In this paper, Campbell and co-authors explore how changes in seasonality and CO2 influence water isotopes during the PETM. To achieve this, they employ iCESM1.2 – this model is the ideal choice because it is able to closely replicate other key climate metrics during the early Eocene/PETM, including global temperatures, climate sensitivity, and the hydrological cycle (i.e., MAP, P/E – see Cramwinckel et al., 2023). The authors find that orbital variability may exert a greater control on water isotopes than a doubling of CO2. This is a really important finding. They also compare model output alongside d18O and d2H proxy indicators for the PETM – they find that del18O proxies (esp. siderites) are typically summer biased. The data-model agreement for leaf wax d2H is more variable with values typically similar or higher than annual values.

This is a useful and important contribution to the paleoclimate community. It also has relevance to both the proxy and modeling community,. The findings are likely to have major implications for future work, especially during other (longer) climate intervals (e,g, the Eocene). 

As such, I have relatively few comments – my suggestions are instead focused on improving the clarity of text/figures, adding additional literature, and exploring some of the implications for other climate intervals moving forward.

Major comment: It is interesting that changes in orbit have a reduced impact at higher CO2 (Page 11, L269) – does this mean that studies from the EECO or Cretaceous are less likely to be influenced by orbital controls. instead, CO2 is the key driver? Conversely, for relatively cold climates (e.g,. Oligocene, Pliocene), are orbital controls more important than CO2? Would be good to discuss this a bit more at the end of the paper. Will be useful for future work!

Other comments:

Page 2, Line 32 – in addition to citing Rae 2021, probably good to cite original studies (i.e., Anagnostou et al., 2020) and give the range of proxy values.

Page 2, Line 35 – see also Inglis et al., 2020 for an independent approach (…that arrives at a similar conclusion)

Page 2, Line 50 – see also Piedrahita et al., 2022 EPSL for insights into orbital variability during  PETM

Page 2, Line 58 – would probably use d2H rather than dD (as recommended by IUPAC)

Page 4, Line 115 – CESM is also in close agreement with hydrological precipitation proxies from the PETM and early Eocene in Cramwinckel et al., 2023 Figure 12 (this was only just published, so I am aware why you could not include this paper!).

Page 5 – note that the carbon preference index is quite low (<1.5) at Mar2X, perhaps suggesting a diagenetic control (see nice review by Alex Sessions in org. geochem for more on this…)

Page 9, L213 – 13’c global warming between 6x and 3x CO2 simulations seems like a LOT and much higher than inferred in proxy synthesis (e.g, Tierney 2022, Inglis 2020 etc). I also don’t remember such high values in Zhu 2019 either – it was more like 5-7 degrees, right? Can you clarify the 13c value?

Page 10, L226 – this is a comment that applies throughout the manuscript – you need to be a little more specific with your language when describing changes in data. For example, “the colder subpolar region does not experience much of an increase…” – but how much of an increase? …and then later “…contributed to the slight increase in relative humidity” – how much did it increase? Another later example is on Page 17, L402 = “one record is closer…”, but what record do you mean? Please add this detail in where appropriate througout, 

Page 10, L228-229: re: narrowing and strengthening of ITCZ during PETM, Cramwinckel et al, 2023 also shows that “…the width of the ITCZ decreases with increased CO2 in five (CESM, COSMOS, HadCM3B, HadCM3BL and MIROC) of the six [DeepMIP] models that provided the meridional wind field variable required to perform ITCZ width calculation” but that… “an ITCZ narrowing signal is not clearly evident within the proxy [MAP] data”.

Page 17, L405-410: as mentioned above, can also be other controls on leaf wax del2H during PETM – e.g., diagenesis. Although a seasonal bias might be important, is this likely to be visible in marine sediments that record a catchment-integrated signal? or will it all be smoothed out?

