Please find our response in italic below.

Dear Editor,

I have read entitled “Climate change drove Late Miocene to Pliocene rise and fall of C4 vegetation at the crossroads of Africa and Eurasia (Anatolia, Türkiye)” detailed in all my good intentions for the study. This study was focused on using duraylı isotope data, mammal and palaeobotany/paleoclimate based on previous studies. It includes major corrections noted on the manuscript (given below). The main problem is that approaches have been made on the Anatolian and Aegean scale based on data from narrow areas with sedimentological characteristics, which are not specified in the results and abstract. The need for narrowing and simplification of these approaches has been tried to be pointed out in the manuscript. In conclusion, this manuscript is suitable for publication in your journal after the major corrections and edits indicated.    

We thank the Reviewer for their constructive comments and suggestions. Our response can be found below.

Major Corrections:

• Table S2 should be updated as Karacaören- early Pliocene, Bingol-Halifan-late Pliocene, Erzurum-Horosan-late Pliocene (Original Time Slice) as stated in the related manuscript. In the related references, the ages of the samples from these locations are given as listed. Comments and figures related to this correction should be revised.

In our manuscript, we refer to the ages that were provided for sites in the most recent publications while also citing the original publications. Therefore, the age ranges of the three sites above are larger (and as such more conservative) than in the original publications. We will add this information to the column ‘Age Remarks’ in Table S2.

• It is not clear how the concepts of Plateau and Coastal given in Table S2 are evaluated within the interpretations of the manuscript.

We subdivided the sites in Coastal and Plateau sites because a decrease in e.g., calculated cold month temperatures (CMT) and mean annual temperatures (MAT; see Fig. 2B,C) at sites located on the modern plateau may result from surface uplift of the CAP (Fig. 2D) rather than from regional climatic changes.

Unfortunately, the low number of sites in the coastal regions (i.e. the regions that should not be affected by surface uplift-induced cooling) and the large uncertainties in CMT and MAT estimates do not allow us to draw firm conclusions (see lines 253-255). We are happy to clarify this in the manuscript.

• The locations marked in the Figure 1 are all in Central Anatolia, Rhodes, Ermenek, Adana are in the Mediterranean and Pikermi, and Samos is in the Aegean and Mediterranean belt. Aegean is used for Western and GB Anatolia. For this reason, the term Anatolian-Aegean region should be replaced by another term to include all locations. Otherwise, it is understood that the data used in the manuscript belong to the Aegean region of Anatolia. Also, it is useful to evaluate it later in the manuscript.

Yes, indeed, we should have been more precise and explicitly labeled the sites on the Mediterranean coast. We will change this throughout the manuscript.

• In the Miocene and Pliocene, as in the present day, different microclimatic conditions (such as topography, elevation, distance to the sea, etc.) were effective from region to region within Anatolia. For this reason, the interpretations made in the Aegean region throughout the manuscript are confusing due to the inadequacy in the definition of Anatolia-Aegean region.

Please see our replies to the previous two comments. Resolving these two comments should make it clear to the reader which local variations may have affected microclimate.

• The multilayered presence of Central Anatolian volcanic activity must have had a more localized impact on the vegetation in CAP region than the general climatic conditions. The notion that this influence was such that C4 vegetation dominated the whole of Anatolia should be used with caution.

First, as detailed in section 4.2 (lines 214-233), we conclude that intramontane floodplains in Anatolia were dominated by C₄ vegetation between 7.1 and 4.9 Ma, while the mountain ranges were dominantly covered with C₃ vegetation (as reconstructed from paleobotanical datasets). We also address the preservation bias in floodplain sediments, which leads to an overrepresentation of C₄ vegetation in our soil carbonate d13C-based vegetation reconstructions. At no point do we claim that the whole of Anatolia was dominated by C₄ vegetation, we consistently refer to the floodplains (see, e.g., Abstract).

Second, there is independent evidence for the (increased) presence of C₄ vegetation during the Late Miocene in areas surrounding Anatolia, which were not affected by extensive volcanism (including the Aegean/Crete, Iran, Cyprus; lines 201-206). As such, this refutes the hypothesis that volcanism played a decisive role in shaping the dominant vegetation type (with respect to photosynthetic pathway) on geological time scales.

Third, we are not aware of studies that suggest a direct link between volcanism and the rise of C₄ vegetation (in contrary to studies that show that fire activity led to the expansion of (C₄) grasslands in many parts of the world, see e.g., Guarino et al. (2024, Biogeosciences) and Loughney et al. (2023, Geology) for contemporary and Middle Miocene case studies). As such, including volcanism as a potential driver would be speculative.

Best regards

Funda AKGÜN

Please find our response in italic below.

In this manuscript, the authors present a new record of pedogenic carbonate d13C data from the Anatolian floodplains. This paper fills in a critical regional gap in the geologic record and provides contrast with African and Asian records of C3/C4 turnover. My largest concern with this manuscript is that the way the information on samples and data are currently published make it challenging to evaluate the quality of the d13C data. There is useful data in the supplemental tables that should be moved to the main text to make the manuscript easier to evaluate.

