Mega-environment analysis and breeding for specific adaptation

Mega-environment (ME) analysis is analysis of multi-year, multi-location crop variety trial data conducted in a target region of a crop to understand the magnitude and nature of genotype-by-environment interaction (GE) of the crop in the region. If repeatable GE patterns are identified, then the target region must be divided into subregions or MEs. Breeding and utilizing ME-specific cultivars will convert the repeatable GE into genotypic main effect (G) within ME, thereby improving heritability (selection reliability) and selection gain and maximize regional and overall productivity. If no repeatable GE is found, then the target region must be treated as a single ME and the GE must be accommodated by testing adequately, that is, at a sufficient number of locations in a sufficient number of years. This paper presents a theoretical framework of ME analysis, describes graphical tools to reveal the which-won-where patterns in a genotype-by-environment dataset, and demonstrates LG (location-grouping) biplot analysis for revealing repeatable GE patterns and delineating MEs. The concept of G + GE or GGE, that is, GE relative to G, is emphasized. It is the relative GE that is the basis for ME analysis and breeding for specific adaptation; absolute magnitude of GE has little relevance for these purposes. Breeding ME-specific oat cultivars in Canada is demonstrated with a real-world example.