Notes. Each layer (each curve) is summarized by the average fitness over a run, yielding a sample of twelve values, one value for each of the twelve runs of a layer. We then compare the fitness curves for all possible pairs of layers, using a two-tailed Welch t-test for samples with unequal variance (heteroscedastic variance). All the pairs of curves in Figure 6 are significantly different, except for Layers 2 and 3 (the variable asexual layer and the sexual layer).