the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Feb 2026
Rapid Communication: Two-phase Arctic cryosphere patterns associated with delayed Norwegian Sea warming peak during the Last Interglacial
Abstract. The Last Interglacial (LIG; ~129–117 ka), when global temperatures were comparable to today, provides a valuable testbed for understanding how Arctic cryosphere–ocean interactions may shape regional climate responses. By synthesizing multiproxy records from the Norwegian Sea, North Atlantic, and Southern Ocean, we identify two previously unrecognized phases of delayed Norwegian Sea warming during the early LIG. Phase I (~129–128 ka) was marked by widespread winter sea ice and freshwater input from the retreating Eurasian ice sheets, and was likely associated with large-scale reorganizations of the Atlantic Meridional Overturning Circulation (AMOC). Phase II (during 128–124 ka) featured a localized delay in Norwegian Sea warming peak, likely associated with enhanced Arctic sea-ice melt and freshwater export rather than residual deglacial meltwater. This two-phase framework suggests that sea ice-driven feedbacks, rather than lingering Eurasian ice sheets, were linked to the Phase II delay. Importantly, Phase II does not necessarily imply a synchronous central Arctic cooling, and may instead reflect a localized “warming hole” in the Norwegian Sea. These findings refine the context for the 127 ka Coupled Model Intercomparison Project (CMIP) paleoclimate simulations and further highlight the potential role of Arctic sea ice dynamics in modulating the AMOC and subpolar climate anomalies.
- Preprint
(1331 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2026-732', Anonymous Referee #1, 07 Apr 2026
-
AC1: 'Reply on RC1', Mohamed M. Ezat, 11 May 2026
We thank the reviewer for the assessment of our manuscript. Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. As far as I can see, the present manuscript does not present any new data or information than that already appearing in Ezat et al. (2024). The only difference is that now the start of the Last Interglacial has been moved from 128 to 129 ka and thus the so-called Phase I (129-128 ka) now includes the impact of the last part of glacial meltwater originating from remnants of the Fennoscandian Ice Sheet (FIS) and shows winter sea ice extending in the southern Nordic Seas. Phase II (128-124 ka), is associated with enhanced Arctic sea-ice melt and freshwater export, as suggested by Ezat et al. (2024).
Response
We agree that the study does not provide new data, as we stated at the beginning of the Methods section: “In this study, we synthesize published data based on high-latitude marine sediment records that capture the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean temperature development.”
We also agree that the novelty of the manuscript was not sufficiently emphasized in the original version. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in the literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework (see also response to Comment #2).
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea ice-related processes (see response to Comment #2).
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies (see response to Comment #3).
4) Introduction of the “warming hole” concept in a paleoclimate (LIG)-context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.We will clarify these aspects more explicitly in the revised manuscript. To our knowledge, these elements have not previously been formulated within a single, coherent framework.
Comment #2. The evidence for the switch from glacial meltwater to Arctic sea-ice meltwater remains difficult to assess. The terminal Heinrich stadial HS11 would have led to extreme cooling in the North Atlantic and downstream Europe and may have arrested melting of the FIS. The absence of IRD from 128 ka merely shows that FIS had retreated from the coastline. The decline in Ba/Ca values in N. pachyderma does indeed provide evidence of decreased glacial runoff, though it is not clear (at least to me) whether the distance from the Norwegian coast to the southern Norwegian Sea sites could allow for continued runoff from a small residual ice sheet that remained undetected by the Ba/Ca proxy at 19PC core. The increase in d18O values of N. pachyderma is used as further evidence for a cessation of glacial meltwater influx, presumably because the ice sheet d18O is typically lighter than sea ice d18O. Missing however from the data in the present MS (and in Ezat et al. (2024)) is a discussion of the temperature component of the d18O of meltwater originating from residual FIS and from sea ice; a deconvolved record of SST and d18O of seawater may have been of assistance here.
Response
We agree with the reviewer that distinguishing between freshwater sources (residual ice-sheet meltwater vs. sea-ice-derived meltwater) is inherently challenging. Of course, the reviewer is right that foraminiferal δ¹⁸O reflects both temperature and seawater composition. Our intention was to refer to reconstructed seawater δ¹⁸O rather than foraminiferal δ¹⁸O, but we acknowledge that this distinction was not sufficiently clear in the manuscript.
Reconstructed seawater δ¹⁸O (based on paired foraminiferal δ¹⁸O and Mg/Ca; Ezat et al., 2016) shows higher values during Phase II compared to Phase I and to the later LIG, supporting a reduced relative contribution of isotopically light continental meltwater. Our interpretation is therefore based on the combined behaviour of multiple independent proxies:
- Ba/Ca indicates a decline in freshwater of continental origin.
- IRD absence indicates reduced iceberg discharge at the area (while not fully excluding residual meltwater).
- Reconstructed seawater δ¹⁸O indicates a shift toward higher values.
We fully acknowledge that proxy sensitivities (including transport distance and signal dilution) may affect detection. In the revised manuscript, we will:
• show the calculated seawater δ¹⁸O record in the figures, and clearly distinguish it from foraminiferal δ¹⁸O;
• clarify that our interpretation reflects a shift in dominant freshwater source rather than a binary switch;
• and emphasize that the proposed mechanism is consistent with the available proxy evidence, but not uniquely proven.Comment #3. Finally, the main premise of the study based on inferred melting of Arctic sea ice rests on contentious evidence, as the authors also concede here (e.g. Stein et al., 2017; Vermassen et al., 2023). In the absence of any new material to address these issues, I am afraid I have no choice but to recommend rejection.
Response
We hope that our response to the previous comment clarifies the basis for our interpretation of changing freshwater sources. Importantly, our manuscript does not aim to resolve the debated state of Arctic sea ice during the LIG definitively. Rather, it proposes a mechanistic interpretation consistent with the available multiproxy evidence from the Norwegian Sea and places it within a broader Arctic–Atlantic framework.
We agree that the state of Arctic sea ice during the LIG remains debated, as highlighted by the reviewer. However, a central contribution of this study is to explicitly raise the question of whether the delayed warming observed in the Norwegian Sea during Phase II implies a synchronous delay in the central Arctic Ocean.
In principle, this question would best be addressed using central Arctic records. However, such records are currently limited by large chronological uncertainties – often on the order of tens of thousands of years – and by ongoing debates regarding stratigraphic attribution (e.g., differing assignments of the same intervals to MIS 5e vs. MIS 11 in Vermassen et al., 2023 and Razmjooei et al., 2023, respectively). These limitations hinder robust assessment of phase relationships at the temporal resolution required here.
In this context, our approach is to use well-constrained Norwegian Sea records to formulate a testable hypothesis regarding Arctic–Atlantic coupling, rather than to provide a definitive reconstruction of central Arctic conditions. To clarify this, we will revise the manuscript to explicitly frame the sea-ice mechanism as a hypothesis consistent with available evidence, and emphasize that it is testable, for example using model simulations and additional proxy constraints. We will also strengthen discussion of alternative interpretations and uncertainties.Comment #4. PS I have to say that the quality of Fig. 1 and Fig. B1 is not up to standard. The x-axes are too short and the lines too thick to discern details. The use of dark blue, black and brown colors for the three marine sites in Fig. B1 are ill-chosen as it becomes near-impossible to distinguish them.
Response
We will increase x-axis lengths and reduce line thicknesses for improved readability. We will also use more distinct and colorblind-friendly color schemes.Citation: https://doi.org/10.5194/egusphere-2026-732-AC1
-
AC1: 'Reply on RC1', Mohamed M. Ezat, 11 May 2026
-
RC2: 'Comment on egusphere-2026-732', Anonymous Referee #2, 07 Apr 2026
Ezat and Bakker come up with a notion that the post-glacial oceanographic evolution in the Norwegian Sea during the last interglacial was affected by a 2-phase “Arctic cryosphere pattern” which led to a delayed warm peak in the Norwegian Sea facilitated by a so-called “warming hole”. They refer to 3 nearby sediments cores located at the Iceland-Faroe-Ridge.
The paper is a summary of some results from the last interglacial of the Nordic seas, but basically only looking at the surface water changes of a rather constraint small area in the southernmost Norwegian Sea. The paper ignores almost completely the substantial data that is available from the wide ranges of important parts of the Nordic Seas and in particular along the pathway of Atlantic-derived waters towards the Arctic Ocean through Fram Strait. Especially published cores from farther north, for example on and around the Voring Plateau area, give clear evidence of the various phases during 5e. Regardless of the actually published age models – age models beyond radiocarbon are indeed subject to change anyway – but many published core records could be easily compared on the basis of existing proxy records with those from the IFR.
The one and only claim of the paper is actually that the late delay of the 5e-development is due to enhanced Arctic sea-ice melt contradicting previously proposed long-lasting deglacial effects as prime cause. The authors now claim to have identified a 2-phase development. Apart from the fact that there are more than just 2, in general, it is nothing really new as others have already shown and discussed such phases in detail, too.
In addition, the authors somehow misinterpret the main finding of others: For one, it has already been stated previously by others that the “delay” was caused by residing meltwater after the main deglaciation was concluded – ie. disappearance of iceberg IRD. But icebergs only indicate calving activity on land-ocean margins not how long the abounding western Eurasian continental areas (eg, Norway and Arctic archipelagos farther north) were still glaciated long after the icebergs had vanished from the ocean. And second, the main thing of the previous findings on the delay is that meltwater in vast areas of the Nordic Seas suppressed the inflowing Atlantic water at the very surface, forcing it to flow at greater depth for a considerable time. Only after that meltwater had ceded to exist, did the warm Atlantic water affect the actual ocean surface thus causing the “delay”.
The authors now suggest that enhanced melting of Arctic sea ice is the lone cause for the delay. I wonder how does that work in an overly warm interglacial that apparently had hardly any sea-ice left in the summer and thus could not build up substantial amounts of thick-enough sea-ice during ensuing winters? Work in the Fram Strait clearly show that the delay is found there too, and seemingly much more drastic than further south (see previous work by Zhuravleva et al. from the eastern side, and 2025 by Zehnich et al. for the western Fram Strait); there is also a new work on the last interglacial by Sicard et al. which might be useful.
As mentioned, all data shown now were already published, very few by others, but all relevant ones by Ezat (mostly in the 2024 publication). That being said, however, is no proof that all the proxies employed previously are justified tools. Using, for instance, Ba/Ca in Np as indicator of sea-ice meltwater, contrary to iceberg meltwater, is a far shot from having been properly validated.
This paper is not a study which provides anything new in terms of data. Just the opposite, it is more of a contemplation that muses about a “potential” subject of interest to some. I don’t see the merit of this manuscript for the wider paleoclimate community. As a “rapid communication” it should be rejected.
