Table 5.

REML output of the fit of different mixed models to the CIMMYT maize stress trials.

Model 7			Model 8			Model 9
Fixed	Wald (DF)	P	Fixed	Wald (DF)	P	Fixed	Wald (DF)	P
E	10265.3 (7)	<0.001	E	9759.4 (7)	<0.001	E	6268.8 (7)	<0.001
Random	Estimate	SE	Random	Estimate	SE	Random	Estimate	SE
σ2G	0.297	0.036	σ2G	0.125	0.017	σ2C1	0.439	0.053
σ2ε	0.553	0.020	σ2ε1	0.551	0.057	σ2C2	1	–
			σ2ε2	0.692	0.071	σ2C3	0.042	0.013
			σ2ε3	1.399	0.140	σC1C2	0.551	0.077
			σ2ε4	0.672	0.069	σC1C3	0.109	0.019
			σ2ε5	0.704	0.072	σC2C3	0.115	0.032
			σ2ε6	0.135	0.018	σ2ε1	0.446	0.051
			σ2ε7	0.152	0.019	σ2ε2	0.445	0.052
			σ2ε8	0.761	0.078	σ2ε3	0.736	0.169
						σ2ε4	0.428	0.050
						σ2ε5	0.508	0.057
						σ2ε6	0.145	0.018
						σ2ε7	0.138	0.017
						σ2ε8	0.740	0.080
Deviance (DF)	1077.9 (1678)		Deviance (DF)	838.4 (1671)		Deviance (DF)	619.9 (1667)

Model 7 assumes compound symmetry, model 8: assumes heterogeneity of genetic variance across environments, and model 9 assumes heterogeneity of genetic covariance between groups of environments and heterogeneity of genetic variance across individual environments. Environments are indexed as: 1 = SS92a, 2 = IS92a, 3 = NS92a, 4 = IS94a, 5 = SS94a, 6 = LN96a, 7 = LN96b, 8 = HN96b. Groups of environments are indexed as: C1 = SS92a, IS92a, IS94a, SS94a, HN96b; C2 = NS92a; C3 = LN96a, LN96b.