IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis Processes Driven by Climate, Atmospheric Composition and Land Use Change Climatic regulation of terrestrial carbon exchange

Ecosystem responses to environmental drivers (sunlight, temperature, soil moisture) and to ecological factors (e.g., forest age, nutrient supply, organic substrate availability; see, e.g., Clark, 2002; Ciais et al., 2005b; Dunn et al., 2007) are complex. For example, elevated temperature and higher soil water content enhance rates for heterotrophic respiration in well-aerated soils, but depress these rates in wet soils. Soil warming experiments typically show marked soil respiration increases at elevated temperature (Oechel et al., 2000; Rustad et al., 2001; Melillo et al., 2002), but CO2 fluxes return to initial levels in a few years as pools of organic substrate re-equilibrate with inputs (Knorr et al., 2005). However, in dry soils, decomposition may be limited by moisture and not respond to temperature (Luo et al., 2001). Carbon cycle simulations need to capture both the short- and long-term responses to changing climate to predict carbon cycle responses.

Current models of terrestrial carbon balance have difficulty simulating measured carbon fluxes over the full range of temporal and spatial scales, including instantaneous carbon exchanges at the leaf, plot or ecosystem level, seasonal and annual carbon fluxes at the stand level and decadal to centennial accumulation of biomass and organic matter at stand or regional scales (Melillo et al., 1995; Thornton et al., 2002). Moreover, projections of changes in land carbon storage are tied not only to ecosystem responses to climate change, but also to the modelled projections of climate change itself. As there are strong feedbacks between these components of the Earth system (see Section 7.3.5), future projections must be considered cautiously. Effects of elevated carbon dioxide

On physiological grounds, almost all models predict stimulation of carbon assimilation and sequestration in response to rising CO2, called ‘CO2 fertilization’ (Cramer et al., 2001; Oren et al., 2001; Luo et al., 2004; DeLucia et al., 2005). Free Air CO2 Enrichment (FACE) and chamber studies have been used to examine the response of ecosystems to large (usually about 50%) step increases in CO2 concentration. The results have been variable (e.g., Oren et al., 2001; Nowak et al., 2004; Norby et al., 2005). On average, net CO2 uptake has been stimulated, but not as much as predicted by some models. Other factors (e.g., nutrients or genetic limitations on growth) can limit plant growth and reduce response to CO2. Eleven FACE experiments, encompassing bogs, grasslands, desert and young temperate tree stands report an average increased net primary productivity (NPP) of 12% when compared to ambient CO2 levels (Nowak et al., 2004). There is a large range of responses, with woody plants consistently showing NPP increases of 23 to 25% (Norby et al., 2005), but much smaller increases for grain crops (Ainsworth and Long, 2005), reflecting differential allocation of the incremental organic matter to shorter- vs. longer-lived compartments. Overall, about two-thirds of the experiments show positive response to increased CO2 (Ainsworth and Long, 2005; Luo et al., 2005). Since saturation of CO2 stimulation due to nutrient or other limitations is common (Dukes et al., 2005; Koerner et al., 2005), it is not yet clear how strong the CO2 fertilization effect actually is. Nutrient and ozone limitations to carbon sequestration

The basic biochemistry of photosynthesis implies that stimulation of growth will saturate under high CO2 concentrations and be further limited by nutrient availability (Dukes et al., 2005; Koerner et al., 2005) and by possible acclimation of plants to high CO2 levels (Ainsworth and Long, 2005). Carbon storage by terrestrial plants requires net assimilation of nutrients, especially N, a primary limiting nutrient at middle and high latitudes and an important nutrient at lower latitudes (Vitousek et al., 1998). Hungate et al. (2003) argue that ‘soil C sequestration under elevated CO2 is constrained both directly by N availability and indirectly by nutrients needed to support N2 fixation’, and Reich et al. (2006) conclude that ‘soil N supply is probably an important constraint on global terrestrial responses to elevated CO2’. This view appears to be consistent with other recent studies (e.g., Finzie et al., 2006; Norby et al., 2006; van Groenigen et al., 2006) and with at least some of the FACE data, further complicating estimation of the current effects of rising CO2 on carbon sequestration globally.

