is to build upon the comparative advantages of each knowledge tradition, leading to a participatory and collaborative strategy for agroforestry technology development and assuring that technologies are client-oriented and grounded in local dynamics of socio-economic and agroforestry development (Scherr, 1993).
