Climatic plant power

Levels of atmospheric carbon dioxide constrain vegetation types and thus also non-biological uptake during rock weathering. That's the reasoning used to explain why CO2 levels did not fall below a certain point in the Miocene.

The world is currently at risk of overheating in response to all the carbon dioxide being pumped into the atmosphere from the use of fossil fuels: the current atmospheric concentration of CO2 is about 385 parts per million (p.p.m.), compared with a 'pre-industrial' level of around 280 p.p.m. But overheating is an atypical menace in the recent history of the Earth. Over most of the past 24 million years, it was the possibility of cooling that posed the main threat to life. Cooling, however, did not reach the levels of severity that might have been expected, and on page 85 of this issue Pagani et al.1 put forward a thought-provoking case as to why that was so.

Since the end of the Eocene, around 40 million years ago, Earth's climate has been naturally getting colder. In temporal terms, the cooling has sometimes occurred in discrete steps, sometimes as a long-term trend2. Over the same interval, levels of atmospheric CO2 have fallen from around 1,400 p.p.m. at the end of the Eocene to possibly as low as 200 p.p.m. during the Miocene3 — the geological period between around 24 million and 5 million years ago.

This long-term history of atmospheric CO2 is the result of the interplay between several processes. The degassing of the Earth through magmatic activity (for instance volcanic eruptions) is the main source of CO2, and the dissolution of continental rocks captures atmospheric CO2, which is eventually stored as marine carbonate sediments4. The efficiency of the dissolution process — chemical weathering — is heavily dependent on climate, but also depends on vegetation and physical erosion. The last two parameters boost CO2 uptake by rock weathering. In particular, land plants promote rock dissolution through the mechanical action of roots, and by acidifying the water in contact with rocks5. Acidification occurs through the release of organic acids and the large-scale accumulation of CO2 in soil through root respiration. Removing plants, particularly trees, may strongly decrease the dissolution rate of rocks and the ability of this process to consume atmospheric CO2.

In their paper, Pagani and colleagues1 consider the potential role of the rise of many mountainous regions (orogens) over the past 40 million years, especially in the warm and humid low-latitude areas. In these mountain ranges, physical erosion would break down rocks and expose them to intense chemical weathering. The uptake of atmospheric CO2 would consequently increase, as indicated by the levels of CO2 measured, which could have declined to the lowest levels since multicellular life evolved on Earth some 500 million years ago.

But what could stop this CO2 uptake pump? Degassing through magmatic activity was probably declining at the same time (at best it remained constant), and tectonic activity accelerated mountain uplift over the past 24 million years. According to this line of evidence, CO2 levels should have plunged to below 200 p.p.m., with ice ultimately covering large surfaces of the Earth as a consequence. But that was not the case. Earth even experienced a warm spell between 18 million and 14 million years ago2.

Pagani et al.1 propose an exciting hypothesis to explain why, 24 million years ago, CO2 might have levelled off at about 200 p.p.m., and then stuck there. They suggest that, when CO2 levels became too low, forests became starved and were progressively replaced by grasslands, particularly in the low-latitude orogens. Grasslands exert a much less vigorous effect on rocks than do trees. In consequence, runs the thinking, CO2 consumption due to weathering declined because of changing terrestrial ecosystems, which in turn stabilized atmospheric CO2 at around 200 p.p.m.

Overall, the authors' model provides an elegant twist on several ideas about the Earth system that emphasize the role of vegetation in dynamically regulating and fixing the lower limit of atmospheric CO2. But it also raises contentious issues.

First, in the model1, forest starvation is triggered by the low level of atmospheric CO2, and by elevated temperature. But do proxy estimates of conditions at that time confirm this paradoxical combination? Proxy measurements of CO2 levels include marine carbonate boron isotopes6, carbon-isotope values of alkenones produced by oceanic algae3 and the density of stomata — a measure of gas exchange — in fossil leaves7. Unlike the geochemical proxy records3,6, the more recent estimates based on the stomatal index7 depict a highly variable CO2 trend over the Miocene (in good agreement with climatic fluctuations), rather than a CO2 level stuck at 200 p.p.m. Furthermore, the estimates show that CO2 concentrations are above the forest-starvation level most of the time, oscillating between 300 and 500 p.p.m.

