Keywords: Nitrogen fertiliser, Fertiliser recommendations, Soil N supply, Soil mineral nitrogen, Modelling.
In the early days of fertiliser use, advisers encouraged farmers to rely on their own experience when assessing the need for nitrogen applications to cereals. Since then, advisers have also become experimenters and make a large number of experiments each year to develop and improve N recommendations. The optimum N amount in each experiment is best determined by fitting to the results a function which closely describes the diminishing returns in grain yield obtained as the N application increases, but even with such a function, N optima can seldom be estimated precisely. Recommendation schemes which have evolved over recent decades in the UK have not been based on field measurements of soil or crop N but have rather categorised field histories and soil types in order to infer amounts of soil-derived N, crop yields and N uptakes, and thus the best amounts of fertiliser N to apply. Results of experiments have been used to refine advice for the future but seldom to check past advice. Our attempt at testing current systems for N recommendations shows that their success depends more on accurately predicting the average optimum than on identifying and accounting for deviants. The best prospect for further improvement lies in predicting deviant sites, especially those where the optimum amount of applied N will be either near zero or very large. Measurements of soil mineral N have shown promise where small optima are suspected. More sophisticated crop observations may allow optimum amounts of N to be matched more closely and thereby minimise N wastage and unrealised profits.
Judgements which have been arrived at by experiment on the best forms of fertiliser N and the best times for application largely conform with current understandings of behaviour of N in soil and its uptake by the crop. However, a quantitative and dynamic description of these processes, as is now possible with computer models, should help to prevent both waste of N and shortages of N for crop growth. Such models are shown here to give good simulations of the amounts of soil mineral N and crop N in the spring. They currently start from an autumn measurement of mineral N, but work in progress on the modelling of the longer-term turn-over of N through the soil organic matter should make this unnecessary. Several uncertainties remain, but if aberrant sites can be satisfactorily identified through the simple trial system proposed, we can progress towards more accurate and more site-specific on-farm advice. This, however, depends on the necessary research being supported.
R Sylvester-Bradley1, T M Addiscott2, L V Vaidyanathan1, A W A Murray3 and A P Whitmore2.
1ADAS, Brooklands Avenue, Cambridge CB2 2DR, UK.
2AFRC Institute of Arable Crop Research, Rothamsted Experimental Soation, Harpenden AL5 2JQ, UK.
3formerly as2, now: British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK.
36 Pages, 13 Figures, 4 Tables, 82 References.