Table 3. Output on adding extraneous error component to the selected Model 1 or Base for DM, FYLD, SHTWT, and HI.

Data	Trait	Model	Variance					h2	LLk	χ-square
			${\sigma }_g^2$	${\sigma }_s^2$	${\sigma }_r^2$	${\sigma }_R^2$	${\sigma }_C^2$
Ibadan_2013_C1	FYLD	Base	7.425	NA	19.791	NA	NA	0.7	−1359.31	21.34
		Model2	6.876	NA	18.171	NA	2.049	0.68	−1348.64	(3.80E−06)
	SHTWT	Model1	9.022	1.729	11.875	NA	NA	0.59	−1275.65	6.48
		Model2	8.686	1.763	11.419	NA	0.645	0.6	−1272.41	(0.01)
	HI	Base	0.005	NA	0.009	NA	NA	0.44	1092.16	7.89
		Model2	0.005	NA	0.009	NA	0.001	0.44	1096.1	(0.005)
Mokwa_2013_C1	FYLD	Model1	1.649	1.113	19.218	NA	NA	0.86	−1475.81	2.88
		Model2	1.362	1.101	18.609	NA	0.824	0.87	−1474.37	(0.089)
	HI	Base	0.005	NA	0.009	NA	NA	0.62	1270.87	10.15
		Model2	0.005	NA	0.009	0.001	NA	0.66	1275.94	(0.001)
Mokwa_14_C1	HI	Base	0.007	NA	0.006	NA	NA	0.59	561.33	3.66
		Model2	0.006	NA	0.005	0	NA	0.54	563.15	(0.055)
Ikenne_2014_C1	DM	Base	31.949	NA	0.719	NA	NA	0.44	−579.46	21.87
		Model2	36.116	NA	0.033	NA	1.379	0.46	−568.52	(2.9e−06)
	SHTWT	Base	721.089	NA	29.191	NA	NA	0.5	−1109.42	3.27
		Model2	752.485	NA	15.792	10.999	NA	0.49	−1107.79	(0.071)

Base is the model having only the genetic variance component. Model 1 has a spatial variance component in addition, and Model 2 has extraneous error component added to the best model selected between Model 1 and Base for a particular trial and trait. Narrow sense heritability (h2) is calculated from BLUP values and genotypic variance. Chi-square statistic is calculated from the log likelihood values (LLk) of the Base/Model 1, and selected Model 2 is given with p-value in brackets. The table shows results from trial-trait analysis with significant improvement in model fit of Model 2 over the best of Base/Model 1 at α = 0.1.