Few further notes (there could be many more):
Because the relevant cores are from the shallow and rather restricted entrance gateway into the Nordic Seas, they undoubtedly are related to the surface waters in the southern Norwegian Sea. But these cores alone have little to say about the entire Nordic Seas, and certainly even much less about Arctic proper which is even farther north.
Phase 1 is described as 1 ky long (129-128ka), in main figure 1 it starts much earlier and clearly is nothing but part of deglacial Termination 2.
The increasing re-occurrence of polar Np after 124 ka (making up almost entirely the high carbonate content; thus, this proxy reflects shell abundance not warmth as stated(!) which is quite typical for the Nordic Seas in general and rather specific near the end of 5e. Together with increasing IRD means melting icebergs, and hence a surface salinity drop and potentially winter sea-ice. That BPIP25-Ip25 etc. does not show up is likely due to decreased sedimentation rates, causing massive degradation of organic matter, the principal bearer of biomarkers. This might indeed indicate a limitation of this so-called sea-ice proxy and associated biomarkers.
Fig. 1: f and c have faulty y-scales
Fig B1: it misses the “i” label
Note to the Editor:
This so-called paper strikes me as being a bit odd considering both its intention and what’s in it for the paleo-community in terms of novelty.
1- The latter certainly is hardly there – data-wise it’s just a reiteration of Ezat’s own recent publication (2024 in Nat comm) – for a Rapid Communication I would expect something surprisingly new. Instead what I see is a negligence of other peoples’ work and interpretations who actually have generated data on the very subject in the Nordic Seas too.
2- And the former relates to the co-author Bakker, by record mainly a modeler. His contribution is basically zero, at least nothing in the manuscript relates to any of his work, according to the references listed. That makes me wonder if the authors use the rapid communication platform as a pretext, hoping for a quickly citable reference for something that is already in their pipeline, namely a LIG-CMIP simulation using their suggested “warming hole” as the principle novel idea? Of course, I may be wrong...just a thought.
Citation: https://doi.org/10.5194/egusphere-2026-732-RC2 -
AC2: 'Reply on RC2', Mohamed M. Ezat, 11 May 2026
Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. Ezat and Bakker come up with a notion that the post-glacial oceanographic evolution in the Norwegian Sea during the last interglacial was affected by a 2-phase “Arctic cryosphere pattern” which led to a delayed warm peak in the Norwegian Sea facilitated by a so-called “warming hole”. They refer to 3 nearby sediments cores located at the Iceland-Faroe-Ridge.
The paper is a summary of some results from the last interglacial of the Nordic seas, but basically only looking at the surface water changes of a rather constraint small area in the southernmost Norwegian Sea. The paper ignores almost completely the substantial data that is available from the wide ranges of important parts of the Nordic Seas and in particular along the pathway of Atlantic-derived waters towards the Arctic Ocean through Fram Strait. Especially published cores from farther north, for example on and around the Voring Plateau area, give clear evidence of the various phases during 5e. Regardless of the actually published age models – age models beyond radiocarbon are indeed subject to change anyway – but many published core records could be easily compared on the basis of existing proxy records with those from the IFR.
Response
We respectfully disagree with the reviewer’s assessment and clarify the scope and selection criteria of our study. As stated in the Methods, the primary objective of this study is to assess the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean. This requires records that can be placed on a consistent and independently constrained chronological framework. For this reason, the identification of tephra layer 5e-Low/BAS-IV, which is also identified in the North Atlantic record ENAM33, is essential for establishing robust inter-regional correlations. Without such constraints, lead–lag relationships cannot be assessed reliably. We do not ignore other Nordic Seas records; rather, we explicitly evaluate their suitability for this specific objective (see Methods). While additional records from, for example, the Vøring Plateau or Fram Strait provide valuable regional context, many lack the independent chronological constraints required for robust phase comparisons across basins. While we agree that age models are subject to uncertainty, broader inter-core comparisons without consistent chronological control introduce additional uncertainties that are difficult to quantify for the specific timing question addressed here.
We also note that proxy-based alignments (e.g., using the relative abundance of N. pachyderma) can lead to substantially different phase relationships depending on the proxy and alignment strategy used. For example, aligning the studied records (~62.7°N; 4°W) to the very nearby North Atlantic record ENAM33 (~61.27°N; 11.16°W) using such approaches would obscure the distinction between the phases identified by this and previous studies. This highlights that proxy-based correlations are inherently dependent on methodological choices and assumptions, and may obscure rather than resolve phase relationships. Also, while additional Nordic Seas records from the Fram Strait and the Greenland Sea could, in principle, be incorporated through proxy-based alignment, a key limitation is that such approaches do not allow robust quantification of chronological uncertainties associated with tie points. This is particularly critical for the objectives of our study, which focus on assessing the relative timing of peak warming between regions. Without well-constrained and quantifiable age uncertainties, apparent lead–lag relationships cannot be evaluated with confidence and may not be statistically meaningful. In contrast, the use of independently identified stratigraphic markers such as the tephra layer 5e-Low/BAS-IV, which is present at the beginning of the LIG warming peak in the southern Norwegian Sea records, provides a more robust basis for inter-regional correlation. We therefore prioritize records where such constraints are available, rather than relying on less constrained proxy alignments. As part of this study, we also revisit the selection of Nordic Seas records used in previous compilations (e.g., Capron et al., 2014, 2017). In particular, we emphasize that the tephra layer 5e-Low/BAS-IV was previously assumed to be present in some records (e.g., HM71-19), whereas geochemical evidence shows that these ash layers are not equivalent (Wastegård and Rasmussen, 2001). This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas, and represents an additional contribution of the present study.
This approach does not reflect an omission of available records, but rather a careful selection of cores that are suitable for the specific question addressed here. Importantly, our interpretations are restricted to the studied region, and we do not generalize the findings to the entire Nordic Seas.
We also note that the reviewer elsewhere emphasizes that the studied cores primarily reflect conditions in the southern Norwegian Sea and may not be representative of the wider Nordic Seas or Arctic Ocean (see comment #5). We agree with this point, and this possible regional heterogeneity is in fact one of the reasons why we are cautious about applying unconstrained proxy-based alignments between distant Nordic Seas records. This further supports our emphasis on independently constrained chronostratigraphic markers for inter-regional temporal comparisons.
We will revise the manuscript to clarify these points more explicitly, in particular the rationale for record selection, the limitations of proxy-based alignments, and the importance of robust independent chronological constraints for assessing lead–lag relationships.
Comment #2. The one and only claim of the paper is actually that the late delay of the 5e-development is due to enhanced Arctic sea-ice melt contradicting previously proposed long-lasting deglacial effects as prime cause. The authors now claim to have identified a 2-phase development. Apart from the fact that there are more than just 2, in general, it is nothing really new as others have already shown and discussed such phases in detail, too.
Response
To our knowledge, no previous study has identified two temporally successive and mechanistically distinct cryosphere processes during the early LIG in the Norwegian Sea within a single integrated framework. We agree that climate evolution during the LIG may involve more than two phases in a broader sense; however, our focus is on two phases that are robustly expressed in the available chronologically constrained records and directly relevant to the timing of the delay of the peak warming in the Norwegian Sea during the early LIG.
Nevertheless, we agree that the novelty of the manuscript could have been emphasized more. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in the literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework.
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea-ice-related processes (see also comment #3).
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies.
4) Introduction of the “warming hole” concept in a paleoclimate (LIG) context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We explicitly frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas (see also comment #1).Comment #3. In addition, the authors somehow misinterpret the main finding of others: For one, it has already been stated previously by others that the “delay” was caused by residing meltwater after the main deglaciation was concluded – ie. disappearance of iceberg IRD. But icebergs only indicate calving activity on land-ocean margins not how long the abounding western Eurasian continental areas (eg, Norway and Arctic archipelagos farther north) were still glaciated long after the icebergs had vanished from the ocean. And second, the main thing of the previous findings on the delay is that meltwater in vast areas of the Nordic Seas suppressed the inflowing Atlantic water at the very surface, forcing it to flow at greater depth for a considerable time. Only after that meltwater had ceded to exist, did the warm Atlantic water affect the actual ocean surface thus causing the “delay”.
Response
We respectfully clarify that this interpretation does not reflect the way the mechanisms are framed in our manuscript. Our interpretation does not assume a single, persistent meltwater mechanism throughout the early LIG, but instead evaluates changes in the dominant processes through time based on multiproxy evidence. Our interpretation is based on the combined behaviour of multiple independent proxies:
- Ba/Ca indicates a decline in freshwater of continental origin in phase II compared to phase I.
- IRD absence indicates reduced iceberg discharge in the study area (while not fully excluding residual meltwater) in phase II compared to phase I.
- Reconstructed seawater δ¹⁸O indicates a shift toward higher values during phase II compared to deglaciation, phase I and the latter part of the LIG, supporting a reduced relative contribution of isotopically light continental meltwater.
We fully acknowledge that proxy sensitivities (including transport distance and signal dilution) may affect detection. In the revised manuscript, we will:
• show the calculated seawater δ¹⁸O record in the figures;
• clarify that our interpretation reflects a shift in dominant freshwater source rather than a binary switch;
• and emphasize that the proposed mechanism is consistent with the available proxy evidence, but not uniquely proven.
Regarding the proposed mechanism of subsurface Atlantic water intrusion, we do not think available proxy evidence supports a sustained subsurface warming during Phase II. Planktic Mg/Ca data from N. pachyderma indicate relatively higher subsurface temperatures during Phase I, consistent with subsurface Atlantic water influence, whereas Phase II is characterized by lower subsurface temperatures (Ezat et al., 2016). In addition, benthic foraminiferal assemblages indicate “interglacial” assemblages during what we describe as Phase II (e.g., Rasmussen et al., 1996), without clear evidence for the sustained subsurface warming required by this mechanism. We will clarify this point in the revised manuscript. We acknowledge that the processes governing stratification and vertical heat distribution in the Nordic Seas during the LIG remain complex, and we will expand the discussion of alternative interpretations in the revised manuscript.
Comment #4. The authors now suggest that enhanced melting of Arctic sea ice is the lone cause for the delay. I wonder how does that work in an overly warm interglacial that apparently had hardly any sea-ice left in the summer and thus could not build up substantial amounts of thick-enough sea-ice during ensuing winters? Work in the Fram Strait clearly show that the delay is found there too, and seemingly much more drastic than further south (see previous work by Zhuravleva et al. from the eastern side, and 2025 by Zehnich et al. for the western Fram Strait); there is also a new work on the last interglacial by Sicard et al. which might be useful.
