The Onset of the Last Interglacial and Fingerprints of Transient Atlantic Meridional Overturning Circulation

Transitions from glacials to interglacials are the largest climate shifts that occurred during the Quaternary. These glacial terminations are characterized by transient changes in the Atlantic Meridional Overturning Circulation (AMOC) and associated alterations in the northward heat transport. It has been a challenge to differentiate between early last interglacial or late penultimate glacial climate conditions at 129-131 ka (thousands of years before present). Neither simulations with a stable interglacial-type nor with a freshwater perturbed AMOC state have reproduced the reconstructed sea surface temperature (SST) ngerprint in the North Atlantic. As previous approaches failed to consider the highly transient nature of the climate system at ~130 ka, the potential of transient, deglaciating AMOC responses and the corresponding impact on North Atlantic SST has yet to be examined. In this study, we employ a fully coupled Atmosphere-Ocean General Circulation Model (AOGCM) equipped with a stable-oxygen isotope module to investigate the underlying AMOC dynamics at the onset of the Last Interglacial (LIG). We demonstrate that successfully capturing both the SST patterns and the calcite δ18O signature in planktonic foraminifera from North Atlantic marine sediment cores necessitates a transiently recovering AMOC. Furthermore, this critically depends on capturing the cold, glacial ocean state prior to the onset of the interglacial. mean summer surface (SSTs) 0.8˚C cooler than PI 12 . To try and explain these cold SSTs, various studies have explored a potential involving remnant meltwater injection into the North Atlantic originating from Northern Hemisphere ice sheet melting throughout the penultimate deglaciation 14,15 , which has been explored with several modelling studies . Such meltwater typically pattern of anomalies, with in proxy record syntheses indicate LIG exactly such 1,9,17,18 . An alternative to in North transition out glacial Termination


The Last Interglacial
Quaternary Interglacials are periods of Earth's history during which continental ice sheet volume and distribution were relatively similar to the Pre-Industrial (PI). They provide excellent testing grounds for our understanding of the Earth system's responses to orbital and greenhouse gas forcing 1

. The Last
Interglacial (LIG) is the most recent interglacial prior to the Holocene. At this time, high-latitude surface ocean temperatures were warmer by at least 1˚C, and polar surface air temperatures by >3˚C relative to pre-industrial temperatures 2 , and the global mean sea level was likely higher than present day by 6-9 meters 3,4 . Strong seasonal shifts in solar insolation changes as well as the preceding deglaciation played a major role in determining the high-latitude climate 2,3 . Attaining a full understanding of the LIG climate has been hampered by a series of challenges both from the perspective of paleoclimate simulations and from proxy-based reconstructions 1 . Modelling the LIG's climate has been thwarted by uncertainties in the con guration of continental ice sheets [5][6][7] , the absolute timing of the prevailing atmospheric GHG concentrations 8 , and, ultimately, the corresponding AMOC state and associated oceanic heat transport 9 . Several proxy-based ocean surface temperature syntheses are available for model-data comparison exercises across the LIG, but until recently they have had strong limitations. In particular, these syntheses did not realistically represent any speci c time interval during the LIG, since they consisted either of one single snapshot 10,11 centered on the peak temperature or represented averaged conditions across the entire LIG interval 12 . In addition, they were based on paleorecords kept on their original timescales causing large dating uncertainties attached to the temperature estimates. Age discrepancies of up to 6 ka can be observed between records from various archives taken from different locations when the age scales are not harmonized 8 . However, the latest temperature syntheses are now based on a coherent temporal framework between paleorecords, and provide quantitative error estimates 12,13 . These syntheses provide a spatio-temporal description of the climate throughout the LIG and consist of multiple time slices of sea surface temperature anomalies relative to recent reference periods (from the recent Industrial Period or PI). Here, we focus on the time slice 129-131 ka (hereafter 130 ka). Data-based reconstructions indicate that the North Atlantic mean summer sea surface temperature (SSTs) at this time was 4.2 ± 0.8˚C cooler than PI 12 . To try and explain these cold SSTs, various studies have explored a potential mechanism involving remnant meltwater injection into the North Atlantic originating from Northern Hemisphere ice sheet melting throughout the penultimate deglaciation 14,15 , which has been explored with several modelling studies 14,16 . Such meltwater events typically manifest as an asynchronous hemispheric pattern of temperature anomalies, with cooling in the North Atlantic accompanied by warming in the Southern Ocean (SO); proxy record syntheses indicate warming in the SO of 1.8±0.9˚C at the onset of the LIG. These types of signals are often recorded in proxy records during the transition from glacials to interglacials. A prominent example relevant for the onset of the LIG is Heinrich Event 11 (HE-11) showing exactly such a seesaw pattern 1,9,17,18 . An alternative scenario to explain the colder temperatures in the North Atlantic may be a transition out of the cold, glacial ocean state that prevailed during Termination II.
Considering the transient nature of the Earth System, climate and ice sheet changes during the LIG were in uenced by the changes occurring throughout the penultimate deglaciation. In this context, adequately capturing the conditions at 130 ka with paleoclimate models is essential to better constrain the starting state for investigations of the LIG's transient evolution. Furthermore, clarifying the state of the ocean circulation at 130 ka is also critical to establish a clear de nition of "full interglacial conditions". Such a de nition would enable intercomparison between the different time periods, and ultimately allow us to deduce commonalities, differences, and the underlying mechanisms during the various Quaternary interglacials. Newer studies highlighted the di culty in establishing a precise de nition of an interglacial 17 , and as such, it is di cult to determine if these cooler conditions in the North Atlantic region occur during the deglacial or interglacial interval. When de ning the length of an interglacial using a threshold value on the ice dD isotope pro le from the EPICA Dome C ice core 19 , the 130 ka interval falls into the interglacial. However, the time period could also be categorized as part of the deglaciation interval when a de nition based on global sea level is applied 20 . Thus, climate changes are apparently not occurring synchronously at 130 ka, thereby favoring a transient interpretation. Such a possibility has not yet been explored by climate simulations -previous modelling studies have examined the LIG from the perspective of a quasi-equilibrated climate state, either with 14 or without 21 freshwater in uence. A conclusive understanding of the dynamics at work during this time could not be reached by these studies.

