The potential for land-based biological CO₂ removal to lower future atmospheric CO₂ concentration

Abstract

A combined approach of deliberate CO₂ removal (CDR) from the atmosphere alongside reducing CO₂ emissions is the best way to minimize the future rise in atmospheric CO₂ concentration, and the only timely way to bring the atmospheric CO₂ concentration back down if it overshoots safe levels. Here, land-based biological CDR and storage methods are reviewed, including afforestation, biomass burial, biochar production and bioenergy with CO₂ capture and storage. The current and future CDR flux they could generate and their total storage capacity for CO₂ are quantitatively assessed. The results suggest that there is already the potential to counterbalance land use change CO₂ emissions. By mid-century, the CDR flux together with natural sinks could match current total CO₂ emissions, thus stabilizing atmospheric CO₂ concentration. By the end of the century, CDR could exceed CO₂ emissions, thus lowering atmospheric CO₂ concentration and global temperature.

Figure 1.  Contrasting the atmospheric CO₂ balance on key time horizons following a conventional, mitigation-only (reducing CO₂ emissions) policy path, or a combined approach of mitigating emissions and CO₂ removal.

The es represent CO₂ in the atmosphere (size gives approximate concentration) and the vertical arrows indicate the major fluxes of CO₂ to and from the atmosphere (width gives approximate magnitude). Illustrative numbers for CO₂ concentration in 2050 and 2100 are based on the modeling discussed in this article, but should not be taken as forecasts.

Figure 2.  The main pathways of land-based biological CO₂ removal.

Carbon capture and storage occurs either from fermentation or combustion. Return fluxes of CO₂ to the atmosphere are not shown. The reservoirs on the right have different residence times for carbon and corresponding leakage rates back to the atmosphere.

CCS: Carbon capture and storage.

Figure 3.  Model results illustrating the potential for land-based biological CO₂ removal to lower (A) future atmospheric CO₂ concentration and (B) future global warming.

Each plot shows baseline scenario with no CDR (solid line), afforestation CDR scenario (dotted line), and bioenergy CDR scenario (dashed line). Scenarios are described in the text. In the afforestation case, carbon removed from the atmosphere is added to the vegetation pool in the model. In the bioenergy case, carbon removed from the atmosphere is taken out of the system.

CDR: CO₂ removal.

Figure 4.  CO₂ fluxes to and from the atmosphere in (A) baseline scenario with no CO₂ removal, (B) afforestation CO₂ removal scenario and (C) bioenergy CO₂ removal scenario.

Total emissions is fossil fuel burning plus land use change (identical in all three cases). Natural sinks is land sink plus ocean sink (a dynamic response of the model). CDR is the prescribed scenarios. Total removal is natural sinks plus CDR. When total removal matches total emissions, atmospheric CO₂ concentration is stabilized; when total removal exceeds total emissions, atmospheric CO₂ concentration is lowered.

CDR: CO₂ removal.

The global carbon cycle is currently perturbed by human fossil fuel burning and land use change activities, with atmospheric CO₂ concentration rising (at ∼2 ppm yr^-1) and carbon also accumulating in the ocean and on land [1]. In order to minimize the risk of dangerous climate change, as enshrined in Article 2 of the UN’s Framework Convention on Climate Change (UNFCCC), the rise of atmospheric CO₂ must be halted and potentially reversed. In simple terms, stabilizing CO₂ concentration demands that sinks match sources, and lowering CO₂ concentration demands that sinks exceed sources (Figure 1). The accepted policy approach to achieving stabilization is to rapidly reduce CO₂ emissions to match natural (i.e., land and ocean) sinks, and then to slowly reduce CO₂ emissions to zero, at the same rate that natural sinks decay. This should maintain an approximately constant CO₂ concentration followed by a stabilization of global warming. However, once the necessary rapid reductions in CO₂ emissions are underway, a safer strategy would be to carry on until they are eliminated, thus limiting the cumulative carbon emission. This will result in a ‘peak then slow decline’ of atmospheric CO₂ concentration, with the cumulative carbon emission determining the corresponding peak in global temperature, termed the ‘cumulative warming commitment’ [2]. Either way, we face an immediate and profound collective challenge to transform the current exponential increase in CO₂ emissions (∼2% yr^-1 over the past 25 years [3] and >3% yr^-1 over the past decade [1]) into a comparable or greater rate of decrease in CO₂ emissions. This transition must start soon and be completed within decades, if global warming is to be restricted to less than 2°C above preindustrial levels [2,4,5]. Already it demands rates of technological and economic change that may simply be unachievable [3]. So, what else can we do?

Rather than just trying to reduce anthropogenic sources of CO₂, if we can also actively create significant CO₂ sinks, then we can halt the rise of CO₂ concentration sooner, lowering the peak CO₂ concentration (Figure 1). Subsequently, if we can make the sum of created and natural (i.e., land and ocean) CO₂ sinks exceed anthropogenic sources of CO₂, we can bring the CO₂ concentration down, and we can do so faster than by reducing emissions alone and relying on natural sinks (Figure 1). Creating sinks is termed ‘CO₂ removal’ (CDR) and it is already implicit in several scenarios for CO₂ stabilization at relatively low levels [6]. Often it takes the form of afforestation and reforestation, which has long been recognized by the UNFCCC. CDR effectively reduces the cumulative carbon emission and hence should reduce the corresponding cumulative warming commitment [2]. An additional long-term role for CDR is that it could allow some ‘essential’ or ‘unavoidable’ fossil fuel CO₂ emissions to continue, without increasing the CO₂ concentration, by counter-balancing them.

Whilst all this sounds promising in principle, it depends critically on how large the potential sink from CDR is in practice. Two factors are critical to determining the potential of CDR. First, the rate of CDR (the flux) that can be achieved at a given time and, second, the total storage capacity for removed CO₂ is of importance. The achievable CDR flux, together with the anthropogenic emissions flux and natural sinks flux, determine whether CO₂ concentration can be stabilized, reduced or will continue rising at a given time (Figure 1). The total storage capacity for removed CO₂, together with the total cumulative CO₂ emission, determines how much anthropogenic CO₂ will remain in the atmosphere–ocean system in the long term and, therefore, the long-term concentration of CO₂ and the corresponding warming [2,7–9]. A third important consideration, especially in the long term, is whether there is leakage of CO₂ from the storage reservoirs back to the atmosphere and, if so, at what rate.

The various methods available for CDR have been summarized in previous work [10] and a Royal Society review [11]. They can be categorized into physical, chemical or biological approaches, and land- or ocean-based approaches. Current assessments suggest that land-based methods of CDR either via biological (photosynthesis) or physical and chemical means have greater potential than ocean-based methods [10,11]. Furthermore, existing economic assessment suggests that land-based biological CDR has a better cost–benefit ratio than air capture of CO₂ using physical and chemical means [12]. Consequently, the chosen focus here is on quantifying land-based biological methods of CDR. At the outset, it is worth noting that physical/chemical air capture would take up far less land space than biological methods and could, in principle, remove as much CO₂ as societies were willing to pay for, but there is a distinct shortage of future projections of the CDR flux it could generate (i.e., very little to review).