As mentioned, all data shown now were already published, very few by others, but all relevant ones by Ezat (mostly in the 2024 publication). That being said, however, is no proof that all the proxies employed previously are justified tools. Using, for instance, Ba/Ca in Np as indicator of sea-ice meltwater, contrary to iceberg meltwater, is a far shot from having been properly validated.
This paper is not a study which provides anything new in terms of data. Just the opposite, it is more of a contemplation that muses about a “potential” subject of interest to some. I don’t see the merit of this manuscript for the wider paleoclimate community. As a “rapid communication” it should be rejected.
Response
We respectfully clarify that our manuscript does not propose enhanced Arctic sea-ice melt as a lone or exclusive mechanism for the delayed warming. Rather, we evaluate changes in dominant processes through time, and explicitly distinguish between a deglacial meltwater-driven Phase I and a subsequent Phase II in which sea-ice-related processes may have contributed to the localized delay in Norwegian Sea warming. We agree that the state of Arctic sea ice during the LIG remains debated. Importantly, our interpretation does not require extensive or persistent summer sea ice, but rather considers the potential role of seasonal sea-ice processes and associated freshwater and/or heat fluxes. Even under relatively warm conditions, changes in sea-ice seasonality, export pathways, and atmospheric circulation can influence freshwater distribution, heat fluxes and upper-ocean stratification (e.g., Sevellec et al., 2017). We will revise the manuscript to clarify the underlying assumptions of our interpretation and to better articulate the range of mechanisms that may explain the observed patterns, including their uncertainties and testability.
We also note that the reviewer raises alternative interpretations regarding sea-ice conditions during Phase II. While the reviewer suggests that Arctic sea ice during the LIG may have been too limited to sustain substantial freshwater export in this comment, the reviewer elsewhere proposes that sea ice in the southern Norwegian Sea may have been more extensive during phase II than indicated by the biomarker proxies due to preservation effects (Comment #7; see also our response there).
Regarding additional records from the broader Nordic Seas and Fram Strait regions, these studies provide valuable regional context; however, as noted in our response to Comment #1, differences in chronological control currently limit robust assessment of phase relationships across regions. We also emphasize that the manuscript synthesizes the available proxy data from the study area where the tephra layer 5e-Low/BAS-IV is identified, including diatom assemblages (Hoff et al., 2019), dinocyst assemblages (Van Nieuwenhove et al., 2011), tephra chronology (e.g., Wastegård and Rasmussen, 2001; Abbott et al., 2014), and planktic assemblages and IRD (e.g., Abbott et al., 2014; Rasmussen et al., 2003). Relevant studies from the broader Nordic Seas are also cited to provide regional context.
We acknowledge that individual proxies, including Ba/Ca in N. pachyderma, are subject to uncertainties and ongoing evaluation and calibration. Our interpretation does not rely on any single proxy, but rather on the combined behaviour of multiple independent indicators: Ba/Ca as an indicator of freshwater of continental origin; IRD as an indicator of iceberg discharge; and reconstructed seawater δ¹⁸O as an indicator of freshwater source characteristics. We will revise the manuscript to more clearly emphasize the limitations of each proxy and to frame our interpretation as one consistent with the available multiproxy evidence, rather than uniquely constrained by any individual proxy. For more details, see our response to Comment #3.
We respectfully emphasize that the contribution of this study is not the presentation of new data, but the integration and reinterpretation of existing records within a coherent chronological and mechanistic framework (see our response to Comment #2), and as we stated at the beginning of the Methods section: “In this study, we synthesize published data based on high-latitude marine sediment records that capture the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean temperature development.”
Comment #5. Because the relevant cores are from the shallow and rather restricted entrance gateway into the Nordic Seas, they undoubtedly are related to the surface waters in the southern Norwegian Sea. But these cores alone have little to say about the entire Nordic Seas, and certainly even much less about Arctic proper which is even farther north.
Response
We agree that the studied cores primarily reflect conditions in the southern Norwegian Sea, and we explicitly frame our interpretation within this regional context. This is consistent with the design and objectives of our study, which focus on this region where robust chronological constraints are available. As noted in the manuscript, individual paleoclimate records represent local conditions unless a broader regional interpretation is supported by appropriate constraints. Our conclusions are therefore restricted accordingly. For more details on the core selection and its suitability to the study’s main objective, see our response to comment #1. We would also like to note that this possible regional heterogeneity is in fact one of the reasons why we are cautious about applying unconstrained proxy-based alignments between distant Nordic Seas sites. If spatial variability in surface ocean conditions was substantial, then correlations based primarily on similarities in proxy behaviour become increasingly dependent on assumptions regarding synchronicity between regions. This further supports our emphasis on independently constrained chronostratigraphic markers, such as tephra layer 5e-Low/BAS-IV, for assessing inter-regional phase relationships. Without this tephra layer, occurring near the end of Phase II, we would not be able to design or fulfill the main objective of this study. Again, for example aligning the studied records (~62.7°N; 4°W) to the very nearby North Atlantic record ENAM33 (~61.27°N; 11.16°W) using such approaches would obscure the distinction between the phases identified by here and previous studies.Comment #6. Phase 1 is described as 1 ky long (129-128ka), in main figure 1 it starts much earlier and clearly is nothing but part of deglacial Termination 2.
Response
Our definition of Phase I is based on the timing of delayed peak warming in the North Atlantic relative to the Southern Ocean, as established in previous studies (e.g., Govin et al., 2012; Capron et al., 2014, 2017). Within this commonly used chronological framework, the interval ~129–128 ka corresponds to the timing of this delayed warming response.
We acknowledge that in the Nordic Seas literature, this interval is often described as part of the late stages of Termination II (including Ezat et al., 2024). However, when records are placed on a consistent age scale with North Atlantic and Southern Ocean records, it corresponds to the phase of delayed peak warming identified in the North Atlantic (e.g., Capron et al., 2014; Stone et al., 2016). Our use of the term “Phase I” therefore follows this inter-regional framework rather than a strictly local deglacial classification.
We further acknowledge that age uncertainties (on the order of ~1–2 kyr) limit the precise delineation of phase boundaries. For this reason, our interpretation places greater emphasis on Phase II, where the identification of the tephra layer 5e-Low/BAS-IV provides a robust cross-basin chronostratigraphic marker and allows more confident assessment of the relative timing between Norwegian Sea and North Atlantic records.
Our intention is not to redefine the deglaciation, but to describe the structure of the delayed warming signal within a consistent inter-basin chronological framework. We will revise the manuscript and figures to more clearly define the phase boundaries.Comment #7. The increasing re-occurrence of polar Np after 124 ka (making up almost entirely the high carbonate content; thus, this proxy reflects shell abundance not warmth as stated(!) which is quite typical for the Nordic Seas in general and rather specific near the end of 5e. Together with increasing IRD means melting icebergs, and hence a surface salinity drop and potentially winter sea-ice. That BPIP25-Ip25 etc. does not show up is likely due to decreased sedimentation rates, causing massive degradation of organic matter, the principal bearer of biomarkers. This might indeed indicate a limitation of this so-called sea-ice proxy and associated biomarkers.
Response
We note that the comment presented by the reviewer refers primarily to the interval after ~124 ka, whereas our manuscript focuses on the early LIG (Phases I and II; 129-124 ka) and the timing of delayed peak warming. Within the interval we define as Phase II, IRD values are low to near-zero and do not show the increase suggested by the reviewer. Importantly, our interpretation of sea-ice-related processes is not based on a single proxy, but on the combined behaviour of multiple independent indicators including IP25 and sterol concentrations as well as diatom and dinocyst assemblages (see e.g., Supplementary Figure B1), which show consistent patterns during the interval of interest. In addition, the biomarker data (IP25 and sterol concentrations) are derived from two sediment cores with substantially different sedimentation rates, yet yield consistent signals, suggesting that the observed patterns are unlikely to be primarily controlled by differential degradation effects. We will still clarify these points more explicitly in the revised manuscript.
We would like also to note that the reviewer raises alternative interpretations regarding sea-ice conditions during Phase II. On the one hand, the reviewer suggests that biomarker-based sea-ice proxies may underestimate sea ice due to degradation effects and low sedimentation rates, potentially implying more extensive sea ice in the Norwegian Sea (this comment). On the other hand, the reviewer argues elsewhere that Arctic sea ice during the LIG may have been too limited to sustain significant sea-ice-related freshwater export (comment #4).Comment #8. Fig. 1: f and c have faulty y-scales
Response
We will correct the y-axis scales in the revised figure.Comment #9. Fig B1: it misses the “i” label
Response
We will correct the missing label in the revised figure.Comment #10. Note to the Editor: This so-called paper strikes me as being a bit odd considering both its intention and what’s in it for the paleo-community in terms of novelty.
1- The latter certainly is hardly there – data-wise it’s just a reiteration of Ezat’s own recent publication (2024 in Nat comm) – for a Rapid Communication I would expect something surprisingly new. Instead what I see is a negligence of other peoples’ work and interpretations who actually have generated data on the very subject in the Nordic Seas too.
2- And the former relates to the co-author Bakker, by record mainly a modeler. His contribution is basically zero, at least nothing in the manuscript relates to any of his work, according to the references listed. That makes me wonder if the authors use the rapid communication platform as a pretext, hoping for a quickly citable reference for something that is already in their pipeline, namely a LIG-CMIP simulation using their suggested “warming hole” as the principle novel idea? Of course, I may be wrong...just a thought.
Response
We note that these comments relate primarily to authorship and perceived intent rather than the scientific content of the manuscript. We respectfully emphasize that all authors meet standard authorship criteria and have contributed substantially to the conception, interpretation, and development of the study – and we do not have currently another manuscript planned on the subject. The manuscript presents a synthesis that integrates proxy-based evidence within a process-oriented framework, and its contribution is therefore conceptual rather than data-driven (as outlined in our responses above).
We also respectfully note to the editor that some reviewer comments appear to reflect differing assumptions regarding both spatial heterogeneity and sea-ice conditions during the LIG. For example, the reviewer suggests in Comment #1 that records from across the Nordic Seas could be easily aligned and compared with our studied cores from the southern Norwegian Sea using proxy alignments, while elsewhere emphasizing possible substantial regional heterogeneity between the southern Norwegian Sea and the wider Nordic Seas in Comment #5 (see our responses to both comments). Similarly, the reviewer argues in Comment #4 that Arctic sea ice during the LIG may have been too limited to sustain substantial freshwater export, while suggesting in Comment #7 that sea ice in the southern Norwegian Sea may have been more extensive than indicated by the biomarker proxies due to preservation effects (see our responses to both comments). We interpret these comments as further illustrating the uncertainties surrounding spatial and temporal sea-ice variability during the LIG, and the importance of independently constrained chronostratigraphic markers and multiproxy approaches in evaluating such questions.Citation: https://doi.org/10.5194/egusphere-2026-732-AC2
-
AC2: 'Reply on RC2', Mohamed M. Ezat, 11 May 2026
-
RC3: 'Comment on egusphere-2026-732', Anonymous Referee #3, 11 Apr 2026
Ezat et al. synthesize existing records from the Norwegian Sea, North Atlantic, and one Southern Ocean site to propose a two-phase pattern of Norwegian Sea surface warming during the Last Interglacial (LIG): (1) an early phase associated with increased IRD and a weakened AMOC, and (2) a delayed warming of the Norwegian Sea relative to North Atlantic SSTs during ~128–124 ka. The authors attribute this pattern to enhanced southward export of Arctic sea ice meltwater.
