|Firstly, thanks to the authors who have engaged with many of my comments. Overall, I think that the dataset presented here will be a useful addition to the Cenozoic paleoclimate community. However, it still remains unclear to me why TEX86 values should fluctuate so widely during the Oligocene (i.e. from 10 and 21’C).|
I have explored some of these comments in detail below:
1) Subsurface production over glacial/interglacial timescales
I am glad to see that both companion papers (Bijl et al. and Salabarnada et al) have now been accepted and/or published (congratulations!). As such, I have more confidence in the authors interpretation re: glacial/interglacial cycles.
However, I would like to point the authors towards an elegant paper published in 2016 (Hertzberg et al; EPSL). Here, the authors reconstruct TEX86 SSTs during the last glacial and interglacial cycle. These estimates are then compared alongside Mg/Ca-derived SST estimates from: 1) upper mixed layer- (0 to 50m), 2) upper thermocline- (25 to 85m) and 3) lower thermocline (60 to 150m)-dwelling foraminifera. During interglacials, TEX86 SSTs closely match Mg/Ca values from mixed layer dwellers. However, during glacials, TEX86 values are much colder (up to 5°C) and appear to represent thermocline temperatures. In other words, there is deeper export of GDGTs during glacial periods.
Therefore, is the glacial/interglacial TEX86 variation at Wilkes Land reflecting a true temperature signal? Or is it just migration of thaumarchaeota in the water column and deeper export?
2) TEX86 variability
There still seems to be large variability in TEX86 values which are unaccounted for by the authors. For example, there can be ca. 5 degrees of variation in the glacial deposits over a very short time interval (e.g. 31 Ma). In your reply, you state that you will discuss this in the manuscript; however, I could not find any reference to this. Could you please point me towards the correct section?
Also, 5 degrees of warming is not insubstantial. The Paleocene-Eocene Thermal Maximum is characterised by a similar magnitude of surface warming (e.g. Sluijs et al., 2006).
The authors state that they are unable to conduct a data-model comparison. However, there is a wealth of literature which they could draw on (e.g. Sijp et al., 2014; Global and Planetary Change).
Figure 2: The colour used to distinguish glacial and mixed facies looks indistinguishable (both look green).
Figure 2: is the key colourblind safe? Check here: http://colorbrewer2.org
Figure 3: Why do you only show error bars for just Wilkes Land?
Figure 3: It may be more useful to plot a LOESS regression through ALL of the Southern Ocean data to understand “Southern Ocean” temperature variability during the Oligocene and Miocene. It would also add more value to your dataset.
Figure 4: this figure has a LOT of information on here. Most of it is essential, but there could be room for improvement. For example:
a) The datapoints which are used to denote changes in IRD, sea-ice dinocysts etc currently overlap within the x-axis for SST.
b) “TEX86-based glacial and interglacial SSTs” has the subheading “linear calibrated SSTs”. However, the corresponding “BAYSPAR SSTs” has no x-axis. Please amend.
Alternatively, just get rid of TEX86 (Kim et al.) and just show BAYSPAR. As the trends are identical (i.e. Figure 3), you only need to show one.
Also, please note that BAYSPAR is LINEAR calibrated!...it just treats uncertainty better.
c) In the Miocene, what are the straight lines in the "linear" and "BAYSPAR" TEX86 SST plots? This really confused me.
Line 240: how do you define “too low” when referring to concentrations of GDGTs?
Line 285: The latest global core-top calibration is Tierney and Tingley.
Line 355: you appear to have said 132 samples earlier.
Line 388: It might be nice to show histograms of the glacial and interglacial TEX86 values…this would illustrate the offset between the two datasets.
Line 373: they can also produce GDGT-1, 2 and 3…just in minor abundances
Line 463: some of your paleolatitudes are VERY precise. I am not sure we can be accurate to one decimal place!
Line 510: I would argue that “striking” is too strong here.
See also new paper by Super et al. in Geology with Miocene and latest Oligocene SSTs