Intriguing. They are probably going on a presumed rate of
mutation to make the estimate. There is some controversy about
whether we can rely on a steady rate of mutation in maing these
estimates, as there could be a more erratic pattern.

The abstract doesn't mention when type O gene(s) would have
probably developed, although it hints that it might be in
the full length article.

If our genome as Homo sapiens has probably been
pretty stable for 100,000 + years, it would suggest that
the spread of one blood type or another is more a matter
of survivorship, without really seeing 'new' genes, just
local selection among the genes available.

So we could theoretically, and do in fact,
find geographical areas in which most or all of the
B's thrived while the A's died out, and so on. It could
explain the predominiantly type A native American group.