A Two-Stage Method for Improving the Prediction Accuracy of Complex Traits by Incorporating Genotype by Environment Interactions in Brassica napus

Table 2

The decrement and its percentages of the MSE between the null model and the alternative model.

Trait	Type	Null model (intercept)	Alternative model (adding components as below)
			E	Q	Q × E	E + Q	E + Q + Q × E
FT	MSE^a	198.4734	17.0725	196.5320	9.1477	3.3914	2.6336
	Decrement^b	—	181.4009	1.9414	189.3257	195.0819	195.8397
	% of decrement^c	—	91.3981	0.9781	95.3910	98.2912	98.6731
OC	MSE	6.2661	4.4863	4.1189	1.6027	1.5870	0.7472
	Decrement	—	1.7798	2.1473	4.6635	4.6792	5.5189
	% of decrement	—	28.4040	34.2680	74.4235	74.6738	88.0751
SY	MSE	0.3503	0.2259	0.3284	0.1360	0.1331	0.1167
	Decrement	—	0.1245	0.0219	0.2143	0.2172	0.2336
	% of decrement	—	35.5276	6.2476	61.1769	62.0056	66.6784

^aMean square error; ^bthe decrement of MSE between the null model that only includes the intercept term and the alternative model that includes both intercept and the corresponding components; ^cthe percentage of the decrement of the MSE between the null model and the alternative model or, equivalently, the adjusted R-squared; FT, flowering time; OC, oil content; SY, seed yield per plant; E, the environment effect; Q, the main effect of locus; QE, the interaction effect between the locus and the environment.