Soil Carbon Sequestration or Biofuel Production: New Land-Use

Oct 13, 2009 - To generate 10% of the World energy demand by 2050 through .... We also define the marginal carbon benefit of biofuels Cmarg,bio = Cbio...
1 downloads 0 Views 1MB Size
Download PDF
Environ. Sci. Technol. 2009, 43, 8678–8683

Soil Carbon Sequestration or Biofuel Production: New Land-Use Opportunities for Mitigating Climate over Abandoned Soviet Farmlands N I C O L A S V U I C H A R D , * ,†,‡ PHILIPPE CIAIS,‡ AND ADAM WOLF§ Centre International de Recherche sur l’Environnement et le De´veloppement-CNRS/EHESS, Nogent sur Marne, France, Laboratoire des Sciences du Climat et de l’Environnement, IPSL-LSCE CEA/CNRS/UVSQ, Saclay, France, and Department of Global Ecology, Carnegie Institution for Science, Stanford University, Stanford, California

Received June 5, 2009. Revised manuscript received September 15, 2009. Accepted September 24, 2009.

Although the CO2 mitigation potential of biofuels has been studied by extrapolation of small-scale studies, few estimates exist of the net regional-scale carbon balance implications of biofuel cultivations programs, either growing conventional biofuel crops or applying new advanced technologies. Here we used a spatially distributed process-driven model over the 20 Mha of recently abandoned agricultural lands of the Former Soviet Union to quantify the GHG mitigation by biofuel production from Low Input/High Diversity (LIHD) grass-legume prairies and to compare this GHG mitigation with the one of soil C sequestration as it currently occurs. LIHD has recently received a lot of attention as an emerging opportunity to produce biofuels over marginal lands leading to a good energy efficiency with minimal adverse consequences on food security and ecosystem services. We found that, depending on the time horizon over which one seeks to maximize the GHG benefit, the optimal time for implementing biofuel production shifts from “never” (short-term horizon) to “as soon as possible” (longerterm horizon). These results highlight the importance of reaching agreement a priori on the target time interval during which biofuels are expected to play a role within the global energy system, to avoid deploying biofuel technology over a time interval for which it has a detrimental impact on the GHG mitigation objective. The window of opportunity for growing LIHD also stresses the need to reduce uncertainties in soil C inputs, turnover, and soil organic matter stability under current and future climate and management practices.

Introduction Biofuels are being promoted as an important part of the global energy mix needed to meet the climate change challenge. In particular, the use of liquid biofuels has * Corresponding author phone: +33 1 69 08 42 26; fax: +33 1 69 08 30 73; e-mail: [email protected]. † Centre International de Recherche sur l’Environnement et le De´veloppement-CNRS/EHESS. ‡ Laboratoire des Sciences du Climat et de l’Environnement, IPSLLSCE CEA/CNRS/UVSQ. § Department of Global Ecology, Carnegie Institution for Science, Stanford University. 8678

