General comments:

The manuscript “Multi-Proxy speleothem-based reconstruction of mid-MIS 3 climate in South Africa” presents new records of speleothem stable isotopes, strontium concentrations and temperature reconstructions based on fluid inclusion methods from Bloukrantz Cave, South Africa. In my opinion this manuscript addresses relevant scientific questions within the scope of Climate of the Past. Overall I think this manuscript would be a valuable publication in Climate of the Past after some major additions outlined below.

This is the first record of speleothem trace element compositions and fluid inclusion-based temperature reconstructions from this region. Since this is a very short and highly resolved record comparison to existing proxy records (which are usually much lower resolved) and the conclusions that can be reached are somewhat limited, nevertheless, I think this is a very relevant and important dataset and that the conclusions are substantial. The methods and assumptions used in the manuscript are valid and clearly outlined and their descriptions are sufficiently complete to allow for traceability of results. The results also support the interpretations and conclusions.

There is a clear distinction between previous work (including proper citations) and the addition of the current work. The title and abstract are clear and reflect the content of the article. I only have minor comments about the overall presentation and language (see below); symbols and abbreviations are correctly defined and used and references are appropriate except for a few minor additions/comments below.

Specific comments:

I have on major technical correction regarding the dating data tables in the supplementary materials. In the current form there is not enough information about the U-Th ages to judge their quality or to re-calculate them in the future e.g. if half-lives of some of the involved isotopes are updated. Ideally the following parameters (and their uncertainties) should be reported for each dating analysis: depth in sample (this is provided already), weight, uncorrected age result, 238U concentration, 232Th concentrations, 230Th concentration, 230Th/232Th ratio, 230Th/238U ratio, 234U/238U ratio, corrected age, reference (are ages before 1950, 2000, or relative to the year when the age was measured [give the year in this case]).

I also have two main points that I think could be improved in the discussion

First, I think the section ‘Hydroclimate reconstructions’ is well written, but in it’s comparison to other proxies it focusses a lot on very distant records in the summer rainfall region. The authors mention other speleothem records from the South African south coast in the Introduction and I understand that these records are very low resolved at the time when the sample used here was formed preventing a direct comparison. There are also two records from the Little Karoo that are not mentioned in the paper (Talma, A. S., & Vogel, J. C. (1992). Quaternary Research, 37(2), 203–213. https://doi.org/10.1016/0033-5894(92)90082-T; Chase, B. M. et al. (2021). Geology, 1. https://doi.org/10.1130/G49323.1). I think that something might might be learned from comparing the range of d18O and d13C values at the different caves sites – do they overlap or not and what might be the reasons? I also think that the Talma & Vogel (1992) record has a decent resolution for the Holocene section that overlaps with this new record and a comparison of that section might be feasible (the Chase record is a composite that mostly replicates the Talma & Vogel record in the Holocene). I think it is also worth mentioning that the pattern of the presented record with increasing stable isotope values during phases of global cooling is also found in previously published records (e.g. most of the published records from this region have somewhat higher values for d18O and d13C during cooler phases like MIS 4 than the warmer MIS 5), despite the very different resolution. The previously published records have been interpreted in a very different way, but this similarity in the relation to global change might suggest that the processes mentioned here also affected the other records.

Second, I think the section on “Significance for the archaeological record” should be rephrased. I think this section is not very clear in its current form and it could be improved. E.g. The Howieson’s Poort is a complex microlithic technology dating to late MIS 4 that can be found at many archaeological sites in southern Africa. It’s disappearance at the beginning of MIS 3 and the decrease of site use intensity on the south coast may suggest a decrease of interactions between different hunter-gatherer groups and possibly a population decline and/or shift of activities towards inland locations. 

Technical corrections:

Line 33-35: in sentence “Homo sapiens was anatomically modern as early as …” I think it is convention to italicize species names, also I would delete the words ‘behavioral proxies in’ later in this sentence

Line 38: “Episodes of significant cultural changes, …” there is an ‘e.g.’ before the second citation at the end of this sentence, either delete it or move it in front of the first one.

Lines 68ff: “In the speleothem record, this is illustrated…” add some information and references to Talma & Vogel (1992) and Chase et al., (2021) here.

Line 85: “The theoretical background of this approach dates back to the 1960s…” – all the citations that follow are from the 40s’ and 50s’

Line 114: Chapter 2.2 Sample description: the authors mention the calcite fabrics throughout this chapter, but only one image of the fabric is included in the Supplementary Materials. I think a few examples of the different fabrics mentioned here in the Supplement would be useful.

Starting in Line 215: sections 3.2 Trace elements and 3.3 Stable isotopes as well as Figure 2: the text in the results describes the Sr/Ca record and both stable isotope records in terms of their temporal changes, yet, the figure that presents the results (Figure 2) plots them against depths. And the Sr/Ca and d13C records are not plotted against age in any of the figures. I think this should be harmonized, either by changing figure 2 to be plotted against age (the plot against depth could still be presented in the supplement if needed) or by referring to depths here (ages could be mentioned along with the depths maybe in parentheses). I also think that the order of Sr/Ca, d13C and d18O in Figure 2 from top to bottom should be the order in which they are mentioned in the text.

