Systems analysis of wheat production on low water-holding soils in a Mediterranean-type environment I. Yield potential and quality

Wheat grain yields, grain protein and grain size are often variable in a Mediterranean-type environment due to large rainfall variability. Grain yields are often low due to low N inputs as a result of large uncertainties of rainfall, particularly during the latter part of the season. Larger amounts of N fertiliser might increase yields and also grain protein in such environments but could have negative effects on grain size. A systems analysis approach was taken, using the APSIM-Nwheat crop model, to assess potential yield, grain protein concentration and grain size of wheat across a range of conditions. Rainfall and temperature varied with location. Soil type, different amounts of initial stored water and a range of management options, including varying sowing dates and N fertiliser applications, were also considered. In Australia, penalties for small grain size are based on 'screenings': the percentage of grain to pass through a 2 mm sieve. A new routine was developed for the model describing the relationship between screenings percentage and average simulated grain size. At low rainfall locations, 30 mm of plant-available soil water at seeding gave a significant increase in grain yield (> 20%) above that for no soil water. At high rainfall locations there was little effect (< 7%). Three typical coarse textured soils had similar yields at low N supplies but soils with higher plant-available water-holding capacity responded more to N applications and thus at high N applications achieved higher grain yields. Above average rainfall from May to September in combination with below average temperatures resulted in the highest simulated yields. However, seasons with an average amount of in-seasonal rainfall but with rainfall well distributed throughout the season also had above average yields. Applications of small amounts of N (30 kg N/ha) caused a slight reduction in grain protein concentration, but higher N applications (> 30 kg N/ha) resulted in increased protein and screenings. Nevertheless, with moderate N (60 kg N/ha) applications and early sowing, the proportion of years with low grain protein (< 10%) and low screenings (< 2%) was still more than two-thirds. Simulated average yields of a location increased with average rainfall but declined with increasing average temperature of a location. This allowed the development of a simple rule of thumb in which average grain yields can be related to the rainfall/temperature index of a location. The simulation analysis showed that there is potential to increase current farmer yields through higher and split N applications on light soils in Mediterranean-type environments without increasing average screenings above 3%. However, grain protein concentrations are likely to be low in most seasons on low water-holding soils. (c) 2007 Elsevier B.V All rights reserved.