Reflections on the past two decades of organized research in tropical
agroforestry raise several issues. Research efforts started with an in
ductive and experiential approach but have subsequently followed a ded
uctive and experimental approach that includes hypothesis testing and
the development of predictive capability; agroforestry research is thu
s being transformed into a rigorous scientific activity. The research
agenda, so far, has given high priority to soil fertility and other bi
ophysical interactions, less priority to anthropological and sociologi
cal aspects, and little priority to evaluating costs and returns, pest
s and diseases, and the so-called non-timber forest (tree) products. M
oreover, larger-spatial-scale issues, such as carbon sequestration, wa
ter quality, and biodiversity conservation, have been neglected becaus
e of the emphasis on field- and farm-scale studies. Overall, the high
expectations that were raised about the role and potential of agrofore
stry as a development vehicle have not been fulfilled. In order to ove
rcome this, it is imperative that research be focused on the generatio
n of appropriate, science-based technologies of wide applicability, es
pecially under resource-poor conditions and in smallholder farming sys
tems. Future research agendas should entail a judicious blending of sc
ience and technology. Applied research should build upon the findings
of basic research to generate technologies for application at the farm
, regional and global levels. Such research should place increased foc
us on previously neglected subjects, for example, the exploitation of
indigenous fruit-producing trees, the agronomic components of agrofore
stry systems, and the global issues mentioned above. Furthermore, an a
ppropriate methodology that embodies economic, social, and environment
al costs and benefits needs to be developed to realistically assess th
e impacts of agroforestry, and an enabling policy environment that wil
l facilitate agroforestry adoption needs to be made available. Agrofor
estry research of the 21st century should strive to build bridges from
the inductive phase of the past, through the deductive phase of the p
resent, to the future phase of harnessing science and generating techn
ologies for the benefit of the land and its present and future users.