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 22, 2009

experienced a very strong growth in recent years. Although the carbon mitigation potentials of biofuels produced with current technologies at small scales have been studied (1, 2), little research has yet been done on the consequences of regional biofuel cultivation programs, using in particular advanced technologies to generate fuel from the cellulose in crops grown for fiber. To generate 10% of the World energy demand by 2050 through large-scale cultivation of biofuels would require an extra 10 × 106 km2 of suitable land, that is a near doubling of the current area of the world under arable agriculture (3, 4). This will constitute the largest and fastest land use change in human history. Regional vegetation and soil carbon stocks have been shown to be highly vulnerable to agriculture, with losses than can last for several decades after initial cultivation (5-7). This has brought up the issue that a “carbon debt” is generated when clearing natural ecosystems for cultivating biofuels (8). This carbon loss will have to be repaid by fossil CO2 substitution with biofuels up until a positive carbon balance can be restored. In the sensitive case of biofuel-induced deforestation in tropical regions, which may be a serious threat for the Amazon (9), it may even take hundreds of years to recover the carbon debt by the biofuels benefit, not even accounting for the loss of biodiversity and other ecosystem services associated with deforestation. In this context, Field et al. (10) and Campbell et al. (11) aimed to estimate at the global scale the area suitable for biofuel production, which would be both climate beneficial and have minimal adverse consequences on food security and ecosystem services. They suggested to look first for already abandoned agricultural lands as an opportunity to grow biofuels. Over such marginal lands indeed, biofuels would compete less with food production. Moreover, the soil carbon stocks of these lands are generally already depleted after past cultivation practice to minimize the carbon debt in case of land reutilization for biofuels. However, even arable lands no longer used for food production may not be well suited for conventional biofuel production, unless perhaps high inputs are prescribed, thus leading to high N2O emissions. To overcome that limitation, Tilman et al. (12) suggested that low-input/high-diversity (LIHD) grass-legume prairies offer a good opportunity for producing biofuels on degraded lands. In that preliminary study, LIHD grasses were indeed found to provide more usable energy, greater greenhouse gas reductions, and less agrochemical pollution than corn grain ethanol or soybean biodiesel. Thus, cultivating LIHD systems on abandoned low-fertility arable lands appears to be a reasonable option for producing biofuels with an environmental benefit. However, we can still wonder, as did Righelatto et al. (13), if it would not be better for mitigating CO2 emission and climate change, to sequester carbon into the soils by a proactive land conversion to natural ecosystems (like reforestation) rather than to produce biofuels. In this paper, we focus on the case study of agricultural lands of the former Soviet Union (FSU) that were abandoned during the 1990s. This abandonment of up to 20 million hectares (Mha) in total (14) constitutes the largest regional land use change of the last 50 years north of the equator. For comparison, this area is approximately the total land area used for small grain production in Canada (maximum 22.1 Mha in 1986), or to 15% of the current U.S. cropland area. Over abandoned collective farmlands in Russia and Ukraine, grass and sage-dominated steppe vegetation is now recovering the land. Such a widespread and unintended land-use change has important implications on the terrestrial C sink, 10.1021/es901652t CCC: $40.75

 2009 American Chemical Society

Published on Web 10/13/2009

particularly by altering soil C dynamics. Vuichard et al. (15) concluded for instance that during the period 1990-1999, about 64 Tg C have been sequestered into the soil of these abandoned lands. The present study uses a spatially distributed process-driven model named ORCHIDEE (15, 16; see also Supporting Information (SI) for more details) to examine the regional GHG balance of the abandoned agricultural lands of the FSU and its future evolution, and in particular (i) to quantify the GHG mitigation by soil C sequestration and (ii) compare it to the GHG mitigation by biofuel production based upon LIHD system.

TABLE 1. Description of the Twenty Factorial Scenarios Related to the Timing of Grassy Biofuel Implementation, Tillage Management, and Future Climate Change Impacts driver timing

