the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Apr 2026
Surface warming in the low-latitude Indo-Pacific Ocean during peak interglacials: A major challenge for Earth System Models
Abstract. The tropical Indian and Pacific Oceans play a crucial role for global climate due to their extensive coverage across the low latitudes. However, our understanding of the regional sea surface temperature (SST) dynamics during major interglacial periods - when Earth’s climate was warmer than the preindustrial period - remains limited. Here, we compare proxy-based SST reconstructions from the low-latitude Indo-Pacific with Community Earth System Model (CESM) simulations for three key interglacial intervals: the mid-Holocene, Marine Isotope Stage (MIS) 5e, and MIS 11c. Proxy data show an overall warming during MIS 5e and MIS 11c relative to the preindustrial period. Moreover, smaller zonal SST gradients across the equatorial Pacific indicate a weakening of the Pacific Walker circulation. We attribute these findings primarily to extratropical warming and its influence on ocean circulation, particularly the Atlantic Meridional Overturning Circulation and shallow meridional overturning circulation cells. In contrast, CESM simulations indicate lower-than-preindustrial SSTs during MIS 5e and MIS 11c peaks and a larger zonal gradient. We perform individual forcing experiments to disentangle the roles of orbital forcing, greenhouse gas concentration and vegetation cover in shaping SST anomalies. We find that the implementation of paleo-vegetation in CESM reduces the discrepancies between proxy and model data. However, it does not improve the zonal SST gradients, suggesting that model representations of tropical ocean and climate dynamics remain insufficient. Possible causes for these shortcomings are discussed. Our findings underscore the need for refined model physics and improved paleo-proxy integration to better simulate tropical climate behavior during interglacials.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Climate of the Past.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1220 KB) - Metadata XML
-
Supplement
(1706 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on egusphere-2026-1891', Jo-Jo Eumerus, 19 Apr 2026
-
RC1: 'Comment on egusphere-2026-1891', David Fastovich, 10 Aug 2026
GENERAL COMMENTS
Hollstein and colleagues present a manuscript that pairs an impressive new proxy SST compilation for the tropical Pacific and Indian Oceans with CESM simulations across several interglacial time slices. In my view the three most valuable contributions are (1) the compilation itself, (2) the finding that land surface feedbacks influence the simulated tropical response, and (3) the extension of the "Holocene Conundrum" beyond the Holocene. The data-side work is careful and the model component is a genuine strength of the study. However, the modeling strength is currently underused, and several of the paper's central mechanistic claims are asserted rather than demonstrated. My comments below are organized into major comments, figure comments, and line-by-line comments. I hope the authors find these comments helpful.
1. The model simulations are underused, particularly with respect to the proposed circulation mechanisms.
The AMOC is introduced as a control on L228–L259 but is supported with just a few papers but no model analysis. The same is true of the weakening of the Pacific Walker Circulation (PWC), which is raised repeatedly through the manuscript and culminates in a proposed connection to the AMOC on L258–L267, where it is left as a hypothesis rather than evaluated against the simulations at hand. L345 illustrates the problem well: "The different SST anomalies lead to contrasting inferences about the atmospheric circulation." Why infer when the raw model output is available? At present the model component is limited to temperature comparisons, essentially plotted SST fields, when the authors have the opportunity to look under the hood and test the AMOC hypothesis directly. This is the single largest missed opportunity in the paper. In its current form the mechanistic discussion reads as proposed rather than tested.
2. The equivalence between the zonal SST gradient and PWC strength is assumed rather than established.
Throughout the manuscript, a reduced zonal SST gradient is treated as equivalent to a weakened PWC (see L214–L217, L435). The logical step connecting the two is never presented, and it is needed to support claims about the PWC based solely on SST records. Either the link should be substantiated (ideally with the model output, which would resolve this directly), or the language should be softened consistently: without a direct measure of the PWC, a reduced zonal SST gradient suggests rather than indicates a weakening, since "indicates" implies direct evidence.