Line 119: “After the hiatus, the Sr/Ca signal drops markedly and shows little variation with an average of 348,…” This sentence sounds like there is little change for the whole Holocene section and that the values stay below what they were before the hiatus. I don't think this is entirely true. I would rephrase to something like: "After the Hiatus, the Sr/Ca signal drops markedly and shows little variations with an average values of 348 between depths of 200 and 150 mm/ages of __ to__ ka. Between 150 and 70mm (__-__ka) values gradually increase to ~500. The top 70mm (__ to __ ka) show some of the highest variability of the record with averages around 500 (not sure about that value).

Line 336 ff: “The inferred overall drying observed in our record…” I think the publication by Engelbrecht, F. A., et al. (2019. Quaternary Science Reviews, 226, 105879. https://doi.org/10.1016/J.QUASCIREV.2019.105879) could be cited here. They show that a northward shift of the westerlies during the LGM might reduce the amount of winter rainfall along a very narrow stretch of the south coast due to downwind effects along the Cape Fold Mountains.

Line 383: “Here, the water content displays little variation…” refer to Fig 4 along with Fig. S7 in this sentence.

Figures:

I have two general suggestions:

1. Remove the line breaks/paragraphs from the figure captions, I think it is not standard practice to do this.
2. I would add subfigure denominations(a, b, c) to figures 2 and 3 instead of referring to ‘top’, ‘middle’ and ‘bottom’. Especially in Figure 3 the authors refer to the bottom for the plot for ages as well as temperature reconstructions which is not very clear. I would advise the same for the supplementary figures.

Line 217: Figure 1: blue shading indicates bathymetry of the surrounding oceans; is the current shown as SAC (South Atlantic Current) not the Antarctic Circumpolar Current? The SAC would be the section in the south Atlantic that also represents the southern branch of the subtropical gyre that then is deflected north into the BC, the ACC is what continues east.

Figure 2: Line 726: “Dashed Lines indicate the isotope transects.” Add ‘along the main growth axes of the speleothem’ since in the text this is what they are said to indicate.

Line 730: “Dashed lines indicate the onset of darker layers…” It looks like the dashed lines were taken from the depth along the isotope profile but were not adjusted for the difference in depth in the Sr profile; are the depths of the dark layers known for the TE profile and could this be adjusted similar to what is shown with the grey boxes?

Line 744: Figure 4: in the text the authors often refer to the outlier data cluster in this plot and that they represent a specific age range. I think these outliers should be marked here, maybe just with a simple circle around the younger samples that are more offset from the meteoric water lines or by using different symbols for them?

Line 749: Figure 5: I think more recent versions of the EDML d18O record do not have the gap between 45 and 44 ka, see here: EPICA Community Members (2010): Stable oxygen isotopes of ice core EDML. doi:10.1594/PANGAEA.754444

Supplements:

I recommend adding a title page to the file that states the title of the paper and names of authors and corresponding author.

Caption to Fig S2: correct ‘Mars’ to ‘March’.

General comments

In their manuscript Maccali and coauthors present a new multiproxy speleothem record of MIS3 climate variability in South Africa. The manuscript covers two bases, a climate record of AIM12, and a high-resolution comparison of two different fluid inclusion-based speleothem temperature proxies. The manuscript does very well not to ‘fall between two stools’ and covers both components well. The record itself is good, but relatively short, and while the growth phases show regional coherency and is worth remarking on, the climatic conclusions are necessarily limited by the time span. However, the additional comparison between microthermometry and fluid inclusion water isotope-based temperature reconstructions is strong and novel, and elevates the study. It is well worth consideration for Climate of the Past.

The novel result of the paper is the lack of significant temperature variability in the YRZ during AIM12, despite millennial (and centennial) scale hydroclimate variability. The result is disputed by the two temperature proxies, with FIWI method showing warming and the microthermometry showing no change within error. The authors reason that the microthermometry is more reliable, as the FIWI signal is influenced by enhanced evaporation due to the dry conditions. If true, then this marks a significant moment, where the relatively nascent microthermometry technique appears to outperform the more established FIWI method as it is less influenced by in-cave hydroclimate variability. I think the argument made in the manuscript for this being the case is reasonable as the temperature change using FIWI is unreasonably large.

My main issue with the result is the lack of presented consideration of nuance, error bars and reliability of microthermometry. The Indian Ocean cooling from 48-46ka is less than 2C while the mean-to-mean temperature warming of the microthermometry is as high as 3C. Therefore, absence of evidence for change (within error bars no change) here is not sufficient for the evidence of absence concluded by the paper. I’m not sure I would expect such a large temperature change in the subtropics to millennial scale variability, unless a major front was involved. This research group is doing tremendous work to improve microthermometry, but this result is perhaps overstated and needs nuance.

