the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Reconstructing 15,000 years of southern France temperatures from coupled pollen and molecular (brGDGT) markers (Canroute, Massif Central)
Léa d'Oliveira
Lucas Dugerdil
Guillemette Ménot
Allowen Evin
Serge D. Muller
Salomé Ansanay-Alex
Julien Azuara
Colline Bonnet
Laurent Bremond
Mehmet Shah
Odile Peyron
Abstract. Climatic changes in southern Europe during the Holocene are characterised by a strong spatial and temporal heterogeneity whose patterns are still poorly understood, notably the presence or not of a Holocene thermal maximum (HTM; 10,000–6,000 cal. BP). The reconstructed climatic patterns also differ according to the proxies used (e.g., pollen, chironomid) and the latitude of the data. Here, a multi-proxy approach combining pollen and lipid biomarkers (branched Glycerol Dialkyl Glycerol Tetraethers, brGDGTs) is applied to the Canroute sedimentological sequence to reconstruct the climatic variation over the last 15,000 years in southern Massif Central, France. This area is poorly documented in terms of vegetation and climate. To provide reliable climate reconstructions, we have (1) performed a multi-method comparison based on transfer functions applied to pollen (MAT, WA-PLS, BRT) and molecular biomarkers (brGDGTs), (2) investigated the role of modern databases/calibrations in climate reconstructions. Three different calibration databases were tested for pollen data: one global based on a Eurasian Pollen Database, and two regional databases corresponding to Mediterranean/Temperate Europe and Temperate Europe/Scandinavian databases respectively. Nine global calibrations were tested for lipid biomarkers including eight for soil and one for peat. The use of different modern databases highlights the importance of considering environmental and ecological constraints when using transfer functions on pollen sequences. Pollen and brGDGT-inferred climate trends are consistent, notably for the Lateglacial, the Early and Late Holocene. However, the reconstructions notably differ concerning the presence of a Holocene thermal maximum with the Modern Analogue Technique (MAT) pollen-based method but not apparent with the BRT pollen method nor brGDGT. The temperature reconstructions estimated from the two independent pollen and lipid proxies are then compared to regional published climate signals (chironomids, pollen, molecular biomarkers) to better derive global regional climatic patterns in South Europe. Altogether, our results from the Canroute sequence and those already available in southern Europe reveal that for the Lateglacial and Early Holocene, the regional climate trends are coherent between sites and proxies, supporting the reliability of their reconstructions despite some discrepancies. During the Holocene, the temperature signal of Canroute does not indicate the clear presence of a pronounced mid-Holocene thermal maximum, but rather stable and warmer temperature compared to Lateglacial ones and overall negative anomalies compared to modern annual temperatures.
Léa d'Oliveira et al.
Status: final response (author comments only)
-
RC1: 'Comment on cp-2023-15', Anonymous Referee #1, 10 May 2023
The authors present a multiproxy study for reconstructing past mean annual air temperatures based on two independent and complementary approaches based on pollen record and a specific bacterial-sourced biomarker based on brGDGTs for the last 15 kyr in the southern France. They have evaluated different existing temperature-calibrations for each proxy and some reasons for biases in derived-temperatures.
They have also documented other previously published studies from Europe using different proxies to derive temperature reconstruction including those studied, to explain similarities and differences with climate signal during the last deglaciation and the Early and Mid-Holocene. Therefore, the authors have made a nice work including different proxies and evaluating different calibrations, and an updated review of other works related with the topic. Thus, the idea and methods are well, I find the manuscript well-written, and figures are appropriated.
However, I have some general comments and some specific revisions that I suggest reviewing for the final version.
One concern is about the use of different modern pollen database. It is intriguing that you select the Scandinavian calibration, given the study site location. Do you have a climatically explanation for that?
It is true that the MEDTEMP calibration gives slightly low R2 values (0.91 for the BRT method), but the EAPDB calibration gives comparable R2 values for the BRT method (0.92), although with higher RMSE values. In my view, this EAPDB calibration gives a reasonable MAAT profile and is comparable to the TEMPSCAND one. Why do you do not include it in the MAAT profile with the brGDGTs-MAAT reconstructions? Please add some discussion about that.
