Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Hadean silicate differentiation preserved by anomalous 142Nd/144Nd ratios in the Réunion hotspot source

Abstract

Active volcanic hotspots can tap into domains in Earth’s deep interior that were formed more than two billion years ago1,2. High-precision data on variability in tungsten isotopes have shown that some of these domains resulted from differentiation events that occurred within the first fifty million years of Earth history3,4. However, it has not proved easy to resolve analogous variability in neodymium isotope compositions that would track regions of Earth’s interior whose composition was established by events occurring within roughly the first five hundred million years of Earth history5,6. Here we report 142Nd/144Nd ratios for Réunion Island igneous rocks, some of which are resolvably either higher or lower than the ratios in modern upper-mantle domains. We also find that Réunion 142Nd/144Nd ratios correlate with helium-isotope ratios (3He/4He), suggesting parallel behaviour of these isotopic systems during very early silicate differentiation, perhaps as early as 4.39 billion years ago. The range of 142Nd/144Nd ratios in Réunion basalts is inconsistent with a single-stage differentiation process, and instead requires mixing of a conjugate melt and residue formed in at least one melting event during the Hadean eon, 4.56 billion to 4 billion years ago. Efficient post-Hadean mixing nearly erased the ancient, anomalous 142Nd/144Nd signatures, and produced the relatively homogeneous 143Nd/144Nd composition that is characteristic of Réunion basalts. Our results show that Réunion magmas tap into a particularly ancient, primitive source compared with other volcanic hotspots7,8,9,10, offering insight into the formation and preservation of ancient heterogeneities in Earth’s interior.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Figure 1: Neodymium isotope data for Réunion OIB.
Figure 2: Correlations between measured μ142Nd values and other isotopic ratios from Réunion basalts.
Figure 3: μ142Nd–ε143Nd systematics of Réunion basalts and dunites, with curves representing single-stage evolution models.
Figure 4: Preferred model of differentiation 4.37 Gyr ago at uppermost lower-mantle pressures.

References

  1. 1

    Cabral, R. A., Jackson, M. G., Rose-Koga, E. F., Koga, K. T., Whitehouse, M. J., Antonelli, M. A., Farquhar, J., Day, J. M. D. & Hauri, E. H. Anomalous sulphur isotopes in plume lavas reveal deep mantle storage of Archaean crust. Nature 496, 490–493 (2013)

    CAS  ADS  Article  Google Scholar 

  2. 2

    Delavault, H., Chauvel, C., Thomassot, E., Devey, C. W. & Dazas, B. Sulfur and lead isotopic evidence of relic Archean sediments in the Pitcairn mantle plume. Proc. Natl Acad. Sci. USA 113, 12952–12956 (2017)

    ADS  Article  Google Scholar 

  3. 3

    Mundl, A. et al. Tungsten-182 heterogeneity in modern ocean island basalts. Science 356, 66–69 (2017)

    CAS  ADS  Article  Google Scholar 

  4. 4

    Rizo, H. et al. Preservation of Earth-forming events in the tungsten isotopic composition of modern flood basalts. Science 352, 809–812 (2016)

    CAS  ADS  Article  Google Scholar 

  5. 5

    Jackson, M. G. & Carlson, R. W. Homogenous superchondritic 142Nd/144Nd in the mid-ocean ridge basalt and ocean island basalt mantle. Geochem. Geophys. Geosyst. 13, (2012)

  6. 6

    Horan, M. F., Carlson, R. W., Walker, R. J., Jackson, M. & Garçon, M. Tracking Hadean processes in modern basalts. Earth Planet. Sci. Lett. 484, 184–191 (2018)

    CAS  ADS  Article  Google Scholar 

  7. 7

    Vlastélic, I., Lewen, E. & Staudacher, T. Th/U and other geochemical evidence for the Réunion plume sampling a less differentiated mantle domain. Earth Planet. Sci. Lett. 248, 379–393 (2006)

    ADS  Article  Google Scholar 

  8. 8

    Bosch, D. et al. Pb, Hf and Nd isotope compositions of the two Réunion volcanoes (Indian Ocean): a tale of two small-scale mantle “blobs”? Earth Planet. Sci. Lett. 265, 748–765 (2008)

