Preliminary Evaluation of the Feasibility of Using Geospatial Information to Refine Soil Fertility recommendations

IFDC carries out fertilizer field trials at hundreds of (georeferenced) locations throughout eastern and southern African countries. Yield data analyses from these trials typically focus on average yield responses and economic returns of specific fertilizer treatments with respect to control treatments. However, a deeper analysis of yield data might generate information about the spatial pattern of the response and provide insight in factors that cause differences in yield responses. These insights might be taken into account to make future fertilizer targeting more location or region- specific, as an alternative to blanket recommendations. The information may also be valuable for fertilizer companies and other actors in the chain at regional scale where to sell which fertilizer composition. The objective of this exploratory study was to gain insight into factors that cause differences in fertilizer response. which might aid in tailoring fertilizer recommendations to local or regional conditions. This is done through (1) a linear mixed model analysis of fertilizer field trial data using spatial soil and climatic data as covariates and (2) an exploratory geostatistical analysis of soil and fertilizer field trial data. The latter included the development of digital soil maps for eleven soil fertility parameters with random forest modeling and kriging. Soil and crop yield data for Burundi were used for these purposes. The results of the modeling exercises indicate that there is a general lack of correlation between soil type and climatic variables and yield. Based on these results, recommendations may not to be refined for Burundi, indicating that fertilizer formulations used in the omission trials may be valid throughout the study zone. The soil nutrient maps do not appear to be indicative of crop nutrient response, but are indicative of nutrient deficiencies that might be anticipated. We note, however, that eliminations of mistakes in the current datasets and used of more spatially and temporally synchronized covariate data would add value to the analysis. These findings may also prove different over more variable environments such as Rwanda, where soils are considerably more variable. The added value of geospatial analysis for improving fertilizer recommendations is not immediately evident from these exploratory analyses but the results and the lessons leamt can be used to scope a more robust and thorough strategy for further geospatial data gathering and analyses of fertilizer response trials and for anticipating required nutrient composition of fertilizers.
Fertilizers, Nutrients
Kempen, B., Vereijken, P.F.G., Keizer, L.P.C., Ruiperez González, M., Bindraban, P., Wendt, J., 2015. VFRC preliminary evaluation of the feasibility of using geospatial information to refine soil fertility recommendations. VFRC Report 2015/6. Virtual Fertilizer Research Center, Washington, D.C. 67 pp.; 20 tables; 31 fig.; 30 ref.; 4 appendices. nbsp;