Tremendous uncertainty surrounds the extent to which global changes over the next century will change carbon sequestration by terrestrial ecosystems. A recent cross-comparison of the output of dynamic global vegetation models (DGVM), which incorporate understanding of physiology, phenology and elemental cycles with predicted changes in atmospheric composition and temperature, predicted a widely distributed terrestrial carbon sink sequestering 260-530 Pg C by 2100 (Cramer et al., 2001), which translates to 16-34% of expected anthropogenic emission over the same time period. However, a more recent assessment has concluded that terrestrial ecosystems will provide a much smaller sink of up to 20 and 150 Pg C over the next 20 and 100 years, respectively (Gruber et al., 2004). One key to this revised estimate was the incorporation of constraints to CO2 fertilization of terrestrial ecosystems by nitrogen availability into the DGVM runs (Hungate et al., 2003).
These findings are informative. In addition to highlighting the massive uncertainty associated with how carbon cycling through terrestrial ecosystems will change in the coming century, they highlight: (i) the need for experimental results to continue to inform model prediction; and (ii) for models to be used to highlight the key process that will control carbon sequestration by terrestrial ecosystems in the future. It is a certainty that new experimental insights into carbon cycling through ecosystems will result in important revisions to model predictions in the coming years.
Uncertainty in predicting the magnitude of carbon sequestration in terrestrial ecosystems over the next century should not detract from the fact that impressive advances in our understanding of how key global changes affecting biological processes have been made in the past decades. A detailed review of these advances is beyond the scope of this chapter; however, it is necessary to say that they encompass new understanding from the gene to the ecosystem, and have been facilitated by research performed at scales from highly controlled growth chambers to large-scale open-air manipulations. What is now required is an effort to understand the extent to which our extensive understanding of the action of single-factor manipulations on biological processes can be used to predict the effects of multifactor global change on ecosystem carbon cycling in the coming century. Specifically, we urgently need to know whether the optimism of Norby and Luo (2004) is appropriate when they suggest that a thorough understanding of the modes of action of single-factor manipulations, when factored into models, should provide the appropriate multifactor outputs required for predicting into the next century, or whether the caution of Osmond et al. (2004) is more appropriate when they highlight how 'our mechanistic understanding of feedbacks between the biosphere and atmosphere is vague at best, often provided by models, which themselves lack mechanistic understanding gained from experiments made at appropriate scale'. Clearly the only way to determine the more valid point of view is the urgent development of targeted, appropriately scaled multifactor experiments, to assess whether multifactor model predictions can be derived from single-factor inputs (Beier, 2004). If it turns out that our largely single-factor understanding is sufficient for predicting into the multifactor future, well and good. If not, a more general requirement for large-scale, multifactor experiments in key ecosystems must be undertaken. Ideally these studies will run in tandem with the sort of controlled large-scale, multifactor experiment described by Osmond et al. (2004). The rational behind this approach is to identify generally applicable mechanisms that will drive future carbon cycling within and between biomes, while minimizing the possibility that findings from individual studies are useful only as case studies.
Of course it must be acknowledged that numerous multifactor manipulations, typically elevated Ca and a second factor (e.g. ozone or temperature) have been performed at scales of enquiry from growth chambers to in-field open-topped chambers and increasingly within free-air CO2 enrichment (FACE) studies. However, care must be taken to make sure that results coming from such studies are either directly relevant to what is likely to happen in the coming century or can be weighted in some way to take account of potential artefacts. These artefacts could be due to experiments being performed at small scale, or vegetation at a specific growth stage. As an example, single-factor CO2 manipulation experiments have shown that effects of increases in CO2 on key physiological and whole-plant processes of numerous species do differ depending on the experimental conditions (Norby et al., 1999; Ainsworth and Long, 2004). In addition, care must be taken that treatments are applied appropriately, particularly heating treatments. It is highly likely that commonly used heating techniques such as soil heating, or infrared (IR) lamps that directly heat a vegetation canopy, will uncouple the normal controls on carbon flows through the plant into the soil. Lastly, experiments must be conscious of the inherent difficulties associated with scaling results onwards and upwards in time and space. Given these limitations to experimentation to date, it is unfortunate that funding constraints continue to prevent troubleshooting and ultimately setting up of a single, large-scale CO2 and temperature, FACE-type manipulation in a forest ecosystem. Such an experiment is essential for providing the first opportunities for validating our current understanding of CO2 and temperature interactions on carbon cycling through forest ecosystems.
Was this article helpful?