[Version 3.0. Final draft. Last updated by Paul Falkowski 14 December 1999]
Integrated understanding of the global
carbon cycle: A test of our knowledge
IGBP Carbon Working Group[1]
We are just beginning to understand how Earth's complex systems operate to sustain life on this planet. Despite important gaps in knowledge, the scientific community is expected to provide reliable models to predict safe limits for human intervention, while simultaneously human activities are altering global systems in unprecedented ways. A traditional, disciplinary approach to science cannot provide the Earth system models that are requisite for formulating rational resource and energy policies. Although climatology, oceanography, geochemistry, and terrestrial and marine ecology all provide essential information about how the Earth functions, the ensemble of information is insufficient unless these discrete facets of the global system can be mechanistically linked. A truly interdisciplinary research effort is needed to develop even a minimal integrated model. As a step towards that goal, the authors of this article met in Stockholm in October 1999 to explore how the known links between the carbon and nutrient cycles serve to integrate the terrestrial, freshwater, atmospheric, and oceanic components of the global system. We present here a conceptual framework to guide the synthesis of current knowledge and to inform the debate on international management of the carbon cycle.
Biogeochemical feedback and control mechanisms in the biosphere
Biological processes, most of which originated over 2 billion years ago, together with the geological and climatological processes with which they coevolved, determine the distribution of the elements of the biosphere (1). The biological assimilation, biochemical transformation, physical transport, and geological mobilization and sequestration of many elements are self-perpetuating and self-regenerating, leading to a ‘cycling’ of the elements. Like life itself, biogeochemical cycles are far from thermodynamic equilibrium. Rate-limiting reactions within key cycles have modified the tempo and mode of evolution, promoting biological diversity and ensuring the continuity of biogeochemical cycles throughout Earth's history. The resulting network of interlinked cycles is responsible for regulating the chemistry and climate of the ocean, atmosphere, and terrestrial ecosystems in the contemporary world.
Natural selection, acting on metabolic processes[NG1], imposes elemental requirements for life that is manifested at all scales from the organism to the whole planet. The carbon cycle is central to all life on Earth, since about half of all biomass consists of carbon, but the carbon cycle did not evolve in vacuo. It is coupled to the dynamics of the approximately 20 other elements necessary for life (Fig. 1). This coupling occurs at the biochemical and cellular level in organisms, and constrains the rates of processes and the sizes of storage pools of carbon. Consequently, carbon and nutrients cycle through ecosystems within a relatively narrow range of proportions. While there is variation in this ‘ecosystem stoichiometry’, the underlying causes of the variations are largely understood, and therefore predictable (2).
Figure 1. The cycles of carbon, nitrogen, phosphorus and several other biologically essential elements are closely coupled in terrestrial, freshwater, and marine ecosystems. These three subsystems of the biosphere are in turn linked to each other via transport through the hydrological cycle and the atmosphere. Ocean-atmosphere gas exchange, which is driven by marine biology in the long term, largely establishes the atmospheric concentration of CO2 and thus the global climate. Terrestrial plant physiology is very sensitive to climate and theconcentration of CO2 in the atmosphere.The state of the vegetation controls the rate of transfer from the land to the sea of elements essential to marine organisms, thus closing the control loop.
Like physiological processes of which they are a part, biogeochemical cycles form interdependent, self-regulating systems, replete with feedback controls and processes[NG2]. The stoichiometry of bioessential elements becomes unbalanced when supply processes do not match demands, leading to growth limitation[NG3]. The metabolic machinery for some biological processes, such as photosynthesis and aerobic respiration, is widely distributed across diverse aquatic and terrestrial organisms. In contrast, other key biogeochemical processes, such as biological nitrogen fixation (BNF) and nitrification, are found in a relatively small number of taxa. For example, most oceanic BNF is performed by a small subset of cyanobacteria and terrestrial BNF is carried out by a subset of heterotrophic bacteria. The paucity of nitrogen fixers in marine and terrestrial systems leads to a biogeochemical bottleneck that co-regulates photosynthetic carbon fixation in both ecosystems (3). This bottleneck, in turn, affects the supply of organic carbon for heterotrophic respiration and BNF itself. This is a simple illustration of a negative feedback that helps drive global biogeochemical cycles towards a quasi-steady state.