We thank the Reviewer for their constructive comments and suggestions. Please find our response below.

Largest concerns:

1. The amount of data published in the supplement does not make it possible to evaluate the quality of the d13C data published.
1. In the methods section the authors should describe what kinds of corrections were done on the data (i.e. linearity, drift, scale corrections). I suggest creating a new supplementary table where the authors publish at least the raw measured values of both samples and standards, as well as the corrected values. Best practice would be to keep a record in that spreadsheet of each correction step. That way, if standard values are updated in the future, it is possible to bring this data into a new reference frame. If there is an associated R or Python script for these corrections, this should be published as well.
2. Were the data all collected in one session? Multiple sessions? Please be clear about how that influenced data correction.

The samples, and as such the data, were acquired over multiple years and multiple sessions that included unrelated carbonate samples from other projects. As mentioned in the manuscript, we follow the procedures in Spötl & Vennemann (2003), which includes scale, linearity, and drift corrections. We included international reference material NBS 18 (IAEA) in all our measurements. We also use a synthetic Merck standard (d13C= -35.69‰ (V-PDB), d18O= 12.38‰ (V-SMOW)) and an in-house Carrara marble standard (d13C= 2.01‰, d18O= -1.74‰), which was weighed in for four different sample sizes for the linearity correction. We will communicate this information more clearly in the manuscript.

The IAEA has established high-precision values for their NBS18 standards over the past decades. As such, the standards are not expected to further improve such that the data need to be reprocessed in the future. Therefore, we do not deem it necessary to upload all measurement files as supplementary materials.

More detail is need about the nodules themselves is needed. What drove sampling choices? Why did the authors run more nodules from some samples and not others? More info on diagenetic evaluation is needed – are there any microscopy or CL images? Please include any images of nodules in the supplement. What diagenetic signs were you looking for? (line 168). How did that influence drilling choices? Given burial depth (line 170), I think the authors are probably right on diagenesis, but just a little more detail would put readers at ease.  

Sampling choices were entirely based on the availability of pedogenic carbonates in outcrops, which were taken over several months of fieldwork. When drilling the samples, only micritic parts devoid of clasts were drilled (i.e. sparry calcite in veins -if present- was not sampled).

We do not have microscopic or CL images of the samples. However, we will provide pictures of representative pedogenic samples for the supplementary materials. The d18O values associated with the d13C values presented in this manuscript do not show any signs of diagenetic alteration. The manuscript with these d18O values was accepted for publication in Earth and Planetary Science Letters on May 01 2025 (see also reply to next comment). Additionally, the accepted publication that includes the d18O data includes two dual clumped isotope (cap47, cap48) temperatures for ca. 5.4 Ma pedogenic carbonate (10CKK64) and for ca. 5.4 Ma lake carbonate (sampled in the Adana basin in southern Turkey). Both analyzed samples plot within error of the dual clumped isotope equilibrium line, which implies that their isotopic compositions are devoid of significant kinetic biases. Second, their cap47 temperatures are 24.1 ± 3.5°C and 15.6 ± 3.3°C for the lacustrine (AVA08) and the pedogenic (10CKK64) carbonate, respectively. The obtained temperatures show that the carbonates precipitated under conditions conformable with soil formation rather than with elevated diagenetic temperatures. More evidence for the primary origin of the sampled carbonates comes from the d18O values and thin sections of lake carbonate that was sampled in the region (in some cases in the same sections, see Meijers et al., 2018, Earth and Planetary Science Letters), as well as the associated d13C values (Meijers et al., 2020, Geosphere), which show no evidence of diagenesis.

1. The authors should also strongly consider putting the d18O data in the supplement, even if it’s not discussed in the manuscript so those data are freely available to future users. Improvements in pedogenic carbonate modern calibrations (using 17O and cap47) are showing that those data might be more useful than previously thought. Make sure that it is clear how those data were corrected as well.

The manuscript with the d18O results was recently accepted for publication in Earth and Planetary Science Letters. The data were interpreted in terms of surface uplift (stable isotope paleoaltimetry). We can therefore include the d18O values in this manuscript, such that the entire dataset is available in one location. The d13C and d18O results were, of course, measured and evaluated simultaneously.

Line by line comments:

Line 23-24: This sentence is confusing, please clarify

Yes, indeed, we will change this to: ‘The transition to C₄in Anatolia coincides with a similar shift from C₃ to C₄vegetation in southern Asia and is generally attributed to the Late Miocene Cooling in response to decreasing atmospheric pCO₂.’

Figure 1: The hexagon is hard to differentiate, consider adding color (maybe something in a cool color spectrum?) or changing the shape so it stands out.

Yes, we will do this.

Lines 64 – 75: I recognize that this is a stylistic choice, but I find it jarring to put the conclusions into the introduction of a paper in this format. I think it makes sense for shorter format papers, but here it feels like putting the cart before the horse. Consider removing or reframing

We will reframe this.

Line 88: did you measure any rhizoliths? If so, need to clearly differentiate that data.

