Climatic Change 33: 63-68 © 1996 Kluwer Academic Publishers
An Editorial Comment
1. The Missing Carbon Sink
Accounting for the "missing" carbon sink has become a required endeavor for global carbon cycle modelers, given policy questions about storing future fossil fuel combustion emissions of carbon in the terrestrial biosphere or the oceans, and concern that if we cannot account for past and current sources and sinks, we cannot make accurate predictions about the future. The term missing sink was coined by ocean modelers [1,2] of the carbon cycle who were unable to account for all the carbon released to the atmosphere from fossil fuel burning and land-use changes in oceanic sinks, but were uncomfortable attributing the unaccountable carbon to a terrestrial sink, without having a detailed model to explain terrestrial biosphere carbon storage mechanisms. Historically, ocean modelers have taken the lead in analyzing the global carbon cycle, both through the development of mechanistic global circulation ocean models and the use of isotropic tracers to constrain and verify their models.
Given the magnitude of the gross annual fluxes of carbon between atmosphere and oceans, atmosphere and terrestrial biosphere, currently around 90 to 100 Gt C/yr respectively , and the uncertainties associated with the estimation of these fluxes, it would appear to an outside observer of the missing carbon sink debate that accounting for a sink almost two orders of magnitude smaller than the overall fluxes would be a hopeless exercise. Even when we consider only the anthropogenic perturbation to the carbon cycle, with sources from fossil fuel combustion of 5.5 ± 0.5 Gt C/yr and tropical land-use change of 1.6 ± 1.0, buildup in the atmosphere of 3.2 ± 0.2 Gt C/yr, sinks in the oceans of 2.0 ± 0.8 Gt C/yr and Northern Hemisphere forest regrowth of 0.5 ± 0.5 Gt C/yr, the remaining "missing" carbon sink of 1.4 Gt C/yr is smaller than its estimated uncertainty (±1.5 Gt C/yr) .
To complicate matters, the cornerstone in the oceanic carbon budgeting effort has been the use of radiocarbon tracers, produced mostly during nuclear weapons testing. However, there are large gaps in the radiocarbon data . In addition, recent work [6,7] indicates that there is controversy with respect to the size of the radiocarbon pool in the ocean, the air-sea gas exchange rate and the bomb-radiocarbon penetration depth. Since the ultimate storage of most of the excess carbon released to the atmosphere is the deep ocean, this additional uncertainty in the size of the annual ocean sink implies a greater uncertainty in the size of the "missing" sink than indicated above. Although ocean models of CO2 uptake will still need to be constrained by radiocarbon tracer distribution for validation, it appears that other tracers must be used to reduce the uncertainty in the ocean sink and thus the size of the missing sink.
Terrestrial biosphere models as well as detailed inventories of terrestrial carbon pools and fluxes can provide an independent estimate of the terrestrial sink. The issue is to what level of process description and geographical resolution must we bring the models and databases to be certain that we have captured the essential features of the terrestrial component of the carbon cycle. At the simplest level, lumped box models can be used to study exchange rates, without much regard for vegetation type, nutrient limitation or climate effects. Although we can use these models for order of magnitude estimates of carbon fluxes, they may be insufficient for determining the size of the terrestrial carbon sink.
Another approach is to discretize the terrestrial biosphere into thousands of georeferenced pixels, with detailed physiological processes in each pixel to transport and store carbon within the pixel, and between the pixel and the atmosphere and lithosphere. The significant gain in resolution is tempered by the fact that we must still use pixel averaged values for key parameters such as altitude (and thus climate and vegetation type), soils and land-use. These models must consider not only actual vegetation, but the historical changes in land-use and vegetation type at each pixel, for which the database is sparse. In addition, the effect of simultaneous carbon, nitrogen and climate fertilization must be captured in the physiological process description, but there are still no long-term field studies of these effects from which to gather data for the models. A different conceptual framework  is to build-up from the individual plant level description to the landscape level and on to a biome and terrestrial biosphere response.
Ultimately, the size of the terrestrial sink will have to be determined using the discretized models. Generating the
needed databases and experimental data from the field will still require several years at best. Policy issues such as
land-use changes, carbon taxation, incentives for alternative energies, etc. have to be addressed in the near future,
with or without the availability of the additional information. Thus, a balance has to be achieved between the
results from the simpler models and the discretized models.
2. The CO2 Fertilization Effect
Kheshgi et al. [this issue, 9] provide an interesting approach for studying the potential magnitude of the CO2 fertilization effect using a simple box model. As the authors point out, the CO2 fertilization effect cannot account for all of the missing sink, but it is potentially a significant mechanism. Their six-box globally-aggregated terrestrial biosphere sub-model has been enhanced to consider both a temperature dependence of the exchange rate coefficients and a logarithmic dependence of net productivity on atmospheric CO2 concentration. Their results indicate that the terrestrial biosphere shifts from a net source of carbon prior to about 1980 to a net carbon sink from 1980 onward, with increasing carbon accumulation in the terrestrial biosphere despite a continuation in land-use changes.