Pathways & constraints

Plants and all other organisms performing photosynthesis are solar-powered carbon-capture devices. Photosynthesis is actually a rather inefficient way of converting sunlight into usable energy – approximately 0.5% efficient at best [13]. Solar thermal or solar photovoltaic methods are capable of approximately 20% efficiency [14], but crucially these methods capture carbon at the same time. So, whilst on purely physical grounds biomass would only be expected to play a modest role in future energy supply [13,14], on chemical grounds it could play a valuable role in various CDR pathways (whilst also supplying carbon-based fuels), if sufficient area is available to be devoted to it.

Currently, global terrestrial net primary production (NPP) is approximately 60 PgC yr^-1[15,16], whilst fossil fuel emissions (including cement production) are approximately 8.5 PgC yr^-1 and land use change emissions are approximately 1.5 PgC yr^-1, totaling approximately 10 PgC yr^-1[1]. If land-based productivity is going to be used to generate a carbon sink to match current total emissions, it will require at least in the order of 15% of the world’s productive land surface. The productive (i.e., ice and desert free) land surface is approximately 10 Gha, so somwhere in the order of 1.5 Gha will be required. If this sounds a lot, for reference, global cropland currently totals approximately 1.5 Gha [17] and managed grazing land more than 3.3 Gha [18]. The area required could be considerably more, because we have assumed that all carbon captured can be converted to a permanent carbon store, whereas present global NPP is mostly counterbalanced by heterotrophic respiration (and fires), leaving a net sink of approximately 2.5 PgC yr^-1[1]. The challenge is to lock carbon away in permanent storage without causing carbon to be lost elsewhere from ecosystems.

Figure 2 summarizes the main land-based biological pathways of CDR, the conversion processes involved and the destination reservoirs for carbon. The simplest land-based biological CDR pathway is to accumulate carbon in woody biomass through permanent afforestation, perhaps augmenting the sink by harvesting some of the biomass as wood products and, thus, maintaining the corresponding forestry plantations in a high growth phase [19,20]. Alternative suggestions are to deliberately bury wood [21] or crop residues [22,23] to store carbon. None of these pathways make use of the chemical energy in biomass, hence they are referred to as ‘biomass CDR’. Alternatively, if energy is extracted from biomass, some of the associated carbon can, in principle, be captured and stored as CO₂ (from fermentation processes or combustion flue gases) [24,25] or as biochar (from pyrolysis of biomass) [26]. The feedstocks for these ‘bioenergy CDR’ pathways could include deliberately grown energy crops, forestry wood that is surplus to other uses, and residues (i.e., waste products) from agriculture, energy crops and forestry. Finally, there is the possibility that some of the biomass consumed in deliberate anthropogenic vegetation fires could be converted to biochar (rather than released as CO₂) [26], which is referred to as ‘slash and char CDR’.

Before getting into the specifics of the different CDR pathways, let us note some key overarching constraints on the potential for management of the land biosphere to generate a removal flux of CO₂. These are the supply of available land area, the yield of carbon (per unit area and time) and the conversion efficiency to permanently stored carbon. Of these, the supply of land area is probably the strongest constraint on the achievable CDR flux. Following others (and consistent with Article 2 of the UNFCCC), it is assumed that natural ecosystems should be protected because they provide valuable services to humanity. Hence, for example, replacing native forest with managed plantations is not a permissible land use change in the pursuit of CDR. Many studies assume that abandoned agricultural land will be the key source of land for afforestation and/or bioenergy crops. At first sight, this seems surprising: with a growing global population and changes in diet towards more land-intensive meat consumption, should we not expect expansion of agricultural land? Historically, since the early 1960s, there has been little net change in the land area under cultivation despite a doubling of population [27]. Yet, even with constant area under cultivation, there can be abandoning of some land, whilst new land goes under cultivation. Looking ahead, several Intergovernmental Panel on Climate Change Special Report Emissions Scenarios (IPCC SRES) [28] project a net decline in cropland (A1, B1 and B2), whilst the A2 scenario projects net growth [29]. All produce a supply of abandoned cropland, which is up to 0.6–1.3 Gha in 2050 and approximately double this in 2100 [29]. Others estimate that up to 3.6 Gha of agricultural land could in theory become available by 2050, if land use patterns are optimized and very efficient agricultural systems adopted [30]. Low-productivity land (including grazed grassland) is projected to dwindle in area and to have negligible potential for deliberate biomass growth [29], although it has been suggested that large areas of desert in Australia and the Sahara could be irrigated by the desalination of seawater and forests grown there [31].

To maximize the CDR flux on a given area, one wants to maximize yield. For a given area of land, the yield (expressed here in MgC ha^-1 yr^-1) varies with plant type, location (different locations have different climatic and soil conditions) and harvesting regime. Yields are often given in units of dry mass rather than carbon, so one also needs to know the carbon content of different biomass types. For tree plantations (e.g., Pinus and Eucalyptus – the two main species globally), achieved yields range over approximately 3–14 Mg ha^-1 yr^-1[32], which assuming approximately 0.5 gC g^-1 average carbon content of wood [33], gives approximately 1.5–7 MgC ha^-1 yr^-1. Yet some projections for woody bioenergy crops have assumed global average yield levels ranging over 1.5–15 MgC ha^-1 yr^-1 (3–30 Mg ha^-1 yr^-1) [32] or 8–10.5 MgC ha^-1 yr^-1 (16–21 Mg ha^-1 yr^-1) [30], which seem ambitiously high [19]. Other energy crops generally have yields less than or equal to woody crops. If carbon is to be maintained in the biomass of standing trees, then the whole life cycle and harvesting regime must be considered and average carbon sequestration correspondingly reduced, with values of 0.8–1.6 MgC ha^-1 yr^-1 being used [20].

The conversion efficiency to stored carbon varies significantly between methods. If one leaves biomass in permanent forests and their soils (where previously the land stored less carbon), conversion efficiency may approach 100%, although natural disturbances such as pests and fire that reduce carbon storage cannot be completely prevented [34]. Similarly, when burying biomass, the conversion efficiency is potentially close to 100%, but decomposition cannot be completely prevented. We generously assume a maximum 100% conversion efficiency for either pathway. Other pathways that involve fermentation, combustion or pyrolysis of biomass inevitably lead to greater losses of CO₂. The energy cost associated with capture and storage of CO₂, together with the price earned on the carbon stored, will economically determine uptake of the associated technologies and the conversion efficiency achieved. Here, we will just consider the fraction of carbon that can in principle be realistically captured. Biomass carbon that is turned into transport fuel represents a dispersed source of CO₂ that cannot be captured, but the process of fermentation to produce ethanol yields a readily captured pure CO₂ stream that contains approximately a third of the carbon in the feedstock [12]. If the remaining biomass carbon is combusted to generate electricity and heat, CO₂ from the flue gasses can be captured with 60–80% efficiency [101]. Together, CO₂ capture from a mixture of fermentation and combustion can capture approximately 50% of the carbon in the feedstock [101]. For pyrolysis, approximately 50% yield of carbon in biochar is also achievable and can be exceeded in some circumstances [26], but where energy output needs to be increased, biochar yield is inevitably reduced. In principle, CO₂ capture could be combined with pyrolysis to boost carbon recovery, but we opt for 50% achievable conversion efficiency for either pathway.

We now turn to consider the specific CDR pathways, their CDR flux potential and total carbon storage capacities.

