Figure 2.

Schematic drawing of the error threshold. If a fitness landscape has a positive minimum fitness (case A), then at a sufficiently high mutation rate all individuals are pushed to this minimum level. The selective strength on the narrow peak is not sufficient to counteract the mutation pressure. If a fitness landscape has no minimum fitness (case B), then the mutation pressure pushes a large fraction of the population to zero fitness. The individuals with zero fitness (shown in gray) are inviable, and thus do not compete with the individuals on the fitness peak. Therefore, a few individuals will always remain on the top of the fitness peak. Note that this conclusion holds only when two assumptions are met: (i) The population is infinite. (Otherwise, stochastic effects push the population away from the peak, and we observe Muller's ratchet.) (ii) Selection is soft, that is, only relative fitness differences matter, and the overall population size is held constant at all times. (If selection is hard, then the population size will decline as the mutation rate is increased, and eventually the population can go extinct. This case is mutational meltdown.)