As a first cut to evaluate the magnitude of this effect, this model provides interesting insights. Their work presents an interesting methodology for analyzing the response of the global carbon cycle, by evaluating the effect of a small perturbation to the carbon cycle against the background of large future anthropogenic perturbations. The resulting variability in the airborne fraction, which has been observed by many other modelers, points out the error made in many global carbon policy models [10-12], which use either a constant airborne fraction of anthropogenic emissions, or a set of impulse response coefficients derived at significantly different initial conditions or pulse sizes than those in which they are applied.
However, several more questions are raised by the model formulation, which must eventually be answered using a more detailed model. It is clear that most land-use changes in past centuries occurred in temperate forests and grasslands, whereas more recent and future land-use changes will occur in tropical climates. The carbon and nutrients released per unit area due to these land-use changes is significantly different. Can we expect the carbon fertilization effect to remain constant over time, while the global composition of biome areas being fertilized is changing with time?
The difference in the response of C3 and C4 plants to increasing CO2 concentrations is also well documented, and different biomes have significantly different compositions of C3 and C4 plants. Temperate and boreal forests are more sensitive to carbon fertilization, whereas grasslands are less sensitive . Even within a biome, between plant species or even genotypes there is a marked differential response to carbon fertilization . A managed temperate forest planted with a highly sensitive species may store larger amounts of carbon than an otherwise equivalent forest planted with less sensitive species, or a comparable tract of old forest. Therefore, the carbon fertilization effect is quite heterogeneous over space as well.
In addition to the diversity in response among plant species to the carbon fertilization effect, there is also an
experimental basis to believe that different species become saturated at different CO2 concentrations, and thus
species that may initially respond well to excess carbon may in the future have an insignificant response to higher
and higher CO2 concentrations. Are we close to the maximum carbon that can be stored in natural and managed
biomes, above and below ground?
3. Climate change effects
Climate change has involves potential global warming and changes in precipitation. Plant growth in natural terrestrial biomes may be positively affected to some extent by increasing temperatures, for example due to an increase in net primary productivity and increased nitrification in soils. While we can consider a globally averaged enhancement in the carbon exchange rates for some physiological processes, at a local and regional level the effect of higher temperatures may result in shifting canopy composition, favoring heat tolerant species over less tolerant individuals. Some biomes may be close to their optimal temperature for plant growth, e.g. tropical rainforests, so a small increase in temperature may actually result in carbon loss to the atmosphere due to increased decomposition . Along with increasing temperatures, precipitation patterns may be significantly altered in coming decades. Regional changes in the hydrological cycle can also result in a selection process, where either drought-resistant or water tolerant species may be favored, depending on conditions. To add to the complexity, water use efficiency of many plants may improve due to stomatal closure, as a response to higher CO2 concentrations. Can we currently speculate on whether climate change will have an overall positive or negative effect on carbon storage in terrestrial biomes?
Shifting climate may eventually result in migration of plant species to more favorable conditions, redistributing the
current biome boundaries. Temperate and boreal forests may shift to higher latitudes. Deserts may expand as well,
reducing the vegetation cover and thus the carbon storage above and below ground. Although these changes may be
slow, they will eventually alter the carbon fertilization effect, due to a resizing of biome areas. Anthropogenic
adaptation to these climate changes may involve additional shifts in land-use patterns, abandoning certain
agricultural areas in favor of more productive regions and altering deforestation and reforestation practices around
the world. The indirect effects of climate change are even less tractable.
4. Other anthropogenic changes
Concurrent with the emission of carbon to the atmosphere, we are pumping out to the environment large amounts of other nutrients. Nitrogen cofertilization may enhance the storage of carbon in temperate and boreal forests which are nitrogen limited . This has been recognized in many managed forests, specially in Europe, resulting in active nitrogen fertilization. On the other hand, excessive nitrogen deposition may be detrimental to forest health, resulting in carbon loss. To capture these processes in a box model would require the linkage of the carbon and nitrogen cycles. However, the nitrogen perturbation is mostly limited to mid-latitudes in the Northern Hemisphere, so a discretized model will be more adequate for answering these issues.
Land degradation and desertification occur as a result of excessive overutilization, environmental changes, and careless management of agricultural areas or lands used for pasture or forestry. Degradation may involve from loss of vegetation cover up to severe soil erosion. After the initial loss of carbon and nutrients due to land-use change, degradation results in additional depletion. Rapid loss of below ground carbon is especially significant because it is considered a longer-term storage. The magnitude of these degradative processes is relevant since they occur in at least 70% of drylands , which occupy almost 40% of the land surface, and are also a factor in humid areas, such as the rain forests and tundra. Is there a carbon fertilization effect on these degraded lands? If antidesertification or land management measures are taken, will the carbon stores return to their original values, will they be higher or lower?
