ENVIRONMENTS FOR SELECTING WIDELY ADAPTED SPRING WHEAT

Breeding widely adapted wheat (Triticum aestivum L.) genotypes with stable and high yields across environments is particularly important for developing countries since yield stabilizing inputs are often limited or not available. To evaluate the screening ability of locations for identification of such genotypes, data collected for 19 yr by the International Spring Wheat Yield Nursery (ISWYN) were analyzed; 1221 trials at 268 locations in 69 countries were involved. To compare single-experiment parameters, i.e., genotypic variance (sigma(g)2)k, error variance (sigma(e)2)k, heritability (h2)k, and coefficient of variation (CV)k, trials without major biotic stresses were divided into three groups according to mean grain yield. Genotypic variance, error variance, and heritability increased and (CV)k, decreased with yield. Group means for the four parameters were significantly (P = 0.01) different. A fourth group containing trials with major biotic stresses had the highest, but not significantly higher, average estimates for (sigma(g)2)k, (h2)k, and (CV)k. The screening ability for each location was calculated as the correlation, r(k), between mean grain yield of genotypes at each location and mean yield across locations. The screening ability was highest for locations with no major abiotic and biotic stress factors apart from leaf rust (Puccinia recondita Roberge ex Desmaz. f. sp. tritici) and stem rust (P. graminis Pers.:Pers. f. sp. tritici Eriks. & E. Henn.). CIMMYT's principal test site, Ciudad Obregon, Sonora, Mexico, was most suitable for screening, with an average r(k), of 0.77. Sensitivity to photoperiod, cold tolerance, need for late maturity, tolerance to problem soils, and resistance to diseases other than rusts were the main adaptation-limiting and location-specific factors.