The manuscript is excellently presented. It is well-written and concise, covering all major bases, with few errors. The number of technical corrections is very small. Congratulations.

After writing my review, I have also had the chance to look at the RC1 comments from Dr. Braun. They seem reasonable and I agree with the majority of them.

Specific comments

Should the title include reference to the microthermometry, as this is one of the highlights of the paper.

Should prior calcite precipitation be changed to prior carbonate precipitation? Can prior calcite precipitation be demonstrated (calcite stalactite, U/Ca information)?

Was the XRF core scanning of the lower sections orientated perpendicular to the sampling axis of the stable isotopes in each individual growth phase, or just to the entire stalagmite. The former is not easy with an ITRAX and I would be curious as to how this was achieved. Was the stalagmite raised above the bed, was the bed adjusted, or was there enough room to reorientate the stalagmite? If the latter then by what technique was the data adjusted to the different depth scale, and could the authors comment to what extent was the data smoothed or otherwise compromised relative to the stable isotopes.

Paragraph starting Line 121: The MIS3 growth phase is mentioned here, for consistency the late Holocene growth phase should also be mentioned here, rather than wait until line 199.

Line 185: I recommend moving the number of replicate measurements for microthermometry up from line 231 into the methods section.

Is the ‘too old’ outlier age excluded due to Uranium loss? Could that indicate any potential bias in microthermometry at those depths?

Line 283: Drip rate is not strictly the control on PCP, rather it is the measure. The control is the rate of infiltration through the karst and cave ceiling.

Line 291: Dissolution is a feature of undersaturation, which can be caused by very wet conditions. There is a need further supporting evidence of dry conditions. Dust is already included, but trends of proxies into may also help.

Line 304: If unresolved hiatus are present then the duration of dry events could be even longer, not shorter.

Figure 3/5: The Indian Ocean SST record should be shown alongside the microthermometry temperature reconstruction. It could either be in Figure 3 or Figure 5, depending on whether the authors view these as ‘results and discussion’ figures or ‘temperature and hydroclimate’ figures respectively.

Figure 3: With a good choice of colour and transparency, overlapping shaded error ranges for both FIWI and microthermometry should be possible.

Figure 5: There should be a better EDML age model through this interval. If not/alternatively, the Antarctic Temperature Stack (Parrenin, Science, 2013) and WAIS Divide (WDPM, Nature 2015) ice core records provide continuous Antarctic records through this period.

Supplement: In the interests of transparency and open science, full age chemistry data should be reported.

I think the authors are correct to go with an Antarctic dominated influence on regional climate variability. The 46.1 kyr BP and 45.5 kyr BP change seem to be suitably distant from Greenland millennial scale events (on the Buizert corrected INTIMATE chronology: GS13 (H5a) starts 48.59. GI12 starts 47.11, GS12 (H5) starts 44.51 kyr BP). The match of growth phase to AIM12 fits better, as does the onset of cooling with the Antarctic Temperature Stack (ATS).

Acknowledging the caveat that there is only so much one can determine from a single specimen, I wonder if there is room to comment on the growth periods of the BL3. 48.1 to 45.3 ka corresponds nicely to Antarctic Isotope Maximum 12. Further, this growth period fits very well with a speleothem growth interval in SW Madagascar at 47.9ka-43.6ka, attributed to the combined impact of high summer insolation and an Antarctic influence of Indian Ocean SSTs (Burns et al., 2022, QSR). There is a reasonable match also to a growth phase in Inland Namibia which starts at 47.3 ka, albeit one which lasts much longer (Railsback et al., Palaeo3, 2016, Railsback et al., QSR in review). Regional coherence of growth phases is suggestive of a genuine climate control, while the differences imply that the Antarctic millennial scale variability is more important at the more southerly latitudes of SW Madagascar and the YRZ, than in inland Namibia where the insolation control seems to be less sensitive to millennial variability.

The Holocene growth phase really picks up around 3.7 ka, again approximately matching growth phases in inland Nambia (4ka onwards) and SW Madagascar (3.1 ka onwards)(Burns et al., QSR 2022, Faina et al. Malagasy Nature 2021), at the most recent summer insolation maxima. Maybe this is too speculative, but it’s a useful regional comparison of well-dated high-res records.

Technical corrections

• Throughout: missing superscripting of 18, 13 and 2 for isotopes.
• Line 8: Misspelling of Sciences
• Line 54/62: SST should be defined at first occurrence.
• Line 67: Specify timeframe of radiation. Mean annual solar/summer/etc
• Line 77: also include soil respiration processes as a major control of speleothem d13C
• Line 164: “subscript c stands for calcite” has been mentioned before.
• Line 224: “As with the Sr/Ca record”
• Line 373: I recommend including the specific dates of these samples (46.7 and 47.7?) here.
• Figure 5: The lines are very flat and do not ‘show off’ the data very well. Is there a way of making the y-axis variability more pronounced. Either by increasing the degree of overlap between panels and/or by making the figure narrower.