On geologically short time scales, human activities have altered many biogeochemical processes to varying degrees, both spatially and temporally, resulting in unbalanced growth conditions relative to biological demand. As a result, biological assimilation of anthropogenic sources of CO2, N, and P are not necessarily coupled, potentially leading to deviations from steady state. How well can we predict the sign of the feedbacks or the outcome of these human activities?
While earth system scientists conceptually understand many of the key processes, we often fail to reproduce critical phenomena in simple models. As human experience has witnessed an infinitesimally small fraction of Earth's history, most of our knowledge is based on proxy data that are reconstructed to form (whenever possible) testable, consistent hypotheses. In some cases, essential information is obscured or absent in our reconstruction of the geological[NG4] record. For example, in spite of its importance, at present we do not fully understand the processes that led to the large increase in atmospheric O2 approximately 2.2 billion years ago (4). The sequestration of organic carbon in the lithosphere is a corollary of the net accumulation of atmospheric O2 by oxygenic photosynthesis. Sequestration requires an imbalance between photosynthesis and respiration; there is no clear consensus on the factors that led to such conditions on a planetary scale. The rapid warming of Earth's atmosphere ca. 55 million years ago appears to correspond with a rapid rise in atmospheric CO2 and CH4. The causes of the rise in these gases are unclear, but the implication is that carbon ‘sequestered’ in one or more reservoirs (e.g. the ocean or lithosphere) was mobilized and partially transferred to the atmosphere (5).
Control of the earth system over the past 400,000 years
There is much more geological and geochemical information about Earth's variability during the past million years, yet our (disciplinary) approach to elucidating processes has been unsatisfactory. A clear example is our lack of understanding of the causes of glacial-interglacial variations in climate and atmospheric CO2. Analyses of air trapped in Antarctic ice reveal that during the past 420,000 years the atmospheric CO2 has varied in concert with climate cycles, alternating between 180-200 ppmv at the peak of glacial periods and 265-280 ppmv during the warm interglacials (6). Although the timing of these climate cycles show remarkable correspondence to cyclic variations in Earth's orbital parameters with typical frequencies of 21, 41 and 100 kyr,
(the so-called ‘Milankovich cycles’; (7)[NG5]; Fig. 2), the direct radiative effects of the changes in energy received from the sun are insufficient to drive glacial-interglacial climate cycles (8). The strong coupling between atmospheric CO2 and temperature suggests that atmospheric CO2 may be the primary amplifier of climate change during glacial terminations, a hypothesis originally proposed by Arrhenius (9).
[insert Figure 2: the 400 kY CO2 record and its spectral analysis – ]
The glacial-interglacial transitions are an example of a highly reproducible ‘flip-flop’ between two quasi-stable states (10). Any mechanism(s) invoked to account for glacial-interglacial transitions must explain the following gross features of the atmospheric CO2 signal: the periodicity of the cycles; the transfer of carbon between terrestrial and oceanic pools (as revealed by changes in 13C content); the remarkable consistency of the upper and lower limits; the apparent fine control for a period of many thousands of years around those limits; the relatively rapid transition from glacial to interglacial; and the initially steep, but punctuated and eventually gradual transition into the glacial.
The upper and lower limits of atmospheric CO2 attained in the glacial and interglacial periods respectively, contain information about the control systems and their feedbacks. We propose that the lower limit is likely to reflect terrestrial ecosystem control and the upper limit primarily oceanic control. Nutrient-mediated interactions between land and ocean transfer the control from one to the other on a periodic basis. We speculate that widespread terrestrial ecosystem failure below 180 ppmv CO2 establishes an active negative feedback control loop. This concentration of CO2 is an ‘ecosystem compensation point’ – the concentration where the photosynthetic assimilation of CO2 isbalanced by autotrophic respiration (note: this compensation point is primarily dictated by C3 plant ecosystems). In the dry climate of the glacial period, water use efficiency, which is strongly influenced by CO2 concentration, would be a particularly important mechanism. We suggest that the upper (280 ppmv) CO2 concentration limit is an upper equilibration point between the atmosphere and the oceans, perhaps resulting from the bottleneck imposed on biospheric carbon fixation by nitrogen fixation, compounded by a loss of dissolved inorganic nitrogen due to enhanced denitrification (11, 12).
