Goodness-of-fit of different models and kinship matrices in explaining phenotypic variation of maize quantitative traits
Flowering time | Ear height | Ear diameter | |||||
---|---|---|---|---|---|---|---|
Method | Kinship matrix | −2 (ML) | BIC | −2 (ML) | BIC | −2 (ML) | BIC |
Simple | NA | 1632.8 | 1643.9 | 2296.0 | 2307.1 | 1282.6 | 1293.5 |
MM | SPAGeDi | 1524.3 | 1541.0 | 2237.7 | 2254.3 | 1254.2 | 1270.5 |
MM | Genotype similarity | 1527.5 | 1544.2 | 2243.1 | 2259.8 | 1266.6 | 1282.9 |
MM | Phylogenetic control | 1521.6 | 1538.6 | 2227.3 | 2243.9 | 1248.9 | 1265.2 |
SA | NA | 1525.7 | 1547.9 | 2248.9 | 2271.1 | 1276.9 | 1298.7 |
SA+MM | SPAGeDi | 1494.9 | 1522.7 | 2220.3 | 2248.1 | 1253.6 | 1280.8 |
SA+MM | Genotype similarity | 1500.9 | 1528.7 | 2227.1 | 2254.9 | 1266.5 | 1293.7 |
SA+MM | Phylogenetic control | 1491.6 | 1519.4 | 2213.2 | 2241.0 | 1248.2 | 1275.4 |
Comparison of the maximum likelihood (ML) and the Bayesian information criterion (BIC) of each model with different kinship matrices for maize quantitative traits. The model with the smaller BIC is preferred. Simple, the simple linear model without adjustment for population effect; SA, the model using the output from STRUCTURE as covariates; MM, the mixed model with different kinship matrices. The descriptions of kinship matrices are the same as in the Figure 1 legend.