Allelic dropout is a crucial problem when working with historical and ancient specimens and although I appreciate the justifications given by the authors in their reply letter I think they should acknowledge that in the main text as a potential problem, even though they feel this is not affecting their results.

Line 148: The authors mention they end up with 25 historical sequences. Did you use the sequences previously published by Rey Iglesia et al 2021? I do not see any historical samples from Madeira in the dataset. It could be a good idea to add those samples to your data, to increase the information fed to the ABC simulations.
We thank the reviewer for his/her relevant suggestion. We originally considered this option. However, the available almost-full mitochondrial genome alignments (bam files) supporting the Rey-Iglesia et al., 2021 paper did unfortunately not include the d-loop region which contains the CR1 (the mitochondrial gene fragment that we are using).

On the contrary, the information is on figure 2A of Rey-Iglesia et al, all samples from Madeira, as well as every other sample processed, both modern and ancient were typed for the 524bp control region fragment used in this study. They all had haplotype MM05. So, this information should have been included in the ABC simulations as it is extremely relevant that this population displayed the same haplotype for all typed specimens, modern and ancient.

Line 203 – “The microsatellite population-structure analyses exhibited, at best fitting K of 2-4 (Figs 1, S14- 15), a strong East-West pattern of differentiation (FST > 0.4, Table S6-7) separating the Atlantic and Mediterranean populations. “

Here I think the authors should just refer to their best fitting K of 4 (??; I guess it is based on figure 1) and not all the values of K where a Mediterranean/Atlantic separation is noticeable. I think it is confusing to mention all three values. With this info the reader still does not know which is the best fitting K for this data.

Line 245 – “In 1997 the Cabo Blanco colony underwent a massive die-off (>2⁄3 of the population), from different potential causes.”

This sentence reads strange it lacks information, as it does not inform the reader what might have cause the die off. Consider editing “… massive die-off caused either by saxitoxins (ref) or by a morbillivirus outbreak (ref) ( over two thirds of the population was wiped out).”

Line 421

I still think Rey-Iglesia et al clearly show connectivity between the two basins, as per shared haplotypes between the two and the authors should refer to this in their manuscript.