Abstract:
The French and Shultz approach that relates seasonal rainfall to potential yield in wheat has yet to be applied to dryland canola. Relationships were derived between grain yield of 42 experimental crops (yield range 0.5?5.4�t/ha) free of weeds, pests, diseases, and nutrient deficiencies in southern New South Wales, and various measures of observed (rainfall, available soil water) and simulated (evapotranspiration) seasonal water supply. April to October rainfall and in-crop rainfall were the poorest predictors of yield (R2 < 0.5). By adjusting in-crop rainfall to account for stored soil water at sowing and that remaining at harvest (termed ?seasonal water supply'), 68% of the variance in yield could be explained. Estimates derived using the APSIM-Canola simulation model or simulated totals of evapotranspiration or transpiration explained 73?82% of the variance. The slope of the regression line between yield of the 42 crops, which simulation indicated had all yielded to their water-limited potential, and seasonal water supply (termed here the water-use efficiency for grain production, WUE) was 11�kg/ha.mm above an intercept of 120�mm. WUE varied from 4 to 18�kg/ha.mm and the upper boundary for WUE in those seasons where rainfall distribution facilitated maximum efficiency was 15�kg/ha.mm. Long-term simulations, conducted at locations with mean annual rainfall of 430?660�mm, confirmed the variability of WUE due to rainfall distribution and also that WUE would be expected to decline, on average, by one-third between sowings in early April and early July. This necessitates caution in accepting a single WUE value as an indicator of agronomic constraints to yield. For the purposes of practical application by farmers and advisors, water-limited potential yield can be calculated in the region as a function of seasonal water supply minus 120�mm up to a limit of 450�mm, beyond which potential yield is not limited by water. Available soil water at sowing can be estimated from summer fallow rainfall above a threshold of 80�mm, and water remaining at harvest can be estimated from post-anthesis rainfall above a threshold of 50�mm. This improved method for estimating water-limited potential yield in canola retains the ease of use of the French and Shultz approach, so that other constraints to yield can be more accurately diagnosed in dryland environments by farmers and advisors.