Afforestation & reforestation

The conversion of unforested land to permanent forest creates a net carbon sink and a store of carbon in the biomass of the trees and in the soil, although there can be transient (and even net) losses of carbon from soil depending on location. Once a forest reaches maturity, the sink is thought to decline to zero, with respiratory carbon losses matching photosynthetic carbon uptake, although recent studies point to a persistent carbon sink in old growth forests [35]. By harvesting carbon in the form of wood products and replanting, forestry plantations can be maintained in a higher average yield state, thus increasing the CDR flux [20]. In addition, by reducing demand for wood products from other land, the total area planted and, hence, the cumulative carbon storage can increase [20].

Large afforestation programs have already been undertaken, with an estimated 264 Mha afforested in 2010, increasing at approximately 5 Mha yr^-1 over 2005–2010 [36]. In China alone, the corresponding CDR flux is estimated to have been 0.19 PgC yr^-1 over 1988–2001 [37]. If the 264 Mha of existing plantations are accumulating carbon at an average rate of 0.8–1.6 MgC ha^-1 yr^-1, as used in global projections [19,20], then the corresponding CDR is already 0.21–0.42 PgC yr^-1. Conceivably, this is an underestimate since yield can be considerably greater in the tropics. If at maturity these plantations have a conservative yield of approximately 100 MgC ha^-1, they will store approximately 26 PgC globally. However, natural disturbances have the potential to significantly reduce carbon storage and the corresponding CDR flux [34].

Many studies of the future CDR potential of afforestation and reforestation have been conducted over the past 20 years, using quite different methods and underlying assumptions (Tables 1 & 2). The key determinants of the future global CDR flux achievable by afforestation (Table 1) are the area that is afforested at a given time and the yield (rate of carbon accumulation per unit area). The afforestation CDR flux grows both as planted trees approach their peak rates of carbon accumulation and as progressively more land is subject to planting. Current forecasts generally start from zero activity at the outset, despite significant afforestation having been underway for decades. Hence, the forecasts tend to underestimate the CDR flux in the short term. Abandoned agricultural land is expected to continue to become available throughout this century, at rates that range over 0–17 Mha yr^-1 across the SRES A2, B2, B1 and A1b scenarios [38]. The A2 scenario gives a low supply of 0–2 Mha yr^-1 throughout the century, but observed rates of afforestation were approximately 5 Mha yr^-1 in 2005–2010 and are already at the upper end of the SRES range. The ongoing supply of abandoned agricultural land potentially allows the afforestation CDR flux to grow continuously to 2100 [20,38].

Forecasts of the potential afforestation CDR flux on different time horizons are mostly of a comparable order of magnitude (Table 1): approximately 0.8 PgC yr^-1 in 2030 [39,40], approximately 1.5 PgC yr^-1 in 2050 [38] and approximately 3.3 PgC yr^-1 in 2100 [20,38]. The one exception is an estimate of approximately 10 PgC yr^-1 in 2035 [41,101], which we have used as an upper limit in previous analysis [10]. On reflection, it appears unrealistic for two reasons. First, a key ‘physical’ constraint is set very high, the average rate of carbon accumulation in new plantations is assumed to be 10 MgC ha^-1 yr^-1[41,101], approximately an order of magnitude greater than some observations [42], or other projections, which use 0.8–1.6 MgC ha^-1 yr^-1[19,20]. Second, there is a lack of consideration of ‘social’ constraints; 1 Gha is assumed to undergo afforestation within 25 years (40 Mha yr^-1) [41,101], approximately an order of magnitude above recent afforestation rates and several times the forecast supply of abandoned agricultural land [20], implying major conflicts with other land uses, notably food production and the preservation of natural ecosystems.

Estimates of the total carbon storage potential of permanent forest plantations are given in Table 2. Historical cumulative carbon emissions from land use change are approximately 200 PgC [40], approximately 150 PgC from deforestation and this has often been used as an upper limit on the amount of carbon that could be recaptured in forest biomass and soils in future. However, in principle, this number can be far exceeded because, in many regions, managed plantations can store more carbon than native vegetation. A recent estimate of the ‘physical potential’ of harvested plantations is that they could store up to approximately 900 PgC by 2100 (on 3.8–4 Gha) after an initial net emission of approximately 200 PgC in establishing them [20]. However, in practice, there are very large constraints on converting land to permanent forest, especially ongoing needs for food production, wood supply and conservation of natural habitats. If only allowing afforestation on abandoned agricultural land, the ‘social potential’ for carbon storage is reduced to 68–133 PgC by 2100 (on 695–1014 Mha) [20]. This estimate is in broad agreement with several other studies, despite very different methods and assumptions (Table 2), including simply extrapolating current rates of afforestation and deforestation [43], or extrapolating increasing demand for wood products [44]. They all suggest that an upper limit of approximately 150 PgC could be stored within 100 years.

There should be potential for significant additional storage beyond the century timescale, because both the CDR flux and the cumulative uptake are projected to be growing in 2100 [20,38]. Many plantations are only forecast to be established late this century (as land becomes available) and would not reach their peak carbon uptake until sometime next century [20,38]. However, the global CDR flux should eventually peak and then decline as the supply of land suitable for afforestation dwindles and the trees in new plantations reach their maximum rate of accumulating carbon [45]. Existing studies do not continue beyond 2100, but approximately 300 PgC would seem conceivable in the long term. Thus, it seems feasible that all the carbon that has been emitted by human land use change activities in the past could, in the long-term future, be recaptured by permanent afforestation.

Biomass burial

If standing plantations are harvested, can the resulting wood supply provide an additional carbon sink? Global wood removals from forests totaled 1.938 Pg dry matter or approximately 1 PgC yr^-1 in the year 2000 [46]. Approximately half of the roundwood extracted goes to wood fuel and half to industrial uses. Most of this wood is soon returned to the atmosphere as CO₂ through combustion or heterotrophic decay and so cannot be considered to be a CDR flux. The global stock of hardwood products is estimated to be 4.2 PgC, with an average lifetime in the order of 10 years [47]. The stock is estimated to be increasing at 0.026 PgC yr^-1[47], representing a small CDR flux if the increase in stock is permanently maintained, but even if this were to all go into wooden buildings, they still have a typical lifetime of less than a century.

The burial of wood deep in soil or of wood products in landfill sites has been suggested as a means of slowing wood decomposition rates under the anaerobic conditions that prevail there. At the extreme, it has been estimated that approximately 10 PgC yr^-1 of dead coarse wood (>10 cm diameter) is produced annually in all the world’s forests, and this could be buried in approximately 25-m deep trenches to create a corresponding CDR flux [21]. Aside from the serious practical constraints of managing all global forests, and the biogeochemical and ecological implications of removing nutrients and habitats in rotting wood, the fact that anaerobic consumption of organic carbon can generate a flux of methane, which is 20–30-times more potent molecule-for-molecule than a greenhouse gas, does not appear to have been considered [21]. This is a well-known problem (or potential energy source) in landfill sites, but only less than 3% of the carbon in solid wood buried in landfill is estimated to be converted to CH₄ and CO₂ in an approximately approximately 1:1 ratio [48]. This means that wood burial could still represent a net sink of CO₂ equivalents. However, the potential for dissolved organic carbon losses also needs to be considered, and the modest global flux of hardwood products limits their CDR potential at present.

