Abstract
Agricultural planning processes at farm to national level are essential for assessing and reacting to land conditions, opportunities, and threats for coffee production. However, lack or uncertainty of information is common during these processes. Bayesian networks can be used to manage these uncertainties. We, therefore, developed the first Bayesian network model for an Agroecological Land Evaluation for Coffea arabica L. (ALECA). A newly developed set of suitability functions was used to populate the nodes in the network. ALECA was then adjusted and validated to Central America. The results show that even without the use of coffee maps as input, ALECA accurately scores the suitability of actual coffee areas for coffee production as higher than that non-coffee areas, and can accurately predict the known order of quality of coffee reference zones in Central America. The results also show that ALECA can be used as a decision-support tool even under data uncertainty.