Use of spatial patterns of soil properties and wheat yield in geostatistics for determination of fertilizer rates

Abstract

Researchers traditionally conduct fertilizer experiments on small, replicated areas at university or governmental research centers. This approach has two problems: (1) extrapolation of the results to large cropped fields requires taking into account the impacts of spatial variability of soil properties on the optimum amount and type of fertilizer to apply and (2) farmers have limited access to the research work, particularly in developing countries, which limits the information transfer needed for farmers to adopt the recommended fertilizer techniques. With these problems in mind, field and laboratory investigations were conducted on a farmer's field to study the spatial patterns of some soil properties and wheat yield using geostatistics for determination of fertilizer rates. The field was divided into four parallel transects, and soil samples were collected at 15‐m intervals along each transect. These soil samples were analyzed for different physical and chemical properties. A uniformity trial was conducted on wheat in the field. Classical statistical and geostatistical techniques were used to describe the variability in measured soil properties and wheat yield. Most of the soil properties and wheat yield had a very high coefficient of variation. Variability in soil properties and wheat yield was best described by a linear model. The potential wheat yield was estimated using the relationship between wheat yield and lime content, and was used in determining nitrogen (N) fertilizer rates for different areas. The effects of soil variability on crop growth and profitability of fertilizer used were also demonstrated to the farmers.