IMPROVING FERTILIZER RECOMMENDATIONS FOR SUBSISTENCE FARMERS IN WEST AFRICA - THE USE OF AGROECONOMIC ANALYSIS OF ON-FARM TRIALS

A large number of zero, half and full rate fertilizer trials were conducted on-farm in Southern Senegal with rainfed lowland rice (n = 24), maize (n = 48), and groundnuts (n = 18). Trial sites were located according to farmer selected criteria: soil texture in the case of rice; compound garden versus outer field in the case of maize; and, previous cropping history in the case of groundnuts. Quadratic fertilizer response curves using all the cases explained only 16-29% of the variance. Subsequent stratification of the fields by soil organic matter, texture, and pH permitted the identification of fertilizer responsive and non-responsive fields. Response curves using only the tests conducted on soils without a limiting constraint explained 36 to 47% of the variance. At half rate fertilization levels VCR's of 3.8 (maize), 5.8 (rice) and 6.9 (groundnuts) resulted. Within productive fields, level of weed control, percent barrenness and final stand at harvest explained much of the remaining variation in yields for rice (82%), maize (61%) and groundnuts (76%) respectively. Response curves were then used in an economic analysis to address on-farm fertilizer allocation issues. Based on survey results and field trial data, partial budgets for small and medium-sized farms were developed. This analysis showed marginal rates of return of 400 and 165 percent to half and full rate fertilization, respectively. This type of fertilizer validation program, conducted on farmer-selected sites, improved targeting of recommendations, and helped to identify agronomic practices that should result in reduced economic risk and increased fertilizer adoption by farmers.