Abstract
The finding that more than 50 percent of the variation in grain yield of breeding lines in two different drought stress treatments could be attributed to variation in yield potential and time of flowering was used to develop a drought-response index (DRI) based on the residual variation in grain yield, adjusted for experimental error. DRI was positively correlated to measured yield in the drought treatments, and independent of both yield potential and time to flowering. DRI in both midseason and terminal stress treatments was unrelated to yield component structure in the irrigated control treatment, indicating that selection for plant type under non-stressed conditions will not influence drought response. DRI was correlated to both grain number per panicle, and grain yield per panicle in both stress treatments, suggesting differential ability to maintain normal grain number, and (or) that grain yield per panicle was an important factor in response to stress. Maintenance of panicle number did not seem to be important for maintenance of yield under drought stress. The correlation of DRI and individual panicle yield was of sufficient magnitude for the latter to serve as a selection index in terminal stress.