We certainly analyzed not only carbonate nodules but also carbonate concretions that formed from rhizoliths (root casts), as they were always collected when found in the field. Whenever identifiable, we will add that information to the data table.

Line 88: It would be more appropriate to use Munsell soil colors here

Unfortunately, no Munsell soil color chart was available during field work and field pictures do not allow for proper color analysis. In general, we believe that color information would benefit our study of soil carbonate d13C values.

Line 109: Because so much of the manuscript relies on the geochronology being right, it is worth it to describe in a little greater detail in the manuscript. I know its published, but make it easier for the readers to evaluate the quality of the geochron data. What kind of radiometric dating? What kind of age model?

The details (i.e. type of radiometric dating, minerals dated etc.) are all available from Supplementary Table S1. However, we will add a supplementary text file which gives a clearer overview of the detailed chronologies and the underlying ages for each of the sites.

Line 123: Stronger justification of the exclusion of this data is needed since so much of the final conclusion is based on hydrologic change in the region.

We will do this for the one Asian site that we excluded (out of eleven) based on the original multi-proxy interpretation of the authors (Zhuang et al., 2011, Earth and Planetary Science Letters) that the high soil carbonate d13C values are a direct result of low soil respiration rates.

Results section (lines 131 – 159):

1. I would suggest that the authors consider significantly shortening the results, and instead translate much of the text into a table. There is a lot of useful information hidden in table S1 that should be moved to the main text (for example, moving some of the geochron details would help alleviate the line 109 comment).

We will move the information to a table. As suggested for the comment on line 109, we can move the geochronological information to a supplementary text file, as including the information in the main text would fill too much space (ca. 1.5 page).

1. It may be more useful to describe the salient trends that we can see in the data rather than the range of isotope values.

We will put more emphasis on the trends after moving the details to a table (see previous comment).

1. The authors present sample means for each of their samples, but the underlying data structure is clearly not a normal distribution (some samples are surely biased by outliers), and so the authors should report median values as well as mean values. The authors should strongly consider calculating and reporting the 2 standard error value for their samples. This is becoming the more common-place error statistic to report in the stable isotope community.

We will certainly do this and test our data accordingly. However, this does not change any of the trends we describe in the manuscript.

1. Be clearer as results are presented if you continue to just report 1 SD (again, this would be easier in a table)

We are not sure we understand this comment, but we hope that adding the table will resolve the issue.

1. Reference figure 2 earlier in the results

Yes, we will do this.

Line 170: please also edit this sentence to reflect the burial depth for Mustafapaşa (400 m).

Consider it done.

Line 176: are there any paleosol characteristics that support or refute this? Color, texture, morphology, etc.

Paleosol classification was not systematically recorded for each sampling site. The data available do not permit to support or refute this hypothesis, which is based on paleobotanical data.

Figure 2:

1. In A: please add a better minor grid and make the average values solid to make them easier to see and differentiate?
2. In B & C – please make the data points a size or two larger to make it easier to read and see.

Yes, we will take care of both suggestions above.

Line 194: include citations here

We could add them. However, they are already listed in the Table S2 and include 20 citations, which might be a bit much for a caption.

Lines 214 – 232: The way this paragraph is written (juxtaposing the contradicting data) is a little confusing, and needs a little reorganization and refining.

We will do this.

Line 244: is this a decrease of up to 10°C?

The reviewer refers to line 254. Yes, indeed, a decrease of up to 10°C. We will point this out more clearly.  

Line 264: indent new paragraph

Consider it done.

Figure 3:  please make it clearer what AMOC + and MSC in c) are in the caption. Strongly consider swapping the x axis direction, with 0 Ma on the right side. This ultimately is stylistic, but it is quite standard in paleoclimate (e.g. Westerhold et al., 2020). It would make your plot easier for other scientists to use.

Yes, we will improve the explanation of AMOC+ and MSC. We agree that swapping the horizontal axis is appropriate and will do so.

Line 314: This mechanism has been shown in modern experimental grasslands. The authors should make reference to Knapp et al., 2020 (https://doi.org/10.1073/pnas.1922030117)

Thank you very much for pointing out this study. We will gladly add a summarizing line and the reference.

Table S1:

• Please make the headings for latitude and longitude clearer (rather than N / E), please also include the reference frame (e.g. WGS84)

Yes, we will do this.

• Please justify the merging of the Bakali sites more clearly. It is also unclear where the age for this sample came from. Please clarify.

The samples of the sites were all taken from time-equivalent deposits in a small area (± 2000 m horizontally), hence we grouped them. The Quaternary age of the samples was derived from the 1:500,000 geological map of Adana, as referred to in line 111 in section 2.2 (‘Pedogenic carbonate chronology’).

• Are there any estimates of paleoelevation? If so, please include

The paleoelevation estimates are part of a manuscript that was accepted for publication in Earth and Planetary Science Letters (see our reply further above).

• Chronologies: If there are samples where adjustments for sedimentation rate were made, please make that clear

In terms of uncertainty of the assigned ages, we regarded the entire interval between dated ignimbrites. We will point this out more clearly.