Abstract
Eight sweet potato cultivars were grown at three locations during four seasons, and five distinct measurements of yield were taken. Analysis of variance revealed significant differences between varieties, locations, and seasons. Regression analysis of varietal means on overall means for the 12 locations permitted the identification of the most yield-stable cultivars. Two novel techniques were used to find a key environment among the 12. In the first, the mean of each environment was regressed on the means for all environments using the eight cultivars as test points. In the second, the means for each environment were compared with the means of all environments by correlation. In both cases, one environment was identified as a key environment. Comparison by ranking showed that results from the key environment identified were closely related to those of the 12 environments, and thus analysis in one key environment would have revealed essentially the same information as analysis of 12 environments. The technique should be useful with other crops as well.
