Contents - Index

Mega-environment classification based on the QQE pattern


Previous: Determining the number of QTL


From the above biplot, it is already obvious that the environments fall into two groups: the Eastern (on the right) and the Western (on the left) mega-environments. Change the singular value partitioning (SVP) to environment-focused (SVP =2), which is appropriate for visualizing the similarities among test environments, this pattern becomes even more obvious:



This mega-environment classification is strikingly similar to that based on the GGE biplot of the yield data per se, suggesting that the GGE pattern is well explained by the QQE pattern, which in turn suggests that barley yield can be effectively manipulated by selecting for these QTL. 


Marker assisted selection in the Eastern mega-environment

The following QQE biplot contains only environments from the eastern provinces of Canada: Manitoba, Ontario, Quebec, and Prince Edward Island. It can be seen that the Harrington alleles of mwg502 and mwg655c, and the Tr306 alleles of mwg914, bcd269, and abc261 had positive associations with barley yield in all environments in this mega-environment. Therefore selecting for appropriate alleles of these 5 QTL should improve barley yield in this mega-environment. QTL represented by mwg838 and mwg626 had inconsistent effects on yield and therefore can not be used in marker assisted selection in this mega-environment.

Note: The cosine of the angle between the vector of a QTL and that of an environment determines the direction of the effect of that QTL in the environment.


Marker assisted selection in the Western mega-environment

The following QQE biplot contains only environments from the western mega-environment, including Canadian provinces Alberta and Saskatchewan and some states of the USA, such as Alaska, Washington, Montana, and North Dakota. Unlike the Eastern mega-environment, no QTL had consistent effects on yield in this mega-environment. One exception is that (the Harrington allele of) the QTL represented by mwg838 had no effects or negative effects in all environments within this mega-environment and, therefore, may be useful in marker-assisted selection.


Note: Other utilities of the QQE biplot.