The savannas of South America: towards a research agenda for CIAT

This paper provides a framework for decision making, to facilitate the process of prioritization in the elaboration of a research agenda for the new savanna program at CIAT. Holistic and rapid appraisals undertaken during 1990-91 in the acid savannas of Brazil, Colombia, and Venezuela strongly support the view that this ecosystem has great potential for ecologically sound resource use and management. Accordingly, CIAT research involvement in savannas should address both the more intensively intervened savanna regions, particularly of Brazil, and the natural unexploited sites. The study also suggests that, to be cost-effective and ensure impact, the research strategy should identify and incorporate existing knowledge and technologies to manage the resources (In-reach activity). To guarantee success, field activities must be built on a holistic understanding and close relationships with different types of partners. This implies the creation of an institutional set-up where close interactions may be possible with private and public organizations at both local, national and international institutions, CIAT should maintain a careful balance in terms of basic and strategic research versus adaptive research, training and institutional strengthening. Setting priorities for the savanna research program at this stage must concentrate on site selection. Alternative sites selection should follow an iterative process, where key aspects of the demand and supply of technology and logistic considerations must be taken into consideration. A joint decision with the interested partners must determine the final sites. Simustaneously CIAT should prompt several activities aimed at: a) producing an institutional inventory of environmental and resource management oriented organizations in the Cerrados of Brazil; and b) starting a broader screening of germplasm of grasses, legumes and rice as crucial components in the system by using more mixed cropping systems criteria than in the past.