Table 3. Cross-validation correlations obtained with BRR and BayesB models by trait and model.

				Proportion of times that the model in row gave a higher correlation than the model in columns
		CV-Correlation		BRRa			BayesBa
Priora	Model # (label)b	Averagec	SDd	M5 (A)	M6 (A+D)	M7 (G)	M5 (A)	M6 (A+D)	M7 (G)
Late Blight
BRR	M5 (A)	0.258	0.023		0.96	0.00	0.33	0.91	0.00
	M6 (A+D)	0.241	0.023	0.04		0.00	0.04	0.57	0.00
	M7 (G)	0.312	0.017	1.00	1.00		1.00	1.00	0.5
BayesB	M5 (A)	0.260	0.024	0.67	0.96	0.00		0.94	0.00
	M6 (A+D)	0.240	0.024	0.09	0.43	0.00	0.06		0.00
	M7 (G)	0.313	0.017	1.00	1.00	0.50	1.00	1.00
Common Scab
BRR	M5 (A)	0.268	0.025		0.81	0.99	0.1	0.55	1.00
	M6 (A+D)	0.259	0.023	0.19		0.99	0.07	0.20	0.99
	M7 (G)	0.218	0.022	0.01	0.01		0.00	0.02	0.63
BayesB	M5 (A)	0.278	0.026	0.9	0.93	1.00		0.91	1.00
	M6 (A+D)	0.265	0.025	0.45	0.8	0.98	0.09		0.98
	M7 (G)	0.216	0.022	0	0.01	0.37	0.00	0.02

^aBRR uses a Gaussian prior for effects, BayesB uses a prior that has a point of mass at zero and a scaled-t slab.

^bA: Additive model, A+D: additive+dominance; G: general model (with up to 4 degrees of freedom per locus).

^cAverage from 100 cross-validations.

^dStandard deviation.