Skip to main content

Thank you for visiting 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.

Global warming

Growing feedback from ocean carbon to climate

The finding that feedbacks between the ocean's carbon cycle and climate may become larger than terrestrial carbon–climate feedbacks has implications for the socio-economic effects of today's fossil-fuel emissions.

Global warming reduces the uptake of carbon dioxide by the ocean and the land biosphere. This allows more human-caused CO2 to remain in the atmosphere, where it causes more global warming — an amplifying climate–carbon feedback. The amplifying feedbacks associated with carbon in vegetation and soils are conventionally thought to be several times larger than those associated with carbon in the ocean. Writing in Global Biogeochemical Cycles, Randerson et al.1 report that this idea needs to be revisited and refined: in the long term, the ocean will have a larger role than land in climate–carbon coupling.

The authors combined results from long-term simulations using an Earth-system model — specifically the Community Earth System Model (CESM)2 — with an established analysis3 of feedbacks in the carbon–climate system. They ran the model up to the year 2300 using contrasting set-ups in which the greenhouse-gas forcing of CO2 was turned on or off. They report that the ocean carbon–climate feedback (measured as the difference in the amount of carbon cumulatively taken up by the ocean for the set-up that accounts for forcing from all the main anthropogenic emissions, including CO2, and for the set-up that omits such forcing) is 3% for 1999 and increases to a staggering 23% by 2300. Surprisingly, the ocean–climate feedback becomes larger than the land–climate feedback beyond 2100 (Fig. 1).

Figure 1: Feedback between climate and the carbon cycle.

Randerson et al.1 have used a computational model to show that the ocean's contribution to feedback between climate and the carbon cycle increases considerably over time, and exceeds contributions from land by the year 2100. The bar chart shows, for different periods, the reduction in the uptake of anthropogenic carbon (measured in petagrams of carbon; 1 petagram is 1015 grams) by the ocean and the land biosphere that is attributable to climate change, per increase in global mean surface air temperature of 1 °C. The authors computed these sensitivities from differences in cumulative carbon flux and global mean surface air temperature between a simulation in which CO2 influences climate and another in which the direct influence of CO2 on global warming is suppressed.

This apparent conflict with the conventional view of a smaller ocean than land feedback4 results from differences in timescales. Most carbon cycles through plants and soils within decades, whereas it takes centuries to transport carbon from the sea surface to the deep ocean. Land carbon–climate feedbacks are therefore larger in the near term than marine ones, but this situation will reverse in the long run.

Ocean carbon uptake is primarily a physico-chemical process in which the rate-limiting step is the transport of water from the surface to the deep, and in which the reaction of CO2 with water releases hydrogen ions, causing ocean acidification5. Randerson and co-workers report that warming decreases the solubility of CO2 in water, reduces the density of surface waters and slows the overturning circulation (the movement of surface water from high latitudes to the deep ocean, compensated for by rising waters from other latitudes). These effects reduce the overall uptake capacity of the ocean for, and the surface-to-deep transport of, anthropogenic carbon. The magnitude of the ocean–climate feedback reported by the researchers, and the main proposed feedback mechanisms — including a near-collapse of the Atlantic overturning circulation and a small role for marine biological processes — are consistent with the findings of earlier studies (see ref. 6, for example) that used models with simpler ocean dynamics than those in the CESM.

On land, the authors find that a loss of carbon results from increased rates of soil-carbon decomposition due to warming. This dominates the stimulation of plant productivity caused by increased moisture and growing-season length at high latitudes. In addition, human land use reduces the amount of carbon in vegetation in their scenario, further limiting the potential amplifying feedback between land carbon and climate.

The representation of biogeochemical processes in the CESM is advanced, but many feedbacks to those processes are uncertain or not represented. These include7 the impacts of ocean acidification, warming, increased levels of dissolved CO2 and a potential increase in the volume of low-oxygen zones on marine ecosystems, biological productivity, the production of the greenhouse gas nitrous oxide, the export of biogenic particles from the surface to the deep ocean, and their sinking velocity and decomposition rate. On land, thawing of permafrost carbon may lead to the release of CO2 (ref. 8) and methane4, processes not accounted for in the version of the CESM used by the authors. More work is needed to address changes in the extent of wetlands and peatlands in model projections, and to better quantify emissions of methane, nitrous oxide and organic compounds from the land biosphere, because these changes and emissions all contribute to climate forcing9.

Data for past variations in CO2 levels and climate can help us to avoid relying exclusively on models to project future trends. Although palaeoclimatic information cannot be directly used to assess how climate change affects the flow of anthropogenic carbon, it does reveal how the natural carbon cycle alters. Palaeoclimatic data document that the ratio of the change in atmospheric CO2 concentration to the change in global mean surface air temperature is increasing with the timescale of climatic variations, and that the increase of CO2 on decade-to-century timescales is probably less than 20 parts per million for each degree of warming10.

