Carbon-capture plants provide one way to reduce the amount of atmospheric carbon dioxide. Credit: Rogelio V. Solis/AP

In Paris last December, the 196 parties to the United Nations Framework Convention on Climate Change (UNFCCC) agreed to balance the human-driven greenhouse-gas budget some time between 2050 and 2100. This commitment is intended to limit the increase in global average temperature above pre-industrial levels to “well below 2 °C” — and preferably to 1.5 °C.

A balanced greenhouse-gas budget either requires that industry and agriculture produce zero emissions or necessitates the active removal of greenhouse gases from the atmosphere (in addition to deep and rapid emissions cuts). In most modelled scenarios that limit warming to 2 °C, several gigatonnes of carbon dioxide have to be extracted and safely stored each year1. For more ambitious targets, tens of gigatonnes per year must be removed2.

Policy: Start research on climate engineering

Many CO2-removal techniques have been proposed. Whether any of them could work at the scale needed to deliver the goal of the Paris agreement depends on three things: feasibility, cost and acceptability. A crucial component of all of these approaches is the non-climatic impacts that large-scale CO2-removal could have on ecosystems and biodiversity.

Until now, the UNFCCC's scientific advisory body, the Intergovernmental Panel on Climate Change (IPCC), has paid relatively little attention to such impacts. It has fallen to other groups to review insights and gaps in our understanding of the influence of CO2-removal techniques on ecology3,4,5; to make broad assessments of climate-engineering schemes6; and to carry out comparative modelling studies7.

It is time for the IPCC, governments and other research-funding agencies to invest in new, internationally coordinated studies to investigate the viability and relative safety of large-scale CO2 removal.


Since its establishment in 1988, the IPCC has predominantly involved physical scientists and modellers, rather than ecologists. This, combined with the only relatively recent evidence that emissions reduction alone is unlikely to avert dangerous climate change, could account for why the IPCC's roughly 5,000-page Fifth Assessment Report, released in 2013 and 2014, leaves out one crucial consideration: the environmental impacts of large-scale CO2 removal.

This omission is striking because the set of IPCC emissions scenarios that are likely to limit the increase in global surface temperature to 2 °C by 2100 (the aim of the RCP2.6 'representative concentration pathway', the IPCC climate-change-response scenario that achieves the lowest emissions) mostly relies on large-scale CO2 removal.

Geoengineering: Journey into geopoetry

These scenarios assume that two techniques could be developed to balance the carbon budget later this century: bioenergy with carbon capture and storage (BECCS), and afforestation. BECCS involves growing bioenergy crops, from grasses to trees; burning them in power stations; stripping the CO2 from the resulting waste gases; and compressing it into a liquid for underground storage. Afforestation — planting trees — also relies on photosynthesis to initially remove CO2 from the atmosphere. Storage is achieved naturally, in timber and soil.

Limiting the global temperature rise to 2 °C, with any confidence, would require the removal of some 600 gigatonnes of CO2 over this century (the median estimate of what is needed)8. Using BECCS, this would probably require crops to be planted solely for the purpose of CO2 removal9 on between 430 million and 580 million hectares of land — around one-third of the current total arable land on the planet, or about half the land area of the United States.

Unless there are remarkable increases in agricultural productivity, greatly exceeding the needs of a growing global population, the land requirements to make BECCS work would vastly accelerate the loss of primary forest and natural grassland. Thus, such dependence on BECCS could cause a loss of terrestrial species at the end of the century perhaps worse than the losses resulting from a temperature increase of about 2.8 °C above pre-industrial levels10.

A more fundamental concern is whether BECCS would be as effective as it is widely assumed to be at stripping CO2 from the atmosphere. Planting at such scale could involve more release than uptake of greenhouse gases, at least initially, as a result of land clearance, soil disturbance and increased use of fertilizer. When such effects are taken into account, the maximum amount of CO2 that can be removed by BECCS (under the RCP2.6 scenario) is estimated to be 391 gigatonnes by 2100. This is about 34% less than the median amount assumed to be needed to keep the temperature rise below 2 °C. If less optimistic but not unrealistic assumptions are made about where the land for bioenergy crops would come from, a net release of 135 gigatonnes of CO2 could occur by 2100 (see 'Future unknown')8.

Credit: Source: Ref. 8

Incomplete understanding throws other assumptions of the BECCS-based scenarios into question9. For instance, little is known about the effect of future climatic conditions on the yields of bioenergy crops; what the water requirements of such crops may be in a warmer world; the implications for food security if bioenergy production directly competes with food production; and the feasibility (including commercial viability) of the associated carbon capture and storage infrastructure.

Is the 2 °C world a fantasy?

Less is expected of afforestation in terms of its ability to take CO2 out of the atmosphere. Yet there is a near-universal assumption that increased forest cover is environmentally desirable. This is true in most cases of reforestation, particularly if a mixture of native trees is planted or re-planted, rather than an exotic monoculture. But afforestation can also involve the loss of natural ecosystems. And planting swathes of forest will cause complex changes in cloud cover, albedo (reflectance) and the soil–water balance (through changes to evaporation and plant transpiration), all of which affect Earth's surface temperature.

Counter-intuitively, afforestation at mid-latitudes and in northern, boreal forests may have a net warming effect, despite increasing the storage of carbon7. Also, as with bioenergy crops, it is difficult, if not impossible, to reliably quantify the effects of future climate change during 2050–2100. Increased fires, droughts, pests and disease could jeopardize the stability of carbon storage in newly planted forests.