In general, the climate variability during the Lateglacial has been characterized by warm conditions during the B-A (14.700-12.900 yr) and a cold YD period (12.900-11.700 yr) before the warm Holocene. This trend is likely showed by the MEDTEMP calibration but not really showed with the brGDGTs calibrations. The brGDGT-MAAT calibrations show similar trends to the EAPDB or TEMPSCAND calibrations. Do you have a hypothesis for that?
Specific revisions:
Line 37: delete nevertheless and replace fluctuations by oscillations
Line 38: at millennial timescale
Line 38: indicated
Line 74: please add after palaeotemperatures depending on the type of the archive and the region
Line 92: please add De Jonge et al., 2021 before Robles et al., 2022b
Please review references by Robles et al. along the text.
Robles et al., 2022a is now Robles et al., 2022
Robles et al., 2022b is now Robles et al., 2023
Line 96: there is other compared pollen- and brGDGT-based studies, please add for instance Watson et al., 2018
Line 107: Conroute peatland
Line 137: Add the acronym (OMC), and then use it in line 139
Line 151: Did you monitor the m/z 1303 instead of 1302?
Lines 160-161: I would denote mr and mrs as mr-1 and mr-2
Lines 161 and 162: Bayesian
Line 164: add the calibration error or RMSE for the Bayesian calibration
Line 261: isoGDGTs
Line 299: I would skip mr and mrs to refer multiple regressions, see my previous comment. You can refer as mr-1 and mr-2 and cite the corresponding reference (De Jonge et al., 2014b).
I find confusing the codes of the different soil and peat calibration used, and I would avoid the initial of the references, since they are indicated in Table 1 and along the text. Then I would replace some of them as:
Soil MBT’ (Peterse et al., 2012), Soil MBT (De Jonge et al., 2014b), Soil MBT’5Me-1 and Soil MBT’5Me-2 (Naafs et al., 2017b), Bog MBT’5Me (Naafs et al., 2017a), Index1 (De Jonge et al., 2014a), mr-1 and mr-2 (De Jonge et al., 2014b).
Line 338: WA-PLS
Line 339: 1.9 ºC
Line 548: the presence of the HTM in the Mediterranean region.
Line 570: a Late-Holocene
Table 1: Please add the calibration error of each equation before the reference.
Figure 4 Panel (a): Abundance (%), GDGT-0, GDGT-1, etc. Cren and Cren’ instead of GDGT4 and Crenach’, in consistence with the text, whereas it is referred as GDGT-0/Cren ratio.
Panel (b): you have identified some double prime isomers, please add some discussion about their significance in the manuscript. Are they detected along the whole record or just at some intervals, likely between 6.600-5.000 yr? Perhaps their distribution suggests one of the brGDGT-based calibration.
Figure 7: Please add different symbols for the different profile in each panel for better reading in a black and white printed version. In panel b, replace as WA-PLS
Figure 8: Please rename the codes of soil and peat calibrations avoiding the initial of each reference, as Bog, Index1, Soil, mr-2, Soil MBT, mr-1, mr-2, etc. See my previous comment.
Figure 9: Please add different symbols for the different profile in each panel for better reading in a black and white printed version. Also, in panel (b), I would rename as Soil Bayesian, mr-2 and Index1 (see my previous comment). Accordingly in the figure caption, please rewrite as: Soil Bayesian (XX symbol and dark blue line; Dearing Crampton-Flood et al., 2020), mr-2 (XX symbol and light blue line; De Jonge et al., 2014b), and Index1 (XX symbol and red line; De Jonge et al., 2014b).
