This referee’s comments offer constructive suggestion for the improvement of the manuscript regarding clarifications of poorly expressed or not in enough detail explained descriptions. These comments and suggestions will help to improve the final version of the manuscript.

Responses to comments are made directly in the pdf version of the manuscript annotated by the reviewer.

We thank the referee for the comments and helpful propositions!

In general, this referee, like the first, also offers some comments with constructive suggestion for the improvement of the manuscript regarding clarifications of poorly expressed or not in enough detail explained/ illustrated descriptions.

The detailed comments, suggestions and concerns are answered below (comments in italic).

1. “One of the premises for the preprint is that there are "regionally different relationships between assemblages and environmental conditions." (line 78). My interpretation is subtlety different: that the relationship between assemblage and the environment is global, but regionally, different variables are important, because they have more variance (perhaps after an appropriate transformation) in that region. This interpretation changes little of the analysis, but does make the possibility of different environmental conditions having similar assemblages (the multiple analogues problem, line 94) less likely.”

• I agree with the reviewer that the relationship between the assemblage and the environment is global, because a specific taxon does not respond differently to a given environmental parameter - meaning its behavior stays within its environmental gradient. But, as also mentioned by the reviewer, different environmental parameters are important in different regions because of a higher variance of some of these environmental variables. This causes the assemblage to “react” mostly to these variables. In other regions, other parameters may have larger gradients, causing the assemblage to “react” more to these variables. Theoretically, this may lead to situations where the assemblages are the same, although some parameters in the environment were different. This will result in a seemingly "regionally different relationship between the assemblages and the environmental conditions".
• In my opinion the reviewers and my interpretation are not really different, although this should be expressed more explicitly in the final version.

1. “The preprint argues that areas with suspected advected dinocysts could be removed from local calibration sets. True, but they could also be removed from hemispheric calibration sets (and obviously dinocyst stratigraphies from cores in such regions needed to be treated with caution).”

• The reviewer is correct and I will add a couple of sentences about that into the final version. It would be interesting to test how the statistics turn out, when we remove areas with suspected advected assemblages from the hemispheric calibration dataset, but this will be done in another study.

1. “The preprint can, and should, be shortened by not reporting numbers in both tables and the text, and removing some unnecessary methods.”

The reviewer mentions also other parts that should be shortened or changed to achieve a better understanding.

• After some reflection, I think that the referee is right. Although, all of the added analyses and data are informative, not all are necessary to illustrate the essential messages of the study.
• It is feasible to remove the multiple factor analysis (and the q-mode factor analysis) as suggested by the referee without losing any important information.
• We will also consider shortening of other parts.

1. “Fig. 1A is a bit busy; I didn't notice the core names initially. Perhaps swap panels A and B, and show the site map without the bathymetry and increase the labels' contrast.”

- This will be changed in the final version

1. ”Some methods use log-transformed per mil data, others use Hellinger transformed data. It is not clear what transformation is used for the transfer function (hopefully none for ML).”

• In general, we applied Hellinger transformation to quantify the relationship between species variances and forcing parameters because the species data do not indicate linear distribution although they are not clearly unimodal either. The Hellinger transformation preserves the Euclidean distance in the linear ordination in case the data “behaves” rather unimodal.
• We did not give the detailed information for the transfer functions as it is given in Hohmann et al. (2020). For WA-PLS and ML no transformation was applied. For MAT the estimates were determined using a log-transformed distance calculation for the analogue selection as we followed the method in de Vernal et al. (2005).
• In the final version, this will be presented more explicitly.

Hohmann, S., Kucera, M., de Vernal, A.: Identifying the signature of sea-surface properties in dinocyst assemblages: Implications for quantitative palaeoceanographical reconstructions by transfer functions and analogue techniques, Mar. Micropaleontol., 159, 101816, https://doi.org/https://doi.org/10.1016/j.marmicro.2019.101816, 2020.

de Vernal, A., Eynaud, F., Henry, M., Hillaire-Marcel, C., Londeix, L., Mangin, S., Matthiessen, J., Marret, F., Radi, T., Rochon, A., Solignac, S., Turon, J.L.: Reconstruction of sea-surface conditions at middle to high latitudes of the Northern Hemisphere during the Last Glacial Maximum (LGM) based on dinoflagellate cyst assemblages, Quat. Sci. Rev., 24, 897–924, https://doi.org/10.1016/j.quascirev.2004.06.014, 2005.

1. ”Line 440, fig 8 The preprint seems to use zero crossing of the PCA axes to delimit the core into zones. I'm not convinced this is a good method - there is nothing particularly special about zero. There are several ways to zone microfossil, for example CONISS which is implemented in the rioja R package.”

• Thank you for mentioning this! As this is a common practice (e.g., Allan et al., 2018, 2021), we followed this procedure. For future studies, the suggestion to use CONISS is relevant, but for our study the “0-method” is still appropriate. All PC-axes of the four cores center around the zero crossing and a shift from one side to the other clearly depicts a change in the assemblage composition which is related to a change in the environment. Nevertheless, I understand the concearn about “zones” and it is probably better to keep the expression “assemblage shifts”.

Allan, E., Vernal, A. De, Knudsen, M.F., Hillaire-Marcel, C., 2018. Late Holocene sea-surface instabilities in the Disko Bugt area, west Greenland, in phase with d18O-oscillations at Camp Century. Paleogeography and Paleoclimatology 33, 227–243. https://doi.org/10.1002/2017PA003289

Allan, E., de Vernal, A., Seidenkrantz, M. ‐S., Briner, J.P., Hillaire‐Marcel, C., Pearce, C., Meire, L., Røy, H., Mathiasen, A.M., Nielsen, M.T., Plesner, J.L., Perner, K., 2021. Insolation vs . meltwater control of productivity and sea surface conditions off SW Greenland during the Holocene. Boreas. https://doi.org/10.1111/bor.12514

1. “In several places, the preprint states that the code and data will be available on publication. The preprint is published. So the data and code should also be published, at least with a private link for review. The zenodo archive linked to has been removed as spam.”

• Thank you for bringing this to my attention. I have no idea why this happened. Apparently, my whole account has been removed. I contacted the support from zenodo. Now the link is active again and my accound has been restored. The code is available now. The data will be published in PANGAEA as soon as possible in September.

We thank the referee for the comments and helpful propositions!