Other options

There is no shortage of other ideas for CO2 removal by biological, geochemical and chemical means (see 'Take your pick'). For all such schemes, modelling the theoretical potential of a proposed approach can give a completely different picture from that obtained when environmental impacts — not to mention practicalities, governance and acceptability — are considered.

The roughly 25 years of discussion, research and policymaking on ocean fertilization, another CO2-removal technique, is a case in point. Since the link was first made between natural changes in the input of dust to the ocean, ocean productivity and climatic conditions, there has been a dramatic scaling-down of expectations of how effective ocean fertilization might be as a way to avoid human-driven global warming11.

During the 1990s, researchers postulated that for every tonne of iron added to seawater, tens of thousands of tonnes of carbon (and hence CO2) could be fixed by the resulting blooms of phytoplankton. This quantity has been whittled down over the years with the realization that most of the CO2 absorbed by such blooms — stimulated either by adding iron or other nutrients to seawater, or by enhancing upwelling through mechanical means — would be released back into the atmosphere when the phytoplankton decomposed. Moreover, a large-scale increase in plankton productivity in one region (across the Southern Ocean, say) could reduce the yields of fisheries elsewhere by depleting other nutrients, or increase the likelihood of mid-water deoxygenation. Such risks have resulted in the near-universal rejection of ocean fertilization as a climate intervention, through bodies such as the Convention on Biological Diversity (CBD)3.

More recently, other, potentially more controllable, ocean-based CO2-removal techniques have been suggested, such as the cultivation of seaweed to cover up to 9% of the global ocean12. The specific environmental implications of this method have yet to be assessed. Yet such an approach would clearly affect, and potentially displace, existing marine ecosystems that have high economic value. (Shallow and coastal waters currently provide around 90% of global fish catches.)

Back on land, other techniques include those to increase the amount of carbon sequestered in the soil, for example by ploughing in organic material such as straw, reducing ploughing (to limit soil disturbance) or adding biochar (a form of charcoal). Another idea is to enhance weathering, which involves the absorption of CO2 from the atmosphere by certain silicate rocks. Existing insights from agriculture, geoscience and mineral extraction enable more informed assessments of the feasibility and acceptability3,4,5,6 of these approaches. Yet it is crucial to know more about the permanence of carbon storage for biologically based methods, and the environmental impacts that might result if such approaches are used at vast scale4,5,6.

Action should focus on urgent emissions reductions.

For example, the use of biochar raises land-use issues. In addition, millions of hectares of soil darkened by the application of biochar would decrease albedo, increasing heat absorption. The addition of pulverized rock to the soil surface, by contrast, would increase reflectivity. Yet to reduce the amount of CO2 in the atmosphere by around 50 parts per million (a roughly 12% decrease from current levels), 1–5 kilograms per square metre of silicate rock would need to be applied each year to 2 billion to 6.9 billion hectares of land (15–45% of Earth's land surface area), mostly in the tropics13. The volume of rock mined and processed would exceed the amount of coal currently produced worldwide, with the total costs of implementation estimated to be between US$60 trillion and $600 trillion. And the chemistry and biology of rivers and adjacent ocean areas would be radically altered.

The most environmentally benign option for large-scale CO2 removal may be direct air capture (DAC). This can be done by passing air through anion-exchange resins that contain hydroxide or carbonate groups, which, when dry, absorb CO2, and release it when moist. The extracted CO2 can then be compressed, stored in liquid form and deposited underground using carbon capture and storage technologies6.

The operational costs for DAC cover a similar range to those estimated for enhanced weathering. The extraction process would also need land and probably water, and, as for BECCS, there is a risk of CO2 leaking out of geological reservoirs. Such risks can be minimized by storing the liquid CO2 beneath the sea or by using geochemical transformation, which involves in situ reactions between CO2 and certain rock types. In theory, cooling (rather than chemistry) to liquefy out the CO2 could also be used to remove CO2 from ambient air14. The technical feasibility, costs and potential environmental impacts of this approach — which could involve setting up plants in remote places such as Antarctica — have yet to be investigated.

Urgent action

As well as a major step up in research, urgent attention must be given to clarification at the UN level of what is considered geoengineering and what is climate mitigation. Once considered distinct approaches, the meaning of these terms has become fuzzier in recent years. CO2 removal is frequently included in both categories, generating confusion and contradiction.

This is crucial to resolve because mitigation and geoengineering have very different psychological connotations. Mitigation is universally considered to be a good thing that reduces risk or damage. Geoengineering frequently elicits suspicion, or is dismissed as a 'high-risk, high-tech' approach that may itself be harmful.

CO2 removal was not specifically discussed in Paris. However, the large-scale extraction of CO2 does seem to be a requirement to meet the goal of the Paris agreement. The CBD considers most, if not all, techniques for CO2 removal to be climate geoengineering, which it has repeatedly rejected as a policy option for addressing climate change. With a few exceptions, the same 195 or so governments make up both the UNFCCC and the CBD.

One solution would be to abandon the term climate geoengineering and simply assess the various methods for mitigating climate change on a case-by-case basis.

The Paris agreement shows where we want to go — the brave new world of a balanced carbon budget — but not how to get there. For now, action should focus on urgent emissions reductions and not on an unproven 'emit now, remove later' strategy. But the unwelcome truth is that, unless a lot more effort is made to cut emissions, significant CO2 removal will need to begin around 2020, with up to 20 gigatonnes of CO2 extracted each year by 2100 to keep the global temperature increase “well below 2 °C”2.

Is that feasible? What environmental risks and constraints are involved? We need to know.