I agree with the other posted reviews that the manuscript does not present sufficiently novel insights. Both the hypothesis and much of the dataset have already been published or synthesized in Ezat et al. (2024, Nature Communications), and the current study does not substantially advance beyond that work.
I also find the proposed mechanism linking increased Arctic meltwater export to suppressed convection in the Norwegian Sea to be speculative. If the Arctic experienced warming and reduced sea-ice extent during the LIG, it is counterintuitive and against my experience why meltwater export to lower latitudes would increase. Moreover, since the authors cite Guarino et al. (2020)'s modeling work as support for higher SSTs and reduced sea ice during Phase II, the associated changes in freshwater export should be testable. Analysis of this model or similar LIG simulations from PMIP4 would strengthen the argument but is currently missing.
Finally, the use of the tephra layer as a key criterion for excluding certain records requires further clarification. More detail is needed about what this tephra layer is. Also, it is unclear why the same criterion is not applied to the North Atlantic records. If the goal is to compare records chronologically, consistent selection criteria should be used across regions.
A figure showing the core locations, at least those ones from the North Atlantic and Norwegian Sea would be helpful.
Citation: https://doi.org/10.5194/egusphere-2026-732-RC3 -
AC3: 'Reply on RC3', Mohamed M. Ezat, 11 May 2026
We thank the reviewer for the assessment of our manuscript. Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. Ezat et al. synthesize existing records from the Norwegian Sea, North Atlantic, and one Southern Ocean site to propose a two-phase pattern of Norwegian Sea surface warming during the Last Interglacial (LIG): (1) an early phase associated with increased IRD and a weakened AMOC, and (2) a delayed warming of the Norwegian Sea relative to North Atlantic SSTs during ~128–124 ka. The authors attribute this pattern to enhanced southward export of Arctic sea ice meltwater.
I agree with the other posted reviews that the manuscript does not present sufficiently novel insights. Both the hypothesis and much of the dataset have already been published or synthesized in Ezat et al. (2024, Nature Communications), and the current study does not substantially advance beyond that work.
Response
We emphasize that the contribution of this study is conceptual rather than data-driven, focusing on the integration and reinterpretation of existing records within a consistent chronological framework. Nevertheless, we agree that the novelty of the manuscript was not sufficiently emphasized in the original version. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework.
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea-ice-related processes.
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies.
4) Introduction of the “warming hole” concept in a paleoclimate (LIG)-context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We explicitly frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.
While aspects of the dataset and individual interpretations have been presented previously (e.g., Govin et al., 2012; Ezat et al., 2024), the present study integrates these elements into a unified temporal and mechanistic framework that was not explicitly developed before. We will clarify these aspects more explicitly in the revised manuscript to better highlight the conceptual advance beyond previous work.Comment #2. I also find the proposed mechanism linking increased Arctic meltwater export to suppressed convection in the Norwegian Sea to be speculative. If the Arctic experienced warming and reduced sea-ice extent during the LIG, it is counterintuitive and against my experience why meltwater export to lower latitudes would increase. Moreover, since the authors cite Guarino et al. (2020)'s modeling work as support for higher SSTs and reduced sea ice during Phase II, the associated changes in freshwater export should be testable. Analysis of this model or similar LIG simulations from PMIP4 would strengthen the argument but is currently missing.
Response
We thank the reviewer for raising this important point regarding the interpretation of Arctic freshwater export during the Last Interglacial. We agree that the relationship between Arctic warming, sea-ice extent, and freshwater export is not straightforward, and we will revise the manuscript to clarify both the physical interpretation and the level of inference supported by the proxy data. For example, we emphasize that our interpretation does not rely solely on increased freshwater flux, but more broadly on changes in surface buoyancy forcing, which may arise from a combination of freshwater and/or heat fluxes associated with sea-ice variability (c.f., Sevellec et al., 2017). In addition, changes in atmospheric circulation and wind-driven transport may modulate the magnitude and routing of freshwater export. We also acknowledge that the temporal persistence of such processes on millennial timescales remains uncertain and requires further evaluation using climate-model simulations. Further, we clarify that our study does not aim to establish a direct causal relationship between Arctic sea-ice melt, Nordic Seas convection and regional/local climate anomalies. Given the nature of proxy records, we instead interpret the data in terms of associations between observed surface conditions and plausible cryosphere–ocean processes. We will revise the manuscript to explicitly clarify these points.
We agree that climate model simulations provide an important avenue to test this hypothesis. In the revised manuscript, we will expand the discussion to include relevant results from existing LIG simulations (e.g., including Guarino et al., 2020 where applicable).Comment #3. Finally, the use of the tephra layer as a key criterion for excluding certain records requires further clarification. More detail is needed about what this tephra layer is. Also, it is unclear why the same criterion is not applied to the North Atlantic records. If the goal is to compare records chronologically, consistent selection criteria should be used across regions.
Response
We thank the reviewer for raising this important point regarding the use of the tephra layer 5e-Low/BAS-IV as a selection and correlation criterion. We agree that the rationale for using this tephra layer and its application across regions requires clearer explanation, and we will revise the manuscript accordingly.
First, we would like to clarify that the tephra layer 5e-Low/BAS-IV is not used exclusively for the Nordic Seas records, but is also identified in a key North Atlantic record (ENAM33; e.g., Wastegård and Rasmussen, 2001). This provides a rare and valuable direct chronostratigraphic link between the Nordic Seas and the North Atlantic, independent of proxy-based alignment.
The motivation for using this tephra layer as a key criterion stems from the asymmetry in chronological constraints between regions, as discussed in detail by Capron et al. (2014). In the North Atlantic, age models can be more robustly established through alignment of SST records with Greenland ice-core δ¹⁸O and methane variability within the AICC2012 framework. This allows relatively consistent synchronization of Atlantic Ocean records with Greenland and Antarctic climate evolution. In contrast, establishing chronologies in the Nordic Seas is substantially more challenging for several reasons:
• Common SST proxies (e.g., planktic foraminiferal assemblages) lose sensitivity at low temperatures typical of the Nordic Seas, limiting their utility for precise alignment.
• Foraminiferal δ¹⁸O records during Termination II are strongly influenced by freshwater inputs and stratification effects, complicating their use in correlation.
• Alternative quantitative SST reconstructions are sparse, and preservation effects may further bias proxy signals.
As a result, age models in the Nordic Seas often rely on multiple indirect tie points (e.g., benthic δ¹⁸O, biostratigraphy), associated with relatively large uncertainties (up to some thousand years). Because the primary objective of our study is to assess the relative timing of peak warming between regions, such uncertainties critically limit the robustness of inferred lead–lag relationships when unconstrained records are included. In this context, the identification of tephra layer 5e-Low/BAS-IV in both Nordic Seas and North Atlantic records provides a unique and independent chronostratigraphic anchor, allowing more robust inter-regional correlation than proxy alignment alone. For this reason, we prioritize records in which this layer is clearly identified.
Importantly, in this study, we also revisited the selection of Nordic Seas records used in latest compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.
We therefore do not apply a stricter selection criterion to the Nordic Seas than to the North Atlantic; rather, we use the tephra layer where available to strengthen cross-regional synchronization, while North Atlantic records benefit from additional independent alignment strategies that are not equally reliable in the Nordic Seas. In the revised manuscript, we will further clarify this rationale.Comment #4. A figure showing the core locations, at least those ones from the North Atlantic and Norwegian Sea would be helpful
Response
We agree and will include a figure showing the locations of the Norwegian Sea, North Atlantic, and Southern Ocean cores used in this study.Citation: https://doi.org/10.5194/egusphere-2026-732-AC3
-
AC3: 'Reply on RC3', Mohamed M. Ezat, 11 May 2026
Status: closed
-
RC1: 'Comment on egusphere-2026-732', Anonymous Referee #1, 07 Apr 2026
As far as I can see, the present manuscript does not present any new data or information than that already appearing in Ezat et al. (2024). The only difference is that now the start of the Last Interglacial has been moved from 128 to 129 ka and thus the so-called Phase I (129-128 ka) now includes the impact of the last part of glacial meltwater originating from remnants of the Fennoscandian Ice Sheet (FIS) and shows winter sea ice extending in the southern Nordic Seas. Phase II (128-124 ka), is associated with enhanced Arctic sea-ice melt and freshwater export, as suggested by Ezat et al. (2024).
The evidence for the switch from glacial meltwater to Arctic sea-ice meltwater remains difficult to assess. The terminal Heinrich stadial HS11 would have led to extreme cooling in the North Atlantic and downstream Europe and may have arrested melting of the FIS. The absence of IRD from 128 ka merely shows that FIS had retreated from the coastline. The decline in Ba/Ca values in N. pachyderma does indeed provide evidence of decreased glacial runoff, though it is not clear (at least to me) whether the distance from the Norwegian coast to the southern Norwegian Sea sites could allow for continued runoff from a small residual ice sheet that remained undetected by the Ba/Ca proxy at 19PC core. The increase in d18O values of N. pachyderma is used as further evidence for a cessation of glacial meltwater influx, presumably because the ice sheet d18O is typically lighter than sea ice d18O. Missing however from the data in the present MS (and in Ezat et al. (2024)) is a discussion of the temperature component of the d18O of meltwater originating from residual FIS and from sea ice; a deconvolved record of SST and d18O of seawater may have been of assistance here.
Finally, the main premise of the study based on inferred melting of Arctic sea ice rests on contentious evidence, as the authors also concede here (e.g. Stein et al., 2017; Vermassen et al., 2023). In the absence of any new material to address these issues, I am afraid I have no choice but to recommend rejection.
PS I have to say that the quality of Fig. 1 and Fig. B1 is not up to standard. The x-axes are too short and the lines too thick to discern details. The use of dark blue, black and brown colors for the three marine sites in Fig. B1 are ill-chosen as it becomes near-impossible to distinguish them.