Page 17, L410 – you are right that fractionation factor a big unknown – one way this could be addressed is to apply a pollen corrected fractionation factor (see Feakins 2013 for first applicaiton, and Inglis et al,. 2022 Paleoceanography & Paleoclimatology for an early Eocene example). This can change absolute values substantially – for example, when applied to sediments from the East Antarctic margin during the early Eocene, the “...pollen-corrected fractionation factor ranges between −107 and −113‰ and is higher than assumed in some Eocene studies (−130‰) (Handley et al., 2012; Pagani et al., 2006). I would therefore try and emphasise how important the fractionation factor could be when undertaking a data-model comparison! Could shift values by 20-30 permil potentially.

Also, note that there is also quite a lot variability in the Inglis 2022 leaf wax de2H record (-30-40 permil) – given your findings, perhaps this could be attributed to orbital variability?

Figure 9/10 – I found it hard to figure out what the top and bottom dashed lines were – could you give these a different colour and/or add these to the key? There are also no error bars on the leaf wax del2H values - can these be added?

Appendix A: please add site names to this table.

Gordon Inglis

General comments:

The aim of the paper is to demonstrate the importance of the insolation variations’ impact during early Eocene. More specifically, the manuscript presents the results of a series of simulations using the CESM water isotope-enabled version for 2 levels of pCO2 (3 and 6 PAL), which mimic the transient evolution of pCO2 during LPTM. Moreover, they compute the impact of different insolation contexts on water cycle to investigate seasonal response and compare it to the water isotopic records.  They attribute the large variation of recorded seasonal water isotope values to insolation changes, which are much larger for weak CO2 values.

The paper is interesting but, unfortunately, globally not convincing on many important points:

1. The simulations are performed in a LPTM context that is largely a transient event (100 Kyears). Therefore, the eccentricity and obliquity may vary. The way the authors intend to capture this variation is done through sensitivity simulations, which are supposed to mimic the envelope of this event. This may be a serious limitation of the exercise, both for the modeling process and for the comparison to data to be sure that the comparison is meaningful.

All these aspects are not seriously discussed at this stage.

1. The isotopic response interpretation in this paper is very limited. There are many reasons to record variations in water isotope (dynamics of the atmosphere: transport of air masses, dynamics of the ocean: origin of evaporation of water...). All these processes are well described in many papers, including C. Risi for example. The authors’ references are interesting from an historical point of view (Craig), but certainly a bit short compared to the large literature and more importantly to the complexity of the water isotope response. Indeed, the authors only evoked temperature and precipitation impact (essentially through amount effects) on water isotope changes. Whereas there are much more processes involved in water isotope changes that are not seriously discussed.

1. The comparison with different proxies recording water isotopes is unclear to me. The authors did claim LPTM only last 100 kyr, but this duration corresponds to a whole eccentricity cycle and to several obliquity and precession cycles. Moreover, the PCO2 increases from two different values that are here fixed to 3 and 6 Pal; therefore, there are two different issues:

First, do we compare water isotope records to 2 different equilibrated climate simulations, which only represents an envelope of the transient LPTM event?

Then, what are the meaning of the isotopic record in terms of model age and duration to be compared with sensitivity model results?

Another important issue that the author avoids completely here is the vegetation impact on water isotope fractionation, especially through evapotranspiration. During LPTM, there is a large change of temperature and hydrologic cycle and therefore, an adaptation on the terrestrial biosphere that has to be accounted for or at least discussed.

For these main reasons, I am not in favor of publishing this manuscript at this stage in Climate of the Past. However, the study is interesting in many aspects, the authors should answer my major and minor comments below, before publication.

Detailed comments:

Abstract:

The limitations of the manuscript should be clarified as far as the point is to compare to LPTM proxies, the equilibrated snapshot simulation using different values of pco2 and insolation is not fully appropriate.

The model / data comparison is also limited, as described above.

Moreover, the proxies used are not easy to interpret and there is no real validation of these proxies in a more constrained context, to be sure that the model data comparison is valid.

The LPTM is a transient event which lasted, as the authors claims, 100 00 yrs. The approximation of this transient event with the difference between two snapshots simulations is therefore not straightforward. It is important to clarify for both modeling and proxy data. From a modeling point of view, it is certainly important to use a water-isotope enabled model to directly compare with data, but what are recording these data during the LPTM transient event?

