Variability of optimum sowing time for wheat yield in Western Australia

Sowing wheat (Triticum aestivum L.) at the right time is one of the most important means of maximising grain yield in dryland agriculture. Objectives of this study were to understand the variation in estimates of optimum sowing time as influenced by cultivar and environmental characteristics, and to assess the relative importance of location, season, sowing time, and cultivar factors in maximising grain yield in Western Australia. Twenty-seven cultivar x time of sowing experiments were conducted over three seasons (2003-05) at a range of locations (annual rainfall 300-450 mm, lat. 28-35 degrees S). There were four types of cultivar x sowing time responses, namely, quadratic, linear declining, flat, and linear increasing, associated with opening rains before mid-May, opening rains after mid-May, low-yielding sites, and good spring rains, respectively. Regression-tree analysis revealed that differences among cultivars in T-max (sowing time when maximum grain yield was achieved) were much less in the eastern sites (mostly drier seasons). A biplot differentiated cultivars for T-max across the range of environments used, while the subset regression analysis specifically indicated an association of average temperature and growing-season rainfall with variation for T-max of individual cultivars. The yield penalty for sowing before the optimum time in quadratic-type responses was clearly greater for shorter season cultivars but no clear relationship was apparent between maturity class of cultivars and the penalty for late sowing, possibly due to differential plasticity of cultivars for grain weight under harsh finishing conditions. The duration of the optimum sowing window at a given location was inversely proportional to the yield potential, implying that it is critical to sow at or close to the optimum time when the yield potential is high, most common when the season breaks early. Yield component analysis showed that the relative change in grain yield over sowing dates was significantly correlated with relative changes in grain numbers/m(2) in the late May sowings but other yield components were also important in the early May experiments. Sowing time accounted for 10% of grain yield variation compared with cultivar (1%), while the rest was due to uncontrollable factors of location and season.