What causes the terrestrial system to switch between these two quasi-stable domains? The onset of glacial periods is systematically keyed to decreased solar radiation due to Milankovich forcing (13). The direct radiative forcing is amplified by a reorganization of oceanic circulation, increased sea ice cover in the Southern Ocean, and a decrease in the hydrological cycle. Positive feedbacks, through the a reduction in greenhouse gases such as H2O, CO2 , and CH4, and the increased planetary albedo accompanying the formation of continental and sea ice, would further cool the Earth (6). Superimposed on these processes is an increase in aeolian iron, which can stimulate oceanic primary productivity in the Southern Ocean (14)and as well as oceanic nitrogen fixation at low latitudes (3). Declining biotic activity on land would initially facilitate the mobilization and transport of nutrients, such as P and Si, from land to the sea (note: the C:P ratio of organic matter on land is ca. 145:1; the supply of P via this mechanism is not insignificant). Eventually, with increased aridity, this nutrient flux would reach an equilibrium such that the net uptake of carbon by the oceans balances the oceanic emissions. That lower limit is primarily controlled by the effect of atmospheric CO2 on terrestrial carbon balance. The glacial terminations are also tied to externally imposed radiative forcing, but are of opposite sign. Enhanced warming leads to an initial outgassing of oceanic CO2 and a stimulated hydrological cycle that corresponds with a marked decline in aeolian mineral fluxes (6, 14, 15). Retreating ice sheets and the corresponding flooding of continental margins accelerates denitrification, while the simultaneous decline in aeolian iron would lead to a reduction in oceanic productivity and nitrogen fixation. Terrestrial biological activity, stimulated by increased the moisture and warmth, accelerates the mobilization of soluble nutrients such as P, and Si from the geosphere. These elements initially build up in soils and are utilized by terrestrial biomass. The enhanced terrestrial nutrient cycle would lead to increased river loading. The flux of nutrients from the land stimulates oceanic net primary production and burial, which begins to stabilize and, subsequently, upon an external forcing set by a Milankovich cycle, draws down atmospheric CO2.
Our simple analysis suggests that the glacial-interglacial transitions reflect a nutrient-driven control and feedback of the carbon cycle involving terrestrial, oceanic and atmospheric realms. The ‘normal’ state of the biosphere during the past 420,000 years is glacial, with low atmospheric CO2 and resultant terrestrial vegetation stress. There is no historical evidence for a stable state or domain of attraction substantially above 280 ppmv. The point of our speculation is not to provide the ‘answer’ for the causes of the glacial-interglacial transitions, but rather to illustrate how an interdisciplinary, systems-based approach can generate a rich set of mechanisms to address a problem which has eluded disciplinary solutions.
Disturbance of global biogeochemistry during the past 200 years
There is even more abundant and less ambiguous biogeochemical data for the period of observational record, but prognostic models, based on a disciplinary approach, are not necessarily more accurate. For example, we know that global biogeochemical cycles of all the bioessential elements have been significantly altered due to human activities within the past 200 years (Table 1). CO2 released by the burning of fossil fuels and from changes in land use have increased atmospheric CO2 concentrations from the interglacial ‘setpoint’ of ca. 280 ppmv to more than 370 ppmv at present. Not only is this concentration is unprecedented over the last 420,000 years of Earth's history, but the rate of increase in atmospheric CO2 is about two orders of magnitude lower than that resulting from human activities since the Industrial Revolution (IPCC, 1995).
The production of synthetic fertilizers, the cultivation of nitrogen-fixing crops, and deposition of fossil fuel-associated nitrogen are together of the same order of magnitude as natural BNF, thus doubling the nitrogen inputs to the biosphere(16). Mining of phosphorus for fertilizer use has increased P inputs to the biosphere by approximately four-fold. The cycling of these elements is normally kept in proportion through the production and subsequent oxidation of organic matter in the biosphere. There are large uncertainties in how alterations in N and P cycles of this magnitude and rapidity interact with human perturbations to the global C cycle. Most global models deal with one element, thus ignoring the potential linkages and feedbacks among the others. At first glance, one might conclude that the simultaneous increases in N fixation and P production would stimulate the biological sequestration of carbon in terrestrial and marine ecosystems. Will such stimulation provide salvation from the continued emissions of CO2 to the atmosphere as a consequence of human activities?
