| When the environments are heterogeneous, the region can be subdivided into sub-regions, with more homogenous conditions that will reduce the genotype x location interaction. A sites regression model was used to study the association among locations and their representativeness for loaf volume (LV) and grain protein content (GPC). The optimum combination of locations and years to achieve a desired precision for each quality trait was calculated. A genotype x trait biplot was used to study the association between 12 quality traits, looking for highly associated ones. The data used in this study were from the state of North Dakota had red spring wheat (Triticum aestivum L.) unreplicated large drill strips conducted in seven locations across North Dakota from 2001 to 2004. This study showed that: (1) the groups of locations for LV and GPC were non-repeatable across years; therefore, they can be treated as a single complex mega-environment; (2) Langdon site was the least representative location of the overall environment; thus, it could be dropped from the testing sites; (3) if dropped, the use of six locations and three years would not reduce the precision; (4) GPC could be used in early generation testing to increase wet gluten content, water absorption, and decrease mixing time; (5) rainfall after flowering (Rainf) was the main factor behind specific adaptation, and average maximum temperature (maxTf) for general adaptation of cultivars for GPC; (6) the location Langdon could be used to screen for cultivars adapted to high Rainf and low maxTf; (7) Rainf was associated with general adaptation for LV. |
