General comments:

This is an excellent paper that addresses the relatively under-studied eolian history of the Red Sea. The subject is of interest to the readership of the paper and will probably draw attention and referenced by future works in the region. It focuses on the central part of the Red Sea around the Tokar Gap. An interesting region that experiences large climatic variations also through winter-summer conditions, and on a larger scale of glacial-interglacial conditions it had shown large variation in the regional climatic conditions. Dust proxies usually present a messy record as there are several sources – specifically in a site that accumulated both eolian and fluvial sediments. Yet, the proxies examined here were discussed appropriately, showing an overall agreement and the text clearly demonstrates this matter. Some differences may arise regarding the interpretation, and I added the fundamental issues below, however, the paper should be published addressing few minor revisions.

Specific comments:

The authors argue that a fluvial origin of palygorskite is unlikely, however, there seem to be a correlation between rising proportions of it and EM3, and the occurrences of S events.

EM3 is considered a fluvial source and has robustly been shown to be correct in previous works in the Red Sea, even by one of the authors. How then, does this alleged discrepancy settle?

Values of 87Sr/86Sr that are higher than 0.707 accompanied with eNd values lower than -2 are not likely to reflect products of basalt weathering. Perhaps there is another source rock in the area that can provide these values? Could it be related to the palygorskite from the previous comment?

Technical comments:

Line 178: According to Fig. 3 the smectite minima at MIS5 occurred at 118-119ka and not at 125ka as stated.

Lines 396-398: The argument, to my understating, is that the alluvial fans channels and wadis stopped providing dust due to their increased hydrological activity, whereas the alluvial plains continued to do so. Thus, the process is wetting rather than desiccation, I advise to use this terminology as the revers is harder to grasp and not chronologically accurate.

General comments: This is a solid paper that presents detailed grain-size, clay mineralogy, sXRF, and Sr and Nd isotope data for a core in the Red Sea. The authors persuasively interpret these downhole data in the context of climate-driven changes in eolian and fluvial sediment supply. Connections are drawn to regional humid periods as interpreted from sapropels and other dust flux records. Though I am not a specialist in this field area, the paper was interesting to read and I recommend publication with some minor revision. My only substantive issue is that some additional information should be provided on the core age model and implementation of the grain-size endmember mixing model. I also have suggestions for expanded discussion on a few points.

Age model: More details on the age model are needed, most importantly estimates of uncertainty. Authors cite Hartman et al 2020 for the upper ~900 cm of core but in quickly looking at this source I was unable to locate any sort of discussion of the age model uncertainty (other than acknowledging that it exists). A quick age-model plot would let interested readers know which parts of the age-scale are based on the new (and essentially undescribed) age model below ~900 cm. Ideally, authors would be able to provide some estimated values for age model uncertainty through various core intervals. Obviously this is critical when assessing dust flux records between sites or comparing the various proxies to sapropel timing.

Similarly, please provide more details on EMMA implementation. What were the model parameters for things like convexity error, weighting exponent, etc?

A few discussion/results items popped out in terms of potentially deserving more discussion or elaboration (though authors may disagree that these are important):

- There’s a distinct and persistent drop in the loading of EM1 during MIS 6, ~160 ka. Since periods of low EM1 loading are discussed prominently, could authors speculate on this?

- The maxima in EM3 seem to consistently predate the timing of sapropels (which authors indicate—in introduction and Fig 7 caption—are associated with AHPs). On the other hand, the EM3 maxima are better aligned in time with the Ti/Al monsoon humidity index in Fig 7d from ODP Site 968. Maybe this is all just age model uncertainty instead of real leads/lags. But could consider elaborating on this in the text.

-I’m not familiar with the Sr and Nd isotope characterization of endmembers in this field area. But my own experience with compiling endmember compositions from the literature and applying those to terrigenous sediments prompts me to ask if it’s worth discussing potential grain-size and/or whole-rock biases when it comes to interpretation of the data. This could arise from (1) bulk terrigenous sediment digestion from the KL11 sediments, which would include both clays and coarse silt (and even fine sand), and (2) potential biases from comparing whole rock vs various sediment size fractions for the endmember characterization. Sr isotopes, in particular, may be subject to strong grain size biases. I’m not suggesting this is the case here, but I think it’s a point worth addressing.

Minor comments/suggestions (numbers refer to line numbers):

53: Maybe more reasonable to state that AHPs led to ecosystem and hydroclimate change that facilitated open pathways for human dispersal?

57/58: “maximum” in what?

87-92: Can you provide some citations to lead interested readers to primary sources for the core (e.g. a cruise report)?

220: A comparison is made to dust accumulation data from KL09. Perhaps useful to plot these data (Fig. 7?) to facilitate the comparison for readers.

231: “…local source for the riverine sediment input…”?

244: specify the callout to Fig 2B (the time series for the EMs)

250: Interesting that EM1 and EM2 are generally anti-phase (except for during AHP 5, as noted). Is this worth elaborating on?

271: Very minor point, but “entrained” seems better here than “emitted”…