A related proposal is to bury agricultural crop residues in the deep ocean [22,23] (which should minimize the problem of a return flux of CH₄ to the atmosphere). In a calculation of the CDR flux potential, it has been estimated that 5 Pg yr^-1 of crop residues are produced globally, corresponding to approximately 2 PgC yr^-1 and that approximately 0.6 PgC yr^-1 (30%) of this could be removed and buried in the ocean without drastically affecting soil carbon stores (Table 3)[23]. However, more rigorous published estimates of global biomass flows in 2000 show that approximately 2.9 Pg yr^-1 of crop residues are already harvested (i.e., go to some other use) and only approximately 1.5 Pg yr^-1 are unused [46]. Using the same assumption with carbon content, and assuming that 30% is buried, produces a CDR flux of 0.18 PgC yr^-1(Table 3). The residues from future bioenergy crop production might provide a significant additional carbon source for burial. However, the removal of crop residues could lead to a counteracting erosion of soil organic carbon, as well as compromising other ecosystem services [49].

An alternative approach might be to bury the residues from forestry felling losses, either deep in soil or in the deep ocean. Current felling losses in forests total 0.65 Pg yr^-1[46], which, if all buried, represents a maximum CDR flux potential of approximately 0.33 PgC yr^-1(Table 3). In the future, if large-scale afforestation with harvesting is undertaken, there is considerable potential for the felling losses to increase. However, removing all of this carbon from the forest floor could conceivably lead to counteracting losses in soil carbon.

The capacity of deep ocean sediments to store additional biomass carbon is deemed large [23], but it is worth noting that a flux of approximately 0.5 PgC yr^-1 is comparable to that currently reaching deep ocean sediments from all marine productivity. An additional flux of pure organic carbon without associated carbonate will tend to dissolve the carbonate already in the sediments, adding alkalinity and carbon to the ocean in a 2:1 ratio and, thus, increasing the long-term capacity of the ocean to store dissolved inorganic carbon and lower atmospheric CO₂[50].

Biochar production & soil carbon

Whilst burying biomass locks carbon away, it makes no use of its energy content. We now turn to CDR methods that liberate some of the energy in biomass and retain some of it to capture part of the carbon from the parent fuel. Biochar and CO₂ capture pathways for bioenergy can both use the same feedstocks; hence, one must be careful to avoid double counting. They also both have approximately 50% potential to capture carbon, so the choice of pathway does not greatly alter the CDR fluxes that are achievable. However, the end products and their destinations are different, leading to different ultimate constraints on the total amount of carbon that can be stored. Furthermore, biochar can be produced in shifting cultivation without yielding energy.

To illustrate the present CDR potential, we choose biochar as an end product. It has been estimated that if the currently available flows of carbon in biomass waste from agriculture and forestry, biomass energy production and human-induced fires were pyrolysed, then 0.56 PgC yr^-1 of biochar could be sequestered: 0.16 PgC yr^-1 from agricultural and forest wastes, 0.18 PgC yr^-1 from deriving all ‘modern’ biomass energy by pyrolysis, 0.21 PgC yr^-1 from ‘slash-and-char’ shifting cultivation, and 0.01 PgC yr^-1 from wastes of charcoal production (Table 3)[26]. However, assessment of global biomass flows in the year 2000 [46], combined with life cycle analysis of biochar production [51], suggests that the potential from agricultural wastes is greater than originally estimated [26]; using 50% of the 1.5 Pg yr^-1 (dry mass) of unused crop residues currently produced could create a sink of 0.18 PgC yr^-1(Table 3)[51]. This would be instead of (rather than additional to) the 0.18 PgC yr^-1 estimated for deep ocean burial of 30% of crop residue, but would come with the added benefits of yielding some useful energy and improving soil. In addition, the 0.65 Pg yr^-1 (∼0.325 PgC yr^-1) of felling losses from forestry [46] (if 50% can be converted to biochar) could produce a CDR flux of approximately 0.16 PgC yr^-1(Table 3). This is less than the effect of burial in the deep ocean, but again comes with co-benefits rather than potential costs. The estimate for slash-and-char CDR may also be greater. Human-induced vegetation fires release approximately 2 PgC yr^-1, approximately a third of which (0.5–0.7 PgC yr^-1) is from shifting cultivation [52], and at least 50% of this carbon can be converted to biochar using simple kilns [26], giving approximately 0.25–0.35 PgC yr^-1(Table 3). The revised total for the current physical potential of biochar CDR is 0.77–0.87 PgC yr^-1.

There are few forecasts of the future CDR flux potential of biochar (Table 1). Assuming a supply of 2.5 PgC yr^-1 of woody biomass from 2035 onwards and a 48% conversion efficiency yields 1.2 PgC yr^-1 of biochar [41,101]. Assuming 1.5% per annum growth, 1.74 PgC yr^-1 could be produced in 2060 and 3.15 PgC yr^-1 in 2100 [10]. However, this is predicated on the unrealistic afforestation scenario discussed previously, covering 1 Gha by 2035. A separate estimate is that if a forecast 180–310 EJ yr^-1 bioenergy supply in 2100 were all produced by pyrolysis then 5.5–9.5 PgC yr^-1 of biochar would be removed [26]. We reconsider this type of estimate in more detail below.

What about the storage capacity for biochar in soils? It has been argued that loadings of up to 140 MgC ha^-1 are not detrimental and, therefore, the current 1.6 Gha of global cropland and 1.25 Gha of temperate grasslands can together accommodate approximately 400 PgC (Table 2)[26]. In addition, the 0.7–1 Gha of cropland forecast to be abandoned this century and potentially subject to either afforestation or bioenergy cropping could, at the same loading, accommodate 98–140 PgC, increasing the total to approximately 500 PgC (Table 2). Even with the aforementioned upper end estimates of biochar production, it would take a century to fill this capacity, but it might conceivably begin to limit biochar production by 2100. This carbon storage capacity is greater than for afforestation and comparable to the lower estimates for geological CO₂ storage. If achieved, it would represent an approximate 25% increase in the carbon content of the world’s soils. However, it should be carefully researched at to whether loadings of 140 MgC ha^-1 biochar everywhere are really benign (e.g., for plant productivity).

At this point, it is worth briefly considering the potential to also increase the organic carbon content of soil. This is already factored into studies of afforestation, but what about on cropland or other managed land? Switching from conventional tillage to no-till farming has been found to sequester carbon at shallow depths at a mean rate of 0.57 ± 0.14 MgC ha^-1 yr^-1 across 67 long-term experiments, giving rise to an increase in soil organic carbon storage of 7.1 ± 1.75 MgC ha^-1, as a new equilibrium is reached within approximately 15 years [53]. If this switch occurred globally on all 1.5 Gha of cropland, simple arithmetic suggests that a change in soil carbon storage of approximately 11 PgC could be achieved with a maximum CDR flux of approximately 0.9 PgC yr^-1 over approximately 12.5 years. However, the few studies that have looked deeper into the soil suggest that reducing tillage shifts the distribution of carbon to shallower depths, but does not increase the total storage [54]. A switch to no-tillage might be augmented with increased inputs to the soil, and others propose that a more sustained CDR flux of 0.4–0.6 PgC yr^-1 over 50 years could achieve 20–30 PgC storage in cropland soils [55,56]. Together with restoration of degraded soils, a CDR flux of 0.6–1.2 PgC yr^-1 over 50 years achieving 30–60 PgC storage has been proposed [57]. This represents an upper limit, since it is similar to historical losses of carbon from soil due to land use change. It is an order of magnitude smaller than other potential carbon stores that we have identified and, more importantly, the required changes in agricultural practices and land uses would have to be maintained for this to represent permanent CDR. Reversion to earlier practices could readily re-release the carbon as CO₂. Consequently, we do not consider this in our overall estimates of CDR potential.

