The aim of this paper is to describe a methodology for recommending crop species for smallholder farmers in the tropics, targeted to biophysical and socio-economic niches, incorporating both available data and expert knowledge. Although it is important that species are matched to biophysical environments, it is equally important that they are matched to the unique socio-economic situation and management practices of a farmer. Agriculture in the tropics and subtropics is complex and heterogeneous, both biophysically and socio-economically. Bayesian modelling has been identified as the most appropriate method for incorporating sparse and uncertain data with expert knowledge to predict the probability of a given species being suitable for a given environment or niche. As a case study, a spatially enabled decision support tool designed to assist in the decision-making process of targeting forages in Central America is under development. It will use existing data and expert knowledge to mimic the decisionmaking process involved in selecting forage varieties for specific socio-economic and biophysical environments. Validation and verification of the model are discussed, as well as the technological specifics of the tool development and deployment.
[1]
Judea Pearl,et al.
Bayesian decision methods
,
1990
.
[2]
J. R. Wright,et al.
Expert Systems in Environmental Planning
,
1993
.
[3]
M. Goodchild,et al.
Uncertainty in geographical information
,
2002
.
[4]
Jeff R. Wright,et al.
Probabilistic Inferencing and Spatial Decision Support Systems
,
1993
.
[5]
Michael Peters,et al.
Linking research on forage germplasm to farmers: the pathway to increased adoption—a CIAT, ILRI and IITA perspective
,
2003
.
[6]
Claude B. Courbois,et al.
Livestock to 2020: The Next Food Revolution
,
2001
.