Figure 4.

Variance ${\sigma }_1^2$ of a focal trait (cyan; simulated circles and Equation 3a solid line), the magnitude of its covariance $\left|{\sigma }_{1,2}\right|$ with the other trait (magenta; simulated circles and Equation 3b solid line), and the trait’s contribution to adaptation ${\mathit{\upsilon }}_1$ (blue; simulated circles and Equation 1 solid line), according to (A) selection coefficient, (B) mutation rate, and (C) population size, averaged over 1.5 × 10⁶ generations. The y-axis is normalized relative to what the variance of the focal trait would be in the absence of the second trait; observed variance is always greater than this. While on its own, increased variance would accelerate adaptation, negative covariance more than cancels this out $\left(\left|{\sigma }_{1,2}\right|>{\sigma }_1^2-\mathit{\upsilon }\left(U,N,s\right)\right)$ for a net reduction in the trait-specific adaptation rate below the value of $\mathit{\upsilon }\left(U,N,s\right)$ that would be seen in the absence of clonal interference. For the parameter values not being varied on the x-axis, s = 0.02, U = 10⁻⁵, and N = 10⁹.