TABLE 5

The effect of cofactors on the power of detection of QTL × E interaction (βe), and accuracy of QTL location (L)

NModelh2%DgeiL(cM)α%→5βe (%) 10.1
112.40.10162.9 ± 1.6645 (70)27 (58)15 (42)
259.9 ± 0.7496 (96)91 (90)76 (76)
359.2 ± 0.4999 (99)99 (99)96 (97)
3150.01560.2 ± 0.7219 (76)8 (57)1 (32)
259.3 ± 0.3742 (56)21 (37)6 (13)
359.6 ± 0.2241 (40)23 (23)4 (5)
5120.00662.6 ± 1.5214 (62)4 (41)1 (23)
260.5 ± 0.8520 (32)7 (16)2 (4)
359.9 ± 0.7118 (19)8 (8)1 (1)
  • The power of the test was obtained using Monte-Carlo simulations (see text); (corresponding results based on χ2 asymptotic distribution of the test statistic are given in brackets). Data of the situation S6 (see Table 1) were used with the target QTL on chromosome N (N = 1, 3 or 5). Three models of the analysis of the residual variation were employed: (1) the cofactors are totally ignored; (2) the effect of the strongest QTL is fitted using two-QTL mixture model; (3) the genetic component of the residual variation is replaced by the equivalent nongenetic variation. Dgei is the variance of QTL × E interaction for the target QTL; h2% is the averaged heritability over environments attributed to the target QTL.