Lower HPD of t includes 0: can we still estimate migration?

[Peter]

Dear all,

For some microsat data sets I get very recent divergence times (at a scale of decades) with HPDs that include zero. I would have thought that migration (in most cases also with lower HPDs of zero) would be impossible to distinguish from the retention of ancestral polymorphism in such cases, but we've just had a paper rejected because the reviewer seems to think that if there's a peak, there's a value. Would you agree that stating that our populations (or rather, different samples of one and the same population) have not diverged sufficiently to estimate any demographic parameters reliably is the better option in this case?