EGU2020-18410
https://doi.org/10.5194/egusphere-egu2020-18410
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Reconstructing ocean temperatures using coccolith clumped isotopes

Luz Maria Mejia1, Alvaro Fernandez1,2, Hongrui Zhang1,3, Jose Guitian1, Stefano Bernasconi1, and Heather Stoll1
Luz Maria Mejia et al.
  • 1ETH, Geological Institute, Zürich, Switzerland (luz.mejia@erdw.ethz.ch)
  • 2Department of Earth Science, University of Bergen, Realfagbygget, Allégt. 41, Bergen
  • 3Ocean and Earth Science school, Tongii University, 1239 Siping Road, 200092, Shanghai

     Reliable temperature reconstructions of the ocean are often difficult to obtain due to the limitations of widely used proxies. The application of clumped isotope thermometry to coccolith calcite, which is geographical and chronological ubiquitously distributed, and whose production is limited to the photic zone, may provide ocean’s temperature information when and where other proxies have been shown inaccurate or not applicable.

     To evaluate the potential of coccolith clumped isotopes in paleoceanography we compare the temperatures derived from the fine fraction (<11µm), a pure mixed coccolith fraction (2-10 µm), and to a fraction of carbonate fragments from unidentified sources (<2 µm), with coeval alkenone sea surface temperatures (SST) from ODP Site 982 in the North Atlantic covering the last 16 Ma. The similarity in magnitudes and trends from the <11 and 2-10 µm size fractions, and trace element analysis of the <2 µm size fraction, suggest that for this site and time interval, exclusion of small unrecognizable fragments is not necessary to obtain reliable temperatures. The warmer values of alkenone SSTs compared to coccolith clumped isotope-derived temperatures cannot be explained by diagenetic processes, but may be related to temperature overestimations by alkenone calibrations, which assume a warm season and/or shallow production of coccolithophores in the study area.           

     Vital effects in coccolith clumped isotopes potentially associated to carbon limitation may also help to explain the differences in cooling magnitudes compared to the alkenone record. To further investigate vital effects in clumped isotopes, we compare calcification temperatures of three pure coccolith size fractions (3-5, 5-8, and 8-10 µm), and relate them to vital effects observed in their δ13C and δ 18O. The analysis of the fine fraction of Holocene sediments (<10 or <8 µm) showing a range of temperature and CO2 concentrations also provide information on the potential effects of carbon availability in coccolith clumped isotopes, and suggests calcification of coccolithophores may occur in deeper habitats than those considered by alkenone calibrations. Our study shows clumped isotope thermometry applied to coccolith calcite as a promising alternative proxy for calcification temperature of coccolithophores.

How to cite: Mejia, L. M., Fernandez, A., Zhang, H., Guitian, J., Bernasconi, S., and Stoll, H.: Reconstructing ocean temperatures using coccolith clumped isotopes, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18410, https://doi.org/10.5194/egusphere-egu2020-18410, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 2 – uploaded on 04 May 2020
I added CC logo to all slides
  • CC1: Comment on EGU2020-18410, Michael Nairn, 04 May 2020

    Hi Luz, neat work. A couple of questions. 

    The sediment trap calibration to satellite data is an extremely close match, which makes the ~10oC offset in paleoSST strange. Could there be any diagenetic issues associated with vertical transport to the sea floor, and then within the sediment? It's also interesting this offset is consistent, suggesting the long term high latitude cooling is preserved. 


    Do you know how foraminferal Mg/Ca estimates of SST compare at the site? In particular relative to the offset between your alkenone and clumped isotope values? Also, it would be really cool to study a low latitude site; as there is a proposed increase in interhemispheric latidudinal temperature gradient through the last 16 Ma. 

    All the best, Michael

    • AC1: Reply to CC1, Luz Maria Mejia Ramirez, 04 May 2020

      The sediment trap calibration to satellite data is an extremely close match, which makes the ~10oC offset in paleoSST strange. Could there be any diagenetic issues associated with vertical transport to the sea floor, and then within the sediment? It's also interesting this offset is consistent, suggesting the long term high latitude cooling is preserved. 


      There are several reasons that argue against diagenesis heavily impactng our clumped isotope record. We have approached diagenesis in different ways so far. For instance, we calculated how much recristalized calcite we would require in our coccolith samples to "bias" our clumped isotope reconstructions by the magnitude of difference with alkenones, assuming constant bottom water temperatures of 4C. We get huge values, averaging 50%. Although SEM imaging is not the best method to quantitatively determine the amount of recristalized calcite (best is EBSD, which has never been conducted in coccoliths due to their small size), from the SEM we do see that there is no way half of the material was affected by diagenesis. 
      We are also running some models (e.g., Stolper et al. 2018) now applied to clumped isotopes, that suggests there should be much more diagenesis in the oldest material compared to the youngest. If this was is true, the reconstructed temperature cooling trend would be biased to show less cooling (relative to that  of the alkenones). However, SEM show that surface overgrowth is not very different between youngest and oldest samples, so the model doesnt describe our observations. Also, this model suggests diagenesis in the youngest sample would not bias clumped temperatures to be different than alkenones, which does not match our observations, as we also see substantial temperature differences in the youngest sample between both proxies. Moreover, our Sr/Ca values from the pure coccolith size fraction are similar to those reported by several coccolith studies. We would expect Sr/Ca to be very low if there would have been a very strong diagenetic effect, causing a difference of 10 C. 
      We are currently trying to get a more quantitative estimation of diagenesis in our samples from SEM imaging and coccolith thickness and volume, but we strongly suspect it will be as low as to bias our clumped isotope estimates with magnitudes smaller than those that our methods can actually detect. For instance, the diagenesis model we are running suggests that with around 20% recrystalization or lower, clumped isotopes would be biased for around 2C across all samples, which is a small bias, compared to what our method can detect.     

      Also, we know that C. pelagicus from the sediment trao was produced in the surface, because this was a very well studied once in a time monospecific bloom. We are certain of the temperature (in depth and time) at which it was produced. This is not the case for the coccolithophores of 982, which may have lived at deeper habitats or colder seasons compared to the temperatures used by alkenone calibrations. 
       
      Do you know how foraminferal Mg/Ca estimates of SST compare at the site? In particular relative to the offset between your alkenone and clumped isotope values? Also, it would be really cool to study a low latitude site; as there is a proposed increase in interhemispheric latidudinal temperature gradient through the last 16 Ma. 

      We have not compared our records with Mg/Ca records. And yes, the idea for the future is to be able to apply this proxy to low latitudes. Especially because this is the place where we have the most problems, as alkenones are usually saturated here. 

Presentation version 1 – uploaded on 04 May 2020 , no comments