    CAS  ADS  Article  Google Scholar 

  9. 9

    Schiano, P. et al. Osmium isotopic systematics of historical lavas from Piton de la Fournaise (Réunion Island, Indian Ocean). Contrib. Mineral. Petrol. 164, 805–820 (2012)

    CAS  ADS  Article  Google Scholar 

  10. 10

    Peters, B. J., Day, J. M. D. & Taylor, L. A. Early mantle heterogeneities in the Réunion hotspot source inferred from highly siderophile elements in cumulate xenoliths. Earth Planet. Sci. Lett. 448, 150–160 (2016)

    CAS  ADS  Article  Google Scholar 

  11. 11

    Mukhopadhyay, S. Early differentiation and volatile accretion recorded in deep-mantle neon and xenon. Nature 486, 101–104 (2012)

    CAS  ADS  Article  Google Scholar 

  12. 12

    Friedman, A. M. et al. Alpha decay half lives of 148Gd, 150Gd and 146Sm. Radiochim. Acta 5, (1966)

  13. 13

    Caro, G., Bourdon, B., Halliday, A. N. & Quitté, G. Super-chondritic Sm/Nd ratios in Mars, the Earth and the Moon. Nature 452, 336–339 (2008)

    CAS  ADS  Article  Google Scholar 

  14. 14

    Murphy, D. T., Brandon, A. D., Debaille, V., Burgess, R. & Ballentine, C. In search of a hidden long-term isolated sub-chondritic 142Nd/144Nd reservoir in the deep mantle: implications for the Nd isotope systematics of the Earth. Geochim. Cosmochim. Acta 74, 738–750 (2010)

    CAS  ADS  Article  Google Scholar 

  15. 15

    Garnero, E. J. & McNamara, A. K. Structure and dynamics of Earth’s lower mantle. Science 320, 626–628 (2008)

    CAS  ADS  Article  Google Scholar 

  16. 16

    Wen, L., Silver, P., James, D. & Kuehnel, R. Seismic evidence for a thermo-chemical boundary at the base of Earth’s mantle. Earth Planet. Sci. Lett. 189, 141–153 (2001)

    CAS  ADS  Article  Google Scholar 

  17. 17

    Mulyukova, E., Steinberger, B., Dabrowski, M. & Sobolev, S. V. Survival of LLSVPs for billions of years in a vigorously convecting mantle: replenishment and destruction of chemical anomaly. J. Geophys. Res. 120, 3824–3847 (2015)

    ADS  Article  Google Scholar 

  18. 18

    Dubrovinsky, L. et al. Iron–silica interaction at extreme conditions and the electrically conducting layer at the base of Earth’s mantle. Nature 422, 58–61 (2003)

    CAS  ADS  Article  Google Scholar 

  19. 19

    Lee, C. T. A . et al. Upside-down differentiation and generation of a ‘primordial’ lower mantle. Nature 463, 930–933 (2010)

    CAS  ADS  Article  Google Scholar 

  20. 20

    Labrosse, S., Hernlund, J. W. & Coltice, N. A crystallizing dense magma ocean at the base of the Earth’s mantle. Nature 450, 866–869 (2007)

    CAS  ADS  Article  Google Scholar 

  21. 21

    Corgne, A., Liebske, C., Wood, B. J., Rubie, D. C. & Frost, D. J. Silicate perovskite-melt partitioning of trace elements and geochemical signature of a deep perovskitic reservoir. Geochim. Cosmochim. Acta 69, 485–496 (2005)

    CAS  ADS  Article  Google Scholar 

  22. 22

    Chauvel, C., Lewin, E., Carpentier, M., Arndt, N. T. & Marini, J.-C. Role of recycled oceanic basalt and sediment in generating the Hf–Nd mantle array. Nat. Geosci. 1, 64–67 (2008)

    CAS  ADS  Article  Google Scholar 

  23. 23

    Blichert-Toft, J., Arndt, N. T., Wilson, A. & Coetzee, G. Hf and Nd isotope systematics of early Archean komatiites from surface sampling and ICDP drilling in the Barberton Greenstone Belt, South Africa. Am. Mineral. 100, 2396–2411 (2015)

