General comments: This paper presents 29 new 10Be and four 36Cl ages from the relatively under-studied regions in the northern Alexander Archipelago, Alaska. The paper is well-written and presents some interesting results related to the deglaciation of the Cordilleran Ice Sheet. For instance, the fact that previously hypothesized ice-free regions of the southern Alexander Archipelago throughout the local Last Glacial Maximum (local-LGM; 20-17 ka is now rejected by their findings. Their results indicate that these sites, and more generally the coastal Alexander Archipelago, became ice-free by 16.3 and 15.1 ka, from south to north. This was contradictory to the previous findings. Most of the age constraints come from well-dated boulder and bedrock samples from 10Be and 36Cl ages. The paper definitely deserves publication in the Geochronology journal, but requires some minor revisions, as suggested below.

Dear Dr. Sarikaya,

We thank you for taking the time to serve as a reviewer for our manuscript and appreciate your suggestions. We discuss these and our manner of incorporating them below.

Specific comments:

-Generally, their methods are up-to-date, and they present all needed information to recalculate the ages. The only comment that I can make is the systematic younger 36Cl ages (n=4) than the 10Be ages (n=9) from the same island, Suemez Island. Here, the sampled locations are 4-5 km apart from each other, however, the 36Cl ages are about 3-5 ka younger. They tried to explain this inconsistency with the post-depositional erosion, and attain the oldest age is to the closest to true age. This is most probably true, however, requires some field-based evidence of the surface erosional features on the sampled rocks. Do those rocks erode differently than the rocks where 10Be samples come? On the other hand, the uncertainties of the 36Cl ages (0.3-0.5 ka) are generally smaller than the 10Be ages (0.7-1.2 ka)? This is generally the opposite. Is there any specific reason for this? 

The boulders we dated using 36Cl are basaltic, and thus different in lithology than those we dated using 10Be, which are granitic. We observed more weathering features and uneven surfaces on boulders we sampled for 36Cl dating. While qualitative evidence may indicate that these basalts weather differently from the rocks we sampled for 10Be dating, we do not have any firm quantitative weathering rate information. Thus, to calculate erosion-corrected ages for sensitivity purposes, we added to our supplementary ages andused the same 0.3 cm/kyr erosion rate for granites and basalts. This exercise shows that erosion-corrected ages are only slightly different from the zero-erosion ages, and thus our interpretations are not affected.

We agree that often 10Be ages generally have smaller uncertainties than 36Cl ages, although this depends on rock or mineral composition used for 36Cl analyses. However, we were told that the 10Be measurements at PRIME lab were performed at a slightly less-precise level to increase analysis speed and reduce the backlog of samples post-initial COVID lockdown.

Technical corrections:

L14: Please avoid using abbreviations such as CIS, LIS, SE AK, It is hard to follow for those readers who are not familiar to the region.

We will eliminate these acronyms and instead replace them with appropriate wording.

L22: I suggest using LLGM instead of lLGM?

We will change lLGM to LLGM.

L70-L147: 25 or 29 ages?

There are 25 10Be ages from the northern Alexander Archipelago, 4 10Be ages from southern Alexander Archipelago (Suemez Island) and 4 36Cl ages from Suemez Island. This adds up to 29 total 10Be ages from this entire study with 4 additional 36Cl ages.

L194: Was a separate 1.2 g of the aliquot used for total chlorine measurements by isotope dilution separately from the 36Cl/Cl ratio measurement. It is not clear? Please clarify.

Yes, we used a separate ~1 g aliquot for the total chlorine measurements. We have revised the methods section to clarify this point. 

L220: It would be good to indicate the percent difference of both calculators' ages, here.

We will include these calculations.

Figure 2. The current coastline is needed to compare the map units. Make it clear.

The current coastline is shown on the map and in the legend. We will also mention it in the caption for added clarity.

Table 2: Footnote "a" is not related to the lat, long, elevation of the samples. Please correct?

Thank you for pointing this out! We will correct this.

Figure 6: Some of the ages in the figure were written in bold, but not others? Why? please clarify.

The bold ages are boulder ages. This is expressed in the legend but we will further clarify this in the caption.

Figure 6: In the first line of the caption of this figure: add "red and orange marks", not just "red"

Thank you for catching this. We will edit the caption accordingly.

