|Greenop et al. have revised their manuscript substantially, improved the model tests and tailored it to show the consensus with previously published studies, rather than the focus on minor discrepancies of the original manuscript. I find this manuscript very much improved, but it still needs a few minor but important revisions. For the sake of simplicity, I will list these in chronological order, which is not necessarily the order of importance.|
Line 175: spelling of name: Martínez-Botí
Lines 192/238/248…: wuellerstorfi
Lines 205/208: Müller
Lines 274 ff: what makes these “minor effects”? In particular the Miocene is a period of intense organic matter deposition (Monterey Formation!), that indicates systematic and substantial removal of isotopically light carbon from the ocean. Systematic changes in boundary conditions such as d13C-DIC and pCO2 are not captured in the assumed constant uncertainty, which accounts only for random changes in boundary conditions. As such, these new d11Bsw estimates are still not the ultimate answer, which is generally nicely acknowledged in this manuscript, but should be repeated in the conclusions (see comment below).
Line 460: its changes
Lines 474/475: what is the “plausible” range and why? Please define “plausible”.
Lines 584-587: the pH decrease may be at odds with the pCO2 estimates shown in Beerling & Royer (2011), but more estimates have since been published that support the greater pCO2 reconstructed from stomata (as already shown in Beerling & Royer). For instance, Zhang et al. (2013) show Miocene pCO2 from alkenones >400 µatm, Bolton et al. (2016) went a step further and considering calcification changes in coccolithophores, they suggest Miocene pCO2 may have been at least 50% higher compared to the Plio-Pleistocene. Foster et al. (2012) and Greenop et al. (2014) also suggest Miocene pCO2 ~ 400 µatm, although they used an exceedingly low alkalinity of 1292 µmol/kg to yield these estimates, a value that is even more extreme than the alkalinity inventories tested in Fig. 7 (~1800-2600, considering that modern alk is somewhere around 2200 µmol/kg), which are considered “extreme and inconsistent with geological evidence” (Line 1138). Using more reasonable alkalinity values of ~2000 µmol/kg (Ridgwell 2005, Tyrrell & Zeebe 2004), the Miocene boron isotope pCO2 estimates range closer to 500-600 µatm, clearly consistent with global evidence for warmer temperatures, and acidification as implied by models. There is uncertainty in all of these pCO2 estimates but growing evidence for higher Miocene pCO2 should be acknowledged and the argument that “proxy CO2 and surface water pH estimates are not well described by the linear change in pH” should be revised in the light of this growing proxy data evidence.
Line 704: please augment this sentence to “Despite some disagreements, and different uncertainties associated with each approach, the fact that…”. This is just to reiterate that Greenop’s approach, like all others, has large uncertainties as well, albeit due to different factors.
Line 712: controls
Line 1109: please specify where the “modern ocean” data are coming from, i.e. they are globally distributed and exceed the data shown in the Atlantic profiles. The reference to Foster et al. (2012) is not sufficient.
Line 1117: Please clarify that the error bars with 95% confidence due to external reproducibility only apply to the new data by Greenop et al. (this study), but not to Raitzch & Hönisch (2013), where they represent propagated uncertainties of external reproducibilities of time equivalent benthic foraminifer samples from different core sites in different ocean basins.
Line 1122: the color choice for the planktic foraminifera symbols is unfortunate and should be changed - the orange and red colors are nearly indistinguishable.
Figure 8 caption: Please explain the 0.201 line
Figures 2 and 11: The d11Bsw data of Raitzsch & Hönisch (2013) are not plotted correctly, the data are all lower than originally presented, and so are the polynomial fits. The polynomial fits should probably be replaced by something like a 5-point running mean anyway, but I am puzzled why the data are lower than presented in the original study? This is particularly striking in Fig. 11, where the data are shifted even lower than in Fig. 2, below the 39.61‰ modern seawater d11Bsw estimate that the caption claims the data have been adjusted to (Lines 1175/1176). This must be rectified. The original publication presented the data originating at a modern d11Bsw average of 39.6‰, so no adjustment of the original is necessary or justified.