    ADS  Article  Google Scholar 

  24. 24

    Rizo, H., Boyet, M., Blichert-Toft, J. & Rosing, M. Combined Nd and Hf isotope evidence for deep-seated source of Isua lavas. Earth Planet. Sci. Lett. 312, 267–279 (2011)

    CAS  ADS  Article  Google Scholar 

  25. 25

    Caro, G., Bourdon, B., Wood, B. J. & Corgne, A. Trace-element fractionation in Hadean mantle generated by melt segregation from a magma ocean. Nature 436, 246–249 (2005)

    CAS  ADS  Article  Google Scholar 

  26. 26

    Tateno, S., Hirose, K. & Ohishi, Y. Melting experiments and peridotite to lowermost mantle conditions. J. Geophys. Res. Solid Earth 119, 4684–4694 (2014)

    CAS  ADS  Article  Google Scholar 

  27. 27

    Fiquet, G. et al. Melting of peridotite to 140 gigapascals. Science 329, 1516–1518 (2010)

    CAS  ADS  Article  Google Scholar 

  28. 28

    Funamori, N. & Sato, T. Density contrast between silicate melts and crystals in the deep mantle: an integrated view based on static-compression data. Earth Planet. Sci. Lett. 295, 435–440 (2010)

    CAS  ADS  Article  Google Scholar 

  29. 29

    Salters, V. & Longhi, J. Trace element partitioning during the initial stages of melting beneath mid-ocean ridges. Earth Planet. Sci. Lett. 166, 15–30 (1999)

    CAS  ADS  Article  Google Scholar 

  30. 30

    O’Neil, J., Carlson, R. W., Francis, D. & Stevenson, R. K. Neodymium-142 evidence for Hadean mafic crust. Science 321, 1828–1831 (2008)

    ADS  Article  Google Scholar 

  31. 31

    Parman, S. W., Kurz, M. D., Hart, S. R. & Grove, T. L. Helium solubility in olivine and implications for high 3He/4He in ocean island basalts. Nature 437, 1140–1143 (2005)

    CAS  ADS  Article  Google Scholar 

  32. 32

    Sobolev, A. V., Hofmann, A. W., Sobolev, S. V. & Nikogosian, I. K. An olivine-free mantle source of Hawaiian shield basalts. Nature 434, 590–597 (2005)

    CAS  ADS  Article  Google Scholar 

  33. 33

    Füri, E. et al. Helium isotope variations between Réunion Island and the Central Indian Ridge (17°–21°S): new evidence for ridge-hot spot interaction. J. Geophys. Res. 116, B02207 (2011)

    ADS  Article  Google Scholar 

  34. 34

    de Leeuw, G. A. M., Ellam, R. M., Stuart, F. M. & Carlson, R. W. 142Nd/144Nd inferences on the nature and origin of the source of high 3He/4He magmas. Earth Planet. Sci. Lett. 472, 62–68 (2017)

    CAS  ADS  Article  Google Scholar 

  35. 35

    Garçon, M., Boyet, M., Carlson, R. W., Horan, M. F., Auclair, D. & Mock, T. D. Factors influencing the precision and accuracy of Nd isotope measurements by thermal ionization mass spectrometery. Chem. Geol. 476, 493–514 (2018)

    ADS  Article  Google Scholar 

  36. 36

    Peters, B. J., Carlson, R. W., Day, J. M. D. & Horan, M. F. Nd isotope data for Réunion Island basalts and cumulate xenoliths. EarthChem https://doi.org/10.1594/IEDA/100740 (2018)

  37. 37

    Burkhardt, C. et al. A nucleosynthetic origin for the Earth’s anomalous 142Nd composition. Nature 537, 394–398 (2016)

    CAS  ADS  Article  Google Scholar 

  38. 38

    Render, J., Fischer-Gödde, M., Burkhardt, C. & Kleine, T. The cosmic molybdenum-neodymium isotope correlation and the building material of the Earth. Geochem. Persp. Lett. 3, 170–178 (2017)

    Article  Google Scholar 

  39. 39

    Hilton, D. R. et al. Helium isotopes at Rungwe volcanic province, Tanzania, and the origin of East African plateau. Geophys. Res. Lett. 38, (2011)

