Figure 3.

Fittest mutant adaptive path in a winding semismooth fitness landscape. An example on six loci of the winding semismooth fitness landscapes from SA C.2 on which the length of the path followed by fittest-mutant SSWM dynamics scales exponentially with the number of loci. Here, the black arrows are the fittest available mutation, and the adaptive path takes 14 steps to reach the fitness peak at 000011. For the generalization of this landscape to $2n$-loci, it would take $2^{n+1}-2$ steps for fittest mutant dynamics to reach the fitness peak at ${\left(00\right)}^{n-1}11$ (SA Theorem 20). Inset is the selection coefficient ($s\left(t\right)={\text{max}}_{y\in N\left(x_t\right){{\displaystyle \cup }}^{\text{}}\left\{x_t\right\}}\frac{w\left(y\right)-w\left(x_t\right)}{w\left(x_t\right)}$; SA D.2) vs. mutation step number (t) for the fittest mutant adaptive path.