The paper decribes using ddRADseq and RDA to understand more about the distribution of beech, and how this may be affected under climate change. The ms is pretty good and of interest but I have quite a few comments on it as-is that I think need addressing in a revision:

Page 2, line 50 - "must"? maybe "is predicted to" is better
Spell out acronyms in the abstract and first time of mentioining in the Intro

Methods - Im not sure what is meant by 570 individuals in 12 libraries. This sounds like individuals weren't barcoded individually, they were pooled. I think maybe just the nomenclature needs checking.

Very rare alleles were included even after cutoff. Typically one might use 5% not 0.5% as a cutoff. And very rare alleles can affect popgen parameters. I wonder what happens if these very rare SNPs are excluded and a more traditional 5% cutoff is used?

P11,L230 - is this correct? A different q value is used to the section above?

Results
Are fragments the same as 'stacks'? Why 569 here and 570 previously?

L309 - now a different minor allele cutoff is given, not 0.5% which was stated above

Fig S5 - I'm a bit confused about S5A - why is it an unfolded SFS - is this just based on frequency relative to the reference allele? Surely a SNP at frequency 0.1 is 'the same' as one at 0.9?

Line 361 - These are frequencies not percentages I think (eg 0.5 = 50% not 0.5%)

L373-6 - I've not seen this done before and I didnt really see the point, until I got to the next section. Can you move this information into the next section?

Fig 2 - how was 'north' and 'not north' assigned to the individuals. what cutoff was used in their lat/long? Was this in the text? I dont think so.

More generally:

Methods and results are lengthy. I strongly suggest these are cut back and become more focussed.

Why was scenario A1B picked?

Some improvement of the text could reduce the confusion (eg above) but there are also some phrases that could be ambiguous, eg:
P3,L36 "Better assessing the... " - i think you mean "A better assessment of...."
P10,L196 - "extremely" linked (they are either linked or not, they cant be extremely linked)
L479 "The first adaptive genetic variation identified" - I am not sure what this phrase means

Did you do any control for linkage among SNPs from different loci or for multiple SNPs within loci? There should be some level of QC here

I have only the following minor comments to make, the authors have done a very good job with the ms.

Inclusion of rare alleles - the authors make a sound argument for the settings chosen
q-value - apologies, I read 0.01 and 0.0001 instead of both being 0.0001. Thank you for changing both to 10-4
fragments - I think this is better now. I am still unsure about the word fragment because it is not clear what a fragment is. It might help to first clarify how many reads per individual (mean +/- SE) were obtained, and then say 'upon grouping the reads into fragments....' or something.
Fig 5A - apologies, IO am not sure what I was referring to, I must have misunderstood
Fig 2 - I was referring to grey vs black dots in fig 2A, not the barplot and the legend saying "black points coming from northern populations and grey points coming from the populations close to the Mediterranean Sea." - so there must have been some way of defining what these two halves were. They are a pretty continuous distribution in fig 1, so where is north and where is Med? Or are these colours based on ancestry? eg black points are >50% cluster 1 and grey points >50% cluster 2? If so, then the barplot needs to go before the PCA.
Picking scenario A1B - I am not convinced this is a good enough reason, but its up to the editor. Picking the most extreme scenario will give the most significant results, and hence "...simulations of future genetic landscapes that used these findings predicted a severe range contraction" (in the abstract) might want to be tempered with "under an extreme scenario of warming", but I appreciate that the change was made in the methods.