scenario BIO1 BIO2

tillage

NOTILL TILL

Materials and Methods Metric for Comparing Sequestration vs Biofuel Production Carbon Benefits. We define the carbon benefit of a given land use scenario (sequestration or biofuel) as the cumulative amount of C withdrawn from the atmosphere over a 60-year target period. This benefit includes net fossil fuel substitution and C sequestration by biomass production or by sequestration alone in abandoned lands. We name Cseq,nat the benefit of sequestration alone in unmanaged systems, and Cbio the total benefit of biofuel production in LIHD systems. Cbio is the sum of two terms, Cbio ) Cseq, bio + Csub, bio, where Cseq, bio is the below-ground C sequestration and Csub, bio is the net fossil fuel substitution benefit. Net fossil fuel substitution is the carbon benefit of biofuels production as an energy substitute. Both carbon sequestration and biofuel production calculations are detailed in the SI. To compare the biofuel and sequestration benefits, we define the crossing time tcross, as the date after implementation at which biofuels deliver the same carbon benefit as sequestration, i.e., Cbio(tcross) ) Cseq,nat(tcross). In principle tcross can vary between zero (if biofuels take the immediate advantage) and infinity (if biofuels never beat sequestration). We also define the marginal carbon benefit of biofuels Cmarg,bio ) Cbio - Cseq,nat. By definition, we have Cmarg,bio(tcross) ) 0. Scenarios and Sensitivity Studies. Sensitivity to the Timing of Grassy Biofuel Implementation. We define two contrasting scenarios in terms of LIHD grassy biofuel implementation over the former Soviet farmlands. The first scenario called BIO1, considers that biofuel production started at the time of land abandonment (year 1990 in the model, but this could be any other recent year). The second scenario called BIO2, assumes that biofuel production starts 60 years after land abandonment (year 2050 in the model). For both biofuel production scenarios, we calculate in parallel the C balance trajectory associated with the alternative sequestration land use. Sensitivity to Tillage Management under LIHD Cultivation. We test the sensitivity of C benefits to two contrasting tillage scenarios for biofuel production. One scenario (NOTILL) assumes no-till practice as it really occurs in a LIHD system. The second scenario (TILL) assumes that the land is annually tilled, which would accelerate soil C decomposition. In the ORCHIDEE model, this effect is parameterized by prescribing a 25% shorter mean residence time (MRT) of soil carbon pools (active, slow and passive soil organic matter being equally affected). The role of developing these two contrasted scenarios is to assess, by difference, the contribution and the effect of the no-till practice to the biofuel benefit. Sensitivity to Future Climate Change Impacts. We consider five climate change scenarios and their impact on Net Primary Productivity (NPP ) GPP - Rplant) and on soil C MRT. The REF climate scenario assumes present-day climate of 1990-1999 repeated in loop in the future to force ORCHIDEE. In the NPP+ and NPP- climate scenarios, climate change is assumed to alter NPP of LIHD grasses by +50 and -50%, respectively, in comparison to REF values. In the MRT+ and MRT- scenarios, climate change is assumed to impact soil carbon MRT by +50 and -50% in comparison to REF.

climate

REF NPP+ NPPMRT+ MRT-

description biofuel production starts at the time of land abandonment in 1990 biofuel production starts 60 years after land abandonment in 2050 no-till practice (no impact on SOC decomposition) tillage accelerates SOC decomposition (MRT is reduced by 25%) present-day climate of 1990-1999 is repeated in loop in the future climate change alters NPP by +50% in comparison to REF values climate change alters NPP by -50% in comparison to REF values climate change alters MRT by +50% in comparison to REF values climate change alters MRT by -50% in comparison to REF values

Although intuition suggests a decrease in MRT and an increase in NPP, there is currently large uncertainty associated with temperature response of both plant growth (17) and soil carbon decomposition (18). Therefore we choose rather large intervals ((50%) for exploring uncertainty on both NPP and MRT. Thus, we ensure that these extreme values bracket the one that will really occurs. In addition to the uncertainty associated to climate change, these sensitivity tests should be seen as a way to deal with the overall uncertainty on NPP and MRT, especially the one from modeling. Overall, we obtain 20 scenario combinations: 2 timing × 2 tillage × 5 climate (Table 1). These must be seen as a test to investigate the sensitivity of our model results to the economic decision of when to plant biofuel, modulated by different soil tillage management and by uncertain climate change effects. We first present results of our control simulation, NOTILL × REF climate, to analyze how C benefits depend on the timing of biofuel implementation. Then, we contrast the TILL and NOTILL scenarios. Last, we analyze how future climate change modulates the modeled C benefits.

