We simulated positive residual correlations between the traits and studied scenarios with a relatively high and low residual correlation (rEjj = 0.7 and rEjj = 0.3, respectively). The QTL was fixed to explain 0.1% of the trait variances. By varying the sign of a1, we created a QTL induced correlation (rG) between trait 1 and traits 2 and 3 which was either positive or negative, enabling us to study the influence of a negative genetic correlation.