  40. 40

    O’Neil, J., Rizo, H., Boyet, M., Carlson, R. W. & Rosing, M. T. Geochemistry and Nd isotopic characteristics of Earth’s Hadean mantle and primitive crust. Earth Planet. Sci. Lett. 442, 194–205 (2016)

    ADS  Article  Google Scholar 

  41. 41

    Nomura, R. et al. Spin crossover and iron-rich silicate melt in the Earth’s deep mantle. Nature 473, 199–202 (2011)

    CAS  ADS  Article  Google Scholar 

  42. 42

    Brown, S. M., Elkins-Tanton, L. T. & Walker, R. J. Effects of magma ocean crystallization and overturn on the development of 142Nd and 182W isotopic heterogeneities in the primordial mantle. Earth Planet. Sci. Lett. 408, 319–330 (2014)

    CAS  ADS  Article  Google Scholar 

  43. 43

    Labrosse, S ., Hernlund, J. W. & Hirose, K. in The Early Earth: Accretion and Differentiation (eds Badro, J . & Walter, M. ) 123–142 (AGU, 2015)

  44. 44

    Cipriani, A., Bonatti, E. & Carlson, R. W. Nonchondritic 142Nd in suboceanic mantle peridotites. Geochem. Geophys. Geosyst. 12, Q03006 (2011)

    ADS  Article  Google Scholar 

Download references

Acknowledgements

B.J.P. is grateful for conversations with fellow postdocs A. Mundl and J. Reimink. Technical assistance with Nd isotope measurements was provided by T. Mock, and data-reduction routines were modified from an original version by M. Garçon. Funding for fieldwork for this study was provided by the National Geographic Society (NGS 8330-07), the Geological Society of America (GSA 10539-14), and a personal donation from R. Rex. Support for laboratory work was provided by the Carnegie Institution for Science.

Author information

Affiliations

Authors

Contributions

B.J.P. performed chemical separations and isotope analyses and wrote the manuscript. R.W.C. and B.J.P. interpreted the data and developed the modelling. J.M.D.D. and B.J.P. collected rock samples. M.F.H. developed the chemical separation method and contributed to data analysis. All authors reviewed, edited and approved the manuscript.

Corresponding author

Correspondence to Bradley J. Peters.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Additional information

Reviewer Information Nature thanks B. Bourdon and B. Hanan for their contribution to the peer review of this work.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Figure 1 Details of possible single-stage differentiation models.

a, Required differentiation ages (y axis) for given bulk-Earth μ142Nd values (x axis) and 147Sm/144Nd ratios (coloured lines) to best match Réunion average μ142Nd and ε143Nd. Given external precisions as a proportion of data range, minimum bulk-Earth μ142Nd values permitted by the Réunion data are indicated for various bulk Earth 147Sm/144Nd ratios (see text for details). b, Misfit of each scenario to Réunion data. Misfits are calculated as a moving average with a 1 p.p.m. μ142Nd window; coarseness among low misfits is a consequence of model resolution. Bulk Earth μ142Nd values are preserved vertically across the panels. For example, assuming a bulk Earth with Sm/Nd = 0.19 and μ142Nd = −5, the differentiation age that produces the best fit to Réunion μ142Nd and ε143Nd is around 4,430 million years, and the total misfit to Réunion data is less than 1%, meaning that this scenario is permitted within the precision of the data. Ma, million years.

Extended Data Figure 2 Hypothesis testing, showing that the spread of data among Réunion igneous rocks is greater than would be expected from analytical precision alone.

Black dots show arithmetic means and s.e.m. for randomly generated μ142Nd values, with a sample mean equal to the Réunion μ142Nd mean and standard deviation equal to external precision. Blue dots show arithmetic means and s.e.m. for randomly selected Réunion sample data with replacement after selection (that is, a given sample can be chosen one or more times). Each point (black or blue) represents a sample size of n = 22, identical to the number of samples analysed in this study. The two simulated datasets (black and blue dots) do not strongly overlap, implying that the total variation in Réunion μ142Nd is greater than would be expected from external precision alone. See text for further details and interpretation; an example simulation is given in the Source Data. Source data

Extended Data Figure 3 Isotope evolution model for differentiation occurring under pressures representative of the modern uppermost lower mantle.