Line 908 in Figure 10's caption: It says "Relative probability plots of boulder 10Be ages" but the F-G are error bars of the mean ages? Correct or add probability plots. Additionally, these are not just 10Be ages, there are at least on 36Cl age used.

We will correct the mistake – indeed they are mean ages. However, we did not include 36Cl ages in this calculation.

Reviewer 2:

I found this paper to be a significant contribution, of high scientific quality, and it is well written. This adds to Jason Briner's groups dataset of the Alexander archipelago. Obviously, the main finding is that it was glaciated, in contrast to common thinking 20 years ago that had a large refugia during the last glacial maximum. They place their results into a regional context that will be of interest to glacial and marine geologists as well as archeologists that are interested in the timing of glaciation as is related to first peopling along the coastal route. The figures are of a high quality.  and I appreciate the numerous pictures of the boulders and bedrock they sampled. The title is long, but it would be good to somehow add the main conclusion of the paper, no refugia, to it. Maybe “Cosmogenic ages indicate no MIS 2 refugia in the Alexander Archepelagio, Alaska.”

Dear Dr. Ward, 

We appreciate the time you took to review our manuscript and the helpful suggestions you have offered. We will shorten the title and our responses to your other comments are below. Here is our new title: “No glacial refugia found on southern Baranof and northern Kruzof islands, southeastern Alaska”

I have suggested a couple of references to be added with information on relevant ages from the Canadian coastline. I am suggesting that they add an estimate of the boulder age with erosion. The area gets a lot of rain, and big storms, so those boulders and bedrock will be weathered. However given the age, maybe it is not a significant difference.

We will include ages corrected for erosion in our supplementary figure, using a rate of 0.3 cm/kyr, similar to the rate used by Menounos et al. (2017).

I recommend publication with minor revisions. 

Minor comments:

Line 18: Technically Beringia extends into North America, so change to Asia?

We will correct this.

Line 41: remain vast stretches of the Pacific coastline that lack direct age constraints on deglaciation

Well this seems a little extreme, sure there could be more dates, but there is quite a bit of data. This is a good summary with references for relevant sites, and also addresses the coastal migration route:

F.G. Hebda, Duncan McLaren a, Quentin Mackie b, Daryl Fedje , Mikkel Winther Pedersen , Eske Willerslev, Kendrick J. Brown, Richard J. Hebda Late Pleistocene palaeoenvironments and a possible glacial refugium on northern Vancouver Island, Canada: Evidence for the viability of early human settlement on the northwest coast of North America Christopher Quaternary Science Reviews 279 (2022) 107388 https://doi.org/10.1016/j.quascirev.2022.107388

We edited this section to: “there are still large areas of the southeastern Alaskan coastline that lack direct age constraints on deglaciation” 

Line 48: Mention here that all ages are in cal years.

We will add this.

Line 55. This paper also addresses mass loss for the CIS, not the saddle area though. May want to add it.

Menounos, B., Goehring, B.M., Osborn, G., Margold, M., Ward, B.C., Bond, J., Clarke, G.K.C.,

Clague, J.J., Lakeman, T., Koch, J. Caffee, M.W., Gosse, J., Stroeven, A.P. Seguinot, J., Heyman

1. 2017. Cordilleran Ice Sheet mass loss precedes climate reversals near the Pleistocene

Termination. Science. V. 358, p. 781-784.

This paper is a valuable resource – however, here we are focusing solely on the saddle collapse. We included a citation for this paper for their erosion rates used in 10Be and 36Cl age calculations. 

Line 65: Hebda et al paper adds to examination of coastal refugia and the coastal migration route.

We will add this citation. A wonderful paper that came out just before submission.

Line 67 Great clear objectives! Nice!

Thank you!

Line 85. Switch Cretaceous and Triassic, Triassic is older and should come first.

We will correct this.

Line 91: You should give a range here, Sitca has >3,000 mm/year, Not sure what some of the other areas have.

We will report a range of ~2200 mm/yr in Sitka and ~3000 mm/yr on Chichagof Island (https://wrcc.dri.edu/summary/)

Line 116:  Add (figure 2) after Shuka Kaa, since most people will not know where this is.

