This is an interesting paper discussing how changed ENSO characteristics may have affected South American climate during the mid-Holocene. Several model experiments are carried out to probe the sensitivity of the South American monsoon to Holocene ENSO characteristics and how these changes are imparted on the isotopic composition of precipitation across the region. The experiments are set up as sensitivity studies that allow diagnosing the influence of a changed ENSO mean state and changed insolation of ENSO on South American climate. The only drawback I can see with these experiments is that they are based on a rather old and outdated version of the ECHAM model, run at a very coarse resolution. The interpretation and description of the results will need some revisions and improvements as outlined below, but overall I think the paper is worth publishing after some moderate revisions.

Throughout the paper more care needs to be exercised with specific statements that are not clear or unambiguous. In the abstract, for example, there are several statements that need to be clarified as they are inconclusive when read on their own. For example, the authors write that’ ENSO mean state changes suppress winter precipitation’, yet it is unclear for which direction of mean state changes this is valid. Similarly, the statement that ‘both SHSI and ENSO mean state changes directly influence precipitation δ18O, resulting in strong negative δ18O anomalies’ is ambiguous as it is not clear whether positive or negative mean state changes lead to these negative d18O anomalies.

Figure 2: it might be better to refer to the more neutral term ‘precipitation’ and not ‘rainfall’ in all panels. Certainly for the description of the Huascaran ice core the descriptor ‘rainfall’ is inappropriate, as accumulation at that site is exclusively in the form of snow and not rain.

Figure 2: The location of the box termed ‘northeastern Brazil’ is outside the actual region known as northeastern Brazil. The box is located over the mouth of the Amazon river in northern Brazil, while northeastern Brazil is known as the region forming the ‘knee of Brazil’ that extends out toward the southern tropical Atlantic.

Figure 2: How exactly were wet, neutral and dry conditions determined? The Sajama ice core record for example is indicated as ‘drier’, yet the trend over the last 6 ky in that record is flat. In fact Table S2 confirms that the values for MH and present-day are essentially identical. This comment also applies to Figure 3a-d.

The model version used, ECHAM 4.6, is rather old and outdated. The spatial resolution is very low. I don’t know if results with a newer version would be fundamentally different, but it is a bit of a concern. I think the failure of the model to accurately reproduce the mid-Holocene conditions over the western Amazon and tropical Andes (Fig. 3e) may partly be attributed to the low resolution and lack of resolving topography over the region. There is not much the authors can do about this aspect, but it should be mentioned somewhere in the paper.

Why is ERA5 plotted in such low resolution in all Figures? Is it upscaled for better comparison with the model results?

Discussion Figure 3: As the authors correctly state on lines 221-222, ‘The difference between the ‘MidH’ and ‘Control’ simulation, referred to as ‘ΔMidH’ should closely approximate the mid-Holocene conditions as captured by regional proxy records’. Yet the model clearly fails to reproduce the observed mid-Holocene enrichment seen in the proxies over the western Amazon and the Andes (Fig. 3e). Even though the signal is better reproduced in the MHinsol signal, this still points to a model deficiency in reproducing Mid-Holocene conditions over the tropical S. America region. This needs to be acknowledged somewhere in the text. The same comment applies to the La Nina state simulation which appears to produce dry conditions over the western Amazon and tropical Andes (Fig. 3b), even though this region experiences excess precipitation during La Nina events. Hence I am a little bit worried about using these simulations to draw conclusive inferences about the relative roles of La Nina state vs. insolation in affecting precipitation and d18O in the region. I don’t think this completely invalidates the results, but a more cautionary tone in the discussion and conclusions, better acknowledging the model deficiencies and caveats seems appropriate. Also, it is not clear to me why this analysis is carried out on an annual basis. Almost all proxy sites analyzed are located in the monsoon region and heavily biased toward the austral summer season. ENSO is similarly phase-locked seasonally. So why not focus on this season? It would mostly likely provide for a much cleaner diagnosis.

Figure 6: same comment as above. Note that what is plotted here for ‘northeastern Brazil’ is really precipitation over the mouth of the Amazon. Northeastern Brazil should be characterized by a clear MAM precipitation peak.

Discussion section 4.3. the argument that SST feedbacks outside the Pacific also matter for South American climate and d18O signals is well taken. In fact, this was shown in recent analyses by Steinman et al. (2022) and Lyu et al. (2024), both documenting the joint influence of Pacific and Atlantic in modulating past d18O signals over tropical South America. A more  thorough discussion of this aspect seems warranted here, as currently this section is rather speculative and not fully incorporating the latest scientific findings on this aspect.

In the supplement (Section S2.), it is stated that no changes are expected in teleconnections from the Pacific to South America affecting d18O in precipitation over South America during the historical period. While there may be no significant trends, it is well known that Pacific multidecadal variability significantly modulates the d18O signal over South America (e.g. see recent analysis by Orrison et al. 2024). So the choice of the time period used as baseline for this analysis does matter. Furthermore, most d18O records over tropical S. America show a clear increase in the d18O values after 1850 CE (in many papers this is referred to as the Current Warm Period, or CWP, e.g. see Bird et al. 2011), hence in many records the values over this period are significantly more enriched in 18O compared to the prior preindustrial period from 850-1850 CE.

