Fig. 2. Heritability estimates for all nine actual datasets with three BLUP and five Bayesian methods.

The heritability is estimated by following a repeated fivefold cross-validation approach, where the experiment is repeated 100 times. The heritability estimates for the wheat yield datasets were higher for the BLUP methods. However, the Bayesian methods achieved higher heritability estimates for the rest of the five traits. Heritability is computed as the ratio of the additive genetic variance to the phenotypic variance. For the BLUP alphabets, all markers are used to estimate additive genetic variance, whereas only the markers with non-null effect are used for the Bayesian methods.