Hello,

Thank you for the opportunity to review this paper. I found that it was an interesting read with useful data, and with some points addressed will fit well into Climate of the Past. I've attached my comments in the PDF below.

Guitián et al. built two new records of ε_p from the Oligocene to early Miocene based on IODP Site 1406 and ODP Site 1168, along with the new ε_p record from Site 925 to supplement previous published low-resolution record. As ε_p, the carbon isotopic fractionation during the photosynthesis of phytoplankton, is determined by both aqueous CO₂ levels and physiological parameters of phytoplankton. To extract variations in atmospheric CO₂ levels based on changes in ε_p from the Oligocene to the early Miocene, the authors evaluated the influence of varying physiological parameters, including cell size and growth rate, on ε_p evolution. The influence of changing cell size was assessed using measured coccolith sizes and a statistical multilinear regression model developed by Stoll et al. (2019), which shows that ε_p is a function of aqueous CO₂ concentration, light, growth rate, and cell radius. For growth rate, they assumed it is controlled by temperature and used the sensitivity of ε_p to temperature, as derived from the culture experiments by Torres Romero et al. (2024), to represent the sensitivity of ε_p to growth rate. They used changes in biogenic silica (bioSi) concentrations in sediments to indicate variations in nutrient concentrations; higher bioSi concentrations suggest increased nutrients and growth rates. They conclude that size and temperature effects have a negligible impact on the long-term declining trend of ε_p, and that a global CO₂ decline is the most likely cause of the decrease in ε_p.

Overall, the manuscript reads well, but a lot of phrases/sentences are confusing and difficult to understand. Please look at the specific comments below. In addition, I see several major shortcomings of the analysis and presentation of the results.  

First, I do not think it is appropriate to use a single value for temperature-ε_p sensitivity (0.48‰ decrease in ε_p per 1°C warming) to represent the sensitivity during the Oligocene to early Miocene. The 0.48‰ decrease in ε_p per 1°C warming is based on the linear regression of temperature and ε_p data (22 samples) from culture studies by Torres Romero et al. (2024), conducted at 22 different combinations of temperature, CO₂(aq), and light. Torres Romero et al. (2024) demonstrates that the sensitivity of ε_p to temperature varies significantly across different CO₂(aq) ranges: 0.37‰ decrease in ε_p per 1°C warming when CO₂(aq) ranges from 4 to 22.5μmol/kg, and 0.95‰ decrease in ε_p per 1°C warming when CO₂(aq) ranges from 22.5 to 41 μmol/kg. Given that CO₂(aq) likely fluctuated across these ranges during the Oligocene to early Miocene, the temperature-ε_p sensitivity may also have varied. Therefore, relying on a single value derived from the culture experiment is not reliable for representing the entire period.

Second, the sensitivity of ε_p to temperature derived from culture experiments does not necessarily represent the sensitivity of ε_p to growth rate in real geological environments. Growth rate is influenced by a combination of factors, including light, temperature, nutrient availability, CO₂ levels, cell size, and other variables, all of which varied significantly over the geologic past. Although Torres Romero et al. (2024) demonstrates that the temperature-sensitive ε_p variation can be fully explained by the temperature sensitivity of growth rate, this does not imply that the relationship is directly applicable to real-world conditions of the geologic past.

Third, more evidence is needed to justify using changes in biogenic silica (bioSi) content in sediments to represent variations in surface ocean nutrient concentrations, particularly nitrate and phosphate, which are critical for coccolithophore growth. Sedimentary bioSi is primarily linked to the Si biogeochemical cycle, which likely differs from N and P cycles. Additionally, sedimentary bioSi is influenced by factors such as dissolution and preservation, limiting its reliability as a proxy for ocean nutrient concentrations, especially for nitrate and phosphate. Furthermore, the interpretation of bioSi results is inconsistent, with the authors at times referring to bioSi content as the delivery rate (Line 305) and at other times as the burial rate (Line 316).

Fourth, the linear relationship between ε_p and SST (or benthic δ¹⁸O) shown for several time slices in Figure 6 is not statistically meaningful, as the sample sizes for most of these time slices are fewer than 10. Therefore, the conclusions drawn from Figure 6 are unreliable.

Lastly, the relationship between ε_p and benthic δ¹⁸O at orbital scales (Figures 7b and 7c) does not yield a clear conclusion, as variations in benthic δ¹⁸O are influenced by both deep-water temperature and ice volume. A more meaningful comparison would be between ε_p and estimated global mean SST (Gaskell et al., 2022; https://doi.org/10.1073/pnas.2111332119) or surface temperature (Evans et al., 2024; https://doi.org/10.1029/2023PA004788).