Further constraints on carbon–climate interactions arise from isotope data that allow terrestrial and oceanic greenhouse-gas sources to be disentangled. Carbon isotope data show that decadal CO2 variations during the past millennium were predominantly driven by land-carbon variations11, consistent with Randerson and colleagues' finding of a small near-term marine feedback. But isotope data for nitrous oxide question the assumption made in the CESM that nitrogen availability strongly limits plant growth. These data show that the land source of nitrous oxide increased during the last deglaciation, which implies that nitrogen flow through the land biosphere increased and that nitrogen limitation was small12. A long-standing research challenge is to develop Earth system models that perform seamless simulations from the past to the future, consistently integrating palaeoclimatic and modern instrumental information in projections.

Randerson and colleagues' work highlights another research front. The marine carbon–climate feedback increases in proportion to ocean-heat content in their model. Researchers would benefit from observational and modelling studies in which spatial and temporal variations of physical and biogeochemical parameters, such as temperature, carbon and nutrients, are analysed jointly to draw a more complete and self-consistent picture of the changing ocean. Biogeochemical variables are particularly sensitive indicators of change, but ocean physics governs both carbon and heat uptake — and therefore sea-level rise, climate change and their impacts.

The current study acts as a timely reminder of several policy-relevant findings, ahead of the conference to be held by the United Nations Framework Convention on Climate Change in Paris this December. First, the larger the amplifying carbon–climate feedback is, the more stringent will emission reductions have to be to stabilize climate at any given level. Second, shifting the peak in carbon emissions to after 2100 imposes on future generations not only the burden of implementing a carbon-free economy, but also global warming of about 10 °C above preindustrial temperatures1.

Third, carbon emitted today will change our environment irreversibly for many generations to come, and these changes increase hand in hand with cumulative carbon emissions13. Sea level and the ocean's acidity and carbon and heat content — and the associated adverse effects — will continue to increase long after atmospheric CO2 levels have stabilized, underscoring the need for near-term emission reduction. Finally, carbon-emission reductions are urgently needed if we are to limit global warming and ocean acidification to moderate levels. Any delay will narrow and eventually close the currently available window to meet stringent climate targets14.

Footnote 1


  1. 1.

    See all news & views


  1. 1

    Randerson, J. T. et al. Glob. Biogeochem. Cycles (2015).

    Google Scholar 

  2. 2

    Hurrell, J. W. et al. Bull. Am. Meteorol. Soc. 94, 1339–1360 (2013).

    Article  ADS  Google Scholar 

  3. 3

    Friedlingstein, P. et al. J. Clim. 19, 3337–3353 (2006).

    Article  ADS  Google Scholar 

  4. 4

    Ciais, P. et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) 465–570 (Cambridge Univ. Press, 2013).

    Google Scholar 

  5. 5

    Orr, J. C. et al. Nature 437, 681–686 (2005).

    CAS  Article  ADS  Google Scholar 

  6. 6

    Joos, F., Plattner, G. K., Stocker, T. F., Marchal, O. & Schmittner, A. Science 284, 464–467 (1999).

    CAS  Article  ADS  Google Scholar 

  7. 7

    Heinze, C. et al. Earth Syst. Dyn. 6, 327–358 (2015).

    Article  ADS  Google Scholar 

  8. 8

    Koven, C. D., Lawrence, D. M. & Riley, W. J. Proc. Natl Acad. Sci. USA 112, 3752–3757 (2015).

    CAS  PubMed  ADS  Google Scholar 

  9. 9

    Stocker, B. D. et al. Nature Clim. Change 3, 666–672 (2013).

    CAS  Article  ADS  Google Scholar 

  10. 10

    Heimann, M. Nova Acta Leopoldina NF 121, 97–99 (2015).

    Google Scholar 

  11. 11

    Bauska, T. K. et al. Nature Geosci. 8, 383–387 (2015).

    CAS  Article  ADS  Google Scholar 

  12. 12

    Schilt, A. et al. Nature 516, 234–237 (2014).

    CAS  Article  ADS  Google Scholar 

  13. 13

    Steinacher, M., Joos, F. & Stocker, T. F. Nature 499, 197–201 (2013).

    CAS  Article  ADS  Google Scholar 

  14. 14

    Stocker, T. F. Science 339, 280–282 (2013).

    CAS  Article  ADS  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Fortunat Joos.

Related links

Related links

Related links in Nature Research

Climate science: A sink down under

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Joos, F. Growing feedback from ocean carbon to climate. Nature 522, 295–296 (2015).

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.


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