�� 2004 Nature Publishing Group 20. Weber, T. et al. Correlated electron emission in multiphoton double ionization. Nature 405, 658���661 (2000). 21. Schulz, M. et al. Three-dimensional imaging or atomic four-body processes. Nature 422, 48���50 (2003). 22. Walter, M. & Briggs, J. S. Selection rules and isotope effects in the full fragmentation of the hydrogen molecule. Phys. Rev. Lett. 85, 1630���1633 (2000). 23. Walter, M. & Briggs, J. S. Photo-double ionization of molecular hydrogen. J. Phys. B 32, 2487���2501 (1999). 24. Weber, T. et al. Auger electron emission from fixed-in-space CO. Phys. Rev. Lett. 90, 153003-1���153003- 4 (2003). 25. D����ez Muino, �� R., Rolles, D., de Abajo, F. J. G., Fadley, C. S. & Hove, M. A. V. Angular distribution of the electrons photoemitted from core levels of oriented diatomic molecules: multiple scattering theory in non-spherical potentials. J. Phys. B 35, L359���L365 (2002). 26. Feagin, J. M. A helium-like description of molecular hydrogen photo-double ionization. J. Phys. B 31, L729���L736 (1998). 27. Reddish, T. J. & Feagin, J. M. Photo double ionization of molecular deuterium. J. Phys. B 32, 2473���2486 (1999). 28. Joy, H. W. & Parr, R. G. A one-center wave function for the hydrogen molecule. J. Chem. Phys. 28, 448���453 (1958). 29. Kheifets, A. S. & Bray, I. Application of the CCC method to the calculation of helium-photoionization triply differential cross sections. J. Phys. B 31, L447���L453 (1998). 30. Hayes, E. F. Accurate single-center expansions using Slater type orbitals: hydrogen molecule. J. Chem. Phys. 46, 4004���4008 (1967). Acknowledgements We thank Roentdek GmbH (www.Roentdek.com) for support with detectors, and acknowledge helpful discussion with colleagues M. Walter, J. Briggs, J. Feagin, T. Reddish and V. Schmidt. This work was supported by the Deutsche Forschungs Gemeinschaft, the Bundesministerium fur �� Bildung und Forschung, and the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, US Department of Energy (DOE). T.W. thanks Graduiertenforderung �� des Landes Hessen, the Alexander von Humboldt Stiftung and the Herrmann Willkomm Stiftung for financial support. Competing interests statement The authors declare that they have no competing financial interests. Correspondence and requests for materials should be addressed to R.D. (doerner@hsb.uni-frankfurt.de). .............................................................. Ecosystem carbon storage in arctic tundra reduced by long-term nutrient fertilization Michelle C. Mack1*, Edward A. G. Schuur1*, M. Syndonia Bret-Harte2, Gaius R. Shaver3 & F. Stuart Chapin III2 1Department of Botany, University of Florida, Gainesville, Florida 32611, USA 2 Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska 99775, USA 3 The Ecosystems Center, Marine Biological Laboratory, Woods Hole, Massachusetts 02543, USA * These authors contributed equally to this work ............................................................................................................................................................................. Global warming is predicted to be most pronounced at high latitudes, and observational evidence over the past 25 years suggests that this warming is already under way1. One-third of the global soil carbon pool is stored in northern latitudes2, so there is considerable interest in understanding how the carbon balance of northern ecosystems will respond to climate warm- ing3,4. Observations of controls over plant productivity in tundra and boreal ecosystems5,6 have been used to build a conceptual model of response to warming, where warmer soils and increased decomposition of plant litter increase nutrient availability, which, in turn, stimulates plant production and increases eco- system carbon storage6,7. Here we present the results of a long- term fertilization experiment in Alaskan tundra, in which increased nutrient availability caused a net ecosystem loss of almost 2,000 grams of carbon per square meter over 20 years. We found that annual aboveground plant production doubled during the experiment. Losses of carbon and nitrogen from deep soil layers, however, were substantial and more than offset the increased carbon and nitrogen storage in plant biomass and litter. Our study suggests that projected release of soil nutrients associated with high-latitude warming may further amplify carbon release from soils, causing a net loss of ecosystem carbon and a positive feedback to climate warming. The effects of climate warming on ecosystem carbon (C) storage remain uncertain. Despite the low temperatures at high latitudes, C storage in tundra and boreal ecosystems is thought to be con- strained ultimately by carbon���nutrient interactions because plant production is usually nitrogen (N)-limited6,7. As soils warm in response to climate change, nutrient mineralization from soil organic matter is expected to increase8,9, which should, in turn, increase plant production. Total ecosystem C storage, however, depends on the balance between production and decomposition, and the relationship between nutrient availability and decompo- sition remains uncertain. In ecosystems at lower latitudes, natural and manipulated nutri- ent concentrations have had a positive, a negative, or no effect on the decomposition of litter and soil organic C (SOC)10���13. This variable response probably reflects ecosystem differences in form and quality of litter and SOC, but the regulatory mechanism for this is poorly understood13. High-latitude ecosystems are unusual because they store a larger proportion of total ecosystem C in soil compared with temperate and tropical ecosystems14. In arctic tundra, as much as 90% of the total ecosystem C resides in organic horizons and frozen mineral soils15. Thus, the response of SOC to changes in nutrient availability will play a critical role in determin- ing net ecosystem C balance in a changing climate. Previous results from nutrient manipulations suggested that increased nutrient availability should increase the total C storage in tundra ecosystems2,9,15,16. Nutrient addition greatly increases C stored aboveground by stimulating plant productivity