Additional reference:
Watson, B.I., Williams, J.W., Russell, J.M., Jackson, S.T., Shane, L., Lowell, T.V., 2018. Temperature variations in the southern Great Lakes during the last deglaciation: Comparison between pollen and GDGT proxies. Quaternary Science Reviews 182, 78–92. https://doi.org/10.1016/j.quascirev.2017.12.011
Citation: https://doi.org/10.5194/cp-2023-15-RC1 -
RC2: 'Comment on cp-2023-15', Cindy De Jonge, 24 May 2023
I have reviewed the manuscript “Reconstructing 15,000 years of southern France temperatures from coupled pollen and molecular (brGDGT) markers (Canroute, Massif Central)”, submitted to Climate of the Past by Léa d’Oliveira and co-authors. The manuscript describes a multi-proxy approach (i.e. biomarker (brGDGT) and pollen-based temperature reconstructions. I have read through the paper with interest, as it represents a novel datatset with the potential to add to our understanding of Holocene climate variability in Southern Europe in the Holocene. The text is well-written, perhaps a careful look can remove some of the repetitions between the results and discussion section. I have several small comments on the introduction, as it does not match the content of the paper 100% and the references were not complete.
In addition, I still have some major comments on the interpretation that would need to be addressed or clarified before I (or the reader) can fully support the interpretation of the climate record. I hope that the authors agree that these would aid to increase confidence (or identify in which areas there is less confidence) in the interpretation of the XRF data and GDGTs. I have less experience with the pollen models, and as such I was not able to comment on the completeness of references in these sections.
Major comments:
Based on accumulation rate, XRF and pollen, there is a sequence of 3 different wetland/peat types. The presence of a mid-Holocene period with a low pH sphagnum peat has been recognized by the authors, but also the change into the late Holocene wetland is associated with a large change in the pH, as indicated by the CBT5ME value. A change in soil/peat pH can have a significant impact on GDGT based temperature, with potential offsets in the range of 5-10 degrees Celsius (De Jonge et al., 2021, full reference see minor comments). The reconstructed pH (based on brGDGTs) should thus be discussed, to potentially constrain this impact. For this, the CBT’ or IR can be calculated and used to reconstruct a pH variability. The ratios that can be used to constrain confounding factors (CI, CBT’, IR) should reported before variations in GDGTs are interpreted as temperature (i.e. before or in section 3.3.3).
The authors have employed a wide variety of brGDGT proxies, including several that have been considered outdated or less relevant in literature (the same manuscripts that propose the proxies). Afterwards, the selection of the ‘best’ temperature reconstructions seems to rely on which temperature trend matches better with expected temperature variability. I would ask the authors to reduce the number of ratios used, and to not to exclude temperature reconstructions based on whether they match expected variability. The comparison of offsets with current MAAT all falls within the calibration error, and can not be used as a real argument for selection between calibrations.
The PCAs seem to be based on non-standardized counts (for XRF) and abundances (for GDGTs). Please consider recalculating these based on standardized relative abundances to show compositional variability.
Minor comments:
L 14: Latitude of data, I suggest to replace by latitude of record, or even latitude of the lake?
L 17, add the word ‘change’ after ‘vegetation and climate’.
L 39. Specify that this is a temperature optimum.
L 51. Mediterranean basin? Is this the region you would refer to to describe the general area of the peat core?
L 53. What do the authors mean with ‘site effect’? Also the impact of erosion on the interpretation of the terrestrial archives is mentioned too briefly and therefore not clear.
L 62. Indicate that this is not a complete representation of the diversity and amount of lacustrine studies where GDGTs are used for paleoclimate reconstruction.
L 63. The temperature dependency of isoGDGTs is not a subject of this paper. Please remove it from the introduction. Instead, the GDGT0/cren ratio is used in the paper, but not introduced. Include the literature on this ratio in the introduction.
Also, the CI ratio (based on brGDGTs) is used but not introduced. Include the literature on this ratio in the introduction.
L 71. Include the recent studies from Halamka et al. (2023) and Chen et al. (2022) here, that have shown GDGT production in a bacterial culture. Zeng et al. (2021) and Sahonero et al. (2021) are better references to support the statement that the producers of brGDGTs are still a subject of investigation.
- Zeng, Z. et al. Identification of a protein responsible for the synthesis of archaeal membrane-spanning GDGT lipids. Nat Commun 13, 1545 (2022).
- Sahonero-Canavesi, D. X. et al. Disentangling the lipid divide: Identification of key enzymes for the biosynthesis of membrane-spanning and ether lipids in Bacteria. Science Advances 8, eabq8652 (2022).