Citation: https://doi.org/10.5194/egusphere-2026-732-RC1 -
AC1: 'Reply on RC1', Mohamed M. Ezat, 11 May 2026
We thank the reviewer for the assessment of our manuscript. Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. As far as I can see, the present manuscript does not present any new data or information than that already appearing in Ezat et al. (2024). The only difference is that now the start of the Last Interglacial has been moved from 128 to 129 ka and thus the so-called Phase I (129-128 ka) now includes the impact of the last part of glacial meltwater originating from remnants of the Fennoscandian Ice Sheet (FIS) and shows winter sea ice extending in the southern Nordic Seas. Phase II (128-124 ka), is associated with enhanced Arctic sea-ice melt and freshwater export, as suggested by Ezat et al. (2024).
Response
We agree that the study does not provide new data, as we stated at the beginning of the Methods section: “In this study, we synthesize published data based on high-latitude marine sediment records that capture the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean temperature development.”
We also agree that the novelty of the manuscript was not sufficiently emphasized in the original version. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in the literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework (see also response to Comment #2).
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea ice-related processes (see response to Comment #2).
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies (see response to Comment #3).
4) Introduction of the “warming hole” concept in a paleoclimate (LIG)-context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.We will clarify these aspects more explicitly in the revised manuscript. To our knowledge, these elements have not previously been formulated within a single, coherent framework.
Comment #2. The evidence for the switch from glacial meltwater to Arctic sea-ice meltwater remains difficult to assess. The terminal Heinrich stadial HS11 would have led to extreme cooling in the North Atlantic and downstream Europe and may have arrested melting of the FIS. The absence of IRD from 128 ka merely shows that FIS had retreated from the coastline. The decline in Ba/Ca values in N. pachyderma does indeed provide evidence of decreased glacial runoff, though it is not clear (at least to me) whether the distance from the Norwegian coast to the southern Norwegian Sea sites could allow for continued runoff from a small residual ice sheet that remained undetected by the Ba/Ca proxy at 19PC core. The increase in d18O values of N. pachyderma is used as further evidence for a cessation of glacial meltwater influx, presumably because the ice sheet d18O is typically lighter than sea ice d18O. Missing however from the data in the present MS (and in Ezat et al. (2024)) is a discussion of the temperature component of the d18O of meltwater originating from residual FIS and from sea ice; a deconvolved record of SST and d18O of seawater may have been of assistance here.
Response
We agree with the reviewer that distinguishing between freshwater sources (residual ice-sheet meltwater vs. sea-ice-derived meltwater) is inherently challenging. Of course, the reviewer is right that foraminiferal δ¹⁸O reflects both temperature and seawater composition. Our intention was to refer to reconstructed seawater δ¹⁸O rather than foraminiferal δ¹⁸O, but we acknowledge that this distinction was not sufficiently clear in the manuscript.
Reconstructed seawater δ¹⁸O (based on paired foraminiferal δ¹⁸O and Mg/Ca; Ezat et al., 2016) shows higher values during Phase II compared to Phase I and to the later LIG, supporting a reduced relative contribution of isotopically light continental meltwater. Our interpretation is therefore based on the combined behaviour of multiple independent proxies:
- Ba/Ca indicates a decline in freshwater of continental origin.
- IRD absence indicates reduced iceberg discharge at the area (while not fully excluding residual meltwater).
- Reconstructed seawater δ¹⁸O indicates a shift toward higher values.
We fully acknowledge that proxy sensitivities (including transport distance and signal dilution) may affect detection. In the revised manuscript, we will:
• show the calculated seawater δ¹⁸O record in the figures, and clearly distinguish it from foraminiferal δ¹⁸O;
• clarify that our interpretation reflects a shift in dominant freshwater source rather than a binary switch;
• and emphasize that the proposed mechanism is consistent with the available proxy evidence, but not uniquely proven.Comment #3. Finally, the main premise of the study based on inferred melting of Arctic sea ice rests on contentious evidence, as the authors also concede here (e.g. Stein et al., 2017; Vermassen et al., 2023). In the absence of any new material to address these issues, I am afraid I have no choice but to recommend rejection.
Response
We hope that our response to the previous comment clarifies the basis for our interpretation of changing freshwater sources. Importantly, our manuscript does not aim to resolve the debated state of Arctic sea ice during the LIG definitively. Rather, it proposes a mechanistic interpretation consistent with the available multiproxy evidence from the Norwegian Sea and places it within a broader Arctic–Atlantic framework.
We agree that the state of Arctic sea ice during the LIG remains debated, as highlighted by the reviewer. However, a central contribution of this study is to explicitly raise the question of whether the delayed warming observed in the Norwegian Sea during Phase II implies a synchronous delay in the central Arctic Ocean.
In principle, this question would best be addressed using central Arctic records. However, such records are currently limited by large chronological uncertainties – often on the order of tens of thousands of years – and by ongoing debates regarding stratigraphic attribution (e.g., differing assignments of the same intervals to MIS 5e vs. MIS 11 in Vermassen et al., 2023 and Razmjooei et al., 2023, respectively). These limitations hinder robust assessment of phase relationships at the temporal resolution required here.
In this context, our approach is to use well-constrained Norwegian Sea records to formulate a testable hypothesis regarding Arctic–Atlantic coupling, rather than to provide a definitive reconstruction of central Arctic conditions. To clarify this, we will revise the manuscript to explicitly frame the sea-ice mechanism as a hypothesis consistent with available evidence, and emphasize that it is testable, for example using model simulations and additional proxy constraints. We will also strengthen discussion of alternative interpretations and uncertainties.Comment #4. PS I have to say that the quality of Fig. 1 and Fig. B1 is not up to standard. The x-axes are too short and the lines too thick to discern details. The use of dark blue, black and brown colors for the three marine sites in Fig. B1 are ill-chosen as it becomes near-impossible to distinguish them.
Response
We will increase x-axis lengths and reduce line thicknesses for improved readability. We will also use more distinct and colorblind-friendly color schemes.Citation: https://doi.org/10.5194/egusphere-2026-732-AC1
-
AC1: 'Reply on RC1', Mohamed M. Ezat, 11 May 2026
-
RC2: 'Comment on egusphere-2026-732', Anonymous Referee #2, 07 Apr 2026
Ezat and Bakker come up with a notion that the post-glacial oceanographic evolution in the Norwegian Sea during the last interglacial was affected by a 2-phase “Arctic cryosphere pattern” which led to a delayed warm peak in the Norwegian Sea facilitated by a so-called “warming hole”. They refer to 3 nearby sediments cores located at the Iceland-Faroe-Ridge.
The paper is a summary of some results from the last interglacial of the Nordic seas, but basically only looking at the surface water changes of a rather constraint small area in the southernmost Norwegian Sea. The paper ignores almost completely the substantial data that is available from the wide ranges of important parts of the Nordic Seas and in particular along the pathway of Atlantic-derived waters towards the Arctic Ocean through Fram Strait. Especially published cores from farther north, for example on and around the Voring Plateau area, give clear evidence of the various phases during 5e. Regardless of the actually published age models – age models beyond radiocarbon are indeed subject to change anyway – but many published core records could be easily compared on the basis of existing proxy records with those from the IFR.
The one and only claim of the paper is actually that the late delay of the 5e-development is due to enhanced Arctic sea-ice melt contradicting previously proposed long-lasting deglacial effects as prime cause. The authors now claim to have identified a 2-phase development. Apart from the fact that there are more than just 2, in general, it is nothing really new as others have already shown and discussed such phases in detail, too.
In addition, the authors somehow misinterpret the main finding of others: For one, it has already been stated previously by others that the “delay” was caused by residing meltwater after the main deglaciation was concluded – ie. disappearance of iceberg IRD. But icebergs only indicate calving activity on land-ocean margins not how long the abounding western Eurasian continental areas (eg, Norway and Arctic archipelagos farther north) were still glaciated long after the icebergs had vanished from the ocean. And second, the main thing of the previous findings on the delay is that meltwater in vast areas of the Nordic Seas suppressed the inflowing Atlantic water at the very surface, forcing it to flow at greater depth for a considerable time. Only after that meltwater had ceded to exist, did the warm Atlantic water affect the actual ocean surface thus causing the “delay”.
The authors now suggest that enhanced melting of Arctic sea ice is the lone cause for the delay. I wonder how does that work in an overly warm interglacial that apparently had hardly any sea-ice left in the summer and thus could not build up substantial amounts of thick-enough sea-ice during ensuing winters? Work in the Fram Strait clearly show that the delay is found there too, and seemingly much more drastic than further south (see previous work by Zhuravleva et al. from the eastern side, and 2025 by Zehnich et al. for the western Fram Strait); there is also a new work on the last interglacial by Sicard et al. which might be useful.
As mentioned, all data shown now were already published, very few by others, but all relevant ones by Ezat (mostly in the 2024 publication). That being said, however, is no proof that all the proxies employed previously are justified tools. Using, for instance, Ba/Ca in Np as indicator of sea-ice meltwater, contrary to iceberg meltwater, is a far shot from having been properly validated.
This paper is not a study which provides anything new in terms of data. Just the opposite, it is more of a contemplation that muses about a “potential” subject of interest to some. I don’t see the merit of this manuscript for the wider paleoclimate community. As a “rapid communication” it should be rejected.
Few further notes (there could be many more):
Because the relevant cores are from the shallow and rather restricted entrance gateway into the Nordic Seas, they undoubtedly are related to the surface waters in the southern Norwegian Sea. But these cores alone have little to say about the entire Nordic Seas, and certainly even much less about Arctic proper which is even farther north.
Phase 1 is described as 1 ky long (129-128ka), in main figure 1 it starts much earlier and clearly is nothing but part of deglacial Termination 2.
The increasing re-occurrence of polar Np after 124 ka (making up almost entirely the high carbonate content; thus, this proxy reflects shell abundance not warmth as stated(!) which is quite typical for the Nordic Seas in general and rather specific near the end of 5e. Together with increasing IRD means melting icebergs, and hence a surface salinity drop and potentially winter sea-ice. That BPIP25-Ip25 etc. does not show up is likely due to decreased sedimentation rates, causing massive degradation of organic matter, the principal bearer of biomarkers. This might indeed indicate a limitation of this so-called sea-ice proxy and associated biomarkers.
Fig. 1: f and c have faulty y-scales
Fig B1: it misses the “i” label
Note to the Editor:
This so-called paper strikes me as being a bit odd considering both its intention and what’s in it for the paleo-community in terms of novelty.
1- The latter certainly is hardly there – data-wise it’s just a reiteration of Ezat’s own recent publication (2024 in Nat comm) – for a Rapid Communication I would expect something surprisingly new. Instead what I see is a negligence of other peoples’ work and interpretations who actually have generated data on the very subject in the Nordic Seas too.
