The results of this type of research would not usually lead directly to
extension recommendations. However, they should provide (and be designed
to provide!) information to help local innovators identify components and
management regimes which could be tried with farmers on different types of
sites for selected functions. It makes sense to set priorities for this type of
research on the basis of questions identified through consultation and
collaboration with users (extensionists and farmers) already managing related
agroforestry systems in the field.
This approach has several implications for a research programme. First,
the design of experiments needs to be based on priorities established through
careful diagnostic research and identification of research and recommendation
domains. This implies a need for sufficient exploratory trials and onfarm
research prior to designing an experiment. Research directors and
funding agencies need to recognize that this "preliminary" work represents
research in its own right, which researchers should report and publish as
such.
Second, much experimental work on component response and interaction
will need to be multi-locational. The shortage of space on research stations,
the need for special farm niches for evaluating agroforestry technologies (e.g.,
homegardens, boundaries, etc.), and the need for representative agroecological
conditions to determine site/treatment interactions suggest a strategy of
using satellite research plots sited appropriately in farming communities. This
strategy will also facilitate the input of farmers into the design and evaluation
of researcher-managed experiments and studies.
Clearly, effort needs to be invested in the development of more efficient
approaches to on-farm experimentation and field studies. Examples of
innovative work in this direction include Beer et al. [1991], Huxley and Mead
[1988], and Gliessman [1988].