LG biplot: a graphical method for mega-environment investigation using existing crop variety trial data

Abstract

Due to the presence of genotype by environment interaction (GE), no crop cultivar performed the best in all regions. Therefore, the growing regions of a crop must be divided into sub-regions or mega-environments, and specifically adapted cultivars must be bred and deployed in each mega-environment. Meaningful mega-environment delineation must be based on repeatable GE patterns, which can be extracted from multi-year, multi-location crop variety trials. In regional crop variety trials, usually the same set of genotypes are tested across locations within a year, but different sets of genotypes are tested in different years, leading to highly unbalanced multi-year data. Such data are abundant for all crops and regions; but there has been no way to fully utilize them for mega-environment delineation. This paper presents a new method that allows utilization of existing variety trial data to identify repeatable GE patterns, to delineate mega-environments, and to understand the scope of unrepeatable GE at a location and within a mega-environment.