Use of U.S. Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land-Use Change

M. Wang and Z. Haq argue that the results of our study are undercut by the fact that U.S. corn exports have remained this year at around 2 billion bushels a year. However, to maintain these corn exports in 2007, the U.S. planted 16 million more acres to corn in 2007 than it did in 2006, a 22% increase(1). Those acres came mainly out of soybeans, but also some out of cotton. That resulted in decreases in soybean and wheat production, which maintained exports only by decreasing stocks. Meanwhile, diversion of corn to ethanol and higher crop prices reduced the rate of growth in meat production (1). At the same time, data is showing increased deforestation in the Amazon as crop prices rise, after years of decreases when crop prices declined (2).

Focusing on any single crop year is always potentially misleading, particularly any statements that ignore changes in stocks. In any year, circumstances influence what happens to imports and exports, including currency swings and weather around the world. The basic proof that biofuels in the U.S. are influencing world agricultural production is the sharp rise in crop prices, meaningfully although far from exclusively, attributable to biofuels(3, 4) Every serious economic analysis of biofuels increases in the U.S. predicts that increased biofuel production will lower U.S. grain exports and trigger increased production abroad (5, 6).

Wang, in his second paragraph, broadly challenges our study on the vague criticism that it constitutes "speculative, limited land use change modeling" that is "inadequate" to guide policy decisions. The only specifically articulated basis for this criticism is that general-equilibrium modeling is required. By contrast, our study was based on partial-equilibrium modeling of land use effects. The difference is that while our model accounts for changes in cropland within the agricultural sector based on interactive forces of supply and demand that the model allows to reach equilibrium, our modeling does not account for broader effects on the economy, such as rates of economic growth, changes in energy consumption, etc. Use of general-equilibrium modeling has the advantage of greater completeness, but the disadvantage of introducing a broader range of uncertainties. For example, effects on the broader economy will depend on the price of producing biofuels, about which people can argue, which will in turn influence broader liquid fuel prices. According to energy experts, higher liquid fuel prices probably depress greenhouse gas emissions in the U.S. by depressing economic output, but may increase greenhouse gas emissions in the developing world by encouraging greater reliance on coal instead of oil or natural gas for electricity. General equilibrium models also have to build on a broad range of expectations about the economy as a whole. As a result, use of a general equilibrium model, and debates over these broader effects, which themselves depend on government policies, could cloud the results of the somewhat more direct effects of our analysis.

Although we welcome broader general equilibrium analysis, to the extent the modelers highlight the key assumptions and importance of these assumptions and the sensitivity of results to them, the kind of partial equilibrium analysis we did has at the very least an independent utility. Any additional basis for Wang's broad criticism is left unarticulated. Different kinds of models have strengths and weaknesses, but the utility and rigor of our approach can perhaps be judged by comparison with the approach of others, whose life-cycle analyses have been used to guide policy decisions—for example, Wang's GREET model. Like other biofuel life-cycle analyses, the GREET model incorporates not only no general equilibrium modeling but no partial equilibrium modeling. Economic interactions of supply and demand are almost entirely absent from all the other analyses used to guide biofuels policies. In addition, unlike other life-cycle analyses of biofuels, the GREET model for corn ethanol does include an analysis of land use change. Mr. Wang deserves conceptual credit for this important recognition, but that fact also means that the GREET results purport to inform the world of a final greenhouse gas effect that does reflect land use change.

