Lower oceanic δC during the Last Interglacial compared to the Holocene

The last time in Earth’s history when the high latitudes were warmer than during pre-industrial times was the last interglacial (LIG, 129–116 ka BP). Since the LIG is the most recent and best documented warm time period, it can provide insights into climate processes in a warmer world. However, some key features of the LIG are not well constrained, notably the oceanic circulation and the global carbon cycle. Here, we use a new database of LIG benthic δC to investigate these two aspects. 5 We find that the oceanic mean δC was ∼0.2 ‰ lower during the LIG (here defined as 125–120 ka BP) when compared to the mid-Holocene (7–4 ka BP). As the LIG was slightly warmer than the Holocene, it is possible that terrestrial carbon was lower, which would have led to both a lower oceanic δC and atmospheric δCO2 as observed in paleo-records. However, given the multi-millennial timescale, the lower oceanic δC most likely reflects a long-term imbalance between weathering and burial of carbon. The δC distribution in the Atlantic Ocean suggests no significant difference in the latitudinal and depth 10 extent of North Atlantic Deep Water (NADW) between the LIG and the mid-Holocene. Furthermore, the data suggests that the multi-millennial mean NADW transport was similar between these two time periods.

1. Definition of analysed data: Some data analysis covers the whole LIG, some 125-120 ka, some all available data including part of Termination II and of the glacial inception. Similarly for the HOL, with which they compare. This needs to be focused. Define your time interval, but also give reasons for your chosen definition. So far, it is said, that 125-120 ka and 7-4 ka are chosen because δ 13 C is stable. Looking at figure 4c (Pacific in HOL), this does not seemed to be the case, here 5-2 ka is much more stable. Maybe use as has been done in Peterson et al. (2014) the late Holocene 6-0 ka. I also believe taking two time windows which are of the same length might be a valid idea. Furthermore, check on the definition of interglacials (Past Interglacials Working Group of PAGES, 2016) when the community thinks Termination I or II was over and when the last glacial inception started. Please discuss your choice based on such literature widely. Also: I believe somewhere it was written, that only data below 2500m water depth are analysed. Is this always the case? If not, please specify in each and every section, which water depth is considered, also add this information in the figure caption, if this info is not popping up from the figure itself.
2. You are missing one important review on simulating LIG vs HOL carbon cycle, which is Brovkin et al. (2016), which also deals with δ 13 C. Discuss your potential explanations within the framework of that study, which contained results from different models, and which finds some explanations for the carbon cycle in the HOL, but not for the LIG. You might also note, that during the end of LIG / during glacial inception CO 2 and sea level / land ice volume / temperature was decoupled on a multi-millennial timescale, which might indicate towards some processes that are important here (Barnola et al., 1987;Hasenclever et al., 2017;Köhler et al., 2018).
3. line 13: PI is NOT 0.7K cooler than the peak Holocene, this differences in Marcott C2 8. All-in-all, the introduction on climate changes in the LIG needs some revision. Please focus on already existing stacks (which also have regional subdivions), that should also be plotted in Fig 1,  10. line 78: I do not understand how atmospheric δ 13 C is influenced be the total amount of carbon in vegetation and soil, please expend. C3 11. line 80: If you compare atmospheric δ 13 C with modern values you need to include a sentence on the contribution of the 13C Suess effect. Either extend or rewrite to a comparison of the pre-Suess effect values.
12. Introduction: I believe the subsections are not necessary here.
13. line 123 and 133 (maybe elsewhere): Uncertainties are typically going symetrically in both direction, so "±" is not necessary. Also, please state, what these uncertainties are, is this 1σ?
14. Table 1  19. line 215, 222: 3 possible explanations. Maybe there are others which you did not think of so far (e.g. decoupling of CO 2 with other climate records at the end of LIG, see #2). Also, you only in detail investigate AMOC changes, and briefly discuss the others. This should be a bit better balanced. I therefore suggest to move section 3.3 to the discussion, and also ask for some more thoughts on the alternative explanations. C4