Results and Discussion Results of the Control Simulation. Decision to Grow LIHD Biofuel Taken Now (BIO1). The time-course of the C benefit associated with BIO1, as compared with pure sequestration, is shown in Figure 1. By the end of the 60-year target period, sequestration produces a lower benefit than biofuels, with Cseq,nat ) 1500 g C m-2 and Cbio ) 2000 g C m-2, respectively. With an annual C yield of ∼95 gC m-2 yr-1, biofuels cultivation substitutes a net cumulative 1600 gC m-2 of fossil fuel (taking into account N2O and process-related diesel CO2 release). Additionally, biofuels also sequester C in the soils; Cseq,bio reaches up to 400 gC m-2 after following a temporal dynamic similar to Cseq,nat, showing that the soil C increase can represent a non-negligible 20% of the total biofuel benefit. In the first three decades however, sequestration (the continuous line in Figure 1A) has a benefit superior to biofuel cultivation, because of its higher “storage” efficiency at that time. The spatial distribution of the crossing time, tcross, at which biofuel begin to yield a larger C benefit than soil sequestration over abandoned farmland, is given by Figure 2. The average regional value of tcross is 30 years (range 15-60 years). Decision to Grow LIHD Biofuel Taken in 60 Years (BIO2). The benefit of sequestration is lower when a land use decision is made 60 years from now (Cseq,nat ) 200 g C m-2 in BIO2 vs 1500 g C m-2 in BIO1). This is because, when it is decided in 60 years under the scenario BIO2 to continue with the VOL. 43, NO. 22, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8679

FIGURE 1. C benefits by soil sequestration (continuous line) and by different scenarios for producing biofuels (dotted line) taking into account sequestration process (dashed line) and substitution effect (not shown) over abandoned agricultural lands of FSU under the BIO1-NOTILL (A), BIO2-NOTILL (B), BIO1-TILL (C), and BIO2-TILL (D) scenarios. Square lines represent time at which biofuels deliver the same carbon benefit as soil sequestration.

FIGURE 2. Spatial distribution of tcross index over abandoned agricultural lands of FSU obtained with the BIO1-NOTILL scenario. sequestration land use, the steppe soil carbon pools have already been partly refilled, approaching their equilibrium steady state (input balancing decomposition). Averaged over the whole domain, the modeled Cseq,nat value in BIO2 is hence quite small (Figure 1B). In addition, starting to grow LIHD grassy biofuel 60 years after the initial land abandonment (i.e., their cultivation starts by year 2050 in the model) has the negative consequence to deplete soil C stocks via reduced litter input. In consequence, the Cseq,bio benefit in the BIO2 scenario remains negative over the period 2050-2110 (mean value of -800 g C m-2). Over the same interval, the amount of substituted fossil carbon Csub,bio is similar to the BIO1 scenario (Csub,bio ) 1600 g C m-2). In consequence, the total biofuel benefit is positive, Cbio ) 800 gC m-2. In BIO2, the regional mean value of tcross is equal to 30 years, similar to the BIO1 scenario in which biofuel cultivation starts in 1990. The important result here is that tcross is insensitive to the timing of implementation for biofuel cultivation. In fact, it is not only tcross that is independent of when the decision to grow biofuels is taken, but also the marginal carbon benefit 8680

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 22, 2009

FIGURE 3. Evolution of the marginal C benefit of the biofuel production strategy in comparison to the soil sequestration strategy over time after biofuel production implementation obtained under NOTILL scenario (continuous line) and TILL scenario starting producing biofuels right after abandonment (dashed line) and 60 years after abandonment (dotted line) over abandoned agricultural lands of FSU. of biofuel vs sequestration. Figure 3 shows the evolution of this marginal benefit, Cmarg,bio ) Cbio - Cseq,nat (solid line). Sensitivity to Tillage Applied on Grasslands Cultivated for Biofuel. Decision to Grow LIHD Biofuel Taken Now (BIO1TILL). By accelerating soil C decomposition, tillage hence diminishes Cseq,bio. This practice does not affect Cseq,nat or Csub,bio. Note however, that in theory tillage increases the diesel-C emissions in production. With tillage being applied each year, after 60 years, the value of Cseq,bio is less than 100 gC m-2, compared with 400 gC m-2 in the Control no-tillage simulation. Consequently, the biofuel benefit Cbio is reduced by 300 gC m-2 compared to the control simulation, reaching a value of 1700 gC m-2. In that case, tcross increases to 48 years (instead of 30 years). Decision to Grow LIHD Biofuel Taken in 60 Years (BIO2TILL). In comparison with the BIO2-NOTILL scenario, the BIO2-TILL one only differs by the temporal dynamics of Cseq,bio, which is reduced by 400 gC m-2 when tillage is applied to the cultivated grasslands (Figure 1D). In consequence,