Possible Causes For A Cooler North Atlantic During The Early Last Interglacial
We hypothesize that the cooler North Atlantic temperatures are caused by a glacial/interglacial transition leading into the LIG rather than a ngerprint of a quasi-equilibrated AMOC weakening induced by freshwater forcing. To test our hypothesis, we use a temporally aligned synthesis of North Atlantic SSTs together with new reconstructions of the isotopic composition of seawater (δ 18  Importantly, the temperature reconstructions are not based on oxygen isotopes, which allows us to separate temperature anomalies from oxygen isotope anomalies. The dataset shows distinct cooling of -3.4˚C ± 1.6˚C for summer SSTs in the North Atlantic (calculated as the July, August, and September average of 130ka minus core-top SST anomalies where the core-top is assumed to be a temporal average of the late Holocene, 3-0ka.) The uncertainty represents spatially averaged 1σ standard error based upon We are able to disentangle and quantify three mechanisms in uencing the δ 18 O c signature in foraminiferal calcite resulting from a freshwater perturbation (described in detail in the Supplementary Information) using our stable-oxygen isotope enabled simulation. This is achieved by re-performing the 0.2Sv injection experiment, but without modifying the isotopic signature of the seawater. This allows us to separate changes induced by the physical effects of a freshwater perturbation from the changes caused by injection meltwater with a strongly depleted δ 18 O signature. With this technique, we are able to determine δ 18 O c changes purely due to oceanic temperatures, yielding an enrichment of +1.0 to +2.5 ‰ in the Nordic seas (Figure 3-a). Secondly, we can detect changes due to modi ed circulation following the freshwater in uence, which induces a slight enrichment of +0.5‰ in the west Atlantic (Figure 3-b), and lastly changes due to injection of strongly depleted meltwater which causes a dilution of the δ 18 O signature in the seawater (Figure 3-c). In this case, the entire region becomes increasingly depleted, with signature changes of -1.0‰ to -2.0‰. While this entanglement is demonstrated for the 0.2 Sv case, these effects are still present in the weaker injection scenarios. As noted earlier, the perturbation of 0.05 Sv yields a markedly smaller RMSE between the reconstructions and the simulation compared to the other injection experiments and might be a suitable representation of the climate dynamics at work at 130 ka BP. Yet, in order to attain these isotopic and temperature values in the simulation, the injection of freshwater needs to be sustained for at least 300 years. This perturbation strength and duration would already result in a sea level increase of 1.4 meters. The proxy reconstructions cover a time span of 131 to 129 ka BP, and maintaining the injection for the entire period would imply a sea level increase of 9.3 meters; which is at the upper limit of the reconstructed sea level changes for the entire LIG period spanning from 130 to 115 ka BP 31,32 . Furthermore, the dilution effect described above would be continuously active during these 2000 years, leading to increasingly depleted δ 18 O seawater values as the freshwater forcing is maintained. Therefore, although the 0.05 Sv injection scenario provides a possible explanation and yields the smallest model/data mismatch based upon the RMSE values, we also explore an alternative mechanism to explain the ocean state reconstructed in the proxy data.

