Figure 3.

Prediction accuracies and marker effect distributions across all 25 NAM families. [A] Prediction accuracies of $\widehat{\mathit{a}}$, ${\widehat{\mathit{a}}}_1$ and ${\widehat{\mathit{a}}}_2$ are shown as the correlations between the predicted and observed trait values from the joint and independent analyses of 25 NAM families. [B] P and S distributions estimated from the independent analyses were tested using Kolmogorov-Smirnov test and the results are shown as -log₁₀(p). The Bonferroni adjusted threshold of P = 0.05/25 is shown as a red horizontal line. [C] Correlations between ${\widehat{\mathit{a}}}_1$ and ${\widehat{\mathit{a}}}_2$ represent the potential constraints when selecting for both P and S markers in OSGS, although the lack of strong negative correlations suggests these are small.