For those who might be interested I had a go at remaking Figure 1 but including BA.2.86 (as the most extreme example of a second generation BA.2 lineage). As you can see it really stretches what its possible to show in these types of schematics. Also notable among all these lineages is the massive overrepresentation of non-synonymous relative to synonymous mutations (mostly absent except in BA.2.86 which has just 1).