- Halamka, T. A. et al. Production of diverse brGDGTs by Acidobacterium Solibacter usitatus in response to temperature, pH, and O2 provides a culturing perspective on brGDGT proxies and biosynthesis. Geobiology 21, 102–118 (2023).
- Chen, Y. et al. The production of diverse brGDGTs by an Acidobacterium providing a physiological basis for paleoclimate proxies. Geochimica et Cosmochimica Acta 337, 155–165 (2022).
L 72. More recently, Naafs et al. (2021) also support that the structure of the membrane lipids determines membrane fluidity.
- Naafs, B. D. A., Oliveira, A. S. F. & Mulholland, A. J. Molecular dynamics simulations support the hypothesis that the brGDGT paleothermometer is based on homeoviscous adaptation. Geochimica et Cosmochimica Acta 312, 44–56 (2021).
L 78. Dearing Crampton Flood does not discuss lake sediments (but instead soils and peats, as they don’t find a difference between the temperature dependency of these groups). Include a correct reference(s) here (see suggestions below).
- Raberg, J. H. et al. Revised fractional abundances and warm-season temperatures substantially improve brGDGT calibrations in lake sediments. Biogeosciences 18, 3579–3603 (2021).
- Martinez Sosa, P. et al. A global Bayesian temperature calibration for lacustrine brGDGTs. (2020).
- Russell, J. M., Hopmans, E. C., Loomis, S. E., Liang, J. & Sinninghe Damsté, J. S. Distributions of 5- and 6-methyl branched glycerol dialkyl glycerol tetraethers (brGDGTs) in East African lake sediment: Effects of temperature, pH, and new lacustrine paleotemperature calibrations. Organic Geochemistry 117, 56–69 (2018).
L 81. ‘specificity of the brGDGT proxy’ is too vague, consider removing as the confounding factors of brGDGT temperature dependency are explained in more detail below.
L 83: For brGDGTs in soils specifically the impact of pH change is a demonstrated confounding factor. This should be mentioned here with a reference, suggestion given below.
- De Jonge, C. et al. The influence of soil chemistry on branched tetraether lipids in mid- and high latitude soils: implications for brGDGT- based paleothermometry. Geochimica et Cosmochimica Acta (2021) doi:10.1016/j.gca.2021.06.037.
L 85-94. This part can be restructured, it goes back and forth between GDGTs and pollen. Perhaps the authors can add here that the residual error in the most recent brGDGT calibrations is still large, which is part of the reason why brGDGTs have not been used often in Holocene temperature reconstructions, as the expected temperature range is small, compared to the error in the calibration.
L 96. Can the authors comment on whether these studies with combined pollen and GDGTs allowed to reach a more robust interpretation? This would further support the approach used here.
L 109. Can oceanic affinity be rephrased? Would this be typical for coastal environments? Or do the authors mean something else?
L 110. Atlantic Ocean influence from the east instead?
L 128. Is peat material influenced by a reservoir age/ hard water offset when performing 14C measurements? Please add if any corrections were performed.
L 138. A first calcination. Was there a second heating step? IF not, remove ‘first’.
L 151. Was 1303 scanned for GDGT-0? Why is this (I had the impression that 1302 was more commonly used)? Are the GDGTs 1-3 used in the manuscript? If not, you can remove their masses here.
L 153, can the authors add the number of compound (n=xx), when they say ‘all GDGTs’? Usually fractional abundances are reported calculated either relative isoprenoid or branched GDGTs.
Table 1: please write explicitly when 5 and 6 methyl compounds are added in the ratios. (Fi CBT, is this ratio based on Ib+ IIb5ME+IIb6ME? Also, is I in fact Ia + Ib + Ic? Write out in full for clarity. The a, b and c suffixes should not be in subscript.
L 171. Please add magnification of the microscope.
L 173. Please make the complete dataset be available as supplementary materials.
Section 2.5.2. It is not clear what the selection of GDGT based ratios is based on. For instance the MBT’ with CBT correction does not make much sense, as it introduces error that has been resolved with the updated chromatographic method used here. I suggest removing this calibration, and consider narrowing down further the suite of calibrations used. At the least, the use of each selected calibration should be motivated in section 2.5.2, along the lines of section 2.6.2.
