TABLE 4

Estimates of QTL allele effects

Method (%) α 1 α 2 α 1 α 2 α 1 α 2
Values simulated
0.300.300.300.000.000.30
Single-trait analysis
LR1000.30 (0.003)0.30 (0.003)0.30 (0.003)0.00 (0.003)0.00 (0.003)0.30 (0.003)
LR100.31 (0.005)0.43 (0.007)0.31 (0.005)0.28 (0.008)0.00 (0.004)0.17 (0.007)
Multiple-trait analysis
LR1000.30 (0.003)0.30 (0.003)0.30 (0.003)0.00 (0.003)0.00 (0.003)0.30 (0.003)
MG1000.30 (0.003)0.30 (0.003)0.30 (0.003)–0.00 (0.003)– 0.00 (0.003)0.30 (0.003)
LR100.30 (0.005)0.29 (0.009)0.30 (0.005)0.00 (0.009)– 0.01 (0.005)0.30 (0.008)
MG100.28 (0.004)0.25 (0.007)0.28 (0.004)0.02 (0.007)– 0.01 (0.005)0.29 (0.008)
BS100.25 (0.008)0.01 (0.008)0.29 (0.009)

Estimates are means and standard errors of 100 replicates, where estimation is by either logistic regression (LR), the method of Muranty and Goffinet (1997; MG), or the method of Bovenhuis and Spelman (1998; BS), with either all animals genotyped (LR100 and MG100) or 10% of animals genotyped (LR10, MG10, and BS10). α1 is the effect of the first trait, and α2 is the effect of the second trait. Where applied, selective genotyping was on the phenotype of the first trait. The within-QTL genotype variance was 1.0, and the within- QTL genotype covariance between the traits was 0.5. Correlations used in the analyses were estimated from the complete data in each replicate.