3. Age model: BIGSTACK should replace LR04.
At L70 and L108–L109, the new BIGSTACK record is a direct update to LR04 and should be used here, particularly because it incorporates age uncertainties explicitly, which fits naturally with the Monte Carlo framework already adopted (Zhou et al., 2026, Geochronology, https://doi.org/10.5194/gchron-8-85-2026).
4. Uncertainty treatment
Three related points. First, I commend the authors for propagating SST uncertainties, but is 1 °C (1σ) appropriate given the uncertainties BAYSPLINE produces for these records? Second, if a Monte Carlo framework is used to propagate temporal and reconstruction uncertainties throughout, it is not clear why Figure 4 uses bars rather than the box-and-whisker plots of Figures 3 and 5.
5. Framing and readability.
The introduction is missing a sentence establishing why interglacials in particular are worth examining, and the logical step from "the IPWP is important" to "we therefore need to study it on long timescales" (L33) is absent. The connection between the IPWP and the Holocene Conundrum is also never made explicitly, even though the extension of that conundrum is one of the paper's main contributions.
More generally, the text is very dense and difficult to follow. Beyond the specific suggestions below, I would encourage the authors to make a dedicated pass for readability. Terminology contributes here. "upshift" (L243 and elsewhere) is not a term I have encountered in the paleoclimate literature, and I suspect "shoaling" is intended. Likewise, "the shallow overturning circulation" (L244, L246) is never defined.
6. Figures
General. Each figure should include text identifying the time slice ("mid-Holocene," "MIS 5e," "MIS 11c"). The manuscript alternates between time-slice names and experiment names, and consistent usage across figures and text would help the reader considerably.
Figure 2. A point symbology legend is needed and the points should be larger. If overlap is a concern, experiment with opacity. Font size needs to be increased to match Figures 3 and 5.
Figures 3 and 5. The three-panel layout is effective but makes it difficult to compare across regions. Combining these into a single panel and using color to identify the regions would enable that comparison. Currently color identifies the time slices, which the x-axis already accomplishes.
Figure 4. See major comment 4 regarding the choice of bars over box-and-whisker plots and the magnitude of the plotted uncertainties.
LINE-BY-LINE COMMENTS
L23–L25: For the introduction, I would leverage the primary role the tropical Pacific and Indian Oceans play in global climate.
L26: It is not clear what is meant by "most dynamic". please be more specific.
L33: The logical step from "the IPWP is important" to "we need to study it on long timescales" is missing and needed.
L33–L56: These paragraphs appear intended to both motivate the need for more data and provide an overview of the literature, but two issues make them hard to follow. First, the connection between the IPWP and the Holocene Conundrum is not made. Second, seasonal biases and model deficiencies are both raised as causes, but their relationship to the CESM simulations presented here is not drawn.
L115–L118: Rather than a separate sentence on the PMIP alignment, consider appending "to align with Paleoclimate Modelling Intercomparison Project (PMIP) experiments."
L133: Consider changing "The reconstructed SST anomalies vary from site to site. To get a better idea of regional SST anomalies," to "We investigate regional SST anomalies by averaging proxy records in the Western Pacific Warm Pool (WPWP), the South China Sea (SCS), and the eastern equatorial Pacific (EEP)."
L174–L181: Beginning the paragraph with caveats limits the impact of the results that follow. If the text is retained here, adding a new preceding section titled "Proxy Compilation Spatial Characteristics" would partly address this. I would instead suggest moving the text elsewhere, for example to Section 3.4.
L182–L184: Here and elsewhere, giving the proportion of sites showing warming in the Pacific as a parenthetical would be helpful.
L208–L210: This sentence is unclear, different in what respect?
L211: "partly very low" reads as contradictory; either remove "partly" or revise the sentence.