2- And the former relates to the co-author Bakker, by record mainly a modeler. His contribution is basically zero, at least nothing in the manuscript relates to any of his work, according to the references listed. That makes me wonder if the authors use the rapid communication platform as a pretext, hoping for a quickly citable reference for something that is already in their pipeline, namely a LIG-CMIP simulation using their suggested “warming hole” as the principle novel idea? Of course, I may be wrong...just a thought.
Citation: https://doi.org/10.5194/egusphere-2026-732-RC2 -
AC2: 'Reply on RC2', Mohamed M. Ezat, 11 May 2026
Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. Ezat and Bakker come up with a notion that the post-glacial oceanographic evolution in the Norwegian Sea during the last interglacial was affected by a 2-phase “Arctic cryosphere pattern” which led to a delayed warm peak in the Norwegian Sea facilitated by a so-called “warming hole”. They refer to 3 nearby sediments cores located at the Iceland-Faroe-Ridge.
The paper is a summary of some results from the last interglacial of the Nordic seas, but basically only looking at the surface water changes of a rather constraint small area in the southernmost Norwegian Sea. The paper ignores almost completely the substantial data that is available from the wide ranges of important parts of the Nordic Seas and in particular along the pathway of Atlantic-derived waters towards the Arctic Ocean through Fram Strait. Especially published cores from farther north, for example on and around the Voring Plateau area, give clear evidence of the various phases during 5e. Regardless of the actually published age models – age models beyond radiocarbon are indeed subject to change anyway – but many published core records could be easily compared on the basis of existing proxy records with those from the IFR.
Response
We respectfully disagree with the reviewer’s assessment and clarify the scope and selection criteria of our study. As stated in the Methods, the primary objective of this study is to assess the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean. This requires records that can be placed on a consistent and independently constrained chronological framework. For this reason, the identification of tephra layer 5e-Low/BAS-IV, which is also identified in the North Atlantic record ENAM33, is essential for establishing robust inter-regional correlations. Without such constraints, lead–lag relationships cannot be assessed reliably. We do not ignore other Nordic Seas records; rather, we explicitly evaluate their suitability for this specific objective (see Methods). While additional records from, for example, the Vøring Plateau or Fram Strait provide valuable regional context, many lack the independent chronological constraints required for robust phase comparisons across basins. While we agree that age models are subject to uncertainty, broader inter-core comparisons without consistent chronological control introduce additional uncertainties that are difficult to quantify for the specific timing question addressed here.
We also note that proxy-based alignments (e.g., using the relative abundance of N. pachyderma) can lead to substantially different phase relationships depending on the proxy and alignment strategy used. For example, aligning the studied records (~62.7°N; 4°W) to the very nearby North Atlantic record ENAM33 (~61.27°N; 11.16°W) using such approaches would obscure the distinction between the phases identified by this and previous studies. This highlights that proxy-based correlations are inherently dependent on methodological choices and assumptions, and may obscure rather than resolve phase relationships. Also, while additional Nordic Seas records from the Fram Strait and the Greenland Sea could, in principle, be incorporated through proxy-based alignment, a key limitation is that such approaches do not allow robust quantification of chronological uncertainties associated with tie points. This is particularly critical for the objectives of our study, which focus on assessing the relative timing of peak warming between regions. Without well-constrained and quantifiable age uncertainties, apparent lead–lag relationships cannot be evaluated with confidence and may not be statistically meaningful. In contrast, the use of independently identified stratigraphic markers such as the tephra layer 5e-Low/BAS-IV, which is present at the beginning of the LIG warming peak in the southern Norwegian Sea records, provides a more robust basis for inter-regional correlation. We therefore prioritize records where such constraints are available, rather than relying on less constrained proxy alignments. As part of this study, we also revisit the selection of Nordic Seas records used in previous compilations (e.g., Capron et al., 2014, 2017). In particular, we emphasize that the tephra layer 5e-Low/BAS-IV was previously assumed to be present in some records (e.g., HM71-19), whereas geochemical evidence shows that these ash layers are not equivalent (Wastegård and Rasmussen, 2001). This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas, and represents an additional contribution of the present study.
This approach does not reflect an omission of available records, but rather a careful selection of cores that are suitable for the specific question addressed here. Importantly, our interpretations are restricted to the studied region, and we do not generalize the findings to the entire Nordic Seas.
We also note that the reviewer elsewhere emphasizes that the studied cores primarily reflect conditions in the southern Norwegian Sea and may not be representative of the wider Nordic Seas or Arctic Ocean (see comment #5). We agree with this point, and this possible regional heterogeneity is in fact one of the reasons why we are cautious about applying unconstrained proxy-based alignments between distant Nordic Seas records. This further supports our emphasis on independently constrained chronostratigraphic markers for inter-regional temporal comparisons.
We will revise the manuscript to clarify these points more explicitly, in particular the rationale for record selection, the limitations of proxy-based alignments, and the importance of robust independent chronological constraints for assessing lead–lag relationships.
Comment #2. The one and only claim of the paper is actually that the late delay of the 5e-development is due to enhanced Arctic sea-ice melt contradicting previously proposed long-lasting deglacial effects as prime cause. The authors now claim to have identified a 2-phase development. Apart from the fact that there are more than just 2, in general, it is nothing really new as others have already shown and discussed such phases in detail, too.
Response
To our knowledge, no previous study has identified two temporally successive and mechanistically distinct cryosphere processes during the early LIG in the Norwegian Sea within a single integrated framework. We agree that climate evolution during the LIG may involve more than two phases in a broader sense; however, our focus is on two phases that are robustly expressed in the available chronologically constrained records and directly relevant to the timing of the delay of the peak warming in the Norwegian Sea during the early LIG.
Nevertheless, we agree that the novelty of the manuscript could have been emphasized more. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in the literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework.
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea-ice-related processes (see also comment #3).
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies.
4) Introduction of the “warming hole” concept in a paleoclimate (LIG) context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We explicitly frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas (see also comment #1).Comment #3. In addition, the authors somehow misinterpret the main finding of others: For one, it has already been stated previously by others that the “delay” was caused by residing meltwater after the main deglaciation was concluded – ie. disappearance of iceberg IRD. But icebergs only indicate calving activity on land-ocean margins not how long the abounding western Eurasian continental areas (eg, Norway and Arctic archipelagos farther north) were still glaciated long after the icebergs had vanished from the ocean. And second, the main thing of the previous findings on the delay is that meltwater in vast areas of the Nordic Seas suppressed the inflowing Atlantic water at the very surface, forcing it to flow at greater depth for a considerable time. Only after that meltwater had ceded to exist, did the warm Atlantic water affect the actual ocean surface thus causing the “delay”.
Response
We respectfully clarify that this interpretation does not reflect the way the mechanisms are framed in our manuscript. Our interpretation does not assume a single, persistent meltwater mechanism throughout the early LIG, but instead evaluates changes in the dominant processes through time based on multiproxy evidence. Our interpretation is based on the combined behaviour of multiple independent proxies:
- Ba/Ca indicates a decline in freshwater of continental origin in phase II compared to phase I.
- IRD absence indicates reduced iceberg discharge in the study area (while not fully excluding residual meltwater) in phase II compared to phase I.
- Reconstructed seawater δ¹⁸O indicates a shift toward higher values during phase II compared to deglaciation, phase I and the latter part of the LIG, supporting a reduced relative contribution of isotopically light continental meltwater.
We fully acknowledge that proxy sensitivities (including transport distance and signal dilution) may affect detection. In the revised manuscript, we will:
• show the calculated seawater δ¹⁸O record in the figures;
• clarify that our interpretation reflects a shift in dominant freshwater source rather than a binary switch;
• and emphasize that the proposed mechanism is consistent with the available proxy evidence, but not uniquely proven.
Regarding the proposed mechanism of subsurface Atlantic water intrusion, we do not think available proxy evidence supports a sustained subsurface warming during Phase II. Planktic Mg/Ca data from N. pachyderma indicate relatively higher subsurface temperatures during Phase I, consistent with subsurface Atlantic water influence, whereas Phase II is characterized by lower subsurface temperatures (Ezat et al., 2016). In addition, benthic foraminiferal assemblages indicate “interglacial” assemblages during what we describe as Phase II (e.g., Rasmussen et al., 1996), without clear evidence for the sustained subsurface warming required by this mechanism. We will clarify this point in the revised manuscript. We acknowledge that the processes governing stratification and vertical heat distribution in the Nordic Seas during the LIG remain complex, and we will expand the discussion of alternative interpretations in the revised manuscript.
Comment #4. The authors now suggest that enhanced melting of Arctic sea ice is the lone cause for the delay. I wonder how does that work in an overly warm interglacial that apparently had hardly any sea-ice left in the summer and thus could not build up substantial amounts of thick-enough sea-ice during ensuing winters? Work in the Fram Strait clearly show that the delay is found there too, and seemingly much more drastic than further south (see previous work by Zhuravleva et al. from the eastern side, and 2025 by Zehnich et al. for the western Fram Strait); there is also a new work on the last interglacial by Sicard et al. which might be useful.
As mentioned, all data shown now were already published, very few by others, but all relevant ones by Ezat (mostly in the 2024 publication). That being said, however, is no proof that all the proxies employed previously are justified tools. Using, for instance, Ba/Ca in Np as indicator of sea-ice meltwater, contrary to iceberg meltwater, is a far shot from having been properly validated.
This paper is not a study which provides anything new in terms of data. Just the opposite, it is more of a contemplation that muses about a “potential” subject of interest to some. I don’t see the merit of this manuscript for the wider paleoclimate community. As a “rapid communication” it should be rejected.
Response
We respectfully clarify that our manuscript does not propose enhanced Arctic sea-ice melt as a lone or exclusive mechanism for the delayed warming. Rather, we evaluate changes in dominant processes through time, and explicitly distinguish between a deglacial meltwater-driven Phase I and a subsequent Phase II in which sea-ice-related processes may have contributed to the localized delay in Norwegian Sea warming. We agree that the state of Arctic sea ice during the LIG remains debated. Importantly, our interpretation does not require extensive or persistent summer sea ice, but rather considers the potential role of seasonal sea-ice processes and associated freshwater and/or heat fluxes. Even under relatively warm conditions, changes in sea-ice seasonality, export pathways, and atmospheric circulation can influence freshwater distribution, heat fluxes and upper-ocean stratification (e.g., Sevellec et al., 2017). We will revise the manuscript to clarify the underlying assumptions of our interpretation and to better articulate the range of mechanisms that may explain the observed patterns, including their uncertainties and testability.
We also note that the reviewer raises alternative interpretations regarding sea-ice conditions during Phase II. While the reviewer suggests that Arctic sea ice during the LIG may have been too limited to sustain substantial freshwater export in this comment, the reviewer elsewhere proposes that sea ice in the southern Norwegian Sea may have been more extensive during phase II than indicated by the biomarker proxies due to preservation effects (Comment #7; see also our response there).
