Fig. 3.

Model fit with and without aneuploidy. The distribution of time to fixation of $2n^{*}$ (i.e. adaptation time) in $10,000$ simulations using MAP parameters of the model with beneficial aneuploidy (blue; $\delta >0,w_{2n}<w_{2n+1}<w_{2n+1^{*}}<w_{2n^{*}}$) compared to alternative models: a model with the same parameter values but without aneuploidy (gray, $\delta =0$, concentrated at $t=450$); a model fitted to the data assuming no aneuploidy (green, $\delta =0$); a model fitted to the data assuming neutral aneuploidy (yellow, $\delta >0,w_{2n+1}=w_{2n},w_{2n+1^{*}}=w_{2n^{*}}$); and a model with beneficial aneuploidy and an extended prior distribution (pink). In the experiment by Yona et al. (2012), one population lost aneuploidy by generation 1,700 and another by generation 2,350 (dashed lines) but not before generation 450. Thus, the blue distribution has a better fit compared to the other distributions (the gray distribution has a particularly poor fit). The MAP likelihood (equation 4) is 0.84, 0.78, 0.67, and 0.14 for the models represented by blue, yellow, green, and pink distributions, respectively. A) Histogram of the time to fixation of $2n^{*}$. The last bin contains all values equal or greater than 3,000. B) Cumulative distribution of the time to fixation.