to the poorest. In particular much private and public R&D is spent on corn, wheat, maize, and rice, while very little is devoted to cassava, millet, sorghum and potatoes. However, it has not proved easy for research and extension organizations to adapt their established practices (Graham et al., 2001) to the new way of understanding rural development as part of an AKST system based on the idea that knowledge is coproduced by all actors involved. The most important of these issues are summarized in the present subchapter. Thinking on rural development has shifted from the 1960s to the 1990s and has reached a balanced state between the productive and social sectors, and between state and market interventions.

Effectiveness of formal AKST organizations. It is well known that many public research and development (R&D) bodies of national agricultural research systems (NARS) are finding it difficult to deal with poor farmer- and peasant economy-based issues in many countries. The problems range from resource constraints on the one hand to rigid, disciplinary-bound research planning on the other (IAC, 2004). Often there is a lack of engagement with client sectors and unwillingness to exchange and co-generate knowledge with other research bodies in the sector. This is also related to the process of identifying research problems, which is often based solely on perceptions of disciplinarybased researchers with incentive systems usually grounded mainly on the number of publications. The inevitable result is that all too often resource allocation to the NARS does not pay off in terms of economic, social and environmental development possibilities for poor farmers. While a number of countries have initiated some remedial policies for these issues, the relevant literature shows that there is still some way to go. The difficulties of more equality-based engagement with farmers, peasants, or "clients" has also to do with an understanding of the reasons guiding rural actors' decisions, actions and livelihoods that is too narrow (see Yapa, 1993 for Asia; Wiesmann, 1998 for Africa; Trawick, 2003 for Latin America).

Promotion of other stakeholders' AKST. Traditionally, the passing on of results of agricultural research to users was handled by state-funded extension services. Not only have these suffered through structural adjustment measures, but an increasing number of questions have also been raised by the extension systems themselves as operational organizational mechanisms (Farrington et al., 2002; IAC, 2004). There is also evidence of an increased need to engage in partnerships in order to reconceptualize (in theory and practice) the delivery of technology in the context of an AKST system that is based on the paradigms of knowledge coproduced by scientists, policy makers and client groups. These partners include private sector organizations, but they also involve NGOs, community-based organizations (CBOs) and social movements that are able to bring skills and knowledge to bear simply due to the close relationships they have established with specific communities. Today's challenges in community development in developing countries make it more compelling for higher education to reach effective changes of vision and prepare professionals to lead innovative rural development processes. Training, capabilchapter

ity building, and reinforcement of small-scale farmers' skills to enable them to participate in the agriculture supply chain are urgent tasks.

Coproduction of agricultural knowledge

The combination of various forms of exogenous scientific knowledge, e.g., from the natural, agronomic, economic or social sciences, with the many and highly diverse forms of so-called "local", traditional or endogenous knowledge is a basic challenge. These different forms of knowledge are represented by different local (farmers, traders, craftsmen, etc.) and external actor groups (civil servants, extensionists, researchers, service providers, etc.). One can therefore call them "knowledge systems". Combining endogenous and exogenous knowledge is achieved by increased participation of "end users"-including marginalized and poor actors- in the different forms of research and development. While the initial focus of combining knowledge was on increasing participation at local levels, today emphasis is shifting towards upscaling participatory processes into the meso- and macro-levels of social organization (Gaventa, 1998) resulting in multilevel and multistakeholder approaches.

     When taking into account the centrality and value of endogenous, traditional or local forms of knowledge related to agricultural development-e.g., through ethnological approaches in sciences studying agricultural soils, plants and animals (Nazarea, 1999; Winklerprins, 1999)-it is necessary to reflect on the ethical and epistemological implications related to the integration of different knowledge systems (Dove and Kammen, 1997; Olesen et al., 2000; Rist and Dahdouh-Guebas, 2006). Integration of, and cooperation between, different knowledge systems is often hampered by interaction that does not take into account the need for the process of communication to move beyond the practical and generally tangible technological economic, ecological and social effects of innovations. In the long run, innovation can only be successful if it "makes sense" to all those involved, i.e., it needs to be integrated into (and by) the different knowledge systems involved. This is also particularly important for innovations in rural development (Dove and Kammen, 1997; Olesen et al., 2000).

     There is also growing consensus among researchers concerned with sustainable agriculture that no single group of actors should appropriate the right to define what type of combination should exist between scientific and "local" forms of knowledge (Röling and Wagemakers, 2000; Rist and Dahdouh-Guebas, 2006). As a consequence, participatory forms of coproduction of knowledge, based on social learning among actors involved, have become a key feature of sustainable agriculture and resource management (Wollenberg et al., 2001; Rist et al., 2003; Pahl-Wostl and Hare, 2004). This means that the role of science within a process of participatory knowledge production must be redefined. Instead of striving to find and voice the ultimate instance of "truth", the scientific community must complement conventional and generally discipline-based knowledge production with inter- and transdisciplinary approaches. The particularity of a transdisciplinary approach is that it implies examining "real-world problems" from a perspective that (1) goes beyond specific disciplines by combining natural, technical, economic and social sciences, and (2) is