Regarding additional records from the broader Nordic Seas and Fram Strait regions, these studies provide valuable regional context; however, as noted in our response to Comment #1, differences in chronological control currently limit robust assessment of phase relationships across regions. We also emphasize that the manuscript synthesizes the available proxy data from the study area where the tephra layer 5e-Low/BAS-IV is identified, including diatom assemblages (Hoff et al., 2019), dinocyst assemblages (Van Nieuwenhove et al., 2011), tephra chronology (e.g., Wastegård and Rasmussen, 2001; Abbott et al., 2014), and planktic assemblages and IRD (e.g., Abbott et al., 2014; Rasmussen et al., 2003). Relevant studies from the broader Nordic Seas are also cited to provide regional context.
We acknowledge that individual proxies, including Ba/Ca in N. pachyderma, are subject to uncertainties and ongoing evaluation and calibration. Our interpretation does not rely on any single proxy, but rather on the combined behaviour of multiple independent indicators: Ba/Ca as an indicator of freshwater of continental origin; IRD as an indicator of iceberg discharge; and reconstructed seawater δ¹⁸O as an indicator of freshwater source characteristics. We will revise the manuscript to more clearly emphasize the limitations of each proxy and to frame our interpretation as one consistent with the available multiproxy evidence, rather than uniquely constrained by any individual proxy. For more details, see our response to Comment #3.
We respectfully emphasize that the contribution of this study is not the presentation of new data, but the integration and reinterpretation of existing records within a coherent chronological and mechanistic framework (see our response to Comment #2), and as we stated at the beginning of the Methods section: “In this study, we synthesize published data based on high-latitude marine sediment records that capture the timing of peak LIG warming in the Norwegian Sea and its relationship to the North Atlantic and Southern Ocean temperature development.”
Comment #5. Because the relevant cores are from the shallow and rather restricted entrance gateway into the Nordic Seas, they undoubtedly are related to the surface waters in the southern Norwegian Sea. But these cores alone have little to say about the entire Nordic Seas, and certainly even much less about Arctic proper which is even farther north.
Response
We agree that the studied cores primarily reflect conditions in the southern Norwegian Sea, and we explicitly frame our interpretation within this regional context. This is consistent with the design and objectives of our study, which focus on this region where robust chronological constraints are available. As noted in the manuscript, individual paleoclimate records represent local conditions unless a broader regional interpretation is supported by appropriate constraints. Our conclusions are therefore restricted accordingly. For more details on the core selection and its suitability to the study’s main objective, see our response to comment #1. We would also like to note that this possible regional heterogeneity is in fact one of the reasons why we are cautious about applying unconstrained proxy-based alignments between distant Nordic Seas sites. If spatial variability in surface ocean conditions was substantial, then correlations based primarily on similarities in proxy behaviour become increasingly dependent on assumptions regarding synchronicity between regions. This further supports our emphasis on independently constrained chronostratigraphic markers, such as tephra layer 5e-Low/BAS-IV, for assessing inter-regional phase relationships. Without this tephra layer, occurring near the end of Phase II, we would not be able to design or fulfill the main objective of this study. Again, for example aligning the studied records (~62.7°N; 4°W) to the very nearby North Atlantic record ENAM33 (~61.27°N; 11.16°W) using such approaches would obscure the distinction between the phases identified by here and previous studies.Comment #6. Phase 1 is described as 1 ky long (129-128ka), in main figure 1 it starts much earlier and clearly is nothing but part of deglacial Termination 2.
Response
Our definition of Phase I is based on the timing of delayed peak warming in the North Atlantic relative to the Southern Ocean, as established in previous studies (e.g., Govin et al., 2012; Capron et al., 2014, 2017). Within this commonly used chronological framework, the interval ~129–128 ka corresponds to the timing of this delayed warming response.
We acknowledge that in the Nordic Seas literature, this interval is often described as part of the late stages of Termination II (including Ezat et al., 2024). However, when records are placed on a consistent age scale with North Atlantic and Southern Ocean records, it corresponds to the phase of delayed peak warming identified in the North Atlantic (e.g., Capron et al., 2014; Stone et al., 2016). Our use of the term “Phase I” therefore follows this inter-regional framework rather than a strictly local deglacial classification.
We further acknowledge that age uncertainties (on the order of ~1–2 kyr) limit the precise delineation of phase boundaries. For this reason, our interpretation places greater emphasis on Phase II, where the identification of the tephra layer 5e-Low/BAS-IV provides a robust cross-basin chronostratigraphic marker and allows more confident assessment of the relative timing between Norwegian Sea and North Atlantic records.
Our intention is not to redefine the deglaciation, but to describe the structure of the delayed warming signal within a consistent inter-basin chronological framework. We will revise the manuscript and figures to more clearly define the phase boundaries.Comment #7. The increasing re-occurrence of polar Np after 124 ka (making up almost entirely the high carbonate content; thus, this proxy reflects shell abundance not warmth as stated(!) which is quite typical for the Nordic Seas in general and rather specific near the end of 5e. Together with increasing IRD means melting icebergs, and hence a surface salinity drop and potentially winter sea-ice. That BPIP25-Ip25 etc. does not show up is likely due to decreased sedimentation rates, causing massive degradation of organic matter, the principal bearer of biomarkers. This might indeed indicate a limitation of this so-called sea-ice proxy and associated biomarkers.
Response
We note that the comment presented by the reviewer refers primarily to the interval after ~124 ka, whereas our manuscript focuses on the early LIG (Phases I and II; 129-124 ka) and the timing of delayed peak warming. Within the interval we define as Phase II, IRD values are low to near-zero and do not show the increase suggested by the reviewer. Importantly, our interpretation of sea-ice-related processes is not based on a single proxy, but on the combined behaviour of multiple independent indicators including IP25 and sterol concentrations as well as diatom and dinocyst assemblages (see e.g., Supplementary Figure B1), which show consistent patterns during the interval of interest. In addition, the biomarker data (IP25 and sterol concentrations) are derived from two sediment cores with substantially different sedimentation rates, yet yield consistent signals, suggesting that the observed patterns are unlikely to be primarily controlled by differential degradation effects. We will still clarify these points more explicitly in the revised manuscript.
We would like also to note that the reviewer raises alternative interpretations regarding sea-ice conditions during Phase II. On the one hand, the reviewer suggests that biomarker-based sea-ice proxies may underestimate sea ice due to degradation effects and low sedimentation rates, potentially implying more extensive sea ice in the Norwegian Sea (this comment). On the other hand, the reviewer argues elsewhere that Arctic sea ice during the LIG may have been too limited to sustain significant sea-ice-related freshwater export (comment #4).Comment #8. Fig. 1: f and c have faulty y-scales
Response
We will correct the y-axis scales in the revised figure.Comment #9. Fig B1: it misses the “i” label
Response
We will correct the missing label in the revised figure.Comment #10. Note to the Editor: This so-called paper strikes me as being a bit odd considering both its intention and what’s in it for the paleo-community in terms of novelty.
1- The latter certainly is hardly there – data-wise it’s just a reiteration of Ezat’s own recent publication (2024 in Nat comm) – for a Rapid Communication I would expect something surprisingly new. Instead what I see is a negligence of other peoples’ work and interpretations who actually have generated data on the very subject in the Nordic Seas too.
2- And the former relates to the co-author Bakker, by record mainly a modeler. His contribution is basically zero, at least nothing in the manuscript relates to any of his work, according to the references listed. That makes me wonder if the authors use the rapid communication platform as a pretext, hoping for a quickly citable reference for something that is already in their pipeline, namely a LIG-CMIP simulation using their suggested “warming hole” as the principle novel idea? Of course, I may be wrong...just a thought.
Response
We note that these comments relate primarily to authorship and perceived intent rather than the scientific content of the manuscript. We respectfully emphasize that all authors meet standard authorship criteria and have contributed substantially to the conception, interpretation, and development of the study – and we do not have currently another manuscript planned on the subject. The manuscript presents a synthesis that integrates proxy-based evidence within a process-oriented framework, and its contribution is therefore conceptual rather than data-driven (as outlined in our responses above).
We also respectfully note to the editor that some reviewer comments appear to reflect differing assumptions regarding both spatial heterogeneity and sea-ice conditions during the LIG. For example, the reviewer suggests in Comment #1 that records from across the Nordic Seas could be easily aligned and compared with our studied cores from the southern Norwegian Sea using proxy alignments, while elsewhere emphasizing possible substantial regional heterogeneity between the southern Norwegian Sea and the wider Nordic Seas in Comment #5 (see our responses to both comments). Similarly, the reviewer argues in Comment #4 that Arctic sea ice during the LIG may have been too limited to sustain substantial freshwater export, while suggesting in Comment #7 that sea ice in the southern Norwegian Sea may have been more extensive than indicated by the biomarker proxies due to preservation effects (see our responses to both comments). We interpret these comments as further illustrating the uncertainties surrounding spatial and temporal sea-ice variability during the LIG, and the importance of independently constrained chronostratigraphic markers and multiproxy approaches in evaluating such questions.Citation: https://doi.org/10.5194/egusphere-2026-732-AC2
-
AC2: 'Reply on RC2', Mohamed M. Ezat, 11 May 2026
-
RC3: 'Comment on egusphere-2026-732', Anonymous Referee #3, 11 Apr 2026
Ezat et al. synthesize existing records from the Norwegian Sea, North Atlantic, and one Southern Ocean site to propose a two-phase pattern of Norwegian Sea surface warming during the Last Interglacial (LIG): (1) an early phase associated with increased IRD and a weakened AMOC, and (2) a delayed warming of the Norwegian Sea relative to North Atlantic SSTs during ~128–124 ka. The authors attribute this pattern to enhanced southward export of Arctic sea ice meltwater.
I agree with the other posted reviews that the manuscript does not present sufficiently novel insights. Both the hypothesis and much of the dataset have already been published or synthesized in Ezat et al. (2024, Nature Communications), and the current study does not substantially advance beyond that work.
I also find the proposed mechanism linking increased Arctic meltwater export to suppressed convection in the Norwegian Sea to be speculative. If the Arctic experienced warming and reduced sea-ice extent during the LIG, it is counterintuitive and against my experience why meltwater export to lower latitudes would increase. Moreover, since the authors cite Guarino et al. (2020)'s modeling work as support for higher SSTs and reduced sea ice during Phase II, the associated changes in freshwater export should be testable. Analysis of this model or similar LIG simulations from PMIP4 would strengthen the argument but is currently missing.
Finally, the use of the tephra layer as a key criterion for excluding certain records requires further clarification. More detail is needed about what this tephra layer is. Also, it is unclear why the same criterion is not applied to the North Atlantic records. If the goal is to compare records chronologically, consistent selection criteria should be used across regions.
