News & Views | Published:

Global change

That sinking feeling

Naturevolume 408pages155156 (2000) | Download Citation


The land and sea soak up much of the carbon dioxide emitted into the atmosphere. But one set of simulations suggests that global warming could greatly impair this ability.

Burning of fossil fuels and changes in land use — mainly deforestation — are resulting in more CO2 in the atmosphere and, it seems, global warming. Much of that extra CO2 is absorbed in 'sinks' on land and in the oceans. But what effect will future warming have on these sinks? In their paper on page 184 of this issue1, Cox et al. find that in the long run they absorb carbon much less effectively. According to the authors' calculations, the result is 2.5 °C greater global warming over land by the year 2100 than the 5.5 °C predicted if the climate–carbon-cycle connection is not taken into account.

At the moment, the annual increase of CO2 in the atmosphere is less than half of the estimated emissions2. The rest is absorbed by the terrestrial and ocean sinks for carbon. So climate projections have to consider not only future emissions but how those sinks will react3,4. It is no easy matter to couple models of climate change and the carbon cycle, but this is what Cox et al. have done.

In their first simulation, they projected how much carbon would be taken up by the land biosphere and ocean if climate remains constant, as in previous studies. They predefined emissions of CO2 at the 'business- as-usual' (IS92a) emission scenario5. This model predicts that the land biosphere will take up 450 Pg of carbon over the coming century, and the ocean 300 Pg, a grand total of 750 Pg (P is peta, 1015) (Fig. 1). At an average of 7.5 Pg C yr−1, this is about 50% more per year than the estimated present uptake. The primary mechanism for the land uptake is increased photosynthesis resulting from the increase in atmospheric CO2 (CO2 fertilization). In the ocean, it is carbon dissolution of the excess atmospheric CO2 in the surface waters and transport to depth.

Figure 1: Global warming and the results of carbon-uptake simulations. Uptake is the amount of carbon absorbed from the atmosphere by 'sinks' on land and in the oceans.
Figure 1

Previous studies have typically estimated the concentrations of atmospheric CO2 for a given emission scenario with models that do not include the effects of global warming on the carbon sinks. The first simulation indicates that such a model based on the IS92a 'business as usual' scenario5 puts 750 Pg C into the land and ocean. The second shows that the atmospheric CO2 projected by such a model will lead to a warming of 5.5 °C over land. But a cross-check on the carbon budget reveals that only 190 Pg C is taken up, mainly because of the effect of warming on the land sink. The third simulation reveals that if the 'missing' 560 Pg C is put back into the model, most of it ends up in the atmosphere. In consequence, warming over land increases to 8.0 °C. (Adapted from ref. 1 with further results from P. M. Cox.)

Cox et al. then carried out a global warming simulation with atmospheric CO2 predefined at the IS92a concentrations predicted by a model without climate warming5. The reason for doing this was to provide a baseline for how much warming their model would project if there were no feedbacks between climate and the carbon cycle. The projected warming is 5.5 °C over land. A simulation of how much CO2 the land biosphere and ocean take up with this climate change provides a check on the consistency of the approach of separately simulating climate and the carbon cycle. The result is dramatic. The land biosphere is projected to be a source to the atmosphere of 60 Pg C over the next century, and the uptake by the ocean drops to only 250 Pg. The combined uptake is only 190 Pg C instead of the 750 Pg C predicted by the first study.

Cox and colleagues' final simulation coupled the climate and carbon-cycle models to find out where the carbon goes. The answer is the atmosphere. This simulation projects an atmospheric CO2 concentration of 980 p.p.m. in 2100, rather than the 700 p.p.m. that comes from the model without climate change. The land biosphere releases 170 Pg of carbon in this model, whereas the ocean takes up 400 Pg of carbon for a total carbon uptake of 230 Pg, only slightly larger than that in the second model. The impact of this increased atmospheric CO2 on climate is large: warming of 8.0 °C over land, 2.5 °C greater than the climate model with predefined CO2.

How good are these estimates? The simulations of Cox et al. that include climate feedbacks give a larger uptake of carbon by the ocean than those without feedbacks. But this is primarily because the rate of increase of CO2 in the atmosphere dominates the ocean carbon sink6. A comparison of the oceanic uptake between the first and second simulations shows a reduction of more than 20% in the ocean carbon uptake resulting primarily from global warming.

Other models allow further comparisons. A set of simulations with predefined atmospheric CO2 using the IS92a scenario in four different climate models7,8,9,10 gave a range in oceanic uptake of 240–470 Pg during the next 100 years (C. Le Quéré, personal communication). The large range of estimates is primarily due to differences in the transport mechanisms that remove carbon from the surface ocean and carry it to the abyss. But how the ocean circulation and biosphere will respond to global warming is poorly understood. So the finding that these two effects tend to cancel3,8,10 must be treated as provisional.