Specific comments

Line 12-13: The statement “most based on the phytoplankton carbon isotopic fractionation (ε_p) proxy” is not accurate. Between 25 and 16 Ma, most of CO2 estimates are based on boron isotopes, not alkenone carbon isotopic fractionation.

Line 17: Full name of “Ma” is needed here

Line 17-18: “a higher resolution sampling” —higher than what?

Line 20: Please specify “the two sites”.

Line 20: climate dynamics is a broad concept. Please clarify it.

Line 21-22: This sentence is confusing, especially the phrase 'average earth surface temperature evolution.' Are the authors referring to the global mean surface temperature?

Line 22-23: what does the inverse relationship between ε_p and benthic δ18O indicate?

Line 25: what do “specific time intervals” represent?

Line 26: this sentence is incomprehensible.

Line 25-27: Confusing. How does the changing cell size and growth rate explain the divergence between ε_p and benthic δ¹⁸O?

Line 29: “While CO₂ changes likely caused significant changes in radiative forcing” is not connected to the following sentence “SST variation at the examined sites may have been conditioned by regional heat transport”.

Line 31: How does “the relationship between benthic δ18O and ε_p” reflect the phasing between ice growth and global temperature?

Line 34: Please specify “long-term trends”. What trend?

Line 44: please specify what time interval shows “multimillion year warming” and what time intervals shows “cooling trends”

Line 54: delete “globally”?

Line 56: References are needed.

Line 60: Full name of “m. y.” is needed

Line 60: Please specify “two sites on the south American margin”.

Line 61: I would add the name of the Site for the additional North Atlantic record

Line 56-63: please reorganize these sentences. The current sentences are not in logic order. I would put the sentence “In this study, we produce a new long-term record of ε_p over the Oligocene to Miocene time interval at two new, widely separated locations” right after “One approach to evaluate the relative contribution of physiological factors vs CO2 is to produce ε_p records from sites of widely contrasting oceanographic setting…”. The difference of environmental factors (important for physiological factors of coccolithophores) between Site 1406 and Site 1168 should also be clarified in order to make it connected to the previous sentence.

Line 68-69: “an indicator of high-latitude temperature and Antarctic ice sheet extent

and/or volume” is not accurate. Variations in Benthic δ18O are controlled by changes in both deep-water temperature and ice volume.

Line 69: what do “These long-term relationships” indicate? “higher resolution” —higher than what?

Line 70: climate dynamics is a broad concept. Please clarify it.

Line 74: Please add the full name of CO2[aq].

Line 75: what do “These” refer to?

Line 80-81: Please add the equation ε_p = ε_f – b/CO2[aq], which makes it easy to read.

Line 83-87: Some statements are incorrect. Zhang et al. (2013) also applied modern relationships between b and phosphate. Bolton et al. (2016) and Henderiks and Pagani (2007) do not estimate the difference between the modern b value at the site and the paleo-setting b value. Bolton et al. (2016) uses previous formulations of the relationship between cell size and b, which is derived from Henderiks and Pagani (2007).

Line 87-88: Please add references.

Line 88: The sentence “b term is not well predicted by growth rate, light or cell size alone in a diffusive model” is confusing.

Line 93-94: Could the author provide a brief implication of lower Rubisco fractionation?

Line 96: what does “This approach” refer to?  The previous sentence does not mention any approach.

Line 96-97: “the observed slope of ε_p dependence on CO2” is difficult to understand.

Line 103: Change “growth rate” to “growth rate μ_i”

Line 107: Please add references after “While cell size can be estimated from coccolith length”.

Line 116-117: Please clarify how 0.5 ‰ decrease in ε_p per 1°C warming is indistinguishable from the prediction of growth rate effect on ε_p? Krumhardt et al. (2017) only demonstrates the increases in sea surface temperature lead to faster coccolithophore growth rates.

Line 134: Please specify “higher resolution”.

Line 139-140: Replace “The ODP Site 1168 age model” to “The age model of ODP Site 1168”. Similar issues occur throughout this manuscript. Please revise accordingly. 

Line 146: what ages do “the two ODP 1168 samples deeper than the Sr isotope measurements” correspond to?

Figure 1: what is ODSN?

Line 159, 170, and 189: Change the bold text to normal formatting. Similar issues are present throughout the manuscript. Please revise them accordingly.

Line 171: please specify the age of “the young set of samples”

Line 173-176: the sentence flow is not clear. Please reorganize these sentences.