A figure showing the core locations, at least those ones from the North Atlantic and Norwegian Sea would be helpful.
Citation: https://doi.org/10.5194/egusphere-2026-732-RC3 -
AC3: 'Reply on RC3', Mohamed M. Ezat, 11 May 2026
We thank the reviewer for the assessment of our manuscript. Below we respond to each point and clarify the scope, novelty, and interpretation of the study.
Comment #1. Ezat et al. synthesize existing records from the Norwegian Sea, North Atlantic, and one Southern Ocean site to propose a two-phase pattern of Norwegian Sea surface warming during the Last Interglacial (LIG): (1) an early phase associated with increased IRD and a weakened AMOC, and (2) a delayed warming of the Norwegian Sea relative to North Atlantic SSTs during ~128–124 ka. The authors attribute this pattern to enhanced southward export of Arctic sea ice meltwater.
I agree with the other posted reviews that the manuscript does not present sufficiently novel insights. Both the hypothesis and much of the dataset have already been published or synthesized in Ezat et al. (2024, Nature Communications), and the current study does not substantially advance beyond that work.
Response
We emphasize that the contribution of this study is conceptual rather than data-driven, focusing on the integration and reinterpretation of existing records within a consistent chronological framework. Nevertheless, we agree that the novelty of the manuscript was not sufficiently emphasized in the original version. We will revise the manuscript to better highlight the following key contributions:
1) Reconciliation of differing interpretations in literature. We show that mechanisms previously discussed as alternatives (deglacial meltwater, e.g., Govin et al., 2012 vs. sea-ice-related processes, Ezat et al., 2024) can be interpreted as operating sequentially within a unified temporal framework.
2) Identification of two temporally and mechanistically distinct phases. We explicitly distinguish between:
• Phase I, characterized by extensive winter sea ice and freshwater input from retreating ice sheets, with delayed warming in both the North Atlantic and Norwegian Sea relative to the Southern Ocean; and
• Phase II, characterized by a more localized delay in Norwegian Sea warming, occurring after North Atlantic SSTs have reached interglacial conditions, and likely associated with Arctic sea-ice-related processes.
3) Extension to Arctic–Atlantic coupling and model context. We discuss implications for the central Arctic Ocean and for 127 ka model experiments, which have not been explicitly addressed in previous “Norwegian Sea” studies.
4) Introduction of the “warming hole” concept in a paleoclimate (LIG)-context. We propose that part of Phase II may represent a localized “warming hole”-type feature. We explicitly frame this as a hypothesis consistent with available evidence and testable through future data–model comparisons.
5) Reassessment of Nordic Seas record selection and chronostratigraphic constraints.
We revisit the selection of Nordic Seas records used in recent compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.
While aspects of the dataset and individual interpretations have been presented previously (e.g., Govin et al., 2012; Ezat et al., 2024), the present study integrates these elements into a unified temporal and mechanistic framework that was not explicitly developed before. We will clarify these aspects more explicitly in the revised manuscript to better highlight the conceptual advance beyond previous work.Comment #2. I also find the proposed mechanism linking increased Arctic meltwater export to suppressed convection in the Norwegian Sea to be speculative. If the Arctic experienced warming and reduced sea-ice extent during the LIG, it is counterintuitive and against my experience why meltwater export to lower latitudes would increase. Moreover, since the authors cite Guarino et al. (2020)'s modeling work as support for higher SSTs and reduced sea ice during Phase II, the associated changes in freshwater export should be testable. Analysis of this model or similar LIG simulations from PMIP4 would strengthen the argument but is currently missing.
Response
We thank the reviewer for raising this important point regarding the interpretation of Arctic freshwater export during the Last Interglacial. We agree that the relationship between Arctic warming, sea-ice extent, and freshwater export is not straightforward, and we will revise the manuscript to clarify both the physical interpretation and the level of inference supported by the proxy data. For example, we emphasize that our interpretation does not rely solely on increased freshwater flux, but more broadly on changes in surface buoyancy forcing, which may arise from a combination of freshwater and/or heat fluxes associated with sea-ice variability (c.f., Sevellec et al., 2017). In addition, changes in atmospheric circulation and wind-driven transport may modulate the magnitude and routing of freshwater export. We also acknowledge that the temporal persistence of such processes on millennial timescales remains uncertain and requires further evaluation using climate-model simulations. Further, we clarify that our study does not aim to establish a direct causal relationship between Arctic sea-ice melt, Nordic Seas convection and regional/local climate anomalies. Given the nature of proxy records, we instead interpret the data in terms of associations between observed surface conditions and plausible cryosphere–ocean processes. We will revise the manuscript to explicitly clarify these points.
We agree that climate model simulations provide an important avenue to test this hypothesis. In the revised manuscript, we will expand the discussion to include relevant results from existing LIG simulations (e.g., including Guarino et al., 2020 where applicable).Comment #3. Finally, the use of the tephra layer as a key criterion for excluding certain records requires further clarification. More detail is needed about what this tephra layer is. Also, it is unclear why the same criterion is not applied to the North Atlantic records. If the goal is to compare records chronologically, consistent selection criteria should be used across regions.
Response
We thank the reviewer for raising this important point regarding the use of the tephra layer 5e-Low/BAS-IV as a selection and correlation criterion. We agree that the rationale for using this tephra layer and its application across regions requires clearer explanation, and we will revise the manuscript accordingly.
First, we would like to clarify that the tephra layer 5e-Low/BAS-IV is not used exclusively for the Nordic Seas records, but is also identified in a key North Atlantic record (ENAM33; e.g., Wastegård and Rasmussen, 2001). This provides a rare and valuable direct chronostratigraphic link between the Nordic Seas and the North Atlantic, independent of proxy-based alignment.
The motivation for using this tephra layer as a key criterion stems from the asymmetry in chronological constraints between regions, as discussed in detail by Capron et al. (2014). In the North Atlantic, age models can be more robustly established through alignment of SST records with Greenland ice-core δ¹⁸O and methane variability within the AICC2012 framework. This allows relatively consistent synchronization of Atlantic Ocean records with Greenland and Antarctic climate evolution. In contrast, establishing chronologies in the Nordic Seas is substantially more challenging for several reasons:
• Common SST proxies (e.g., planktic foraminiferal assemblages) lose sensitivity at low temperatures typical of the Nordic Seas, limiting their utility for precise alignment.
• Foraminiferal δ¹⁸O records during Termination II are strongly influenced by freshwater inputs and stratification effects, complicating their use in correlation.
• Alternative quantitative SST reconstructions are sparse, and preservation effects may further bias proxy signals.
As a result, age models in the Nordic Seas often rely on multiple indirect tie points (e.g., benthic δ¹⁸O, biostratigraphy), associated with relatively large uncertainties (up to some thousand years). Because the primary objective of our study is to assess the relative timing of peak warming between regions, such uncertainties critically limit the robustness of inferred lead–lag relationships when unconstrained records are included. In this context, the identification of tephra layer 5e-Low/BAS-IV in both Nordic Seas and North Atlantic records provides a unique and independent chronostratigraphic anchor, allowing more robust inter-regional correlation than proxy alignment alone. For this reason, we prioritize records in which this layer is clearly identified.
Importantly, in this study, we also revisited the selection of Nordic Seas records used in latest compilations by explicitly considering the presence of the tephra layer 5e-Low/BAS-IV as a robust cross-basin tie point. We note that this layer was previously assumed to be present in some records (e.g., HM71-19; Capron et al., 2014, 2017), but geochemical work (Wastegård and Rasmussen, 2001) shows that these ash layers are not equivalent. This refinement reduces chronological ambiguity and has direct implications for interpreting the timing and structure of early LIG variability in the Nordic Seas.
We therefore do not apply a stricter selection criterion to the Nordic Seas than to the North Atlantic; rather, we use the tephra layer where available to strengthen cross-regional synchronization, while North Atlantic records benefit from additional independent alignment strategies that are not equally reliable in the Nordic Seas. In the revised manuscript, we will further clarify this rationale.Comment #4. A figure showing the core locations, at least those ones from the North Atlantic and Norwegian Sea would be helpful
Response
We agree and will include a figure showing the locations of the Norwegian Sea, North Atlantic, and Southern Ocean cores used in this study.Citation: https://doi.org/10.5194/egusphere-2026-732-AC3
-
AC3: 'Reply on RC3', Mohamed M. Ezat, 11 May 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 958 | 644 | 96 | 1,698 | 195 | 223 |
- HTML: 958
- PDF: 644
- XML: 96
- Total: 1,698
- BibTeX: 195
- EndNote: 223
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
As far as I can see, the present manuscript does not present any new data or information than that already appearing in Ezat et al. (2024). The only difference is that now the start of the Last Interglacial has been moved from 128 to 129 ka and thus the so-called Phase I (129-128 ka) now includes the impact of the last part of glacial meltwater originating from remnants of the Fennoscandian Ice Sheet (FIS) and shows winter sea ice extending in the southern Nordic Seas. Phase II (128-124 ka), is associated with enhanced Arctic sea-ice melt and freshwater export, as suggested by Ezat et al. (2024).
The evidence for the switch from glacial meltwater to Arctic sea-ice meltwater remains difficult to assess. The terminal Heinrich stadial HS11 would have led to extreme cooling in the North Atlantic and downstream Europe and may have arrested melting of the FIS. The absence of IRD from 128 ka merely shows that FIS had retreated from the coastline. The decline in Ba/Ca values in N. pachyderma does indeed provide evidence of decreased glacial runoff, though it is not clear (at least to me) whether the distance from the Norwegian coast to the southern Norwegian Sea sites could allow for continued runoff from a small residual ice sheet that remained undetected by the Ba/Ca proxy at 19PC core. The increase in d18O values of N. pachyderma is used as further evidence for a cessation of glacial meltwater influx, presumably because the ice sheet d18O is typically lighter than sea ice d18O. Missing however from the data in the present MS (and in Ezat et al. (2024)) is a discussion of the temperature component of the d18O of meltwater originating from residual FIS and from sea ice; a deconvolved record of SST and d18O of seawater may have been of assistance here.
Finally, the main premise of the study based on inferred melting of Arctic sea ice rests on contentious evidence, as the authors also concede here (e.g. Stein et al., 2017; Vermassen et al., 2023). In the absence of any new material to address these issues, I am afraid I have no choice but to recommend rejection.
PS I have to say that the quality of Fig. 1 and Fig. B1 is not up to standard. The x-axes are too short and the lines too thick to discern details. The use of dark blue, black and brown colors for the three marine sites in Fig. B1 are ill-chosen as it becomes near-impossible to distinguish them.