As discussed in our supporting materials, Mr. Wang's land use change analysis started with some informal, domestic modeling attributed to economists at USDA—an unpublished product that was sufficiently informal that when we contacted Michael Price, one of the authors, he told us he could not recall it. As described by the documentation for GREET, that USDA analysis (which also uses partial equilibrium modeling, at best, while leaving out international interactions) produced a highly implausible domestic result that nonetheless still predicted significant reductions in U.S. exports (a result Mr. Wang apparently now challenges). To calculate the land use charge from this approach, Wang, by the explicit admission in his materials, made a purely arbitrary assumption that half of the exports would not be replaced by any increased production at all anywhere, and that all additional cropland would result from conversion of grasslands. Then, the model attributed greenhouse gas emissions per acre for that world grassland conversion—one number for all the world's land—exclusively to a personal communication with Mark Delucchi. When we contacted Mr. Delucchi, he could not remember the conversation and therefore could not recall the basis for this figure. Even so, this number is almost certainly based in part on Delucchi's general approach —also discussed in our supporting materials—which assumes that any area that grows biofuels will do so only for 30 years, after which time the original vegetation will return; an assumption left out of the discussion in the GREET documentation. All in all, this chain of analysis led to a finding by GREET that emissions from land use change were minimal, around 1% of total emissions of corn ethanol. It is curious that Wang appears to consider it appropriate for the world to base biofuel policy decisions on his rather informal analysis of land use change, but inappropriate to base policy decisions on what we expect he would acknowledge is at least a somewhat more sophisticated analysis.

Although not fully articulated, Wang's E-Letter also appears to imply that the key reason for his different result was the modest amount of biofuel production at issue at the time—an argument Mr. Wang has articulated in a public version of his E-Letter to Science. Such an argument confuses the total amount of land use change from biofuels (which certainly are lower for lower total biofuel levels), with the rate of emissions from land use change for each liter of biofuel. Because the amount of world cropland for feed grains was still increasing even when Mr. Wang did his GREET calculations, every unit of reduced production in the U.S. would translate then as now into some increase in world land conversion to replace those feed grains. There is no inherent reason to believe that the land use change emissions per liter of corn ethanol were lower then than now. The reason for Mr. Wang's different result, which was never peer-reviewed, was at least primarily due to the methodological differences described above.

At the broadest level, the key insight of our paper is that when cropland-based feedstocks are used for biofuels, the source of any potential greenhouse gas benefit is the use of the land to take carbon out of the atmosphere. The ultimate benefit of such biofuels therefore depends on whether using this land to grow biofuels provides more carbon reduction benefits than using land in its existing use, as a minimum alternative, let alone using it in some potential use such as reforested land that would sequester even more carbon. Leaving out emissions from land use change for such biofuels assumes that land is a carbon-free asset. We would argue that any calculation that ignores these emissions, however challenging it is to predict them with certainty, is too incomplete to provide a basis for policy decisions.

Timothy Searchinger

Woodrow Wilson School, 403 Robertson Hall, Princeton University, Princeton, NJ 08544-1013, USA.

References

1. Food and Agricultural Policy Research Institute, U.S. Baseline Briefing Book: Projections for Agricultural and Biofuels Markets (FAPRI-MU Report #-3-08, University of Missouri, March 2008).

2. S. Grudings, "Amazon deforestation surging again—scientist," Reuters 16 January 2008.

3. World Bank, "Rising food prices: Policy options and World Bank response" (2008).

4. R. Trostle, Global Agricultural Supply and Demand: Factors Contributing to the Recent Increase in Food Commodity Prices (WRS-0801 Economic Research Service, USDA, 2008).

5. P. C. Westcott, Ethanol Expansion in the U.S.: How Will the Agricultural Sector Adjust (Economic Research Service, USDA, 2007).

6. D. K. Birur, T. W. Hertel, W. E. Tyner, paper prepared for presentation at the Food Economy Conference Sponsored by the Dutch Ministry of Agriculture, The Hague, Netherlands, 18 to 19 October 2007.

The article by T. Searchinger et al. (Reports, "Use of U.S. croplands for biofuels increases greenhouse gases through emissions from land-use change," 29 February 2008, p. 1238; Published online 7 February 2008) provides a timely discussion of ethanol's potential effects on greenhouse gas (GHG) emissions when taking into account direct and indirect land-use changes.

Searchinger et al. used the GREET model developed by one of our team at Argonne National Laboratory. They correctly stated that the GREET model includes GHG emissions from direct land-use changes associated with corn ethanol production. The emissions estimates in GREET are based on land-use changes modeled by the U.S. Department of Agriculture in 1999 for an annual production of 4 billion gallons of corn ethanol. There seems to be no indication that U.S. corn ethanol production so far has caused indirect land-use changes in other countries, since U.S. corn exports have remained at about 2 billion bushels a year. Moreover, exports of U.S. distillers grains and soluables (DGS, an animal feed) from corn ethanol plants have steadily increased in the past ten years.