What about the land sink? Cox and colleagues' results stem partly from a dramatic collapse of the Amazonian rainforest because of increased dryness, and partly from a global increase in respiration of organic matter in warmer soils. Here, however, there are especially large uncertainities. For example, a comparative study of the United States11, using various sets of models, estimated that a doubling of CO2 would change carbon content by anywhere between +32.3% and −39.4%. The lower limit stems from a projected decrease in forested area and increase in water stress; the higher limit from a projected increase in forested area and more vigorous cycling of nitrogen in soils due to warming.

Another group12 has compared the next generation of land biosphere simulations, in which models of vegetation extent are coupled to models of biogeochemistry. The range of carbon uptake is comparable to the results from the ocean models (240–500 Pg of carbon over the coming century). But only one climate model was used. Moreover, these simulations11,12 do not include the direct effects of land vegetation on climate, which tend to increase warming.

But perhaps the greatest source of uncertainty in estimates of the land sink lie in assumptions about CO2 fertilization and the temperature sensitivity of soil respiration. These responses determine Cox and colleagues' conclusions. But we cannot yet be confident that they will be the dominant mechanisms over the coming century.

For instance, a study13 of forests in five US states failed to find substantial growth enhancement from CO2 fertilization, implying that other processes are counteracting the fertilization effect. Without CO2 fertilization, the projected land carbon sink would disappear and soil respiration would dominate. Forest regrowth will help to compensate for this effect. The regrowth sink will decrease with time, however, whereas that due to CO2 fertilization in Cox and colleagues' model remains strong throughout the century. On the other hand, the temperature sensitivity of soil respiration used in these models, which is the main route for release of carbon, may not hold over the longer timescales of global warming projections14.

A final illustration of how complex things are comes from another paper in this issue (page 187)15. Betts suggests that increasing the terrestrial uptake of carbon by planting forests may actually increase global warming in some high-latitude regions. The reason is that darkening of the Earth's surface by trees leads to greater absorption of sunlight.

The bottom line, however, is this. Cox and colleagues' estimate that, by 2100, global warming will result in an extra 600 Pg C in the atmosphere because of a reduction in carbon sinks, an average of 6 Pg C per year. What could this mean in economic terms, given the drive to reduce carbon emissions to the atmosphere? The cost of capturing and storing CO2 from power plants has been estimated to be about $200 per ton of carbon emissions avoided16. Setting aside technological progress that might well reduce this cost, and the possibility of reducing CO2 emissions, the cost of compensating for a 6 Pg C annual loss in carbon sinks would be around $1.2 trillion. This calculation, simplistic though it may be, shows what is at stake.


  1. 1

    Cox, P. M., Betts, R. A., Jones, C. D., Spall, S. A. & Totterdell, I. J. Nature 408, 184–187 (2000).

  2. 2

    Houghton, J. T. et al. (eds) Climate Change 1995: The Science of Climate Change (Cambridge Univ. Press, 1996).

  3. 3

    Sarmiento, J. L. et al. Nature 393, 245–249 (1998).

  4. 4

    Cao, M. & Woodward, F. I. Nature 393, 249–252 (1998).

  5. 5

    Houghton, J. T., Callander, B. A. & Varney, S. K. Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment (Cambridge Univ. Press, 1992).

  6. 6

    Sarmiento, J. L. et al. Glob. Biogeochem. Cycles 9, 121– 138 (1995).

  7. 7

    Bopp, L. et al. Glob. Biogeochem. Cycles (in the press).

  8. 8

    Joos, F. et al. Science 284, 464–467 (1999).

  9. 9

    Maier-Reimer, E. et al. Clim. Dynam. 12, 711– 721 (1996).

  10. 10

    Matear, R. J. & Hirst, A. C. Tellus 51B, 722–733 (1999).

  11. 11

    VEMAP Members Glob. Biogeochem. Cycles 9, 407–437 (1995).

  12. 12

    Cramer, W. et al. Glob. Change Biol. (in the press).

  13. 13

    Caspersen, J. et al. Science (in the press).

  14. 14

    Giardina, C. P. & Ryan, M. G. Nature 404, 858–861 (2000).

  15. 15

    Betts, R. A. Nature 408, 187–190 ( 2000).

  16. 16

    Ishitani, H. et al. in Climate Change 1995: The Science of Climate Change (eds Houghton, J. T et al.) 587–647 (Cambridge Univ. Press, 1996).

Download references

Author information


  1. Program in Atmospheric and Oceanic Sciences, Princeton University, PO Box CN710, Sayre Hall, Princeton, 08544-0710, New Jersey, USA

    • Jorge Sarmiento


  1. Search for Jorge Sarmiento in:

Corresponding author

Correspondence to Jorge Sarmiento.

About this article

Publication history

Issue Date


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.

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