Use of GGE Biplot to Evaluate Interaction of Cotton Varieties with Environments

Cotton seed companies use multi-environment trial (MET) data to determine which experimental strains are advanced towards commercialization. Almost invariably the interaction of genotypes (strains and checks) with environments (locations and years) is significant for lint yield, but, more often than not, breeders simply rely upon over-location and over-year means to drive advancement decisions. Since 2003, PhytoGen Seed Company has used GGE Biplot as a tool to gain an understanding of the nature of GxE interactions observed in the Mid-South and Southeast cotton trials. The “Mean vs. Stability” view of GGE Biplot pivots on the average tester coordinate (ATC) and the deviation from it indicates the degree of both positive and negative interactions between genotypes and environments. A simplistic approach to discerning the primary interaction element of MET is to group the environments on each side of the ATC for averaging. Combined with the breeder’s knowledge of the test environments and of the experimental strains and checks, these differential entry means can be highly informative regarding the nature of the GxE interaction. For the past several years, the maturity of test genotypes has been correlated with deviation from the ATC. In 2007, the percent first harvest at Leland, MS had a correlation of 0.48 and -0.43 with entry means that were averaged across the locations either above or below the ATC, respectively. Even though the maturity of genotypes is correlated with deviation from the ATC, the latitude of locations is not necessarily so well correlated.