Sorting out the various causes for changes in the sediment character is challenging in light of the multiple possible triggering processes, both natural and man-made.

The authors use numerous analytical methods for characterizing the layers suspected as seismites generated by a particular earthquake. The discussion on the depositional processes and how they were resolved is somewhat tedious and hard to follow, but I cannot suggest an improvement. The information is indeed relevant to the conclusions, so I suppose it better remain as is. The elaborate discussion is crucial because the radiocarbon dates cannot bracket the events tightly enough. Therefore, I value the submission as an example for applying multiple considerations in order to reach the best fit solution to a complex geological situation.

My main criticism is that the authors consider only mass transport deposits and do not ignore the option of sediment-water interaction during seismic shaking. This was shown to be significant in many previous works, in particular the ones related to the paleo Dead Sea seismites (e.g., Wetzler et al., 2010). Previous research also addressed the difference between earthquake-triggered in-situ deformation of lake bottom sediments and slope-originated mass transport deposits. However, this extensive body of works (e.g., Lu et al., 2017, 2020) is unfortunately ignored here.

Minor comments:

The authors conclude that their results “suggest that inland lakes can be sensitive recorders of earthquakes”. This is old news since evidence that inland lakes can be sensitive recorders of earthquakes has been around for over three decades. It is not a new revelation of this research.

Line 27: Earlier, much longer earthquake records (220 ka) have been reported from the Dead Sea Basin, where seismites are directly linked to synsedimentary faults and historical accounts of earthquakes (Lu et al., 2020 and references therein).

1. 406: Flood deposits usually sink to the bottom within hours, even in saline lakes, where the debris/brine density contrast is smaller than in fresh water.

L 585: “steep” is too vague, please provide a measure of the slope. Slope failures can occur on less the 1°.

References

Lu, Y., Moernaut, J., Bookman, R., Waldmann, N., Wetzler, N., Agnon, A., Marco, S., Alsop, G.I., Strasser, M., and Hubert-Ferrari, A., 2021, A New Approach to Constrain the Seismic Origin for Prehistoric Turbidites as Applied to the Dead Sea Basin: Geophysical Research Letters, v. 48, doi:10.1029/2020GL090947.

Lu, Y., Waldmann, N., Ian Alsop, G., and Marco, S., 2017, Interpreting Soft Sediment Deformation and Mass Transport Deposits as Seismites in the Dead Sea Depocenter: Journal of Geophysical Research: Solid Earth, v. 122, p. 8305–8325, doi:10.1002/2017JB014342.

Lu, Y., Wetzler, N., Waldmann, N., Agnon, A., Biasi, G.P., and Marco, S., 2020, A 220,000-year-long continuous large earthquake record on a slow-slipping plate boundary: Science Advances, v. 6, p. eaba4170, doi:10.1126/sciadv.aba4170.

Wetzler, N., Marco, S., and Heifetz, E., 2010, Quantitative analysis of seismogenic shear-induced turbulence in lake sediments: Geology, v. 38, p. 303–306, doi:10.1130/G30685.1.

It is clear that a lot of work and methods have gone into this impressive dataset and it contains a promising record. However, I do have concerns with some of the interpretations in the discussion, which can be summarized as: the authors invoke many poorly constrained mechanisms (e.g., lake level lowering, fine particles leaking out of a delta) and events (e.g. additional earthquakes in the historical part of the earthquake) too explain observations that can be more easily explained by widely recognized processes such as a delta failure. Also, some 137Cs dates would be really helpful to pinpoint the 1963/64 depth. My main comments are below and small comments are added to an annotated pdf that is attached.

There are (too) many figures in the manuscript. Please consider to move some to supplement.

Methods. Ideally you present the XRF data as centered-log-ratio (CLR) transformed to reduce the closed sum effect. See Weltje et al. (2015), or application in Schwestermann et al. (2020). This is nowadays routinely done for XRF scanning data.

Discussion.