Additional N supplied through atmospheric deposition or direct fertilization can stimulate plant growth (Vitousek, 2004) and in principle could relieve the nutrient constraint on CO2 fertilization. Direct canopy uptake of atmospheric N may be particularly effective (Sievering et al., 2000). Overall, the effectiveness of N inputs appears to be limited by immobilisation and other mechanisms. For example, when labelled nitrogen (15N) was added to soil and litter in a forest over seven years, only a small fraction became available for tree growth (Nadelhoffer et al., 2004). Moreover, atmospheric N deposition is spatially correlated with air pollution, including elevated atmospheric ozone. Ozone and other pollutants may have detrimental effects on plant growth, possibly further limiting the stimulation of carbon uptake by anthropogenic N emissions (Ollinger and Aber, 2002; Holland and Carroll, 2003). Indeed, Felzer et al. (2004) estimate that surface ozone increases since 1950 may have reduced CO2 sequestration in the USA by 18 to 20 TgC yr–1. The current generation of coupled carbon-climate models (see Section 7.3.5) does not include nutrient limitations or air pollution effects. Fire

Fire is a major agent for conversion of biomass and soil organic matter to CO2 (Randerson et al., 2002a–d; Cochrane, 2003; Nepstad et al., 2004; Jones and Cox, 2005; Kasischke et al., 2005; Randerson et al., 2005). Globally, wildfires (savannah and forest fires, excluding biomass burning for fuel and land clearing) oxidize 1.7 to 4.1 GtC yr–1 (Mack et al., 1996; Andreae and Merlet, 2001), or about 3 to 8% of total terrestrial NPP. There is an additional large enhancement of CO2 emissions associated with fires stimulated by human activities, such as deforestation and tropical agricultural development. Thus, there is a large potential for future alteration in the terrestrial carbon balance through altered fire regimes. A striking example occurred during the 1997 to 1998 El Niño, when large fires in the Southeast Asian archipelago are estimated to have released 0.8 to 2.6 GtC (see Section Fire frequency and intensity are strongly sensitive to climate change and variability, and to land use practices. Over the last century, trends in burned area have been largely driven by land use practices, through fire suppression policies in mid-latitude temperate regions and increased use of fire to clear forest in tropical regions (Mouillot and Field, 2005). However, there is also evidence that climate change has contributed to an increase in fire frequency in Canada (Gillett et al., 2004). The decrease in fire frequency in regions like the USA and Europe has contributed to the land carbon sink there, while increased fire frequency in regions like Amazonia, Southeast Asia and Canada has contributed to the carbon source. At high latitudes, the role of fire appears to have increased in recent decades: fire disturbance in boreal forests was higher in the 1980s than in any previous decade on record (Kurz et al., 1995; Kurz and Apps, 1999; Moulliot and Field, 2005). Flannigan et al. (2005) estimate that in the future, the CO2 source from fire will increase. Direct effects of land use and land management

Evolution of landscape structure, including woody thickening: Changes in the structure and distribution of ecosystems are driven in part by changes in climate and atmospheric CO2, but also by human alterations of landscapes through land management and the introduction of invasive species and exotic pathogens. The single most important process in the latter category is woody encroachment or vegetation thickening, the increase in woody biomass occurring in (mainly semi-arid) grazing lands. In many regions, this increase arises from fire suppression and associated grazing management practices, but there is also a possibility that increases in CO2 are giving C3 woody plants a competitive advantage over C4 grasses (Bond et al., 2003). Woody encroachment could account for as much as 22 to 40% of the regional carbon sink in the USA (Pacala et al., 2001), and a high proportion in northeast Australia (Burrows et al., 2002). Comprehensive data are lacking to define this effect accurately.