These environmental stresses, combined with other factors such as increased fire frequency, the introduction of
exogenous plant and animal species, pests and diseases to natural and managed biomes, contribute to further
magnify the uncertainty in our prediction of the size of the terrestrial sink, and the effect of carbon fertilization and
5. Concluding remarks
Can we address the issue of the missing sink, or the carbon fertilization effect with a simple box model? The answer depends on the degree of accuracy we expect. For a rough order of magnitude estimate, the model presented by Kheshgi et al. may be sufficient. The results are only as good as the assumptions made. Applying a constant carbon fertilization factor, ß, over several centuries is a necessary simplification, given the lack of hard data and a model framework to which the data could be applied. Changes in biome composition and species response to increased CO2 in the atmosphere imply that ß will be spatially and temporally variable, and it is not clear how this can be reduced to a globally averaged value.
How should the value of the carbon fertilization factor, ß, be interpreted? Having derived it from the 1980-89 carbon balance, the value of ß = 0.42 is strictly valid for that period, and has in fact a significant uncertainty associated with it, since it has been forced to match 16 Gt C from land use emissions for the 80's decade. The best estimate of tropical land-use change emissions for that decade is 16 + 10 Gt C , reflecting an uncertainty of 60% around the mean value (no information is available on the confidence interval, as is the case with many estimates of uncertainty). Therefore, forward or backward extrapolation of the carbon fertilization effect using ß can at best be used to estimate the magnitude of the terrestrial sink, and it's shift from a source to a sink, as the authors have pointed out.
Closing the "missing" sink debate will require high resolution models that can address not only the effects of
climate, increased atmospheric carbon, nitrogen availability and soil quality, but also questions about land-use,
land management, and other anthropogenic perturbations to terrestrial biomes. The role of the simpler models is to
point us in the correct direction by defining the magnitude of the response and to provide interim answers to policy
 Broecker, W. S. et al., 1979: Fate of fossil fuel carbon dioxide and the global carbon budget, Science, 206:409-418
 Siegenthaler, U. and H. Oeschger, 1978: Predicting future atmospheric carbon dioxide levels, Science, 199:388-395
 Siegenthaler, U. and J. L. Sarmiento, 1993: Atmospheric carbon dioxide and the ocean, Nature, 365:119-125
 IPCC (1994) Climate Change 1994, Radiative Forcing of Climate Change and an Evaluation of the IPCC IS92 Emission Scenarios eds. J. T. Houghton, L. G. Meria Filho, J. Bruce, Hoesung Lee, B. A. Callander, E. Haites, N. Harris and K. Maskell for the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, Great Britain
 Joos F., 1994: Imbalance in the budget, Nature, 370:181-182
 Hesshaimer V., M. Heimann and I. Levin, 1994: Radiocarbon evidence for a smaller oceanic carbon dioxide sink than previously believed, Nature, 370:201-203
 Broecker, W. S. and T. Peng, 1994: Stratospheric contribution to the global bomb radiocarbon inventory: Model versus observation, Global Biogeochem. Cycles, 8:377-384
 Dale, V. H. and H. M. Rauscher, 1994: Assessing impacts of climate change on forests, Climatic Change 28:65-90
 Kheshgi, H. S., A. K. Jain and D. J. Wuebbles, 1996: Accounting for the missing carbon sink with the CO2 fertilization effect, Climatic Change, in print
 Hulme, M., S. C. B. Raper and T. M. L. Wigley, 1995: An integrated framework to address climate change (ESCAPE) and further developments of the global and regional climate modules (MAGICC), Energy Policy, 23:347-355
 Manne, A., R. Mendelsohn and R. Richels, 1995: MERGE: A model for evaluating regional and global effects of GHG reduction policies, Energy Policy, 23:17-34
 Peck, S. C. and T. J. Teisberg, 1994: Optimal carbon emissions trajectories when damages depend on the rate or level of global warming, Climatic Change, 28:289-314
 Keller, A. A. and R. A. Goldstein, 1994: The human effect on the global carbon cycle: response functions to analyze management strategies, World Resources Review, 6:63-87
 Hudson, R. J. M., S. A. Gherini and R. A. Goldstein, 1994: Modeling the global carbon cycle: nitrogen fertilization of the terrestrial biosphere and the "missing" CO2 sink, Global Biogeochem. Cycles, 8:307-333
 Dregne, H., M. Kassas and B. Rosanov, 1991: A new assessment of the world status of desertification,
Desertification Control Bulletin, UNEP, 20:6-18