Dryland wheat was fertilized with ammonium nitrate or liquid urea-ammonium nitrate at the time of sowing or about 3 months later (generally at the terminal-spikelet stage) on a well-drained site near Harden on the south-west slopes of New South Wales. The experiments continued from the second to the fifth year (1981-1984) of the cropping phase of a crop-pasture rotation. The maximum agronomic efficiencies for yield in the four consecutive years were 19, 4, 23 and 25 kg grain per kg of applied nitrogen (N). The three large responses were obtained in wetter than average seasons and the small response was obtained during drought. In the last three years of the study the yield response to nitrogen at the terminal-spikelet stage was found to be close to but slightly less than that for N applied at sowing. In those years the agronomic efficiencies for the late-applied N were 0, 22 and 22. The apparent recovery of fertilizer N in the above-ground parts of the crop at maturity was up to 70% of the fertilizer applied in the year of sowing, and, after the drought during which there was little uptake of fertilizer N, up to 62% by the subsequent crop. The fertilizer efficiencies in the non-drought years were higher than generally reported in south-eastern Australia, and indicate potential for profitable delayed application of N fertilizer to wheat. Grain-protein responses were variable from year to year and are discussed against a simple theoretical background of the amount of N applied and grain-yield response.