and by shifting species composition from slow-growing species to more productive shrubs that accumulate C in long-lived woody bio- mass4,17���19. In addition, leaf, root and stem litter from shrubs decomposes more slowly than the graminoid-dominated litter they replace9, so conversion to shrub tundra was thought to slow decomposition and increase ecosystem C accumulation19. However, these inferences were based on aboveground and surface soil measurements only. The lack of soil-profile measurements reflects the expectation that the large heterogeneous belowground C pool Figure 1 Effect of fertilization on vascular plant aboveground net primary production (ANPP) in tundra. Fertilized plots in moist acidic tundra near Toolik Lake, Alaska, have received 10 g N m22 yr21 and 5 g P m22 yr21 since 1981. Values are means (^1 standard error, s.e.) means from 1982���95 are reported in ref. 19 the year-2000 data are from this study (n �� 4). Components of ANPP (new leaves and reproductive parts, new stems and secondary growth) are shown in Supplementary Fig. 1. letters to nature NATURE | VOL 431 | 23 SEPTEMBER 2004 | www.nature.com/nature 440

�� 2004 Nature Publishing Group would respond to changes in plant inputs too slowly to be detected in short-term experimental manipulations. To investigate the effects of nutrient availability on whole- ecosystem C balance, we examined C and N pools in a long-term fertilization experiment at the arctic Long-Term Ecological Research site near Toolik Lake, Alaska. To our knowledge, this is the longest-running nutrient-addition experiment in arctic tundra. Fertilized plots in moist acidic tundra (MAT) have received 10 g N and 5 g P m22 year21 since 1981 (ref. 19). This is approximately 5 to 8 times the annual soil N uptake requirement for aboveground production in MAT, and similar to the N uptake requirement in nearby shrub tundra characteristic of warmer sites5,20. Two decades of fertilization greatly increased aboveground net primary productivity (ANPP Fig. 1) over this time about 1,500 g m22 of additional C entered fertilized plots as ANPP, which had shifted from graminoid tundra dominated by the tussock-forming sedge, Eriophorum vaginatum, to shrub tundra dominated by Betula nana19. Because of the unique long-term nature of this experiment, changes in belowground C pools and total ecosystem C balance are now detectable. In our experiment, increased nutrient availability had a larger effect on decomposition than on plant production, resulting in a net loss of almost 2,000 g C m22 from this ecosystem over 20 yr (Fig. 2a P �� 0.04). Carbon storage increased aboveground (P , 0.001) because of the accumulation of woody shrub biomass and litter, but this was offset by a larger decrease of C in belowground pools (P �� 0.02) due to a pronounced decrease in the C contained in deep organic (.5 cm depth) and upper mineral soil layers (Fig. 2b). The decrease in the deep organic layer C pool was the result of a reduction in the thickness of the layer, because neither %C nor bulk density was affected by fertilization (Supplementary Infor- mation). In the upper mineral soil, fertilization reduced %C by 50% (P �� 0.04), whereas the depth to the frozen soil surface and mineral soil bulk density did not change (Supplementary Information). Decreased C storage in the fertilized treatment does not appear to be caused by decreased plant production. As expected, ANPP of vascular plants was higher in fertilized plots (Fig. 1). Root C pools were not different between treatments (Fig. 2b), and in a related study, root productivity tended to be higher in fertilized plots21. The productivity of mosses and lichens was not measured in this harvest, but their production has been estimated as 25���60 g C m22 in unmanipulated MAT5,20,22. Although they were mostly absent from fertilized plots in both 1995 and 2000, the loss of their productivity would be insufficient to offset the large increases in vascular productivity. Because plant production increased total C inputs, the net loss of C from the fertilized plots could only have been caused by accelerated decomposition of C. Several mechanisms could have contributed to the nutrient- induced acceleration of decomposition. First, nutrient additions could have altered the decomposability of fresh plant litter through changes in the species composition or tissue quality of the plant community. However, the increased C in biomass, litter and surface soils of fertilized plots (1,311 g m22) was similar to increased ANPP inputs over the past 20 yr (,1,500 g m22), indicating relatively little decomposition of the increased litter inputs. Furthermore, B. nana leaves, stems and roots are relatively less decomposable than those of the community they replace9, making it unlikely that litter decom- posed more quickly in the fertilized treatment owing to changes in tissue quality alone. These observations argue against changes in litter decomposability having a major role in increased decompo- sition in fertilized plots. Second, the loss of deep-rooted graminoid species from the fertilized plots19 could have altered environmental controls over root decomposition by changing the depth at which root litter was deposited. Although total root biomass C was not different between treatments, fertilization did shift the distribution of root biomass Figure 2 Effects of fertilization on tundra carbon and nitrogen pools after 20 yr of fertilization. a, c, Mean (^1 s.e.) above- and belowground carbon (a) and nitrogen (c) pools in unmanipulated control and fertilized treatments of moist acidic tundra near Toolik Lake, Alaska. Aboveground pools include shoots, standing dead plant material, and rhizomes. Belowground pools include surface litter, roots, and organic and mineral soil. b, d, Mean (^1 s.e.) carbon (b) and nitrogen (d) pools in plant and soil compartments. Pool treatment means were compared with nested ANOVA (n �� 4). Means that are significantly different are indicated with asterisks: *P , 0.05 **P , 0.01 ***P , 0.001. n.s.d., not significantly different. letters to nature NATURE | VOL 431 | 23 SEPTEMBER 2004 | www.nature.com/nature 441