Deforestation: Forest clearing (mainly in the tropics) is a large contributor to the land use change component of the current atmospheric CO2 budget, accounting for up to one-third of total anthropogenic emissions (see Table 7.2; Section; also Table 7.1, row ‘land use change flux’). The future evolution of this term in the CO2 budget is therefore of critical importance. Deforestation in Africa, Asia and the tropical Americas is expected to decrease towards the end of the 21st century to a small fraction of the levels in 1990 (IPCC, 2000). The declines in Asia and Africa are driven by the depletion of forests, while trends in the Americas have the highest uncertainty given the extent of the forest resource.

Afforestation: Recent (since 1970) afforestation and reforestation as direct human-induced activities have not yet had much impact on the global terrestrial carbon sink. However, regional sinks have been created in areas such as China, where afforestation since the 1970s has sequestered 0.45 GtC (Fang et al., 2001). The largest effect of afforestation is not immediate but through its legacy.

Agricultural practices: Improvement of agricultural practices on carbon-depleted soils has created a carbon sink. For instance, the introduction of conservation tillage in the USA is estimated to have increased soil organic matter (SOM) stocks by about 1.4 GtC over the last 30 years. However, yearly increases in SOM can be sustained only for 50 to 100 years, after which the system reaches a new equilibrium (Cole et al., 1996; Smith et al., 1997). Moreover, modern conservation tillage often entails large inputs of chemicals and fertilizer, which are made using fossil fuels, reducing the CO2 benefit from carbon sequestration in agricultural soils. The increase in soil carbon stocks under low-tillage systems may also be mostly a topsoil effect with little increase in total profile carbon storage observed, confounded by the fact that most studies of low-tillage systems have only sampled the uppermost soil layers. Forest regrowth

Some studies suggest that forest regrowth could be a major contributor to the global land carbon sink (e.g., Pacala et al., 2001; Schimel et al., 2001; Hurtt et al., 2002). Forest areas generally increased during the 20th century at middle and high latitudes (unlike in the tropics). This surprising trend reflects the intensification of agriculture and forestry. Globally, more food is being grown on less land, reflecting mechanisation of agriculture, increased fertilizer use and adoption of high-yield cultivars, although in parts of Africa and Asia the opposite is occurring. Likewise, intensive forest management and agroforestry produce more fibre on less land; improved forest management favours more rapid regrowth of forests after harvest. These trends have led to carbon sequestration by regrowing forests. It should be noted, however, that industrialised agriculture and forestry require high inputs of fossil energy, so it is difficult to assess the net global effects of agricultural intensification on atmospheric greenhouse gases and radiative forcing.

Regional studies have confirmed the plausibility of strong mid-latitude sinks due to forest regrowth. Data from the eddy flux tower network show that forests on long-abandoned former agricultural lands (Curtis et al., 2002) and in industrial managed forests (Hollinger et al., 2002) take up significant amounts of carbon every year. Analysis of forest inventory data shows that, in aggregate, current forest lands are significant sinks for atmospheric CO2 (Pacala et al., 2001). Few old growth forests remain at mid-latitudes (most forests are less than 70 years old), in part due to forest management. Therefore, forests in these areas are accumulating biomass because of their ages and stages of succession. Within wide error bands (see Section, the uptake rates inferred from flux towers are generally consistent with those inferred from inverse methods (e.g., Hurtt et al., 2002). Stocks of soil carbon are also likely increasing due to replenishment of soil organic matter and necromass depleted during the agricultural phase, and changes in soil microclimate associated with reforestation; these effects might add 30 to 50% to the quantity of CO2 sequestered (e.g., Barford et al., 2001). It is important to note that at least some of this sequestration is ‘refilling’ the deficits in biomass and soil organic matter, accumulated in previous epochs (see Figure 7.3), and the associated CO2 uptake should be expected to decline in the coming decades unless sustained by careful management strategies designed to accomplish that purpose.