FIGURE 4. C marginal benefits for the NOTILL, BIO1-TILL, and BIO2-TILL scenarios over four time periods after biofuel implementation under the REF, NPP+, and NPP- climate conditions (A) and the REF, MRT+, and MRT- climate conditions (B). Cbio ) 800 - 400 ) 400 gC m-2 in BIO2-TILL, which is only half of the benefit obtained in BIO2-NOTILL. tcross equals 53 years, that is 5 years more than if biofuel production starts in 1990, as assumed in BIO1-TILL. Similarly, the marginal Cmarg,bio curve differs from the BIO1-TILL scenario (Figure 3) because of the nonsymmetric trends of slowly recovering carbon (continuous line of Figure 1C) and of quickly losing carbon (dashed line of Figure 1D). Sensitivity to Climate Change. First we analyze the effects of climate-induced trends in NPP. The marginal benefits of biofuel production obtained for climate enhanced NPP (NPP+) or for decreased NPP (NPP-) scenarios are shown in Figure 4. Future changes in NPP impact the time derivative (slope) of the marginal C benefit. For instance, a climateinduced increase in NPP (NPP+) leads a ∼100% higher slope than in the REF climate scenario where climate is equal to present. The converse is true for the NPP- scenario. We calculated the sensitivity of the marginal C benefit to NPP future changes for all three combinations of tillage and timing of date of starting cultivation (NOTILL being insensitive to timing). In the NPP+ climate scenario for instance, Cmarg,bio is 118%, 221%, and 280% greater than under the REF climate for the NOTILL, BIO1-TILL, and BIO2-TILL scenarios, respectively, after 60 years of biofuel cultivation (Figure 4A). Opposite and symmetric results are obtained with the NPPscenario, thus leading to negative benefits values, for all combinations of tillage/starting date scenario. Second, we analyze effects of climate-induced changes in soil organic carbon decomposition rates. Increasing the MRT of soil carbon worsens the competitiveness of biofuel production in comparison to C sequestration. An opposite result is obtained by decreasing the soil MRT value. For instance, the MRT- scenario results indicate a biofuel marginal C benefit 49%, 142%, and 236% greater than the control MRT value under the REF climate, for the NOTILL, BIO1-TILL, and BIO2-TILL scenario respectively (Figure 4B). While the marginal C benefit of growing LIHD biofuel on former Soviet farmlands is higher than in the REF experiment for both the NPP+ and the MRT- scenarios, it is interesting to note a contrast in the time evolution of Cmarg,bio under these two scenarios. In the NPP+ scenario, the longer one waits to plant LIHD grassy biofuels, the higher the extra benefit above the REF scenario. In contrast, in the MRTscenario, accelerated soil organic matter decomposition gives an advantage over the REF scenario only as long as soil C pools are out equilibrium. When they approach equilibrium, there is no longer an advantage. The tcross index is also highly sensitive to MRT and NPP changes. For the NPP+ and the MRT- scenarios, tcross gets reduced by 8-16 years in comparison with the REF scenario (Table 2). tcross is conversely

TABLE 2. tcross Index Values (Expressed in Years after Biofuel Implementaion) for NOTILL, BIO1-TILL, and BIO2-TILL Scenarios under the REF, NPP+, NPP-, MRT+, and MRTFuture Climate Change future climate change effects