L219–L230: This paragraph presents both a key finding (warm anomalies unrelated to insolation or GHGs) and a key hypothesis analyzed later (the role of the AMOC). Moving "Hence, the warm anomalies are not related to local insolation forcing and varying GHGs, but to the astronomically modulated distribution of insolation and its feedback mechanisms" to after "While local effects might also play a role — such as precession-controlled changes in upwelling intensity (Lückge et al., 2009) — the consistency of surface ocean warming across the entire study area points to common forcings" would make the transition to the AMOC considerably clearer.
L243: Here and elsewhere, I believe "upshift" should be "shoaling," though I am not certain in the context of this sentence.
L244 and L246: "The shallow overturning circulation" is never defined. Is this referring to the subtropical cells?
L281: Here and elsewhere, when presenting modeling results, it is unclear which experiment this sentence refers to, and whether the paragraph discusses the "without expanded NH vegetation" or "with expanded NH vegetation" case.
L297: Briefly restating the proxy-based interpretation of SST anomalies here would keep the reader in the text rather than jumping back to earlier sections.
L345: "The different SST anomalies lead to contrasting inferences about the atmospheric circulation" — the model output could resolve this directly (see major comment 1).
L377–L379 and L390–L391: As written, it is unclear which model output is being used. Monthly minima and maxima? Daily values? Seasonal means?
L434: "the regional SSTs" refers back to the earlier regional averages, but by this point the reader will have forgotten them. Please write them out again.
L435: Without a direct measure of the PWC, reduced zonal SST gradients suggest, rather than indicate, a weakening.
-David Fastovich
Citation: https://doi.org/10.5194/egusphere-2026-1891-RC1 -
RC2: 'Comment on egusphere-2026-1891', Anonymous Referee #2, 13 Aug 2026
The manuscript of Hollstein et al presents a synthesis of sea surface temperature (SST) reconstructions from the low-latitude Indo-Pacific for the mid-Holocene, MIS 5e, and MIS 11c and a comparison with the CESM1.2.2 time-slice simulations for 6 ka, 127 ka and 409 ka. The main finding is that the proxy compilation indicates generally warmer-than-preindustrial SSTs during MIS 5e and MIS 11c and a smaller zonal SSTgradie nt across the equatorial Pacific, whereas the CESM simulations tend to produce cooler SST anomalies and a larger zonal gradient. Further sensitivity experiments show that the implementation of paleo-vegetation in CESM reduces the model-proxy discrepancies but does not improve the zonal SST gradient. Based on these results, the authors suggest that refined model physics and improved paleo-proxy integration are needed to better simulate tropical climate behavior during interglacials. It is an interesting study and addresses a challenging topic on proxy-model comparison over the tropical oceans. I hope the following comments and suggestions will be helpful for strengthening the manuscript and clarifying the interpretation of the main results.
1. The proxy SST anomalies are mainly calculated relative to 1850–1900 COBEv2 SST rather than to late-Holocene or preindustrial values derived from the same proxy records. This approach may introduce site- and proxy-specific offsets related to calibration, ecology, habitat depth, and seasonality. This issue may be important because the proxy–COBE offsets reported in Tables S2 and S3 are comparable in magnitude to the reconstructed MIS 5e and MIS 11c warming. The CE-based sensitivity test is useful, although it is necessarily limited to a subset of the records. It would therefore be helpful if the authors could further quantify how the main results change when only records with a same-proxy CE reference are considered. In particular, it would be useful to examine whether the regional SST anomalies and the WPWP–EEP gradient retain similar signs and magnitudes.
2. The proxy SSTs are averaged over relatively broad intervals (4–8, 123–131, and 405–413 ka), whereas the CESM simulations represent fixed equilibrium states at 6, 127, and 409 ka. The proxy averages and model simulations therefore do not necessarily represent exactly the same climatic state. I wonder whether the proxy averages can straightforwardly be interpreted as "conservative estimates of peak warmth". Temporal averaging may damp local SST maxima, but it may also combine different orbital states and different timings of local SST maxima among sites. It would therefore be useful to test the sensitivity of the reconstructed regional SST anomalies and zonal SST gradient to the width of the selected time windows, using narrower windows where the data resolution allows. In addition, the possible influence of chronological uncertainty in the proxy records may also be considered.