Deposit J. The tail related to deposit J was initially not included to the event deposit, event though from Fig. 14 it is pretty clear that there is a tail (Bouma Te division) that is indeed not included in the deposit. This tail should, however, be included already in the results, so that it can also be taken into account for the age model. Furthermore, I am far from convinced that the silt deposit below J is part of the same event. We know from comparison with well-described events (e.g., Van Daele et al., 2017; Wils et al., 2021) that a long muddy tail means a significant time lag of at least days to weeks.

What about the deposit around 40 cm in SQB5 (Fig. 14), this also seems like an event deposit with a tail reaching until the base of event I.

Deposits G, H, I. I have sincere problems with the proposed interpretations. A lot of new mechanisms are invented (e.g. line 550-555 and section 4.3.3), while there are plenty known mechanisms (rock avalanche, delta failure...) that can much easier explain the observed deposits.

- I: do the authors interpret this as sourced from terrestrial or subaquatic slopes? If subaquatic, why did the 1700 CE earthquake not trigger any (!!!) failures on these slopes? Hence, I suggest the authors clarify in the text that this must be the terrestrial slopes.

The summary further includes a dam collapse and lake lowering for which no further evidence is provided. I have the feeling that a lot of additional mechanisms are invoked for which there is no evidence. In my opinion the authors make the story more complicated then it needs to be.

- H: in my opinion the authors give too much credit to core SBQ9. This is the only core where multiple pulses are observed. Why is the option that it is in fact an amalgamated turbidite considered unlikely? This core is in the depocenter (this is indeed the location where this could be expected), and apart form this core, only in SQB13 and SQB10 there is perhaps some evidence of such amalgamation, which is indeed also in the depocenter and away from the main (deltaic) sources, where also flow partitioning could get more influence. Furthermore, also event deposit J seems to have 2 pulses in exactly these cores, (and event G!) indicating that the presence of multiple "pulses" seems to be related to these locations, rather then to the specific event(s). Also, how do the authors explain these additional earthquakes, while they have not been historically reported?

- G: as reverse grading is observed, could this be a catchment response ("flood") related to events H and I? The authors link it to a documented dam failure, in that case the deposit should be coarser and thicker towards the dam (e.g. 1929 dam collapse in Eklutna Lake; Boes et al., 2018), is this the case?

An alternative interpretation of this sequence would be something similar to what's discussed in Van Daele et al. (2019) (this is anyway a pretty important reference in this paper, as it also deals with the sedimentary imprint of megathrust and intraslab earthquakes and how to distinguish them). As the 1873 earthquake was an intraslab earthquake, the high-frequency content of the shaking could have cause onshore landslides (in contrast to 1700, which would've caused more voluminous deltaic failures due to the longer duration of low frequency shaking). Hence, initially onshore landslides in the schist along the lake could have traveled directly into the lake (event deposit I). The shaking would've also cause delta failures (albeit small ones), which arrive slightly later to the core locations (event deposit H). Finally, onshore landslide in the catchment would've been transported to the lake in the years following the earthquake (event deposit G). UNLESS there is actually background sediment between event deposits I and H...?

Events C-A: Some 137Cs dates seem to be indispensable to locate the 1963/64 atomic bomb peak and thus confidently attribute the corect deposit to the 1964 floods, and probably also to the 1955 floods.

Fig. 23: the ratio is probably organic/inorganic, unlike what is mentioned in the caption. This data should be plotted in the same style as all other figures, and both with the same software.

References: 

Schwestermann, T., Huang, J., Konzett, J., Kioka, A., Wefer, G., Ikehara, K., Moernaut, J., Eglinton, T.I., Strasser, M., 2020. Multivariate Statistical and Multiproxy Constraints on Earthquake-Triggered Sediment Remobilization Processes in the Central Japan Trench. Geochemistry, Geophysics, Geosystems 21, 1–24. https://doi.org/10.1029/2019GC008861