This paper presents a valuable new hydroclimatic record derived from varved sediments in a deep fjord lake at the western edge of the Atlantic Ocean, offering insights into climate variability over the past 1500 years. The authors emphasize the potential of the GL-13 lamination sequence as a proxy for regional hydroclimate and suggest that their findings align with broader teleconnections, particularly with the North Atlantic Oscillation (NAO) and the Atlantic Multidecadal Variability (AMV). While the paper makes important contributions to our understanding of long-term hydroclimatic dynamics, there are several notable gaps in the analysis that warrant further attention. The interpretation of this high-potential record is undervalued.

The authors present a well-written and insightful paper that proposes to use a clastic varve records covering over 1,500 yrs to track past hydroclimates in northern America. By correlating the record of varve thickness to regional precipitation patterns as well as other regional records, the variations observed are attributed to changes in winter snowfall. These interpretations are then placed in a wider context to determine the role of large-scale synotic climate, such as the North-Atlantic Oscillation and Greendand Blockage on winter precipitation. The 1,500 yr long varved record is therefore suggested to provide insights into past NAO/GB variability.

While this is undoubtly a valuable record and a publication that falls within the scope of Climate of the Past, I would like to raise a few issues regarding the predicting power of the record to track winter precipitation and NAO. I do not refute the conclusions per se, but I question the robustness of the statistical methods on which the argumentation relies and try to propose additionnal ways to explore the data.

Varves:

• I know this is standard in varve research, but there is no mention that the sediment blocks were embedded in resin. Since this paper is adressing a larger audience, I would recommend to add a line on that. Also are the SEM pictures taken on the thin sections or sediment blocks?
• I don’t doubt that these laminations are real varves (indeed a textbook example of clastic varves!) but I was wondering whether you had any hint of years with more than two sublayers, i.e., with more than one discharge event?

Data analysis:

• Determination of drivers of sediment input (VT). Correlation of precipitation at Goose A and sediment accumulation (VT) (fig 5): Please provide r² instead of (or in addition to) r. The r value shows that the variables are positively and significantly correlated (aka. yes, total precipitation/snowfall influences the amount of sediment deposited at the site). The r² value shows how much of the variance in the dataset can be explained by the driver considered – since you are using this relationship to make predictions about winter precipitation and the NAO, it is not trivial. In your case, an r² of ca. 0.1 (which is more or less what you must have here) means that most of the variance in the dataset (90%) is explained by something else than total precipitation or snow precipitation. Did you investigate other drivers? I am wondering whether running a multiple linear regression with adjusted r² would help to constrain the role of e.g., rain+snow+temperature on sediment input. In terms of processes, I can well imagine that a combination between how fast the temperature rises in the spring and how much snow has been deposited in the winter might exert a strong control on discharge and flow strength.
• Same comment for Figure 9c: use r2 instead of r if you want to show that GB is a good predictor for winter precipitation. By the way: how is the GBI calculated?
• Minor point: in Fig 5, did I understand correctly that you compare log(VT) to log(precip)? That would make sense since both datasets are constrained (R+, only positive).
• p values need to be reported and discussed (see for instance the recent paper of Benjamin et al., Nature Human Behaviour 2018)
• time-series analyses: the data have been analysed using a wavelet analysis (Fig. S2) (using which software?). I am a bit surprised that the analysis (which is unique as being annually-resolved) shows little high-frequency variability, esp. in the range of those originating from the NAO (7-yr if I am right?), if this is supposed to be an important driver. Do the authors have an explanation for this? Did they also perform a periodogram (e.g., multi-tapper)? Did you run time-series analyses on the xrf records?

In terms of structure, the authors might consider moving Fig. 5 and the associated paragraphs to the discussion, after section 5.1. The attemps to determine the drivers will come more logically after having explained the depositional process of the varves.

Interpretations:

• Line 235: “One of the most notable features in the varve thickness series at GL is the major warm peak between the 1150s-1170s CE.” I don’t understand this sentence. In Fig. 5, the only predictor tested is precipitation, not temperature. The correlations with temperature proxies (not temperature records) only tell us that general trends in hydroclimates might be consistent regionally. Please rephrase accordingly.
• As in most fluvial systems, the VT record might be affected by memory/autocorrelation effects (and nonlinear relationships between precipitation/flood strength/sediment transport). Did you try to use differentiation (y_diff=y-y_-1) to reduce these effects? Or investigate the autocorrelation patterns?
• I found the discussion on weather patterns interesting and have no comments on that part.
• A purely curiosity-driven question: do the authors notice changes in interannual variability of VT (potentially precipitation) throughout the record?

Figures:

Figure 2: pannel b is hard to see in an otherwise well constructed and clear figure. Pannel c: what do the codes mean (abp, abo, ab…?)

In Fig. 5,6,7: please add underlines or boxes to show where the MCA and the LIA are.

Figure 6: please plot the original data with the resampled ones so that one can keep track of the data transformation.

Technical comment: There is ample litterature to emphasize the need to use a log-ratio transformation of xrf data (which are by essence compositional and therefore constrained, see for instance Kucera and Malmgren, Marine Micropaleontology 1998). Data presented as count rates can provide spurious results and should therefore be transformed (see recent publication by Bertrand et al., Earth Science Reviews 2024, with many clear examples).

I would therefore recommend that the authors revise figure 2 to show the iron relative concentration as log ratios (or centered log-ratios), with a clear x-axis and unit given.

Minor mistakes:

Line 156: Three years *were* added