L 208. It would be interesting to see selected other temperature reconstructions, for instance seasonal reconstructions, based on the current discussion on the seasonality of the HTM. Also, precipitation changes should be plotted, as the change between peat and lake (or lake depth) will also influence the distribution of brGDGTs.
L 221. This statistical treatment is not so common and needs to be explained. What is the effect?
The PCA based on XRF shows (unexpectedly?) that all elements plot positively on PC1. What standardization was done on the XRF values before analysis using a PCA? Performing this analysis on the standardized counts can result in visualizing the real variance in elemental composition.
L250. Fig. 4 caption. “Zr element” is not enough information. Are counts plotted? Relative counts?
Fig. 4. Is the compound plotted crenarchaeol or crenarchaeol isomer? These are generally plotted as separate compounds. Don’t write 1-4 as subscript.
The double prime (fi IIa’’) compounds are not mentioned before. Include in introduction and methods.
L 281-285. Please rephrase, the meaning of this sentence is not clear.
L 282. It took me some time to find this figure in the Appendix. Fig. Dd is not self-explanatory, please correct.
Fig. 5e. The interpretation of this panel is not helped by the many temperature reconstructions plotted. Reduce the number (fi all calibrations based on MBT’5ME will show the same temperature trend), or summarize the variability by plotting a 95% confidence range based on all reconstructed temperatures?
L 361. The accumulation rates here are very low. Is it possible that the YD is not recorded in the peat record because of lack of accumulation?
L 365. It is not clear why the expanse of oak compared to hazelnut would have resulted in the observed decrease in detrital material. Is there any reference for this?
L 402: ‘show more reliable reconstructed MAAT anomalies”, what is this based on? Compared to the current MAAT? Compared to the pollen record? Compared to what is expected from literature? Needs to be explained. The same goes for the other sites (next few lines).
L 420. A change in the IIIa/IIa ratio from 0.12 to 0.46 is a large shift, the relative abundance of IIa in this ratio shows a fourfold increase. No change in the BIT index doesn’t mean anything in this context, right? Unless this can be substantiated, please remove.
L 423. In addition to the CI index, the impact of pH needs to be discussed here as well! The calculation of these ratios should preceed the discussion of the MBT’5ME as a temperature proxy in the results and discussion as well.
L 425. The link between this sentence and the next is not clear. Bacteria and archaea have different environmental drivers. Unless the first sentence means that oxic/anoxic conditions have been shown to impact GDGT producer communities, the link with GDGT0/cren implies a link between archaea and GDGT producing bacteria that’s arguably not there?
L 433-442. It is not clear why the authors discuss the seasonality in precipitation here. It seems like there is no indication that this peat environment would be characterized by a lack of moisture? Also, the seasonality of temperature can be mentioned in the methods and materials section, and should not be mentioned this late in the discussion?
L 458. Why was the MEDTEMP database then considered in the first place? Consider removing this calibration from the paper.
L 490. What time period do the authors mean when referring to ‘Mid-Holocene cooling’ here? This is not observed in brGDGTs when the grey area is removed. There is stable temperatures between 12ka and ~7 ka (unless Mid-Holocene falls between 7 and 12ka, but then it should be specified in the sentence), and a warmer period since 5ka. Mind that this is also occurring at a different CBT5ME and that thus the effect of pH needs to be constrained first!
L 514. Please be specific, which events are meant here? What is there expected duration?
L 515. Would the authors argue that the peat sequence lacks the required resolution to see these events? Not perse based on the number of samples, but rather on the impact of bioturbation on the wetland type archive.
L 539. If both these records are based on the MBT, calibration can not cause a disagreement between the trends.
L 540. This sentence is not clear about the ‘cooling trend’. Which cooling trend is supported?
Appendix: Fig. C does not seem necessary.
Citation: https://doi.org/10.5194/cp-2023-15-RC2
Léa d'Oliveira et al.
Léa d'Oliveira et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 319 | 103 | 11 | 433 | 2 | 4 |
- HTML: 319
- PDF: 103
- XML: 11
- Total: 433
- BibTeX: 2
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1