AKST will be to find ways for combining local knowledge with innovations developed in similar other contexts to generate locally adapted new options. The question devel­opment agents will have to address is, under which circum­stances they may scale up innovations per se and when they should focus on scaling-up innovation processes (Franzel et al., 2004). In the scaling-up process, it will be crucial that research and extension act in a careful, empirical and criti­cal way (Tripp, 2006). If wide dissemination of innovations that were successful in a certain context is attempted, this may create exaggerated expectations and hence frustra­tion, if these innovations are not adapted in many other contexts. This happened for example with alley cropping (Carter, 1995; Akyeampong and Hitimana, 1996; Swinkels and Franzel, 2000; Radersma et al., 2004) or the system of rice intensification (SRI) developed in Madagascar (Stoop et al., 2002). Agricultural research and extension still largely works with technologies that rely strongly on external in­puts, even in less favored areas (Stoop, 2002). Potential for innovation in low external input agricul­ture is highest if research focuses on understanding and

building on local concepts of farming such as the exploita­tion of within-farm variation, or intercropping. However, if research and extension work with technologies that rely strongly on external inputs, farmers will seldom adopt the results (Stoop, 2002). A further challenge is the dissemina­tion, as farmer-to-farmer diffusion is less important than commonly assumed for such innovations (Tripp, 2006). Low External Input Sustainable Agriculture (LEISA) comprises organic farming. Organic farming and conven­tional (non-labeled) LEISA can mutually benefit from each other. Organic farming with its stringent rules on external input use has to be even more innovative to solve produc­tion problems, sometimes opening up new avenues. Organic farming has the additional opportunity of deriving benefits from close links between producers and consumers. The challenge, however, is to exploit this potential. New low external input technologies have the poten­tial to improve productivity while conserving the natural resource base, but there is no evidence that they are spe­cifically pro-poor (Tripp, 2006). An important concern in low external input farming is soil nutrient depletion. Across