Line 207-208: References are needed.

Line 209-210: Guitián et al. (2020) does not demonstrate that the bulk carbonate is dominated by Reticulofenestra coccoliths.

Line 212-213: Here the authors assume that δ¹³C of bulk carbonates is equivalent to the δ¹³C of coccolith. However, they do not provide any evidence to support this assumption.

Line 213: The citation should be the original paper, McClelland et al. (2017), rather than Stoll et al. (2019).

Line 215: Guitián et al. (2019) describes the method for measuring stable isotopes of benthic foraminifera, not bulk carbonate. Before the section “Estimation of aqueous carbon dioxide δ¹³C”, a section describing the method for measuring stable isotopes of bulk carbonate is needed, including details on sample preprocessing.

Line 237-238: Do the authors use the value in equation (1) or the linear relationship between ε_p and cell radius? The slope of the linear relationship between ε_p and cell radius, derived from a compilation of culture experiments, is certainly different from the value in equation (1). Please specify the sensitivity of ε_p to cell radius used here and provide justification for its selection.

Line 247: Replace “26 ma” to “26 Ma”.

Line 249: Replace “from 28.8 to 29.6 Ma” to “from 29.6 to 28.8 Ma”. Similar issues occur throughout this manuscript. Please revise accordingly. 

Line 250: “Several ̴ 100 ky orbital scale variations of 0.75 ‰ benthic δ18O and bulk δ18O” is incomprehensible.

Figure 2: Did the authors measure δ¹³C of the bulk carbonate for all three sites? The methods section does not clarify which sites were analyzed for bulk carbonate carbon isotopes.

Line 255: Instead of using solid and transparent lines, I recommend using different colors for the lines.

Line 258: White symbols are not visible on this figure; consider using a more visible color.

Lines 259–260: To maintain consistency, I suggest using either 1σ or 2σ for all the error bars.

Line 267: Change “an overall low and stable early Miocene” to “overall low and stable values in the early Miocene”?

Line 272: Is Curry et al. (1995) the correct citation? Curry et al. (1995) is the Initial Report for Leg 154, covering ODP Site 925 alone. It does not include DSDP 516 or ODP 608.

Line 272: The phrase “As seen in sites 1168 and 1406” is confusing, as this paper does not present the Sr isotopic stratigraphy of Sites 1168 and 1406.

Line 282 and 287: “within age uncertainty of the decrease” and “within the age model uncertainty of the minimum” are difficult to understand.

Line 284: “5‰ peak” is confusing. Do the authors mean “5‰ increase”?

Line 295: the title is not accurate.

Line 296: The authors have not discussed the effect of CO2 on ε_p yet.

Line 296:  replace “cell surface area to volume ratio” with “cell size”

Line 304: Confusing. How does the deeper mixing cause the lower mean light levels?

Line 316-322: The main point of this paragraph is not clear.

Line 318 and 323: Misra and Froelich (2012) do not suggest an increase in erosion and weathering rates from the Oligocene to the early Miocene. In fact, their δ⁷Li_SW​ data show little change from the middle Oligocene to the early Miocene.

Line 399: Is “1 ‰ range” typo?

Line 403: what do “these variables” indicate?  

Figure 7: The numeric labels on the x- and y-axes (e.g., "2,2" and similar) are difficult to read. Please adjust them to a clearer format, such as "2.2." Similar issues occur in other figures. Please revise accordingly. 

Figure 7: please add a, b, and c to each panel of this figure.

Line 409: r2=-0.34 is not possible. R-squared is always a positive value. 

Figure 8: Please add the full name of MMCO, Mi-1, LOW, and MOGI.

Line 443: delete very. what do “a different set of feedbacks” mean? Different from what?

Line 446: Please clarify “a substantially different relationship between ice expansion and CO2.”

Line 451-452:  Please delete “and decline in radiative forcing from the greenhouse effect.”

Line 453: there is no evidence to support the claim “the ODP Site 1168 temperature trend reflects global temperature”. SST change of Site 1168 is likely a regional signal.

Line 462: The term "late Oligocene divergence" is not easy to understand. Please consider replacing "divergence" with a clearer term throughout the manuscript to improve clarity.

Line 465: The conclusion section merely repeats the results presented in earlier sections. In addition to summarizing the findings, the conclusion should discuss the broader implications of the results.

Figures and supplementary figures: The current color scheme, particularly the use of red and green in the same figure, is not color-blind-friendly. Please adjust the colors to enhance accessibility and readability for all readers.