Bioenergy with carbon storage

The present potential for bioenergy CDR from sugarcane-based ethanol production and chemical pulp mills has been estimated at 0.19–0.23 PgC yr^-1[25]. Meanwhile, future projections of the physical potential for bioenergy CDR via capture and storage of CO₂ and/or biochar production are in short supply, with the exception of a few ‘back-of-envelope’-type calculations (Table 1)[26,41,58,101]. However, there are numerous fairly detailed scenarios for future bioenergy production reviewed elsewhere [32,59], which provide a useful starting point in trying to estimate the CDR potential. Estimates of bioenergy potential range over 50–500 EJ yr^-1 in 2050, with the main contributors expected to be energy crops (40–330 EJ yr^-1), surplus forest biomass (60–100 EJ yr^-1) and the ‘residues’ (i.e., waste) that accompany agriculture and forestry (30–180 EJ yr^-1) [59]. To convert bioenergy forecasts in energy units to carbon fluxes requires knowing the corresponding energy density and carbon content of the various fuel types, and these are often not given in the published studies. An excellent source in this regard is a review of 17 studies up to 2003 that details the underlying assumptions regarding planting areas and dry mass yields [32]. Inspection of these assumptions suggests the upper end of the bioenergy ranges should be treated with caution [32].

Underlying the estimates for energy crops is the assumption that the plantation area will range over 390–750 Mha in 2050, with yields of typically 8–15 Mg ha^-1 yr^-1 (dry mass) and global production estimates clustering at around approximately 5–6 Pg yr^-1 (dry mass) [32]. Assuming that (at the upper limit) 0.5 gC g^-1 gives 2.5–3 PgC yr^-1 of energy crop biomass in 2050, and assuming that approximately 50% of this can, in principle, be captured [101], gives up to 1.25–1.5 PgC yr^-1 CDR via energy crops in 2050. A much higher estimate of 6.16 PgC yr^-1 has been made for 2035 (Table 1)[101] based on 1.15 Gha of plantation (0.43 Gha sugar cane and 0.72 Gha switchgrass), but it implies an average yield of approximately 20 Mg ha^-1 yr^-1, which is well above other studies, and a supply of land of 46 Mha yr^-1, far in excess of the projections of abandonment of agricultural land.

The forecast bioenergy supply from surplus forest biomass of 60–100 EJ yr^-1 in 2050 [59], given that wood has a typical energy content of 20 GJ Mg^-1, corresponds to 3–5 Pg yr^-1 dry mass, or 1.5–2.5 PgC yr^-1, suggesting a CDR potential of 0.75–1.25 PgC yr^-1(Table 3). An alternative estimate is 1.8 PgC yr^-1 in 2035 (Table 1)[101], but this relies on 1 Gha of afforestation in 25 years. The implied supply of carbon in 2050 should be critically compared with the yield from existing forests and independent estimates of afforestation discussed earlier. Current global wood removals are approximately 1 PgC yr^-1[46], but only approximately half of this is used as fuel and most of that does not count as ‘modern’ biomass energy amenable to large-scale bioenergy with carbon storage (BECS). Meanwhile, the accumulation of carbon in the biomass of new forests is only forecast to be approximately 1.5 PgC yr^-1 in 2050 (Table 1). Thus, the upper end estimates of bioenergy from ‘surplus’ forest biomass in 2050 appear to be in excess of the likely supply. This can be reconciled if higher yields are being assumed and are achievable. Otherwise, either net removal of carbon from standing biomass or larger areas of afforestation are implied. In the first case, the biomass would not be ‘surplus’ and it should not be considered as CDR (but rather a potential CO₂ source). In the second case, the implied afforestation may conflict with other land uses.

A supply of agricultural and forest residues of 30–180 (mean: 100) EJ yr^-1 in 2050 is deemed to be the most certain source of bioenergy [59]. Assuming these residues have an average energy content of 15 GJ Mg^-1, they correspond to 2–12 (mean: ∼7) Pg yr^-1 dry mass or 1–6 (∼3.5) PgC yr^-1. Currently, all unused crop residues are 1.5 Pg yr^-1 and felling losses in forests are 0.65 Pg yr^-1, totaling 2.15 Pg yr^-1[46] or approximately 1.1 PgC yr^-1; hence, the lower end of the range projected for 2050 is certainly realistic. A growing population will lead to more agricultural waste and, if large-scale afforestation and bioenergy cropping also occur by 2050, then 3.5 PgC yr^-1 seems plausible. If this was all subject to CO₂ capture and storage, then CDR of up to approximately 1.75 PgC yr^-1 may be achievable. However, it may make more sense to devote agricultural and forestry residues to biochar production resulting in a comparable CDR flux and also helping to maintain soil quality.

By 2100, integrated assessments of bioenergy potential constrained by the supply of suitable land area tend to be approximately double what they are in 2050 [29,32]. The greatest ‘geographical’ potential is on abandoned agricultural land, with one study giving a range of 240–850 EJ yr^-1 in 2100 for woody energy crops [29]. Assuming a typical energy content of 20 GJ Mg^-1 and 0.5 gC g^-1, this corresponds to 6–21 PgC yr^-1, which if it was all subject to 50% efficient CO₂ capture gives a potential CDR flux of 3–10.5 PgC yr^-1. This compares reasonably well with an earlier estimate of 5.5–9.5 PgC yr^-1 potential biochar CDR flux [26], from 180–310 EJ yr^-1 biomass energy supply in 2100 [32]; although the assumed conversion efficiency is clearly higher in that study.

The ultimate storage capacity for liquid CO₂ is determined by the size of suitable geologic reserves, assuming deep ocean injection will not be used (because of its finite, although lengthy, residence time and fears about impacts on deep sea ecosystems). Estimates of geologic storage capacity range upwards from approximately 500 PgC to approximately 3000 PgC (Table 2)[60]. The low end of this range could present a significant constraint, as it would only take 100 years to produce 500 PgC with the upper estimates of CDR flux discussed above. Consequently, lack of storage capacity could prevent the upper end CDR flux estimates for 2100 from being realized. The problem would be exacerbated if there is competition for storage capacity between liquid CO₂ captured at the point of emission (conventional carbon capture and storage [CCS], which is a mitigation approach), and liquid CO₂ captured from the free air by bioenergy or by chemical pathways. Consequently, the uncertain but potentially large capacity of saline aquifers to store CO₂(Table 2) is critical in determining whether several centuries of ambitious CDR is feasible, and whether they could accommodate all the carbon from know fossil fuel reserves.