Introduction:

Early Eocene is maybe less appropriate than MMCO concerning the concentration atmospheric CO2 of the end of this century, because the paleogeography is closer for the MMCO. Moreover, the absence of cryosphere for Eocene considerably modified the earth response to the perturbation.

The reference to Berger for the role of insolation at LGM on the hydrologic cycle is maybe inappropriate.

On the same note, the reference to Ruddiman for the ice sheet is not obvious; Ruddiman mostly depicts the role of orography on climate. The impact of insolation of ice sheet growth for Antarctica or Greenland are better described by Ladant et al., Paleoceanography and Paleoclimatology, 2014, or Tan et al., Nature Communications, 2018 respectively, or by De Conto and Pollard, Nature, 2008, for the transient climate evolution.

Craig is certainly a pioneer reference, but here, the author used a coupled model.

Is the water isotope also simulated in the ocean model?

Anyway, there are much more than the temperature effect and the amount effect to account for, such as the changes of water mass transports due to atmospheric dynamics change, or the origin of the evaporated water from ocean associated to the ocean thermal response, as well as the ocean dynamics.

Terrestrial data are difficult to interpret for many reasons:

1. Age models are often very large and spread over several orbital cycle, which make them not very well constrained.
2. The processes involved in water isotopes variations and records are sometimes a bit ambiguous and not unimodal; those isotope records may hide many different physical processes, not only the increased temperature and amount effects. Moreover, the isotope enabled models analyses clearly demonstrated - in a series of paleoclimate studies for different climate changes from mid Holocene to Eocene - that the complexity to associate the water isotope variations to climate changes (see for instance Joussaume, Nature, 1984, for a pioneering study or Camille Risi et al., Reviews of Geophysics, 2016, and Werner, Geoscience, 2016, for more recent studies)

The o18 enable coupled models include a bunch of mechanisms which allow a disentangling of the different processes responsible of the O18 changes. These processes may be associated with thermal, hydrology effects but also dynamics of atmosphere and ocean.

Indeed, there is a crucial role played by the vegetation in the water isotope variations, superimposed to pCO2 levels and astronomical forcing. The evapotranspiration is certainly an important process to account for. The author should stress this point in the introduction.

In summary, there are many issues in this introduction that need clarifications both in the text and references.

2/ Method:

Because the author uses a previous simulation (Tierney 2022), she gave a minimum of information; even the number of vertical levels in atmosphere and ocean are not specified.

Moreover, we expect to have some information of strength and weaknesses on the LPTM simulation.

More importantly, depending on the scientific question, the insolation simulations may be different. Here, the hypotheses derived from Lunt et al. 2017, but for LPTM, it is possible to compute the exact value of astronomical parameters from Laskar, Astronomy & Astrophysics, 2004, and therefore, to get the min and max insolations for a latitude or the separated values of each orbital parameter.

In this paper, the aim is water cycle, so the question is what is the most appropriate insolation? A discussion is lacking on this important point.

It is very classical to prescribe the PI value for methane, concentration, but it is worth to specify that methane could be higher in a warmer and wetter climate.

The problem concerning the use of terrestrial data is the model age, this data encompasses several astronomical cycles. Therefore, it is not easy to have something else than a large envelope.

3/ Results:

This section is interesting, the analysis of the series of simulations are well described in section 3.1 to 3.3 and compared to data in section 3.4, which is less convincing for me.

3.1/ Response to orbit:

The authors refer most of the time to Berger; no problem with that for Quaternary, because Berger and Laskar are similar, but for the Eocene, it may be interesting to compare both values for orbital parameters.

The interpretation of o18 is not restricted to thermal and hydrological effects. Indeed, the atmosphere dynamics and also ocean dynamics, especially for the Eocene paleogeography, play an important role.

The changes of origin of atmospheric air mass or the changes of evaporation area due to SST changes may drastically impact coastal region, for instance. The same is true for the tropics, where the Hadley cell is modified, and in such regions, the atmospheric dynamics are changed, so the isotopes may change only for dynamic reasons.