At this depth, pressures are less than 34–36 GPa (ref. 27); ferropericlase (Fp) is the liquidus phase and Ca-perovskite (Ca-Pv) melts out before bridgmanite (Mg-perovskite, Mg-Pv). Results from a single model run are shown vertically in each column, with melt-out points set to 60% (Ca-Pv) and 80% (Mg-Pv) of melting (left panels), and to 80% (Ca-Pv) and 95% (Mg-Pv) of melting (right panels). a, b, Mantle and melt modes. c, d, Changes in Sm/Nd and Lu/Hf through accumulated fractional melting. e, f, Best-fit differentiation ages and melting degrees for given bulk-Earth μ142Nd and Sm/Nd values, given as moving averages around a 1 p.p.m. window. Bridgmanite and Ca-perovskite partition coefficients are from ref. 21; partitioning of Hf and rare earth elements in ferropericlase is assumed to be nil. Source data

Extended Data Figure 4 Isotope evolution model for differentiation occurring under pressures representative of the modern uppermost lower mantle.

Continuing on from Extended Data Fig. 3, results from a single model run are shown vertically in each column, with melt-out points set to 60% and 80% of melting (left panels), and to 80% and 95% of melting (right panels). a, b, Misfit to data averages for scenarios illustrated in Extended Data Fig. 3e, f. c, d, Selected model fit to Réunion μ142Nd–ε143Nd data. e, f, Same model as in panels c, d, with fit to Réunion ε143Nd–ε176Hf (ref. 8) data. Best-fit models return slightly earlier differentiation ages compared with deeper-mantle models (Extended Data Figs 3, 5 and 7) and require a lower degree of differentiation than does melting occurring at lowest mantle pressures (Extended Data Figs 7, 8). Differentiation at these relatively low pressures produces the steepest negative correlations between ε176Hf and ε143Nd compared with differentiation at higher pressures. Correlations between μ142Nd and ε143Nd appear to be shallow at the scales presented in panels c and d; however, they reflect the small x-axis scale that best represents the distribution of Réunion data. Réunion Nd-isotope data in panels c and d can be found in Supplementary Table 1 or in Source Data. Source data

Extended Data Figure 5 Isotope-evolution model for differentiation occurring under pressures representative of the modern mid-lower mantle.

At these pressures (36–60 GPa; ref. 26), bridgmanite is the liquidus phase and Ca-perovskite melts out before ferropericlase. Left panels: 60% (Ca-Pv) and 85% (Fp) melt-out points; right panels: 80% (Ca-Pv) and 95% (Fp) melt-out points. a, b, Mantle and melt modes. c, d, Changes in Sm/Nd and Lu/Hf through accumulated fractional melting. e, f, Best-fit differentiation ages and melting degrees for given bulk-Earth μ142Nd and Sm/Nd values, given as moving averages around a 1 p.p.m. window. A relatively constant melting degree (71–77%) is predicted over a range of bulk Earth μ142Nd values and differentiation ages when Ca-perovskite and ferropericlase are consumed late during melting (panel f), making these model conditions the most adaptable to possible conditions of Hadean mantle domain formation. Source data

Extended Data Figure 6 Isotope-evolution model for differentiation occurring under pressures representative of the modern mid-lower mantle.

Continuing on from Extended Data Fig. 5, left panels: 60% and 85% melt-out points; right panels: 80% and 95% melt-out points. a, b, Misfit to data averages for scenarios illustrated in Extended Data Fig. 5e, f. c, d, Selected model fit to Réunion μ142Nd–ε143Nd data. e, f, Same model as in panels c, d, with fit to ε143Nd–ε176Hf (ref. 8) data. Somewhat later preferred differentiation times (around 4.40 Gyr ago) produce slightly shallower ε176Hf–ε143Nd correlations compared with those observed in Extended Data Figs 3 and 4, and required contributions from the EER (c, d) are less than for other pressure conditions (Extended Data Figs 3, 4 and Extended Data Figs 7, 8) Correlations between μ142Nd and ε143Nd appear shallow at the scales presented in panels c and d; however, they reflect the small x-axis scale that best represents the distribution of Réunion data. Réunion Nd isotope data in panels c and d can be found in Supplementary Table 1 and in the Source Data. Source data

Extended Data Figure 7 Isotope-evolution model for differentiation occurring under pressures representative of the modern lowest mantle.