Second, in their model Pagani et al. assume that rock weathering generated by mountain uplift would have continuously consumed atmospheric CO2 until it reached the forest-starvation level. But there is evidence that the extra consumption of CO2 due to the Himalayan uplift, the most important orogeny of the recent past, occurred mainly through the burial of organic matter in the Bengal fan, and not through rock weathering8,9. In addition, the tectonic history of the past 24 million years is still subject to debate, and the timing of the uplift of the main mountain ranges, such as the Himalaya and Andes, is far from fully constrained10.

Finally, the link between weathering and continental vegetation is well recognized. But it is complex. Apart from acidifying water and mechanical effects, land plants also control the hydrology of soils. In humid tropical environments, about 70% of the rainfall is absorbed by land plants and then evaporates through their leaves. This effect should inhibit weathering reactions by limiting the amount of water available for rock dissolution. Also, in equatorial uplifted areas, intense erosion occurs through landslides triggered by heavy rainfall11. These landslides bring fresh rock material in contact with water by removing the soil mantle, promoting weathering and CO2 consumption. The role of vegetation cover in these systems might not be as significant as Pagani et al. suggest.

The authors themselves acknowledge some of these limitations, and all in all have put forward a bold and provocative hypothesis. But accounting for all of the processes and constraints involved is probably beyond the capabilities of the first-order global models that Pagani et al. used, and more-complex and process-based modelling12,13 will be required to test their conclusions. Whatever the outcome, that should prove to be a fruitful exercise for carbon-cycle modellers intent on understanding the processes that drove climate and CO2 oscillations during the Miocene.


  1. 1

    Pagani, M., Caldeira, K., Berner, R. & Beerling, D. J. Nature 460, 85–88 (2009).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Zachos, J. C., Dickens, G. R. & Zeebe, R. E. Nature 451, 279–283 (2008).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Pagani, M. et al. Science 309, 600–603 (2005).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Walker, J. C. G., Hays, P. B. & Kasting, J. F. J. Geophys. Res. 86, 9776–9782 (1981).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Berner, R. A. The Phanerozoic Carbon Cycle (Oxford Univ. Press, 2004).

    Google Scholar 

  6. 6

    Pearson, P. N. & Palmer, M. R. Nature 406, 695–699 (2000).

    ADS  CAS  Article  Google Scholar 

  7. 7

    Kürschner, W. M., Kvaček, Z. & Dilcher, D. L. Proc. Natl Acad. Sci. USA 105, 449–453 (2008).

    ADS  Article  Google Scholar 

  8. 8

    Galy, V. et al. Nature 450, 407–410 (2007).

    ADS  CAS  Article  Google Scholar 

  9. 9

    France-Lanord, C. & Derry, L. A. Nature 390, 65–67 (1997).

    ADS  CAS  Article  Google Scholar 

  10. 10

    Sempere, T., Hartley, A. & Roperch, P. Science 314, 760 (2006).

    ADS  CAS  Article  Google Scholar 

  11. 11

    Bhatt, M. P. & McDowell, W. H. Water Resour. Res. 43, doi:10.1029/2007WR005915 (2007).

  12. 12

    Donnadieu, Y. et al. Geochem. Geophys. Geosyst. 7, doi:10.1029/2006GC001278 (2006).

  13. 13

    Donnadieu, Y., Goddéris, Y. & Bouttes, N. Climate Past 5, 85–96 (2009).

    ADS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Goddéris, Y., Donnadieu, Y. Climatic plant power. Nature 460, 40–41 (2009).

Download citation

Further reading


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.