3.2/ Response to CO2 and 3.3/ Orbital sensitivity under different CO2 background state:

No major comment, the analyses are interesting but there are two severe limitations:

• A major limitation in this manuscript: there is no impact of vegetation changes, and during LPTM, when the earth experienced such a change, a major modification / adaptation of the biosphere occurred, which will drastically modify the evapotranspiration and the water cycle. This should be discussed in much more details.
• A minor one: the CO2 increase may be associated to a methane increase; this will also modify the temperature pattern and therefore, the hydrological cycle. Such a possibility should be at least discussed.

3.4/ Comparison with data:

The authors focused on LPTM because more data are available.

Could the authors explain how exactly they account for the “Calendar” effect? (See reference from Barthlein et al. for Quaternary, and it is possible to use Laskar or Berger results, because they are similar). What about the LPTM? I don’t understand neither what the authors mean by microclimate, and the point they raised concerning its duration.

4/ Discussion:

The comparison of model/data for Holocene is not really comparable to the LPTM one for many reasons. The first sentence of the discussion is very general and too vague to be appropriate in this context.

There are indeed caveats and limitation in comparing model and data, especially when this comparison deals with a transient event, the onset of the LPTM, and especially when the target is to quantify, which may be the effect of orbital forcing through a seasonal behavior.

The sensitivity to orbital forcing is higher than the one of CO2 in o18.

Is it a real surprise, there are many papers, especially from the Bristol group, which depict that the effect of CO2 on monsoon for different time scales - geologic or glacial interglacial - is very weak.

Another important issue, which is not only for the proxy used here, is what seasonal temperature it represents; if it represents the warmer season, then it is interesting to account for the effect of insolation.

It is only in the discussion section that there is some reference to the bias of CESM o18 limitation; clearly, this is an important point to be discussed in the model description section 2 but also more in details in the discussion section.

The author assesses at the end of the paper that the low resolution is not appropriate for regional analyses, which is a bit strange when compared to their previous sentence on microclimate.

Cited references :

Ladant, J.-B., Donnadieu, Y., Lefebvre, V., and Dumas, C. (2014), The respective role of atmospheric carbon dioxide and orbital parameters on ice sheet evolution at the Eocene-Oligocene transition, Paleoceanography, 29, 810–823, doi:10.1002/2013PA002593.

Tan, N., Ladant, JB., Ramstein, G. et al. Dynamic Greenland ice sheet driven by pCO₂ variations across the Pliocene Pleistocene transition. Nat Commun 9, 4755 (2018). https://doi.org/10.1038/s41467-018-07206-w

DeConto, R., Pollard, D., Wilson, P. et al. Thresholds for Cenozoic bipolar glaciation. Nature 455, 652–656 (2008). https://doi.org/10.1038/nature07337

Joussaume, S., Sadourny, R. & Jouzel, J. A general circulation model of water isotope cycles in the atmosphere. Nature 311, 24–29 (1984). https://doi.org/10.1038/311024a0

Galewsky, J., Steen-Larsen, H. C., Field, R. D., Worden, J., Risi, C., and Schneider, M. (2016), Stable isotopes in atmospheric water vapor and applications to the hydrologic cycle, Rev. Geophys., 54, 809–865, doi:10.1002/2015RG000512.

Werner, M., Haese, B., Xu, X., Zhang, X., Butzin, M., and Lohmann, G.: Glacial–interglacial changes in H₂¹⁸O, HDO and deuterium excess – results from the fully coupled ECHAM5/MPI-OM Earth system model, Geosci. Model Dev., 9, 647–670, https://doi.org/10.5194/gmd-9-647-2016, 2016.

Laskar, P. Robutel, F. Joutel, M. Gastineau, A. C. M. Correia and B. Levrard, A&A,428 1 (2004) 261-285, A long-term numerical solution for the insolation quantities of the Earth, DOI: https://doi.org/10.1051/0004-6361:20041335