In assessing the carbon sequestration potential of terrestrial ecosystems now and in the future, we need to consider the time scale over which the carbon gain is measured or estimated. In the course of a day, carbon sequestration essentially follows solar radiation (provided there is no major constraint such as frozen or dry soil): Carbon is accumulated during the daylight hours and lost at night. Depending on the balance between these short-term gains and losses, for any vegetation there may be a net gain or loss of carbon over any day (24 hours) in the year. In boreal forest, for example, carbon may be gained in daytime by a stand of trees on about 75 percent of the days through a growing season but lost on 25 percent of the days, the latter being days of particularly low solar radiation input (e.g., days with clouds and rain, or smoke in the atmosphere), high temperatures, or drought (see, e.g., Jarvis et al., 1997; Lindroth et al., 1998).
Over a year, the length of the growing season has a major influence on carbon gain. In boreal coniferous forest, for example, one-third to one-half of the carbon gained in the summer months is lost during autumn and the long winter period when the ground is partly frozen. The length of the growing season in broad-leaved temperate forest is defined by bud burst in the spring and leaf senescence in the autumn, whereas the season of net photosynthetic gain in temperate coniferous forest is conditioned largely by the day length and daily total radiation input in the winter months, thus may be 10 months long in maritime climates. Vegetation in Mediterranean climates is generally sparse and strongly seasonal, with small NEP that is strongly constrained by water availability for a large part of the time. Only in moist tropical forests is the carbon gain nearly continuous throughout the year, reduced only by occasional short periods of low solar radiation (cloudiness), low temperature, or water deficit (Malhi et al., 1998).
Annual dynamics are particularly important in forest systems because their carbon turnover times can be many decades, characterized by major changes in structure through a series of stages in the life cycle. The carbon sequestration potential of a young forest stand that is regenerating or regrowing after a disturbance such as fire or harvesting is critically dependent on the point of time within the life cycle. Initially the disturbed area is likely to be losing carbon to the atmosphere (the length of this period depends on species, site conditions, and degree of disturbance), but the trees that subsequently occupy the site fully will eventually replace the lost carbon (Krankina et al., 1999). Carbon will then again accumulate during a phase of rapid growth that may last for centuries or at least decades, depending on the species of trees and site conditions (Buchmann and Schulze, 1999). Overmature forest stands take up carbon from the atmosphere at slower rates, but even as the growth increment of the trees approaches zero, carbon may continue to be funneled from the atmosphere to the soil via the trees in the form of aboveground and below-ground detritus. Nonetheless, forest stands can become net sources of carbon to the atmosphere-for example, if soil temperature abruptly increases (Peterjohn et al., 1994) or the soil becomes more aerobic after drainage (Lindroth et al., 1998), thus promoting oxidation of soil organic matter.
Over the long term, the sink capacity of any ecosystem is determined by the size of the pools (i.e., the aboveground and below-ground biomass) and their turnover times. Thus, additional carbon can be stored in an ecosystem only if more carbon is kept for the same period of time or the same amounts of carbon are kept over longer periods of time. A reduced rate of disturbance could therefore enhance carbon storage. Increased growth, on the other hand, will not add to the long-term sink if disturbances (or harvests) increase in frequency.
At any time during this life cycle, there may be appreciable interannual variation in NEP as a result of variability in the weather from year to year. Thus, a measurement of carbon uptake made for a single year at some arbitrary time in the life cycle, or even for a few years-such as the first 5-year "commitment period" of the Kyoto Protocol-may give a misleading picture of ongoing carbon sequestration as well as the long-term carbon sequestration potential of the vegetation, unless interannual variation is accounted for specifically. The results obtained from field measurement in recent years (as reported below) are still few, cover only selected areas of about 200 ha, and are not representative of all stages in the life cycle. Extrapolation to infer carbon sequestration over larger areas and longer time periods must be done with great care, and it is not as yet possible to obtain reliable average estimates for biomes as a whole from stand-scale measurements of NEP alone (see Section 1.3.2). Instead, forest inventory methods are generally being used to estimate standing stocks of carbon and the larger area and longer term carbon balances of these ecosystems and then extrapolated into the future using models.
Other reports in this collection