Considering The Ocean State: Glacial Conditions Following Termination Il
A new hypothesis studied here for the rst time is the in uence of the ocean's initial state. This provides an alternative mechanism explaining the proxy reconstructions, which does not rely on freshwater injection. We postulate that the AMOC may still have been in a weak state at 130 ka due to large HE-11 16 meltwater injections occurring earlier during the preceding deglaciation (T-II), and the North Atlantic may thus have still been in a "glacial" state. Circulation characteristics were dominated by the orbital/GHG forcing only after this freshwater source had ceased. To test this hypothesis, we mimic a transition from glacial to interglacial by initializing our model from a cold ocean state, which is an idealized representation of T-II (this experiment runs for a total of 2000 years, see Methods for further details). The  (Figure 4a). However, the changes in the AMOC (Figure 4b) occur without any freshwater in uence, thus no additional sea level changes would accompany such a transition. Therefore, we propose that this transition out of a glacial oceanic state towards an interglacial state is a further possible explanation to understand the features captured by the proxy reconstructions, which may be recording the near-surface temperature and δ 18 O c ngerprints of an AMOC transition rather than a freshwater perturbation.

Implications And Conclusions
In summary, we have shown that the reconstructions at 130ka might be best reproduced by a simulated transition between a weak and a strong AMOC. In our simulations, the ocean circulation remains at a level which mirrors the proxy reconstructions for ~750 years of simulation time if we initialize our model from an idealized T-II/140 ka boundary state (Figure 4). While a freshwater injection of strongly depleted meltwater (Figure 3b) is also able to produce a model/data t; we discard this scenario due to the fact that a relatively long, permanent freshwater injection would result in sea level rise which is higher than suggested by reconstructions, as explained above. Thus, a glacial ocean state should be considered for initialization in order to correctly represent the onset of the LIG.
Finally, although it is beyond the scope of this work to determine a new de nition indicative of "full interglacial conditions", it is clear that such a de nition must include some description of the AMOC state.
When designing PMIP-style model intercomparison projects 18,27 for the LIG, the potential for a transient ocean circulation state should be considered, as such a climate state provides challenges in interpretation. Since simulations of the AMOC are very model-dependent 28 , any deductions made regarding model skill should need to be appropriately contextualized.
Future work needs to examine communalities and differences between T-II and the last deglaciation leading into the Holocene/Anthropocene as well as the onset of other interglacials. While Termination I and the deglaciation out of the Last Glacial Maximum has been well studied [29][30][31][32] , examining the transitions from older glacial periods into the following interglacials have lately gained further interest from the paleoclimate community. Gaining insight into these transitions into past warmer climates will be of key importance in understanding a potential future transition to a warmer climate of the Earth caused by human activities. Figure 1 We use a temporally aligned synthesis of North Atlantic SSTs together with new reconstructions of the isotopic composition of seawater (δ18Osw) and planktic foraminiferal calcite (δ18Oc).

Figure 3
We are able to determine δ18Oc changes purely due to oceanic temperatures, yielding an enrichment of +1.0 to +2.5 ‰ in the Nordic seas (Figure 3-a). Secondly, we can detect changes due to modi ed circulation following the freshwater in uence, which induces a slight enrichment of +0.5‰ in the west Atlantic (Figure 3-b), and lastly changes due to injection of strongly depleted meltwater which causes a dilution of the δ18O signature in the seawater (Figure 3-c).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.