I think extent-based (distribution-range-based) evaluation could be a solution.
Change in distribution range has been a candidate for calculation of indicator 2 in the earlier stage of developing the indicator.
Though it requires additional input for Kobo, the benefit of this approach is that it can avoid difficulties in defining populations.
Another benefit is that it can avoid confusion about differences in the definition of "populations" for calculating indicators 1 and 2.
(consideration for contemporary (indicator 1) vs historical (indicator 2) gene flow)