|The revised version of the manuscript is a major improvement on the initial submission, which itself was already a solid contribution. Overall, the authors have done a superb job addressing the initial round of concerns, particularly those regarding: 1) age model consistency, 2) the lack of a d13C mass balance estimate, and 3) the depth coverage of the compiled records. While the last point remains an issue given the very few records below 3 km depth in the Pacific, it is now much easier for the reader to assess this limitation given the addition of Figure 3. It is also very useful to have an estimate of the amount of terrestrial organic matter that would be required to yield lower mean oceanic d13C during the LIG, even with the assumption of closed system behavior. |
Two other issues, regarding AMOC variability and the driver of lower oceanic d13C, were also addressed but in a less complete way. In regard to the first issue, the authors use a temporal and spatial averaging that make it difficult to see any differences in AMOC between the early and late LIG, despite clear evidence for differences in subpolar North Atlantic d13C. For example, Figure 7 of Hodell et al. (2009) shows a ~1 per mil offset between early and late MIS 5e d13C records at depths shallower than 2 km. Such a large change points to marked shift in the endmember d13C of NADW, which was likely associated with weakened AMOC early in MIS 5e. Note that this is not just centennial-scale variability but rather changes in d13C that span several millennia. In Figure S2 of the revised manuscript, however, there is little apparent difference between the early and late MIS 5e results. It appears that part of the problem is that the mean d13C values in Figure S2 are based on an average of all d13C results from the Atlantic basin, including light d13C data in the South Atlantic that reflect the influence of southern source watermasses. The limited difference in mean d13C between the two time intervals is also likely due to the use of overlapping windows (125-120 ka and 128-123 ka). If the spatial domain were instead limited to the North Atlantic between 1.5 and 3.5 km, for example, and the time windows were non-overlapping, would the resulting mean d13C values still be the same? It is also important to note that most of the LIG data are from the eastern Atlantic, where AMOC perturbations to the d13C tracer field will be less obvious than in the western Atlantic. Given these issues, it would be helpful if the authors used different spatial and temporal constraints for their analysis, or if they were to add caveats to the existing results to alert the reader to potential issues.
Lastly, the authors have added a substantial amount of discussion regarding the driver of lower mean ocean d13C during the LIG. The additions regarding ocean-atmosphere gas exchange and terrestrial carbon storage are very helpful, in particular the details about peatlands and permafrost. Yet after walking through the various possibilities, the authors return to their original statement about a long-term imbalance between isotopic fluxes to and from the lithosphere. Do the authors believe the LIG vs. Holocene contrast is simply part of a longer term trend that these two time windows happen to capture? If this were the case, we should expect to see earlier interglacial intervals to have even more depleted mean ocean d13C that MIS 5e. Is there evidence to support such a trend?
And a couple final details…
Line 144: ‘Intermediate’ typically is reserved for depths less than 1000 m, consider using ‘deep’ instead.
Line 260: Check the slopes listed here, they seem to be reversed from those in the figure.
Line 321: The d13C of mantle carbon is ~ -7 per mil, which is quite different than zero. Do you mean that volcanic CO2 has a d13C similar to the atmosphere?