Figure S5 panel d). In the heading is states that this panel shows the correlation between GNIP d18O and the Nino3.4 index. Yet in the caption it is stated that the panel shows correlations with proxy data. Which is it and how would a correlation based on proxy data be calculated?

Figure S5: What are the gray cells in Figures S5e & S5f showing? I assume they indicate percentages above 200% difference (since they are in the middle of the red ITCZ region and are apparently showing significant changes (the cells are stippled). They should be plotted using saturated red colors, not an unexplained gray color.

Minor edits:

Line 76: a more ‘La Nina–like’ state of the tropical Pacific

Line 241: ‘Numbers next to speleothem sites’. You also show ice core and lake sediment records in this Figure, so you should refer to ‘Numbers next to proxy sites’ here.

Line 465: ‘Lawrence’ is a first name and should be abbreviated

Line 505: no need to capitalize ‘J. Atmos. Sci.’

Line 512: check formatting of the tilde sign (El Nino)

Supplement Line 16: Pacific

Supplement Line 91: is run

Supplement Line 102: ‘(d) ERA5 Reanalysis data’ should be ‘(f) ERA5 Reanalysis data’

Supplement Line 111: ‘Francisco’ is a first name and should be abbreviated

References cited in review

Bird, B.W., et al., 2011: A 2,300-year-long annually resolved record of the South American summer monsoon from the Peruvian Andes. Proc. Nat. Acad. Sci., 108(21), 8583-8588, https://doi.org/10.1073/pnas.1003719108.

Lyu, Z., et al., 2024: South American monsoon intensification during the last millennium driven by joint Pacific and Atlantic forcing. Sci. Adv. 10, eado9543, https://doi.org/10.1126/sciadv.ado9543.

Orrison, R., et al., 2024: Pacific interannual and multidecadal variability recorded in δ18O of South American Summer monsoon precipitation J. Geophys. Res., 129(17), e2024JD040999, https://doi.org/10.1029/2024JD040999.

Steinman, B.A., et al., 2022: North-south antiphasing of neotropical precipitation over the past millennium. Proc. Natl. Acad. Sci., 119(17), e2120015119, https://doi.org/10.1073/pnas.2120015119. 

The manuscript explores how changes in the ENSO characteristics and local solar insolation during the mid-Holocene influenced precipitation patterns and oxygen isotope ratios in tropical South America, using a water isotope-enabled atmospheric general circulation model (ECHAM4.6).

The authors use climate simulations to understand the effects of: (1) Reduced ENSO variability, (2) A La Niña-like mean state in the tropical Pacific, (3) Lower Southern Hemisphere summertime insolation (solar input due to orbital changes).

The major finding includes: (1) Reduced ENSO variability had only minor effects on average precipitation and isotope ratios. (2) A La Niña-like mean state and lower insolation both produced an east-west dipole in rainfall changes (drier western Amazon/southern Brazil, wetter northeastern Amazon/Nordeste region), consistent with a number of paleoclimate records. (3) However, changes in isotope ratios did not always mirror precipitation changes due to differences in how precipitation seasonality and mean state changes affected regional isotope records. (4) The model indicates that western Amazon isotope anomalies during the mid-Holocene are more strongly influenced by weaker insolation (seasonality) than by Pacific mean state changes, whereas both factors reinforce strong negative isotope anomalies in the northeast.

I actually really like this “clean” approach but having some concerns about the experiment design. I hope that the following points can be addressed for better understanding and clarity.

Model configuration and experiment design: The study uses the ECHAM4.6 atmospheric general circulation model with prescribed SSTs and runs several experiments manipulating ENSO characteristics and insolation. The decision not to use a fully coupled ocean-atmosphere model in the main analysis may limit the realism of certain feedbacks, especially given the demonstrated importance of global SST feedbacks in shaping isotope patterns (kind of circular). The rationale and limitations of prescribing only tropical Pacific SST anomalies, versus allowing full ocean dynamical feedback, should be explicitly justified.

Proxy-Model comparison (Western Amazon): The model fails to replicate positive isotopic anomalies recorded in Western Amazon proxies. More discussion or attempted quantification of the likely causes (missing Atlantic SST feedbacks, proxy uncertainties, limitations of isotope parameterizations or even dynamical reasons) would be valuable for both paleoclimate and modeling audiences.

Separation of ENSO magnitude and mean state: The experimental design cleanly separates mean state and ENSO amplitude effects, but real-world ENSO regimes often exhibit co-variation and nonlinearities. Please elaborate on how robust these attributions are, and discuss any residual uncertainty in interpreting main findings as the result of independent factors.

Isotope dynamics and proxy interpretation: The manuscript acknowledges that the δ¹⁸O rainfall signal is complicated by competing effects of precipitation amount, seasonality, and source moisture. This section would benefit from a more dynamical explanation to show how the induced circulation plays a role. Also The ECHAM4.6 runs at a relatively coarse resolution, which may dampen key gradients and convective responses. The implications for hydroclimate signal fidelity should be discussed more explicitly.