Overall potential

Table 3 summarizes the total potential of land biological pathways of CDR based on present biomass flows, and on future projections of afforestation and bioenergy supply for 2050 and 2100. In coming up with the total potential, there is a danger of double-counting carbon if summing apparently independent estimates of CDR flux derived for different methods, because these often (implicitly) take the same biomass carbon source to different end products. A pertinent question, for example, is whether estimates of bioenergy CDR are additional to the potential afforestation CDR? Comparison of studies conducted with IMAGE 2.2 [20,29,38] suggests not; the same land supply is being considered either for permanent afforestation [20,38] or for short-rotation woody energy crops [29]. Given this, the upper estimates of the totals for 2050 and 2100 (Table 3) should be treated with caution, since it is implying the use of land in addition to the supply from abandoning of cropland. Strikingly, there is already 1–1.5 PgC yr^-1 potential for land biological CDR, which is comparable to the latest estimates of the CO₂ emissions due to land use change of 1.5 ± 0.7 PgC yr^-1 over 1990–2005 and 1.2 PgC yr^-1 in 2008 [1]. In 2050, the potential for land biological CDR increases to 4–6 PgC yr^-1, which is approximately 50% of the current total (fossil fuel and land use change) emissions. Hence, if these are cut globally by approximately 50% by 2050, then land biological CDR could match emissions and, with natural sinks present, atmospheric CO₂ concentration would have already passed peak levels and would be declining. In 2100, the potential for land biological CDR levels of 6–14 PgC yr^-1 could match or exceed current total emissions. So, even if mitigation efforts have limited success and emissions are only back to present levels by 2100, the use of land biological CDR could have atmospheric CO₂ concentration declining rather than still rising.

To illustrate this further, we use a simple coupled carbon cycle–climate model [7], with the same set up as in recent work [3]. As a baseline scenario, we assume that fossil fuel emissions are currently rising at the mean rate of 1.9% yr^-1 over the past 25 years, and that concerted global mitigation activity starts immediately in 2010. We further assume that it will take 40 years to undergo the economic and technological transition necessary to achieve a maximum rate of decrease in fossil fuel emissions of -1.9% yr^-1. In 2050, fossil fuel emissions return to today’s level (∼8.4 PgC yr^-1), and from then on they decline at the same rate at which they recently grew, until they cease. This contrasts with our earlier stabilization work [3], but is the same approach used by others [2]. The results in a total fossil fuel emission of 903 PgC after 2000, which added to 282 PgC over 1800–2000 gives 1185 PgC. Cumulative land use change emissions after 2000 are assumed to be 100 PgC (following an exponential decay) and were 180 PgC over 1800–2000, giving 280 PgC. Total emissions are 1465 PgC throughout. We view this as an economically and technologically feasible mitigation scenario, which is still fairly ambitious, but falls short of what will be required to avoid 2°C warming above preindustrial levels. In response, atmospheric CO₂ concentration peaks at 523 ppm in 2100 and global warming peaks at 2.50°C soon after, in 2113 (Figure 3). The 2°C target is exceeded in 2050 as atmospheric CO₂ concentration reaches 490 ppm and cumulative total emissions reach 990 PgC (in good agreement with the ‘trillion tonnes of carbon’ target [2]).

To include CDR in a simple fashion, with as few free parameters as possible, we assume that the flux of CDR as a function of time, R(t), follows a Gaussian curve (a scaled normal distribution). The total storage capacity for carbon (S) constrains the area under the curve, leaving only the year of peak removal activity (the mean, µ) and how narrow/tall to make the distribution (the variance, σ²) to be specified:

From the preceding review, CDR activity has the capacity to increase throughout this century, but if it is being maximized, then it is likely to become constrained by total carbon storage capacity by the end of the century. Consequently, we set the peak CDR flux to occur in µ = 2100. We consider two cases: afforestation CDR, with a total removal of S = 300 PgC (Table 2) and σ = 40 yr; and bioenergy CDR, with a total removal of 1000 PgC (nominally 500 PgC biochar and 500 PgC stored CO₂, Table 2) and σ = 32 yr. The values of σ (together with S) are chosen to produce CDR fluxes comparable to the estimated potential (Table 3): afforestation CDR is 0.24 PgC yr^-1 in 2010 (low end of range), 1.4 PgC yr^-1 in 2050 and peaks at 3.0 PgC yr^-1 in 2100. Bioenergy CDR is 0.24 PgC yr^-1 in 2010 (i.e., it includes current afforestation), 3.7 PgC yr^-1 in 2050, and peaks at 12.5 PgC yr^-1 in 2100 (rather high).

The afforestation CDR scenario lowers peak atmospheric CO₂ by 38 ppm to 485 ppm and brings the peak forward by nearly 35 years to 2067 (Figure 3A). Global warming still exceeds the 2°C target, when CO₂ reaches 480 ppm in 2054 and the ‘cumulative carbon loading’ (i.e., cumulative emissions minus cumulative CDR) is again 990 PgC (only 36 PgC cumulative CDR has occurred by this time). Peak warming is brought forward by 35 years (to 2078) and lowered by 0.34°C to 2.16°C (Figure 3B). By 2100, atmospheric CO₂ has been reduced to 464 ppm (a lowering of 59 ppm) and global warming by 0.42°C to 2.07°C. The calculated drawdown of atmospheric CO₂ due to afforestation is comparable to other estimates of up to 52 ppm in 2100 [20].

The bioenergy CDR scenario lowers peak atmospheric CO₂ by 54 ppm to 469 ppm and brings the peak forward by nearly 50 years to 2053 (Figure 3A). Peak warming is brought forward by 50 years and lowered by 0.51°C, to 1.99°C in 2063, just staying below the 2°C target (Figure 3B). When temperature peaks, the cumulative carbon loading is 970 PgC, having had 123 PgC of cumulative CDR (and emissions in excess of a trillion tonnes). By 2100, atmospheric CO₂ has been reduced to 365 ppm (a lowering of 158 ppm) and global warming by 1.16°C to 1.34°C. At this point, 1283 PgC has been emitted, but 500 PgC has been removed, giving a cumulative carbon loading of 783 PgC.

Inspection of the fluxes of CO₂ to and from the atmosphere (Figure 4) reveals some interesting behavior: CDR suppresses the natural land and ocean carbon sinks and, in the bioenergy CDR scenario, they actually become a net carbon source in 2086 (the land becoming a source in 2073 and the ocean in 2105). This is because any anthropogenic perturbation to atmospheric CO₂, be it an addition or a removal, triggers a counterbalancing response from the land and ocean, as detailed in our previous work [10]. In the bioenergy CDR scenario, with net carbon removal from the system via the atmosphere, and atmospheric CO₂ declining, the land and ocean respond by out-gassing CO₂, as they tend towards a new equilibrium state with lower carbon storage. This behavior emphasizes an inherent limitation for CDR approaches and suggests that the rather high values for bioenergy CDR flux assumed by the end of the century may not be desirable anyway.

