Figure 1.
Figure 1.

Model fitting of simulated populations with different parameters, independent runs. The −2 residual log-likelihood values of mixed models with different relative kinship (K) matrices and variance ratio [Vg/(Vg + VR)] estimates. Dashed lines represent the simulated heritability (h2). QTL, quantitative trait loci.

 


Figure 2.
Figure 2.

Model fitting of simulated individual populations with different relative kinship (K) with quantitative trait loci (QTL) = 20 and m = 400, multiple resampling at intermediate proportions. (a–d) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different K matrices. (e–h) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars. h2, heritability.

 


Figure 3.
Figure 3.

Model fitting of simulated individual populations with different relative kinship (K) with quantitative trait loci (QTL) = 50 and m = 800, multiple resampling at intermediate proportions. (a–d) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different K matrices. (e–h) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars. h2, heritability.

 


Figure 4.
Figure 4.

Frequency distribution of polymorphism information content (PIC) for two types of molecular markers. (a) PIC values of 912 single nucleotide polymorphisms (SNPs) and 89 simple sequence repeats (SSRs) scored on 274 diverse maize inbred lines; (b) PIC values of 471 SSRs scored on 266 crossbred dogs.

 


Figure 5.
Figure 5.

Model fitting for maize quantitative traits with both population structure (Q) and relative kinship (K) being estimated with different numbers of markers. (a–c) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different Q and K estimates based on different proportion of the whole set of background markers (912 single nucleotide polymorphisms [SNPs] or 89 simple sequence repeats [SSRs]). (d–f) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars.

 


Figure 6.
Figure 6.

Model fitting for maize quantitative traits with population structure (Q) being constant but relative kinship (K) being estimated with different numbers of markers. (a–c) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different K based on different proportion of the whole set of background markers (912 single nucleotide polymorphisms [SNPs] or 89 simple sequence repeats [SSRs]). The constant Q was estimated with the whole set of 89 SSR markers. (d–f) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars.

 


Figure 7.
Figure 7.

Model fitting for maize quantitative traits with relative kinship (K) being constant but population structure (Q) being estimated with different numbers of markers. (a–c) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different Q based on different proportion of the whole set of background markers (912 single nucleotide polymorphisms [SNPs] or 89 simple sequence repeats [SSRs]). The constant K was estimated with the whole set of 912 SNP markers. (d–f) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars.

 


Figure 8.
Figure 8.

Model fitting for canine dysplastic traits. (a, b) Bayesian Information Criterion (BIC) and −2 residual log-likelihood values of mixed models with different relative kinship (K) matrices based on different proportion of 471 simple sequence repeats (SSRs). (c, d) Variance ratio [Vg/(Vg + VR)] estimates from mixed models. Standard deviations are shown by vertical bars. DI, distraction index; DLS, dorsolateral subluxation.