NOTILL

BIO1-TILL

BIO2-TILL

REF NPP+ NPPMRT+ MRT-

30 22 >60 57 21

48 36 >60 >60 35

53 43 >60 >60 37

increased by up to 30 years in the NPP- and MRT+ scenarios and without tillage (i.e., most advantageous scenario for sequestration). Based on a climate-carbon coupled model projection (IPSL-CM4-LOOP, (19)), climate change is expected to enhance NPP over FSU that increases by 11.4% per decade relatively to the 1991-2000 mean value while MRT is only reduced by ∼2% per decade. Both trends consequently favor the biofuel strategy: under the LOOP climate projection, tcross gets reduced only by 1 year relative to the REF climate for the BIO1-NOTILL scenario but Cmarg,bio increases by 300 gC m-2 (+54%) after 60 years of biofuel production. Is There an Optimal Timing for Starting Biofuel Cultivation Programs? We showed that thanks to the no-till practice, the marginal carbon benefit of biofuel production under LIHD system is insensitive to the timing of their implementation. Roughly 30 years are required on average for biofuels to outcompete sequestration for mitigating the increase of atmospheric CO2. Oppositely, under conventional tillage, the longer we wait for developing biofuels, the longer it takes to do better than sequestration. For instance, tcross increases from 48 to 53 years between an early implementation right after abandonment (year 1990) and a late implementation 60 years after abandonment (year 2050). Here, we would like to translate these results in terms of optimality, addressing the question: “Is there an optimal time topt for starting to develop biofuel production which would maximize carbon benefits?” While the concept of a repayment time of carbon debt (8) does not consider the differences between near-term and long-term decision making, in practice these different time-horizons are distinct for different stakeholders, specifically nations that must meet short-term treaty obligations, and the global society that desires to minimize long-term emissions for containment of GHGinduced climate change. In our study, we calculate an optimal time to develop biofuel production in the FSU, which varies VOL. 43, NO. 22, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8681

with the target time-window over which stakeholders seek to maximize carbon benefit. We showed that the first years of biofuel production are always associated with a negative C marginal benefit, during the period going from t ) 0 to t ) tcross (Figure 3). In other words, it is never beneficial to implement biofuel production for maximizing carbon benefits over any time-window [0, t1] with t1 < tcross. For meeting such a short-term objective, it is always more favorable to let the steppe recover over the abandoned farmlands. At face value, over a longer time window [0, t2] with t2 > tcross, it is always beneficial to produce biofuels rather than to set aside farmland for sequestration. We have shown in this paper that the biofuel sequestration benefits, Cbio, will always increase with time if t > t2 > tcross. Thus, the earlier the onset date of biofuel production, the bigger the long-term C benefit. In summary, depending on the time horizon over which one seeks to maximize carbon benefits, the optimal time topt for starting biofuel production shifts from “never” (if a too shortterm horizon is chosen) to “as soon as possible” (if a longerterm horizon is decided). The boundary between the shortand long-term horizons is delineated by the crossing time tcross. Interestingly, the value of optimal timing topt for starting LIHD cultivations in Russia and Ukraine abandoned croplands is controlled by two factors. The first factor is related to economic strategy and long-term planning. It defines the time horizon over which a carbon benefit is sought by stakeholders and the beneficiary of the GHG mitigation, which is itself driven by economic return of capital investment, land ownership structure, and institutional constraints (e.g., the duration of emission reduction commitment, such as the Kyoto Protocol). The second factor relates more to soil carbon biogeochemical processes which control tcross. We demonstrated in Figure 2 that tcross varies regionally according to environmental controls on NPP and on decomposition rates. Locally, (e.g., for biofuel production with tillage), the value of tcross can also depend on the initial soil C stocks. Last, we showed that climate change modifications of NPP and MRT have the capacity to significantly impact tcross. The value of tcross is thus difficult to quantify without a spatially explicit ecosystem model like the one we used. Taking the judicious decision of when to start biofuel production over abandoned agricultural croplands of FSU to mitigate climate change thus requires a detailed analysis of economic opportunities, but also a sound knowledge of soil C turnover and stability. In summary, massive biofuel production appears to have a real climate mitigation potential in the case of abandoned collective farmlands of the Former Soviet Union. Because the land that could be used for their cultivation is huge (20 Mha of abandoned lands) Low Input High Diversity (LIHD) grassy biofuels would have the capacity to produce annually 0.23 EJ, i.e., meeting roughly 0.05% of the world energy demand or 0.4% of the FSU energy demand. In terms of environmental benefit, we estimated that biofuel production that would have started in 1990 in that region would sequester 0.40 Gt C after 60 years of production, 0.08 GtC through soil C sequestration and 0.32 GtC through fossil fuel substitution. In comparison, over the same 60-year interval, letting the steppe claim back abandoned farmland would sequester only 0.30 Gt C into the soils. In addition, replacement of cultivated agriculture with no-till agriculture using diversified species mixes can also have positive impacts on soil quality, biodiversity, reduced erosion, and creation of economic opportunities in rural communities that are also low-carbon emitting (12). We deliver here three concluding remarks, two on the future of the energy system and the last on the notion of land marginality. These remarks are of general interest for climate 8682