Our results come with several further caveats. They represent an approximate upper limit on what land-based biological CDR, using abandoned cropland, can achieve this century. Bioenergy CDR appears to have significantly greater potential than afforestation CDR, despite 50% of the carbon being assumed to be lost as CO₂. The main reason is that in the reviewed studies, the productivity of short-rotation woody biomass energy crops is modeled to be far greater (∼10 MgC ha^-1 yr^-1) [29] than the average yield of afforestation (∼1 MgC ha^-1 yr^-1) [20]. This warrants further scrutiny, as does whether high yields of carbon can be removed from bioenergy ecosystems (and only partially returned as biochar), without reducing soil organic carbon storage. In our modeling, we have not calculated non-CO₂ climatic effects. In particular, afforestation in the high latitudes can lead to net warming due to shading snow and lowering surface albedo and, if biochar is exposed on bare soil surfaces (e.g., after cropping), it may also lower surface albedo, causing warming.

Future perspective

Current pledges for national emissions targets under the Copenhagen Accord cannot limit global warming to 2°C (despite this being the stated aim of the Accord) [5,61]. Given this, if the 194 member states of the UNFCCC are genuinely committed to its goal (Article 2), then they will have to give much more policy attention to methods of CDR, as an additional means of trying to avoid dangerous climate change. Our review and modeling suggests that land-based biological methods of CDR could play a significant role in helping limit global warming to no more than 2°C, but early action is imperative, just as it is for efforts to reduce CO₂ emissions.

Encouragingly, significant afforestation is already underway, and recent trends suggest that afforestation activities will continue to grow and deforestation to dwindle [36]. In an optimistic scenario, within 10 years, net land use change activities could be approaching carbon neutrality (i.e., created sinks balancing land use sources). For land-based biological CDR by means other than afforestation to become significant in the future requires that credits be earned for carbon removed, such as through biochar production or CO₂ entering geological storage. Accreditation requires verification that a given amount of carbon has been removed and that it is staying where it is put. Quantifying removed carbon should actually be easier and potentially more reliable than quantifying avoided emissions, especially from deforestation – because it is an exercise in measuring an actual flux rather than an avoided one. The biochar that is produced, for example, can simply be weighed. However, considerable monitoring would be required to establish whether, for example, biochar in soil was losing carbon and at what rate. The accreditation problem for CO₂ in geological storage is being tackled anyway, because of the intended use of CCS at points of fossil fuel emission, and this should aid the uptake of bioenergy with CO₂ capture and storage.

More broadly, carbon markets will need to become widespread and stable, and the price of CO₂ pollution, or conversely the earning from CO₂ removed, will need to be set at a reasonably high level for there to be large-scale uptake of CDR methods. This is also a prerequisite for many methods of achieving meaningful reductions in CO₂ emissions. The precise price of carbon required to trigger significant activity will vary between the technologies and needs further research. The fact that significant afforestation is already happening suggests that the cost is not prohibitive, yet existing studies range over orders of magnitude in their cost estimates [19]. The cost of afforestation will increase with the size of CDR flux being generated, but most of the short term potential can be realized at less than US$100 MgC^-1[38]. Biochar production has been argued to be competitive with biomass combustion (without CCS), at a relatively low cost of US$33–59 MgC^-1, if one factors in the co-benefits of applying biochar to soil [62]. Subsequent life cycle analysis puts the breakeven price for biochar production at only US$7 MgC^-1 from yard waste, but US$147 MgC^-1 from crop residues and US$227 MgC^-1 from bioenergy crops [51]. BECS carries the cost of CCS, making it relatively expensive (e.g., US$84–194 MgC^-1) [25]. Burial of biomass makes no use of embodied energy, yet wood burial could still be relatively cheap at an estimated US$50 MgC^-1[21]; whereas burial of crop residue in the deep ocean may be the most expensive option, at an estimated US$340 MgC^-1[23].

Further research is needed into several of the CDR methods before large-scale deployment is considered, such as long term experiments on the effects of wood burial deep in soil. In addition, the effects of (unavoidable) climate change on photosynthesis-based CDR need to be assessed. CO₂ fertilization would be expected to enhance CDR fluxes (as long as water or nutrients are not limiting), whilst the effects of rising temperatures could be positive or negative depending on the region and the magnitude of warming. Recent trends of increasing forest disturbance and losses of soil carbon suggest that it may become harder to generate and maintain stores of carbon in forests and soils, at least in some regions.

For land-based biological methods of CDR to play a significant future role then they must not threaten food production (Article 2 of the UNFCCC). Hence, to a large extent, their potential rests in large part on the wider scientific and societal challenge of increasing the efficiency of land use for food production. Whilst large increases in the land use efficiency of global food production seem eminently possible in principle [30,63], if they cannot be achieved in practice, then attention should be turned to physical and chemical methods of CDR that make a much smaller demand on land area. Exciting research is underway on these [11] but, as noted at the outset, the costs are comparatively high [12].

Table 1.  Existing forecasts of potential CO2 removal flux from afforestation, biochar production and bioenergy with CO2 capture.

Method	Time (year)	CDR flux (PgC yr^-1)	Key conditions/assumptions	Ref.
Afforestation	2000	0.19	China only (1998–2001)	[37]
	2010	0.21–0.42	264 Mha × 0.8–1.6 MgC ha^-1 yr^-1	[36], this study
	2015	0.53	193 Mha (≡ 2.74 MgC ha^-1 yr^-1)	[45]
	2025	0.83	345 Mha in ∼35 years	[45]
	2030	0.12–0.24	US$20–100 (tCO2)^-1	[39,40]
		0.4–0.8	As above, all forest products	[39,40]
	2035	10.0	1 Gha × 10 MgC ha^-1 yr^-1	[41,101]
	2050	1.0	550 Mha	[56]
		0.2–1.5	SRES A2, B1, B2, A1b range	[38]
		∼0.6–1.2	SRES B2, A1b	[20]
	2055	1.48	345 Mha	[45]
	2100	0.3–3.3	SRES A2, B1, B2, A1b range	[38]
		∼1.5–3.3	SRES B2, A1b	[20]
Biochar production	N/A	0.56	Estimate of present potential	[26]
	2035	1.2	1 Gha, 2.5 MgC ha^-1 yr^-1 to biochar	[41,101]
	2060	1.74	1.5% yr^-1 growth of above	[41,101]
	2100	3.15	Continued 1.5% yr^-1 growth	[10]
		5.5–9.5	180–310 EJ yr^-1 all by pyrolysis	[26]
CO2 capture from bioenergy	N/A	0.19-0.23	Present sugarcane and pulp mills	[25]
	2035	6.16	1.15 Gha sugarcane and switchgrass, ∼10 MgC ha^-1 yr^-1, 53% captured	[101]
		1.8	1 Gha forest, 2.5 MgC ha^-1 yr^-1, 60% captured	[101], corrected
	2060	11.6	1.5% yr^-1 growth of above	[101]

Note that the start years of future projections often differ between studies.

CDR: CO₂ removal; SRES: Special Report Emissions Scenarios.

Table 2.  Estimates of the storage capacity for removed carbon in various forms and reservoirs.

