Abstract
Twelve spring wheat (Triticum aestivum L.) cultivars were planted in 16 trials which received seasonal rainfall (including irrigation) ranging from 230 to 755 mm year-l. Under low rainfall conditions, drought stress occurred during the period of stem elongation when the ears grew. Grain filling was largely free of stress. Variation in precipitation explained 75% of the variation in mean yield among trials, and the main yield component responsible for yield reduction at low precipitation was ear number, as affected by tiller survival rate. The larger-yielding cultivars sustained a higher kernel number per ear in compensation for the reduced number of ears under stress. The yield stability of cultivars under low precipitation was defined by both the joint linear regression method and by the linear regression of yield on precipitation, both of which gave very similar predictions. The magnitude of the intercept of the latter regression was taken as indicator of drought resistance in terms of yield. The variations among cultivars in this intercept was well explained by corresponding variations in osmotic adjustment, heat tolerance and canopy temperature. Grain filling from stem reserve mobilization was not important in this respect, evidently because grain filling was not subjected to drought stress in these trials. Early-heading cultivars had a relative yield advantage only when precipitation were less than about 300 mm and grain yield was less than about 350 g m-2. Both early heading and heat tolerance were positively correlated with osmotic adjustment across cultivars. It was therefore concluded that the main physiological attribute associated with yield stability under conditions of drought stress during the period of ear growth was the capacity for osmotic adjustment, which apparently allowed sustained ear growth under drought stress.