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 22, 2009

mitigation by biofuel cultivation programs, beyond the case study of abandoned Soviet farmland opportunities. First, we have demonstrated that, depending on the time horizon over which one seeks to maximize the carbon benefit, the optimal time for implementing biofuel production shifts from “never” (short-term horizon) to “as soon as possible” (longer-term horizon). Consequently, it is crucial for stakeholders to clearly define and reach agreement a priori on the target time interval during which biofuels are expected to play a role within the global energy system. Globally, biofuels do not have the potential for replacing a large fraction of liquid fuels without competing with food supply. Thus, they should rather be envisaged as temporary components in energy systems mixing bio- and fossil fuels allowing to mitigate the increase of atmospheric CO2, while new low carbon energy systems are developed and deployed at the global scale. It is this time period that should be well evaluated to avoid deploying biofuel technology over a time interval for which it has a detrimental impact on the C mitigation objective. A consequence of this reasoning is that, once a biofuel program is decided, it will have to be fulfilled at least until the crossing time (tcross). Second, identification of the fossil energy type that biofuels will replace now and in the future should be done accurately since it strongly impacts the environmental benefits of producing biofuels. This is of importance because, contrary to the sequestration strategy whose environmental benefit is intrinsically contained in the soil C accumulation process, biofuels production has little carbon benefit by itself for ecosystems, most of its carbon benefit comes from the substitution of fossil fuels. Finally, Field et al. (10) and Campbell et al. (11) stated that presently abandoned agricultural lands roughly represent the maximal area suitable for producing biofuels safely for climate. In their assessments of the global potential for energy supplied by biofuel over such marginal lands, these authors considered all current abandoned lands as a stable opportunity, without including the time-dependency for using the land. In this paper, we have shown that under given conventional management, such as applying normal tillage practice, the marginal C benefit of biofuel production vs sequestration differs according to the time since the land was abandoned (compare dashed and dotted curves of Figure 3). Thus, from a carbon mitigation perspective, it is inappropriate to treat lands that have been abandoned recently or a hundred years ago equally. Furthermore, the “abandoned” status of currently unused agricultural lands may change in the future. Although most of the abandoned lands are seen today as marginal lands on which producing food or feed is not economically viable, they may become economically valuable again and reverted to agriculture in the future, when relative prices of grain and the fossil fuel inputs used to produce it can change. In consequence, the commitment to produce biofuels on an abandoned land over a defined time interval chosen to deliver a net climate benefit may exacerbate the competition with food production in the future. All the three points discussed above show the current need for future global energy production scenarios related to global energy demand, agro-technology systems, and human population growth and development to implement the most suitable biofuel programs, reconciling both energy and environmental benefits.

Supporting Information Available Details on the studied region, the model used in this study, and how the carbon sequestration and biofuel production are calculated. This information is available free of charge via the Internet at http://pubs.acs.org/.

