Figure 2.

Multiple environments. Effect of including phenotypic data from multiple environments before modelling. Along the x-axis is the number of environments used in the pre-processing of phenotypic data, and the y-axis is the fraction of variance explained. For each number of environments, all possible permutations of the available environments were tested. Each permutation was evaluated by a 50-permutation bootstrap of a whole-genome model fitted using ridge regression. Dotted lines are 95% confidence intervals for the mean derived using the t-test.