Large areas of dryland cereals in semi-arid regions of southern Australia grow on alkaline soils that have subsoils containing high levels of salinity, sodicity and boron. Yield and grain protein levels of cereals growing on these soils are generally low and are not increasing as rapidly as those in regions with no subsoil limitations. The severity of the limitations varies across the landscape. The levels of salinity, boron and sodicity are positively correlated and tend to increase with depth in the soil, typically reaching levels that confine the roots of cereal crops to the top 40-90 cm. Areas with the shallowest rooting zones have limited amounts of available soil water, leading to low cereal yields. They give poor yield responses to N fertilizer, despite having generally low levels of N fertility in the region. Two hypotheses are discussed to maximize profit from N applied to cereals growing in this landscape. One strategy is to concentrate N fertiliser on the parts of the landscape with the least subsoil limitations. To test this hypothesis, we applied ~10m-wide strips of urea along the length (~1km) of eight cereal paddocks in the southern Mallee Region of Victoria. The yield response to N were measured and related to apparent Electrical Conductivity (ECa), measured by ElectroMagnetic induction (EM) surveys. In five of the paddocks the average yield response to N decreased with increasing ECa, falling from 20 kg grain/kg N where ECa was zero, to no response where ECa=136 mS/m. Of the other paddocks, responses were masked by frost damage and drought. The results offer a method to increase cereal yield by concentrating N on the most responsive parts of the landscape identified by EM surveys. The other strategy is to manage the placement and timing of N application. We conducted 9 experiments with wheat growing on farms during 1999 and 2000 in the southern Mallee region of Victoria. We measured yield and protein responses to N applied, as urea, in different ways and at different times, including banding, incorporation by sowing, and topdressing. In these experiments the topdressing was relatively late in crop development – between stem elongation and booting. We aimed to apply the fertilizer just before rain, and in most cases succeeded. As shown with the paddock-scale strips, the largest yield responses to N were observed on the sites with the least subsoil limitations. Where the subsoil limitations were severe, the most effective way to increase yield with N fertilizer was by multiple topdressings, followed by mid-row banding, a system in which N fertilizer is injected between every second seed-row in a one-pass operation. Both banding and topdressing led to slow seedling growth. In the case of banding, the slow growth was because of slow nitrification from highly concentrated urea in the bands. The least efficient method of applying N was Incorporation By Sowing (IBS), which is probably the most common commercial method. Seedlings fertilized in this way rapidly took up N, grew rapidly and apparently exhausted soil water before grain filling. This process is known as 'haying-off', a yield decrease in response to applied N in conditions of terminal drought.

Topdressing and banding N led to larger yield responses than IBS where subsoil limitations were most severe. However on soils with severe subsoil limitations, N topdressing at the booting stage before rain increased grain protein with minimum risk of haying off.