At these pressures (greater than 60 GPa; ref. 26), bridgmanite is the liquidus phase and ferropericlase melts out before Ca-perovskite. Left panels: 60% and 85% melt-out points; right panels: 80% and 95% melt-out points. a, b, Mantle and melt modes. c, d, Changes in Sm/Nd and Lu/Hf through accumulated fractional melting. e, f, Best-fit differentiation ages and melting degrees for given bulk-Earth μ142Nd and Sm/Nd values, given as moving averages around a 1 p.p.m. window. In this highest pressure range, the model demands the highest degree of mantle melting (greater than 90% for late consumption of ferropericlase and Ca-perovskite), but returns the shallowest correlations between ε143Nd, μ142Nd and ε176Hf and under the same conditions (see Extended Data Fig. 8).

Extended Data Figure 8 Isotope-evolution model for differentiation occurring under pressures representative of the modern lowest mantle.

Continuing on from Extended Data Fig. 7, left panels: 60% and 85% melt-out points; right panels: 80% and 95% melt-out points. a, b, Misfit to data averages for scenarios illustrated in Extended Data Fig. 7e, f. c, d, Selected model fit to Réunion μ142Nd–ε143Nd data. e, f, Same model as in panels c, d, with fit to ε143Nd–ε176Hf (ref. 8) data. Deep-mantle Hadean domains formed under these conditions are possibly the most susceptible to any overprinting that would mask correlations between μ142Nd and long-lived radiogenic isotope systems. Correlations between μ142Nd and ε143Nd appear to be shallow at the scales presented in panels c and d; however, they reflect the small x-axis scale that best represents the distribution of Réunion data. Réunion Nd isotope data in panels c and d can be found in Supplementary Table 1 or in the Source Data.

Extended Data Figure 9 A non-unique, upper-mantle differentiation scenario occurring in a bulk Earth with present-day values for 147Sm/144Nd of 0.196 and μ142Nd of −10.

Melting of 1.4% by mass at an age of 4.26 Gyr will form an early enriched reservoir (green line) that will evolve to possess a μ142Nd of about −31, and will leave behind a residual depleted mantle (orange line) that will evolve to possess a μ142Nd of 0; this latter value is identical to the terrestrial standard JNdi and identical to estimates for the composition of the upper mantle44. If melting locally progresses to 2.7%, it will produce an enriched reservoir (blue line) with a μ142Nd of about −28 and a conjugate depleted reservoir (red line) with a μ142Nd of about +20, which may subsequently mix to produce a μ142Nd composition similar to that of the Réunion source. Full model inputs are described in the Methods. Note that, unlike in Fig. 4a, the y-axis is normalized to the 142Nd/144Nd composition of the bulk Earth at a given x-axis time (μ142NdT) as opposed to the 142Nd/144Nd composition of the modern bulk Earth.

Extended Data Figure 10 Run information for the terrestrial neodymium standard JNdi throughout the analysis campaign, normalized to average 142Nd/144Nd before correction.

Analyses typically comprised 600–900 cycles of four steps each; steps 1 and 3, and steps 2 and 4, were each paired to calculate two dynamic-corrected 142Nd/144Nd ratios per cycle. Dynamic-corrected values correct for uncertainties in static measurements that are introduced by variable Faraday cup efficiencies. See Methods for details. Dynamic-corrected symbols represent averages and 2 s.e.m. of all dynamic-corrected ratios from that run.

Supplementary information

Supplementary Tables

This file contains Supplementary Tables 1 and 2. Supplementary Table 1 contains data per-sample Nd isotope results. Supplementary Table 2 contains per-run Nd isotope results. (XLSX 73 kb)

PowerPoint slides

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Peters, B., Carlson, R., Day, J. et al. Hadean silicate differentiation preserved by anomalous 142Nd/144Nd ratios in the Réunion hotspot source. Nature 555, 89–93 (2018). https://doi.org/10.1038/nature25754

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing