The Heat Is Online

US Forests Do Not Offset Emissions


CORVALLIS, Oregon, March 17, 2000 (ENS) - New research has found that the massive amounts of atmospheric carbon dioxide generated by fossil fuel use in the U.S. are not completely offset by the storage of carbon in growing forests and other vegetation, as some earlier studies had suggested. The new study, published today in the journal "Science," may have important implications for the role of the U.S. in combating the greenhouse effect and global warming, said author Ronald Neilson, a professor of botany at Oregon State University and bioclimatologist with the U.S. Forest Service. "Some have argued that the U.S. does not need to reduce greenhouse gas emissions because we're not part of the problem," said Neilson.

"Based on this study, we can no longer make that claim." Neilson coauthored the report with U.S. researchers from the Ecosystems Center at Woods Hole, Massachusetts and the National Center for Atmospheric Research, and scientists from the Max-Planck-Institute for Biogeochemistry in Germany.

Researchers have been looking for the "missing sink" of carbon. More carbon, they say, is being injected into the atmosphere by industrialized nations than can be accounted for in the Earth's atmosphere, land, vegetation and oceans. "Some past studies suggested that a big part of the missing carbon sink was in the forests and changing land use practices of North America," Neilson said. In their Vegetation and Ecosystem Modeling and Analysis Project, or VEMAP, the scientists found that atmospheric fertilization and other phenomena would hold only an additional .08 gigatons of carbon within the lower 48 states, and perhaps double that for all of North America. Regrowth of forest vegetation would hold no more than an extra one or two times that amount. In simpler terms, the study suggests at least 70 to 90 percent of the carbon injected into the atmosphere by fossil fuel use in the U.S. is either staying there or being held somewhere besides North America.

Contribution of Increasing CO2 and Climate to Carbon Storage by Ecosystems in the United States

Science, Volume 287, Number 5460, 17 Mar 2000, pp. 2004 - 2006

David Schimel, 1* Jerry Melillo, 2 Hanqin Tian, 2* A. David McGuire, 3 David Kicklighter, 2 Timothy Kittel, 4 Nan Rosenbloom, 4 Steven Running, 5 Peter Thornton, 5 Dennis Ojima, 6 William Parton, 6 Robin Kelly, 6 Martin Sykes, 7 Ron Neilson, 8 Brian Rizzo 9

The effects of increasing carbon dioxide (CO2) and climate on net carbon storage in terrestrial ecosystems of the conterminous United States for the period 1895-1993 were modeled with new, detailed historical climate information. For the period 1980-1993, results from an ensemble of three models agree within 25%, simulating a land carbon sink from CO2 and climate effects of 0.08 gigaton of carbon per year. The best estimates of the total sink from inventory data are about three times larger, suggesting that processes such as regrowth on abandoned agricultural land or in forests harvested before 1980 have effects as large as or larger than the direct effects of CO2 and climate. The modeled sink varies by about 100% from year to year as a result of climate variability.

1 Max-Planck-Institute for Biogeochemistry, Postfach 10 01 64, D-07701 Jena, Germany.
2 The Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA 02543, USA.
3 U.S. Geological Survey, Alaska Cooperative Fish and Wildlife Research Unit, University of Alaska, Fairbanks, AK 99775-7020, USA.
4 National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO 80307-3000, USA.
5 University of Montana, Missoula, MT 59812, USA.
6 NREL, Colorado State University, Fort Collins, CO 80523-1499, USA.
7 Plant Ecology, Lund University, Ekologihuset 223 62 Lund, Sweden.
8 U.S. Department of Agriculture, Forest Service, Oregon State University, Forest Science Laboratory, 3200 Southwest Jefferson Way, Corvallis, OR 97333, USA.
9 Department of Environmental Sciences, University of Virginia, Charlottesville, VA 22903, USA.
* To whom correspondence should be addressed. E-mail: and

Recent analyses of the global carbon cycle suggest a significant role for terrestrial uptake of CO2 in the overall budget (1-4). Analyses of atmospheric CO2 have persistently suggested that this terrestrial uptake is largest in the Northern Hemisphere (2, 3), and one atmospheric analysis suggests that the United States may play a disproportionate role (2). Currently, a number of phenomena contribute to enhanced carbon uptake by ecosystems, including CO2 fertilization of photosynthesis, climate, nitrogen deposition, recovery from historical land use, and erosion/sedimentation (4-6). Although preliminary attempts have been made to partition the terrestrial sink among these processes globally, this quantification is currently extremely crude. It is essential to understand the mechanisms controlling carbon exchange today as a basis for prediction and management interventions (7).

Here we present results from the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP) aimed at understanding the contribution of ecosystem physiological mechanisms to terrestrial sinks in the conterminous United States during the period 1980-1993. Specifically, we consider how changes in climate and CO2 concentration affect ecosystem physiology. Three ecosystem models [Biome-BioGeochemical Cycles (Biome-BGC), Century, and the Terrestrial Ecosystem Model (TEM)] that dynamically calculate net carbon storage at a 0.5° × 0.5° resolution (8) were used. All three models simulated changes to soil and vegetation carbon in natural ecosystems.

Century simulated both natural and simulated agricultural ecosystems. To compute complete regional carbon budgets, we identified those grid cells dominated by agriculture (about 40% of the United States) and used the Century agricultural results to simulate net carbon storage in these grid cells for all models. We analyzed the period 1980-1993 to compare our results with spatial atmospheric inverse calculations (2, 3) and inventory-based estimates of forest carbon storage (4, 9, 10). The period spans a range of climatic conditions and includes three El Niño events and the global cooling that followed the eruption of Mount Pinatubo (June 1991).

We used climate information for 1895-1993. Climate information was derived from the National Oceanic and Atmospheric Administration (NOAA) Historical Climate Network (HCN) database (11). Monthly precipitation and mean minimum and maximum temperature information was derived from the NOAA HCN database and other primary, cooperative, and snowpack telemetry (SNOTEL) station data sets (11). Because few station records spanned the entire period, we estimated the spatial autocorrelation structure around each station for its period of record and created a continuous record for 1895-1993 for all stations by geostatistically computing missing anomalies from nearby stations (12). The station data were gridded at 0.5° (latitude by longitude) with a terrain-following algorithm (12). Forest types and soils data were as described in VEMAP 1995 (7) with forest distributions remapped with a database derived from satellite observations (13). Agricultural regions were determined from a 1990 land cover inventory of the United States (13) and land management practices (fertilization, planting and harvest dates, irrigation, tillage intensity) from U.S. Department of Agriculture National Resource Inventory data (14, 15). Atmospheric CO2 data were from Enting et al. (16).

Our ensemble of means for the U.S. CO2/climate sink in ecosystems for the period 1980-1993 is 0.08 Pg of carbon per year (Table 1). The three models agree within 25% in estimating the continental mean. Comparison of model experiments with observed versus constant CO2 shows that the bulk of the increase is due to CO2 fertilization, with the rate of uptake varying with and modulated by climate. Annual net carbon storage per unit area is relatively evenly distributed over the conterminous United States, ranging from 100 kg ha1 in the Great Plains and the Northeast to 150 kg ha1 in the Southeast (Fig. 1). As expected, intermodel variability is higher at the regional level than in the continental total, but the model results remain similar within a factor of 3 and are comparable to inventory-based estimates (9, 10). Agriculture plays a negligible role in modeled current carbon storage, but Century simulations suggest that, with best management practices, U.S. agriculture can remain a modest sink for decades to come.

The region we simulated is not geographically identical to the regions defined in atmospheric inverse models (2, 3). For perspective, in global simulations with VEMAP models, net carbon storage in the conterminous United States is typically about 60% of the total we model for the region equivalent to the North American domain of Fan et al. (2). Also, the atmospheric signal results from the outcome of all processes, including processes we do not model such as forest regrowth, erosion, and nitrogen deposition. Therefore, we expect, a priori, the atmosphere to show a somewhat larger sink than we model (Table 1). Our estimate is close in magnitude to inventory-based estimates (9, 10) and to some atmospheric estimates (3).

Despite high uncertainty, the inventory estimates tend to be larger than the VEMAP estimate for a CO2/climate sink of 0.08 Pg of carbon per year. For example, Brown and Schroeder (10) estimated 0.17 Pg of carbon per year for eastern U.S. forests (compared with our value of 0.04) (Fig. 1). Birdsey and Heath (9) estimated a U.S. sink of 0.3 Pg of carbon per year, whereas Houghton et al. (4) estimate a range of 0.15 to 0.35 Pg of carbon per year. The effects of intensive forest management and agricultural abandonment on carbon uptake in the United States are probably as large as or larger than the effects of climate and CO2. If the total sink is about 0.3 Pg of carbon per year, and the CO2/climate sink is about 0.1 Pg of carbon per year, other processes such as regrowth on abandoned agricultural and harvested forest lands must cause a sink of about 0.2 Pg of carbon per year.

This is a different perspective from that given in many global analyses (1, 19). A large role for land use effects is consistent with suggestions from the ecological community in the wake of the Kyoto Protocol (18). The relative roles of physiological (climate, CO2) changes compared with the direct effects of human domination of ecosystems need to be reassessed as a basis for understanding how the carbon cycle will change in the future.

Despite the discrepancies, the estimates from the VEMAP models are an order of magnitude less than the high atmospherically based estimates of Fan et al. (2). Inventory data also suggest a sink of the order of 0.3 Pg of carbon per year. Thus, the best current information suggests that CO2 and land use contribute a few tenths of a petagram of carbon uptake each year in the United States. The other hypothesized processes for ecosystem carbon storage (nitrogen deposition and sedimentation) are thought to be of a similar magnitude or smaller in this region (5). Inventory and model results are in conflict with high estimates from atmospheric inverse estimates. The next steps in the quantification of the North American carbon sink will require additional observations (20).


  1. D. S. Schimel, Glob. Change Biol. 1, 77 (1995) [ISI].
  2. S. M. Fan, et al., Science 282, 442 (1998) [ISI] [Abstract/Full Text].
  3. P. J., Rayner, I. G. Enting , R. J. Francey and R. Langenfelds, Tellus 51B, 213 (1999) .
  4. R. A. Houghton, J. L. Hackler, K. T. Lawrence, Science 285, 574 (1999) [ISI] [Abstract/Full Text].
  5. A. Dai and I. Fung, Glob. Biogeochem. Cyc. 7, 599 (1993) [ISI]; E. Holland, et al., J. Geophys. Res. 102(D13), 15849 (1997) [ISI]; R. F. Stallard, Glob. Biogeochem. Cyc. 12(2), 231 (1998) [ISI].
  6. H. Tian, J. M. Melillo, D. W. Kicklighter, A. D. McGuire, J. Helfrich, Tellus 51B, 414 (1999) [ISI].
  7. VEMAP Members, Glob. Biogeochem. Cyc. 9(4), 407 (1995) ; T. G. F. Kittel, et al., J. Biogeogr. 22, 857 (1995) [ISI].
  8. We used three biogeochemical models (Biome-BGC, Century, and TEM) for our analyses here. Detailed description of the three models can be found in Running and Hunt [S. W. Running and E. R. Hunt Jr., in Scaling Processes Between Leaf and Landscape Levels, J. R. Ehleringer and C. Field, Eds. (Academic Press, Orlando, FL, 1993), pp. 141-158] for Biome-BGC, in Parton et al. [W. J. Parton, D. S. Schimel, D. S. Ojima, C. V. Cole, in Quantitative Modeling of Soil Forming Processes, R. B. Bryant and R. W. Arnold, Eds. (Special Publication 39, Soil Science Society of America, Madison, WI, 1994), pp. 147-167] for Century; and in (6) for the TEM.
  9. R. A. Birdsey and L. S. Heath, in Productivity of America's Forest and Climate Change, General Technical Report RM-GTR-271, L. A. Joyce, Ed. (U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station, Fort Collins, CO, 1995), pp. 56-70.
  10. S. L. Brown and P. E. Schroeder, Ecol. Appl. 9(3), 968 (1999) [ISI].
  11. T. C. Peterson and R. S. Vose, Bull. Am. Meteorol. Soc. 78(12), 2837 (1997) [ISI]; T. R. Karl and R. W. Knight, Bull. Am. Meteorol. Soc. 79, 231 (1998) [ISI].
  12. T. G. F. Kittel et al., Proc. 10th Conference on Applied Climatology, Boston (American Meteorological Society, New York, 1997), pp. 222-229; C. Daly, R. Neilson, D. Philips, J. Appl. Meteorol. 33, 140 (1994) [ISI].
  13. T. R. Loveland, J. W. Merchant, D. O. Ohlen, J. F. Brown, Photogramm. Eng. Remote Sens. 57(11), 1453 (1991) [ISI].
  14. The VEMAP data layer for current vegetation cover was developed from the 1990 1-km EROS Data Center (EDC) Conterminous U.S. Seasonal Land Cover Classification (13) ( The 154 classes from the 1-km EDC classification were assigned to 33 VEMAP current vegetation types (22 unmanaged classes, 11 managed classes) by VEMAP participants. The reclassed 1-km data were then aggregated to the standard VEMAP 0.5° grid cells by majority.
  15. Managed vegetation classes were subdivided by regional management practices and growing season length. This was done based on data from the 1995 Cropping Practices Survey ( conducted by the Resource Information System Section of the Resources and Technology Division of the Economic Research Service for all crops except sorghum and hay (USDA, 1991 (r 1995i* Cropping Practices Survey, unofficial USDA data files). Sorghum information is from the 1991 Cropping Practices Survey (, and hay practices are based on yield data from state-level reports (1995 USDA Annual Crop Summary by state, National Agricultural Statistics Service) ( Where management practices (irrigation, fertilization, planting or harvest date, tillage practices, or rotation) differed considerably within a land cover class, the land cover class was spatially subdivided to capture these significant differences.
  16. I. Enting, T. Wigley, M. Heimann, CSIRO Div. Atmos. Res. Technical Pap. 31, 1 (1994) .
  17. M. Goulden, et al., Science 271, 1576 (1996) [ISI] [Abstract]; B. H. Braswell, D. S. Schimel, E. Linder, B. Moore III, Science 278, 870 (1997) [ISI] [Abstract/Full Text]; P. Ciais, P. P. Tans, M. Trolier, J. W. C. White, R. J. Francey, Science 269, 1098 (1995) [ISI] .
  18. IGBP Terrestrial Carbon Working Group, Science 280, 1393 (1998) [Full Text].
  19. M. Cao and I. F. Woodward, Nature 393, 249 (1998) [ISI] .
  20. J. L. Samiento and S. Wofsy, Eds. A U.S. Carbon Cycle Science Plan (University Corporation for Atmospheric Research, Boulder, CO, 1999).
  21. VEMAP is sponsored by the Electric Power Research Institute, NASA, and the U.S. Forest Service. Additional support for VEMAP data set development comes from NOAA, NSF, and the National Center for Atmospheric Research (NCAR). NCAR is sponsored by NSF. We thank NOAA's National Climate Data Center for their assistance with climate station data and USDA Natural Resources Conservation Service for access to snowpack data. We thank C. Daly and W. Gibson (Oregon Climate Service, Oregon State University), A. Royle (currently with the U.S. Fish and Wildlife Service), H. Fisher (NCAR), and NSF's Geophysical Statistics Project at NCAR for their crucial role in data set preparation.
10 November 1999; accepted 17 January 2000 Volume 287, Number 5460 Issue of 17 Mar 2000, pp. 2004 - 2006 ©2000 by The American Association for the Advancement of Science.