We have added this for clarification.

Line 112-end of section A lot if not all of this information is again stated in the discussion. Not sure what the best way to deal with this. Likely best to reduce the background here and keep in discussion.

We eliminated this section, given that this information is included in the discussion section.

Line 135. “few direct ages” well, there are limiting ages, just maybe not than many TCN ages. See references in Hebda et al.

Eliminated per previous comment.

Line 138. Replace draped with present. Unless it actually is present across the whole island.

Eliminated per previous statement.

Line 189: 36Cl section considerably longer than 10Be, can you shorten? Or move into supplemental section. Important, yes, but most readers are not going to read detailed descriptions of TCN methods.

While we appreciate that many readers may not read these detailed descriptions, we choose to retain this section because the 36Cl methodology we followed is different from standard methods (i.e., Stone et al., 1996; Licciardi et al., 2008). We feel these methodological details are important to document and will be of interest to readers, particularly given the scope of this journal. We will, however, edit this section for clarity.

Line 238: Considering the area gets 3000-5000 mm of precipitation and has storms where the rain is driven by >100 km/hr winds, saying the boulders are not weathered or that some areas of bare bedrock were not covered by sediment at some point is likely wrong. I suspect you have lost some grains off the rocks but the question is, is in enough to make a difference to the ages. I would suggest calculating an erosion age as well for the table, even if you want to just use your present ages so readers can compare. Definitely the basaltic boulders are weathered as you state on line 397. Also many of the boulders in yor photos looks like they have undergone granular weathering.

We will include exposure ages calculated with erosion rates in the supplementary information and will add further clarification that our reported rates are minimum ages.

Line 423 Why do you specify this age as Cal ka when other ages are just ka?

This was a copy-editing error. We will change the rest to cal ka.

Line 462 Change dates to ages, you date something to get an age

Thank you, we will correct this.

Line 466 Glacial till! What other kinds of till are there? That is from the Redundancy department of redundancy.

We will edit this.

Line 468: So I am not sure where you got this age from the Coquitlam Valley. The site is in the Vancouver area, ice would have initially covered the area ~21 ka 14C BP (Coquitlam Stade) retreated at 18.5 ka 14C BP (port Moody Interstade) and then covered again 17.5 ka 14C BP (Vashon Stade) with the southern part of the CIS then extending down to its’ maximum extent om the Puget lowland at 14.5 ka 14C BP. So I guess I am not really sure what the relevance is of the Coquitlam Valley is. Something on the west coast of Vancouver Island is likely more relevant.

We have decided to omit the Coquitlam Valley data point.

Line 471: reference figure Sanak Island is mentioned.

We will add a reference to the figure.

Line 509: Add (Figure 1) to Bearing Sea core. But is a core this far away from your study area relevant.

Given the lack of other cores nearby, we believe this core to be relevant, especially as it records the presence of sea ice in the north Pacific.

Line 520: “all within uncertainty of” need a the before Uncertainty.

We have corrected to “all within the uncertainty of”

Line 527: “Fisher et al. (2008) developed an ice core record from Mt. Logan, British Columbia. However, measurements of δ18O likely represent shifting precipitation sources, rather than paleoclimate changes.” Suggest deleting as this is not relevant.

We deleted this and removed its location from our figures.

Line 530: “A growth hiatus in a speleothem from El Capitan Cave (southern Alexander Archipelago) spanning ~41.5 to ~13.4 ka suggests the cave was either overridden by the CIS, experienced permafrost conditions and a mean annual air temperature < 0 C, or lacked drip water (Wilcox et al., 2019).” Isn’t it most important that the youngest age is a minimum for deglaciation?

We agree and will add a sentence to clarify this point: “The youngest date also serves as a minimum limit on deglaciation, as the area was thawed by ~13.4 ka”

Line 570: No period at the end of this sentence.

We will add this period.

Line 850: Those do not look like groves and chatter marks, looks like differential weathering.

We agree in studying the photos a bit more. We have removed this from the caption.

Lime 870: Be nice to add sample height on the boulders.

We will add this to the table.

Line 875: “collected on Suemez on island:” get rid of extra on and capitalize Island. Lesnek et al., 2018). Add bracket “(“.

We will correct this.