Abstract

In precision agriculture, inputs are adjusted to the varying demand across a field in order to optimize net returns and avoid losses to the environment. For site-specific N application, it is useful to know how fertilizer N demand, plant available soil N (Np, i.e. soil N taken up by plants) and potential yield relate to each other and to different soil characteristics within a field. A 3-year field investigation was carried out on a 15-ha arable field with large soil texture differences in south-west Sweden, on which winter wheat and spring barley were grown. Variation in Np was considerable both within the field and between years but could only partly be explained by variations in soil organic matter, clay and elevation. Maps of yield, grain protein content and Np differed between years, partly due to differences in seasonal variation in soil moisture. Together, protein and yield maps indicated where N supply was sufficient and where factors other than N were limiting, which allowed the accuracy of the N fertilization to be evaluated retrospectively. Differences in yield response to N between areas with different soil texture were small when soil moisture was sufficient. In a dry year, yields were smaller at sandy sites, while in a wet year Np, and thereby yield, was lower on clayey sites. Soil moisture is related to soil electrical conductivity (SEC) and elevation, which are easily measured densely within the field. Therefore, these parameters are useful for dividing the field into zones with different risks for drought and waterlogging and can be used for variable N application, assuming that the season can be defined as dry, normal or wet at the time of fertilization. Average values in zones created from a densely measured variable proved to be a better alternative for many variables in this field than interpolation of sparsely collected soil data without respect to distinct borders.