Form of carbon	Location	Storage capacity (PgC)	Key assumptions/basis	Ref.
Biomass	Permanent new forests	50–100		[42]
		104	345 Mha in 100 years	[45]
		60–87	2050	[47]
		51–113	2100, extrapolation of current rates	[43]
		48–147	2105, demand for forest products	[44]
		120	2035, 1 Gha	[41]
		900	2100, physical potential	[20]
		17–146	2100, SRES A2, B1, B2, A1b range	[38]
		68–133	2100, social potential, SRES B2, A1b	[20]
		∼300	Long-term social potential
Biochar in soils	Cropland	224	1.6 Gha, 140 MgC ha^-1	[26]
	Grassland	175	1.25 Gha, 140 MgC ha^-1	[26]
	Abandoned cropland	98–140	0.7–1 Gha by 2100, 140 MgC ha^-1	This study
		∼500	Total
Liquid CO2	Oil and gas fields	184–245		[60]
	Unmineable coal seams	4–55		[60]
	Deep saline formations	273–2700		[60]
		∼500–3000	Total

SRES: Special Report Emissions Scenarios.

Table 3.  Estimates of present, 2050 and 2100 global land biological CO2 removal flux potential by pathway.

Carbon source	Carbon store	CO2 removal flux (PgC yr^-1)	Key conditions/assumptions	Ref.
Present
Afforestation	Standing biomass	0.21–0.42	264 Mha × 0.8–1.6 MgC ha^-1 yr^-1	[36], this study
Forestry residues	Burial	0.33	0.65 Pg yr^-1 of felling losses	[46], this study
	Biochar	0.16	All felling losses from forestry	[46], this study
Crop residues	Burial	0.18–0.6	1.5–5 Pg yr^-1, 30% removed	[23,46], this study
	Biochar	0.18	50% of unused crop residues	[51]
All residues	Biochar	0.16–0.34	Original-revised estimates	[26,46], this study
Energy crops	Biochar	0.18	All biomass energy by pyrolysis	[26]
Shifting cultivation	Biochar	0.21–0.35	All shifting cultivation fires	[26,52], this study
Charcoal making	Biochar	0.01	All waste from charcoal making	[26]
All	Various	1–1.5	Total potential
2050
Afforestation	Standing biomass	0.2–1.5	All abandoned cropland	[20,38,45,56]
Energy crops	BECS CO2	1.25–1.5	390–750 Mha, 8–15 Mg ha^-1 yr^-1 (dm)	[32], this study
Surplus wood	BECS CO2	0.75–1.25	60–100 EJ yr^-1, 20 GJ Mg^-1, 50%	[59], this study
All residues	BECS CO2	1.75	∼100 EJ yr^-1, 15 GJ Mg^-1, 50%	[59], this study
Shifting cultivation	Biochar	0.25–0.35	As today	[52], this study
All	Various	4–6	Total potential
2100
Afforestation	Standing biomass	0.3–3.3	SRES A2, B1, B2, A1b range	[20,38]
All bioenergy	BECS CO2	3–10.5	240–850 EJ yr^-1, 20 GJ Mg^-1, 50% captured	[29], this study
	Biochar	5.5–9.5	180–310 EJ yr^-1 all by pyrolysis	[26]
All	Various	6–14	Total potential

BECS: Bioenergy with carbon storage; SRES: Special Report Emissions Scenarios.

Article 2 of The UN’s Framework Convention on Climate Change (UNFCCC)

States the overarching objective of “stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system. Such a level should be achieved within a time-frame sufficient to allow ecosystems to adapt naturally to climate change, to ensure that food production is not threatened and to enable economic development to proceed in a sustainable manner”.

Geoengineering

Deliberate large-scale manipulation of the planetary environment to counteract anthropogenic climate change. It can be subdivided into methods of reducing the absorption of sunlight (‘solar radiation management’) and methods of removing greenhouse gases from the atmosphere, especially ‘CO₂ removal’.

CO₂ removal (CDR)

Subset of geoengineering methods that involve actively removing CO₂ from the ambient air, by biological, chemical or physical means, and storing the resulting carbon in long-lived reservoirs.

Afforestation

Establishment of forest on land that has not recently been forested; whereas, reforestation is the reestablishment of forest after recent removal. For afforestation to count as a carbon store it must be permanent.

Biochar

Charcoal created by the pyrolysis of biomass that is added to soil to store carbon.

Bioenergy with carbon storage (BECS)

Technologies that use photosynthesis to remove carbon from the atmosphere, making use of some of the energy in the resulting biomass and capturing some of the carbon in long-lived forms (either CO₂ in geological storage or biochar).

Executive summary

Pathways & constraints

	▪ Land-based biological CO2 removal (CDR) involves diverting carbon captured from the atmosphere by photosynthesis to long-lived reservoirs.
	▪ These reservoirs include permanent forests, buried biomass (either deep in soil or in the deep ocean), biochar in soils and CO2 stored in geological formations.
	▪ The CDR flux achievable is constrained by the supply of land area, the yield of carbon on that land area, and the efficiency of converting it to long-lived carbon stores.

Afforestation & reforestation

	▪ Afforestation is already removing an estimated 0.21–0.42 PgC yr-1.
	▪ This could rise to approximately 1.5 PgC yr-1 in 2050 and approximately 3.3 PgC yr-1 in 2100.
	▪ The total storage capacity is approximately 300 PgC in standing trees and associated soil carbon.

Biomass burial

	▪ Burial of all existing forestry residues could remove 0.33 PgC yr-1.
	▪ Burial of 30% of existing crop residues could remove 0.18 PgC yr-1.
	▪ However, the potential for methane generation, erosion of soil carbon, and compromises to other ecosystem services suggest further research is required.

Biochar production & soil carbon

	▪ Biochar production from existing biomass flows could remove 0.77–0.87 PgC yr-1.
	▪ It carries fewer risks and greater benefits than burying biomass.
	▪ The total storage capacity for biochar in soil is approximately 500 PgC.
	▪ The potential to increase soil organic carbon by changing agricultural practices is an order of magnitude smaller and would need to be permanent in order to count as CDR.

Bioenergy with carbon storage

	▪ Bioenergy with capture and storage (as CO2 or biochar) could remove approximately 4 PgC yr-1 in 2050 and more than 10 PgC yr-1 in 2100.
	▪ The achievable CDR flux depends crucially on the supply of abandoned cropland, and could ultimately be constrained by lack of storage capacity.
	▪ The total storage capacity for CO2 in geological formations is approximately 500–3000 PgC, but the upper estimates depend critically on the uncertain capacity of saline aquifers.

Overall potential

	▪ There is already the potential for approximately 1–1.5 PgC yr-1 of CDR, by diverting or altering existing biomass flows.
	▪ This could counterbalance current CO2 emissions from land-use change.
	▪ By 2050, the potential CDR flux is 4–6 PgC yr-1.
	▪ Together with natural sinks, this could match current total CO2 emissions, thus stabilizing atmospheric CO2 concentration and lowering peak global warming.
	▪ By 2100, the potential CDR flux is 6–14 PgC yr-1.
	▪ This could significantly exceed mitigated CO2 emissions, thus bringing down atmospheric CO2 concentration and reducing global warming.
	▪ The total amount of carbon that can be stored probably exceeds 1000 PgC.
	▪ This is sufficient to accommodate at least two centuries of CDR activity and could potentially contain all the fossil and land-use carbon that we will emit.

Acknowledgements

I thank Ed Sears, Nem Vaughan and the anonymous referees for helpful comments.

Financial & competing interests disclosure

This research was supported by the Norfolk Charitable Trust through the GeoEngineering Assessment and Research (GEAR) initiative at the University of East Anglia. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.