Fertiliser use and definition of farmer domains for impact-oriented research in the northern Guinea savanna of Nigeria

One of the options to alleviate soil fertility constraints for sustainable agriculture in the savannas of West Africa is to develop soil nutrient mana gement technologies from an adequate supply and feasible share of organic a nd mineral inputs. This paper makes a diagnosis of farm-level use of organi c and inorganic inputs, as a basis for the development of technologies. The results from the diagnosis are then used to develop a framework for charac terizing farmers for impact-oriented research on soil nutrient management s ystems. The survey was carried out with 200 farmers carefully selected in t wo villages in the northern Guinea savanna of Nigeria. The results showed t hat more than 90% of farmers in both villages used chemical fertilizers. Th is is contrary to a general belief that they are not widely applied to food crops by smallholders in African agriculture. However, up to 81% of the fi elds received less than half of the recommended 120 because of high costs d ue, probably to removal of subsidies and inefficient marketing systems. Org anic inputs such as animal manure were applied in very small quantities (ab out 8% of the requirements). However there is evidence of integrated use of inorganic fertilizers and organic manure on some (24%) of the fields. The problem to be addressed is that of the production (and efficient utilisatio n) of organic inputs in the northern Guinea savanna. Nitrogen deficiency is the most limiting soil nutrient in the cereal-dominated systems of study a rea. On this basis, farmers were classified into two a-priori groups using a threshold of 30 , and multiple quantitative variables were fitted in a di scriminant analysis to validate the typology. Results indicated that more t han 75% of farmers were well classified into two groups that had the charac teristics of the a-priori groups. Two others were atypical and included the remaining 25% of farmers. Thus, there are a total of four groups of farmer s referred to as farmer domains in this paper. The two domains with 75% of well-classified individuals are suitable for the selection of farmers with whom to conduct applied research or for development activities because they represent the general patterns in the supply and use of soil nutrients in the study area. Although basic research can be done in the four domains, th e two atypical groups are most suited for process-level studies to improve the understanding of factors that make the systems either more efficient or less efficient than the two other farmer domains. In either case, represen tative farmers were easily identified by their highest probability of belon ging to a specific domain from the model results. Multivariate models const itute a good framework to make a typology of, and to select farmers for, pa rticipatory research and extrapolation of results in the northern Guinea sa vanna.