Table 3 Analysis of two cycles of selection for resistance to ascochyta blight in Pisum sativum
Base ModelSecond ModelFull Model
REML LogLikelihood−3458.5−3301.0−3296.6
Significance of change between modelsP < 0.001P = 0.012
Fixed effects: Incremental Wald F statistic with 1 df
Site33.3c2.9 ns0.4 ns
Linear row cycle 13.2 ns3.5 ns3.9a
Linear row cycle 2102.4c105.2c105.4c
Linear range cycle 1126.9c41.8c44.9c
Linear range cycle 210.9c9.6b9.4b
Variance components for random effects:
Residual components
Variance cycle 14.730 ± 1.3673.774 ± 1.0153.802 ± 1.021
Range autocorrelation−0.053 ± 0.047−0.063 ± 0.050−0.062 ± 0.049
Row autocorrelation0.004 ± 0.046−0.013 ± 0.049−0.014 ± 0.048
Variance cycle 23.715 ± 0.3013.603 ± 0.2793.533 ± 0.267
Range autocorrelation0.167 ± 0.0460.167 ± 0.0400.173 ± 0.040
Row autocorrelation0.061 ± 0.0440.062 ± 0.0380.064 ± 0.039
Other spatial components
Row effects cycle 10.109 ± 0.0750.114 ± 0.0680.118 ± 0.069
Range effects cycle 20.124 ± 0.0760.096 ± 0.0610.096 ± 0.061
Blocks cycle 10.000 ± N/A0.039 ± 0.1110.033 ± 0.102
Blocks cycle 20.013 ± 0.2860.000 ± N/A0.000 ± N/A
Total genetic effects
Variance cycle 11.802 ± 1.367
Variance cycle 22.975 ± 0.408
Additive genetic effects
Variance cycle 11.864 ± 0.4252.098 ± 0.450
Variance cycle 22.336 ± 0.4082.286 ± 0.392
Correlation0.823 ± 0.111
Nonadditive genetic effects
Variance cycle 11.046 ± 1.0140.970 ± 1.019
Variance cycle 20.537 ± 0.3700.673 ± 0.352
Correlation0.673 ± 0.924
Narrow-sense heritability
Cycle 10.305
Cycle 20.352
Average accuracy of predictions
Cycle 10.5620.8940.815
Cycle 20.6960.8070.834
  • Based on ascochyta blight score (ABS) on single-spaced plants in partially replicated field experiments in cycle 1 (2010) and cycle 2 (2013). Three genetic models were fitted. The base model ignored genetic relationships and covariances between cycles. The second model included genetic relationships but ignored covariances between cycles. The full model included the additive and nonadditive genetic covariances between cycles. The models included fixed effects (experiment, linear row, and linear range effects) and random effects (additive and nonadditive genotype effects, genotype by environment effects, blocking structure, random row effects, and range and row correlations from the AR1:AR1 spatial model). Narrow-sense heritability is estimated in cycles 1 and 2 from the full model, ignoring the fact that the residuals are correlated, and average accuracy of predictions was calculated for genotypes present in each cycle for each model. ns, not significant; N/A, not available.

  • a 0.01 < P < 0.05.

  • b 0.001 < P < 0.01.

  • c P < 0.001.