Figure 7. Mean number of parallel divergent SNPs and orientation skew for nine simulation models.

Confidence intervals are calculated from 100 independent datasets generated under each model. The empirical mean values are significantly higher than the simulated mean values across all models. A moderate excess of parallel divergent SNPs, although not as high as the empirically observed parallel divergence, was generated by three models: constantlow, 3tightestbottle, and OC-EAmig100. These models represent extreme demographic scenarios that are unlikely to accurately approximate the histories of the populations analyzed. Descriptions of the simulated models and their results can be found in Table S4. (A) Mean number of parallel divergent SNPs (B) Mean orientation skew (frequency of major orientation).