U.S. corn ethanol production could expand rapidly to 15 billion gallons a year by 2015. Argonne, and several other organizations, have begun to address both direct and indirect land-use changes associated with future expanded U.S. biofuels production. To do so adequately requires the development and use of general equilibrium models to take into account the supply and demand of agricultural commodities, land use patterns, and land availability (all at the global scale), among many other factors. Present scientific assessment of the land-use changes that could occur globally as a result of U.S. corn ethanol production is inadequate. Therefore, conclusions on the GHG emissions effects of biofuels drawn from speculative, limited land-use-change modeling may lead to misguided biofuel policy actions.

Michael Wang

Center for Transportation Research, Argonne National Laboratory, Argonne, IL 60439, USA.

Zia Haq

Office of Biomass Program, Office of Energy Efficiency and Renewable Energy, U.S. Department of Energy, Washington, D.C. 20586, USA.

T. Searchinger et al. [as well as Fargione et al. (1)] make the case that whether biofuel production has a net greenhouse gas benefit depends on land use changes as large new areas are converted to agricultural uses and a carbon debt is incurred. Alternatively, dietary changes in the United States could free vast areas for biofuel production without converting more land to agricultural use. Beef requires 13 kg of grain and 30 kg of forage per kg of meat produced; broiler chickens, 2.3 kg of grain per kg of meat. Additionally, beef production requires 57 kcal of fossil fuel for each kcal of food gained, while broiler chickens have a 4:1 ratio (2). The FAO estimates cattle rearing and processing accounts for 18% of global greenhouse gas emissions (3). The U.S. grows slightly under 40 million hectares of corn and roughly 25 million hectares of soybeans (4). A little over half the corn we grow goes into animal feed (5). Over half the soybeans grown in the U.S. are crushed for soymeal and oil, and approximately 90% of the soymeal produced goes into livestock feed for meat (6). A lacto-ovo-vegetarian diet (dairy, eggs, plants) also requires animal feed, but about half as much as a meat-heavy American diet (2).

If the U.S. consumed less red meat, millions of hectares in corn or soy production could be used for ethanol or biodiesel, with no undesirable land use alterations and hence no additional carbon debt. Fossil fuel demand would also be reduced if we moved away from the 57:1 ratio mentioned above. Additionally, livestock produce methane, a potent greenhouse gas. Moderating our carnivory could therefore help reduce what are presently abnormally high atmospheric methane levels (7). We don't need to wait for technological advances in lignocellulosic ethanol to make biofuel production an important component of atmospheric carbon reduction. We just need to eat as though we live on a finite planet.

Steven A. Kolmes

Environmental Studies Program, University of Portland, Portland, Oregon 97203, USA.

References

1. J. Fargione, J. Hill, D. Tilman, S. Polasky, P. Hawthorne, Science 319, 1235 (2008).

2. D. Pimentel, M. Pimentel, Am. J. Clinical Nutrition 78, 660S (2003).

3. H. Steinfeld et al., Livestock's Long Shadow, Environmental Issues and Options, Food and Agriculture Organization of the United Nations, ISBN 978- 92-5-105571-1 (2006) (available at http://www.virtualcentre.org/en/library/key_pub/longshad/A0701E00.pdf).

4. National Agricultural Statistics Service, USDA, Report Released June 29, 2007 (available at http://www.usda.gov/nass/PUBS/TODAYRPT/acrg0607.txt).

5. E. Leibtag, Amberwaves (2008) (available at http://www.ers.usda.gov/AmberWaves/February08/Features/CornPrices.htm).

6. Soy Stats, 2007, available at http://www.soystats.com/2007/Default- frames.htm (Sponsored by the Illinois Soybean Association, Indiana Soybean Alliance, Minnesota Soybean Research & Promotion Council, North Dakota Soybean Council, Ohio Soybean Council, Iowa Soybean Association, South Dakota Soybean Research & Promotion Council, and the Kentucky Soybean Board.)

7. J. R. Petit et al., Nature 399, 429 (1999).