3. The regional SST means are weighted by the square root of the number of observations within each time window. It would be helpful to provide more justification for this weighting approach. A higher-resolution core does not necessarily provide more independent information about a regional spatial mean, particularly because adjacent measurements within an individual core are likely to be temporally autocorrelated. It would therefore be helpful to compare the current results with calculations based on equal weighting among sites. A leave-one-site-out sensitivity test could also be informative, particularly for MIS 11c, where the number of available records is relatively small. Such tests would be particularly useful for evaluating the robustness of the WPWP–EEP SST-gradient anomalies, as this gradient represents one of the central results of the manuscript.
4. The spatial distribution of the proxy records is relatively uneven, with many sites located near continental margins or coastal regions, whereas large parts of the central and open Indo-Pacific are less well represented. Coastal SSTs may be more strongly influenced by regional processes such as upwelling, river discharge, monsoon variability, and changes in local ocean circulation, and may therefore not always reflect basin-scale SST variability. I would suggest to discuss more explicitly to what extent the available proxy network can be considered representative of the low-latitude Indo-Pacific as a whole. In particular, the limited coverage of the central ocean basins should be taken into account when interpreting the regional mean SST anomalies and when making broader statements about Indo-Pacific-wide warming.
5. Given that many of the proxy records are located near continental margins or in coastal regions, changes in sea level may have a strong influence on local oceanographic conditions during past interglacials. In the CESM simulations, present-day topography and sea level are prescribed. It would be useful to discuss whether this could influence the proxy-model comparison, particularly for sites located close to continental margins or shallow shelves. This issue may also be relevant when interpreting the reconstructed anomalies in terms of large-scale Indo-Pacific temperature changes.
6. The manuscript discusses a possible mechanism linking extratropical warming and sea-ice loss to changes in the AMOC, reduced cold-water supply and upwelling, EEP warming, and ultimately a weaker Walker circulation. This is an interesting and physically plausible interpretation. However, the proposed mechanism is not directly evaluated in the analyses currently presented. Since the CESM simulations are available, the manuscript could be further strengthened by examining relevant model diagnostics associated with these processes.
7. The conclusion on the proxy-model discrepancies, including the proposed “temperature conundrum” during past interglacials, is based on comparison with time-slice simulations of one specific model. It would be more informative to compare with the results from more models, especially to test to what extent the current conclusion is model dependent, and in addition to vegetation, whether the model results are also affected by simulation type. For example, in a transient simulation (Fig. 5a of Wu et al., 2025, https://doi.org/10.1038/s41467-025-57076-2), the simulated peak warmth of MIS 5 and MIS 7 in the western equatorial Pacific SST is similar to PI, and MIS 11c and MIS 9 are warmer than PI.
Citation: https://doi.org/10.5194/egusphere-2026-1891-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 357 | 165 | 32 | 554 | 80 | 36 | 32 |
- HTML: 357
- PDF: 165
- XML: 32
- Total: 554
- Supplement: 80
- BibTeX: 36
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
A minor consideration is whether models accurately capture the climate response to paleo-vegetation changes. For example, much of the remote effect of a Green Sahara (perhaps the paradigmatic paleo-vegetation change during interglacials) on the Pacific is mediated by equatorial Atlantic SST changes (Pausata et al 2017 - 10.1038/ncomms16020 and Tiwari et al 2025 - 10.1038/s43247-025-02639-w), a region notorious for biases involving, for example, underestimated cooling by monsoon inflow winds (e.g Richter and Tokinaga 2020 - 10.1007/s00382-020-05409-w)