Several models have computed the partitioning of anthropogenic CO2 from 1850 to 1990 (17)(Bruno and Joos 1997; Ver et al. 1999). The results of one such model suggest the existence, for most of this century, of a modest oceanic CO2 sink and a larger terrestrial carbon sink on non-agricultural land(Fig. 3). Without these sinks, the rate of increase in atmospheric CO2 would be twice the observed rate. While it is likely that both sinks will continue to exist and grow over the next few decades, there is increasing evidence for a leveling-off, and even a decline in the strength of these sinks within the next 50 to100 years.
Biotic sinks for CO2 require nutrients for the formation of organic matter. Enhanced net primary production in forests associated with enhanced inputs of N from atmospheric deposition has been suggested as an important sink for anthropogenic CO2 (Schimel 1995). The magnitude of this sink is highly uncertain and depends on a number of factors including the C:N ratio of the stored organic matter and the degree of N-saturation of soils (Nadelhoffer et al. 1999). While N deposition is predicted to continue, and has the potential to enhance the carbon sink in N-limited ecosystems, it will become decreasingly effective at doing so. Future N deposition will largely occur on already N-saturated soils, such as in Western Europe forests, China and India, and on agricultural lands in the tropics whose capacity to sequester C is intrinsically small and where soils are mostly P-limited (18).
Enhanced terrestrial productivity in response to increased levels of atmospheric CO2 has also been proposed as a significant terrestrial sink (19). The relationship between terrestrial Net Primary Production (NPP) and atmospheric CO2 reaches a plateau at high atmospheric CO2. Some experimental evidence suggests that in natural ecosystems it may level off at 10-20% above current terrestrial NPP, at an atmospheric CO2 level as low as double pre-industrial concentrations (i.e. 550-650 ppm), because of negative feedbacks such as nutrient limitation (20). In addition, the associated increased temperature will bring higher respiration, particularly by soil microbes, which may completely counteract or even overtake the CO2 uptake. This may in some ecosystems change current C sinks into C sources. The magnitude and rate of this change is, however, uncertain because plants can acclimate to changes in temperature (Dewar et al. 1999) and functional and compositional shifts may take place in microbial communities (Zogg et al. 1997).
In addition to these physiological considerations, land use change plays a major role in the strength of the carbon source/sink dynamics. Forest conversion to pasture and agriculture, in order to sustain increased food and fiber demands largely in the developing world, simultaneously provides additional carbon sources and reduces the area available for active sinks. Abandonment of agricultural land and regrowth of forests, largely in the temperate northern hemisphere, is likely a significant terrestrial CO2 sink at present, but cannot be sustained indefinitely.
Oceanic net primary production is not directly affected by atmospheric CO2, so nosubstantial increase in the strength of the biological pump is anticipated as a direct consequence of CO2 emissions. For the oceans to become a major ink in the future, one or more of the following conditions must be met: limiting nutrients must be added to the ocean; nutrients that are unused in the Southern ocean must be consumed; or the elemental composition of the organic matter must change (for instance, the ratio of calcite to organic matter in the sinking flux could decrease) (21).
The oceanic sinks of carbon are both abiotic and biotically controlled, and the interactions between the two control processes are poorly understood. The present uptake of anthropogenic CO2 by the oceans is thought to be primarily a consequence of abiotic processes, controlled by the transport of dissolved inorganic carbon from the surface to the deep oceans through circulation. Coupled climate-ocean simulations (22, 23)suggest that CO2-induced global warming may lead to increased vertical stratification of the water column. Should this occur, the physical transport of carbon from the upper ocean to the deep ocean will be reduced with a resulting decrease in the rate of sequestration of atmospheric carbon in the ocean (24) (Joos et al. 1999). The same physical processes that lead to increased stratification will conspire to reduce the efficacy of biological carbon sequestration in the ocean.