GRAIN YIELD AND YIELD STABILITY OF WINTER WHEAT CULTIVARS IN CONTRASTING WEATHER CONDITIONS

Increasing yield without sacrificing yield stability is a major challenge for wheat breeding. This paper presents the yields and several stability parameters of modern Romanian winter wheat cultivars, tested in very diverse weather conditions in South Romania, and compares the information provided by various approaches. Fourteen Romanian winter wheat cultivars were tested along with the long term check Russian cultivar Bezostaya 1, in yield trials in 6 locations during 2002-2007, totaling 52 testing environments. The diversity of conditions included in the study is reflected by the large variation of average yields of the 15 cultivars in the 52 environments, which varied from only 291 kg/ha to more than 8000 kg/ha. Average yields were significantly correlated with water availability (rainfall 9 irrigation), suggesting that water stress was a major factor that influenced yield variation. Average yield of each cultivar correlated significantly with all calculated stability parameters, except minimum yield, the coefficient of variation (CV%) and the regression intercept (a). Plotting phenotypic variances, or coefficients of variation, against average yields allowed identification of cultivars Izvor, Delabrad, Gruia, Faur and Glosa, showing smaller yield variation than expected based on their average yield. Ecovalence (W(2)) did not describe yield stability, but characterized the similarity of response with the average of all tested cultivars. The regression analysis identified cultivars adapted only to favorable conditions with b > 1 and lower value of a (Dor and Alex) and cultivars adapted only to less favorable conditions with b < 1 and higher value of a (Bezostaya 1) Among the cultivars with wide adaptability, with b close to 1 and relatively large a, it was possible to identify those that better use good environments, with b > 1 (Glosa and Gruia), and those with better adaptation to lower yielding environments (in this case drier environments), like Izvor and Delabrad. Deviations from regression (delta(2)) were almost perfectly correlated with the regression slope (b), suggesting that cultivars with higher regression coefficients had different response patterns to the environment than most tested cultivars. Our study suggests that none of the used methods is sufficient for characterizing yield stability and for describing the specific response of each cultivar to the environmental variation. Plotting CV against average yield proved to be most useful in identifying cultivars with high and stable yield. Additional information was obtained by analyzing minimum yields as well as slopes and intercepts of cultivar regression lines on average yield of each trial. Our results showed that high-yielding cultivars can differ in yield stability, and suggest that yield stability and high grain yield are not mutually exclusive.