Actually without herbicide, resistant strains are less fit than wild-types, so the rare variant seems to be a disadvantage. Hence our suggestion that “balancing selection” mechanisms may keep it at low frequency for rapid adaptation to changing environments!

Thanks for putting it simply! That's correct and most interestingly the "rare gene" (or rather a rare variant of a gene) was already present in "normal cyanobacteria", just at a low frequency (they have multiple genome copies)...So they are keeping a backup of rare variants for times of crisis!