Literature Cited (1) International Energy Agency. Biofuels for Transport - An International Perspective; Office of Energy Efficiency, Technology and R&D, OECD/IEA: Paris, France, 2004. (2) Food and Agriculture Organisation. State of Food and Agriculture - Biofuels: Prospects, Risks and Opportunities; FAO: Rome, 2008. (3) Fisher, G.; Schrattenholzer, L. Global bioenergy potentials through 2050. Biomass Bioenergy 2001, 20, 151–159. (4) Hoogwijk, M. On the global and regional potential of renewable energy sources, PhD Thesis, Utrecht University, 2004. (5) Gervois, S.; Ciais, P.; de Noblet-Ducoudre, N.; Brisson, N.; Vuichard, N.; Viovy, N. Carbon and water balance of European croplands throughout the 20th century. Global Biogeochem. Cycl. 2008, 22, 2. (6) Gitz, V.; Hourcade, J. C.; Ciais, P. The timing of biological carbon sequestration and carbon abatement in the energy sector under optimal strategies against climate risks. Energy J. 2006, 27, 113– 133. (7) Parton, W. J.; Gutmann, M. P.; Williams, S. A.; Easter, M.; Ojima, D. Ecological impact of historical land-use patterns in the great plains: A methodological assessment. Ecol. Appl. 2005, 15, 1915– 1928. (8) Fargione, J.; Hill, J.; Tilman, D.; Polasky, S.; Hawthorne, P. Land Clearing and the Biofuel Carbon Debt. Science 2008, 319, 1235– 1238. (9) Morton, D. C.; DeFries, R. S.; Shimabukuro, Y. E.; Anderson, L. O.; Arai, E.; Espirito-Santo, F. D.; Freitas, R.; Morisette, J. Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 14637.

(10) Field, C. B.; Campbell, J. E.; Lobell, D. B. Biomass energy: the scale of the potential resource. Trends Ecol. Evol. 2007, 23 (2), 65–72. (11) Campbell, J. E.; Lobell, D. B.; Genova, R. C.; Field, C. B. The global potential of bioenergy on abandoned agriculture lands. Environ. Sci. Technol. 2008, 42 (15), 5791–5794. (12) Tilman, D.; Hill, J.; Lehman, C. Carbon-negative biofuels from low-input high-diversity grassland biomass. Science 2006, 314, 1598–1600. (13) Righelato, R.; Spracklen, D. V. Carbon mitigation by biofuels or by saving and restoring forests. Science 2007, 317, 902. (14) Food and Agriculture Organisation. FAOSTAT data; http:// faostat.fao.org (posted February 2004). (15) Vuichard, N.; Ciais, P.; Belelli, L.; Smith, P.; Valentini, R. Carbon sequestration due to the abandonment of agriculture in former USSR since 1990. Global Biogeochem. Cycl. 2008, 22, 4. (16) Krinner, G.; Viovy, N.; de Noblet-Ducoudre, N.; Ogee, J.; Polcher, J.; Friedlingstein, P.; Ciais, P.; Sitch, S.; Prentice, I. C. A dynamic global vegetation model for studies of the coupled atmospherebiosphere system. Global Biogeochem. Cycl. 2005, 19, 1. (17) Lobell, D. B.; Burke, M. B. Why are agricultural impacts of climate change so uncertain? The importance of temperature relative to precipitation. Environ. Res. Lett. 2008, 3, 034007. (18) Davidson, E. A.; Janssens, I. A. Temperature sensitivity of soil carbon decomposition and feedbacks to climate change. Nature 2006, 440, 165–173. (19) Cadule, P.; Friedlingstein, P.; Bopp, L.; Piao, S.; Ciais, P. Contrasting response between volcanic and anthropogenic aerosol induced cooling on the carbon cycle. Proc. Natl. Acad. Sci. U.S.A. in revision.

ES901652T

VOL. 43, NO. 22, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8683