At present, there are few carbon flux measurements in tropical forests over periods long enough to provide an annual estimate of NEP, although the number of such measurements are expected to increase appreciably over the next 5 to 10 years. Measurements in pristine, seasonal tropical rain forests in Amazonia indicate NEP of approximately 1.0 t C ha-1 yr-1 (Grace et al., 1995a,b) and approximately 2.0 and 5.9 t C ha-1 yr-1 for dense, moist rain forest (Fan et al., 1990; Malhi et al., 1998, 1999). Sample plot studies also indicate net carbon sequestration rates within this range: 2.9 t C ha-1 yr-1 in selectively logged evergreen rain forest, 2.4 t C ha-1 yr-1 in heavily logged rain forest (Nabuurs and Mohren, 1993), and 0.7 to 1.5 t C ha-1 yr-1 in semi-evergreen tropical rainforest (Phillips et al., 1998; Mahli et al., 1999). Measurements of NEP for seasonally dry forest (savannas) indicate annual carbon sequestration rates of 0.12 and 0.75 t C ha-1 yr-1 for Sahelian and north Australian sites, respectively (Hanan et al., 1998; Eamus et al., n.d.).
Virtually all forests in the temperate region are managed to a greater or lesser extent, and there are only a few patches that might be regarded as pristine. Recent measurements of NEP by eddy covariance over one or more years in managed forests in Europe in the EUROFLUX network (Valentini et al., 2000) and in North America (e.g., Greco and Baldocchi, 1996; Goulden et al., 1996a,b; Baldocchi et al., 1997) indicate rates of carbon sequestration in the range of 2.5 to 7 t C ha-1 yr-1 over approximately 20 sites differing in latitude, altitude, climate, species composition, and management intensity. The highest values are for young (20-50 years), fast-growing coniferous plantations and broad-leaved trees in old-field succession. Sample plots and yield tables, for example, lead to estimates of NPP of up to 4.5 t C ha-1 yr-1 for Sitka spruce in a maritime environment (Jarvis, 1981), 1.7 to 3.6 t C ha-1 yr-1 for European beech and mixed deciduous forests, and 2.5 to 3.4 t C ha-1 yr-1 for Norway spruce forests in Central Europe (e.g., Spiecker et al., 1996) and 3.4 t C ha-1 yr-1 for Douglas fir in the northwest United States.
The extensive boreal forests across Siberia and Canada have been subjected to increasing exploitation over the past 150 years, but parts of the area standing can still be regarded as largely pristine. Most of the 12 or more boreal forest sites investigated are sequestering carbon (NEP) at annual rates of up to 2.5 t C ha-1 yr-1 (Black et al., 1996; McCaughey et al., 1997; Blanken et al., 1998; Jarvis, 1998; Chen et al., 1999). The values obtained depend particularly on latitude, soil type, and successional stage. However, NEP measurements over periods of up to 5 years in northern Canada in the BOREAS experiment (Sellers et al., 1997), in Siberia (Schulze et al., 1999), and in northern Europe (Valentini et al., 2000) have demonstrated that a few old-growth coniferous stands may be carbon neutral (Goulden et al., 1998) and in warm and cloudy years can be a carbon source (Lindroth et al., 1998), losing carbon at a rate of up to 1.0 t C ha-1 yr-1. Sample plots give average values for several species of aboveground and below-ground NPP of 4.7 t C ha-1 yr-1 for deciduous, broad-leaved species and 2.7 t C ha-1 yr-1 for coniferous species (Gower et al., 1997).
Interannual variability in NEP may be considerable. The sequestration potential of tropical forests may vary by 10 percent from year to year depending on the length of the dry period and variation in solar radiation inputs and temperature, such as those caused by the eruption of Pinatubo (Grace et al., 1995a,b) or those that occur during strong El Ni�o years (Tian et al., 1998). The NEP of evergreen temperate forests (largely conifers) is similarly affected by variations in radiation input and temperature over the year and by the length of the photosynthetic season, particularly in oceanic climates (Valentini et al., 2000). The NEP of deciduous temperate forest (mostly broad-leaved trees) is also particularly sensitive to the length of the growing season as defined by the times of bud burst and leaf senescence, which may vary by 2 to 3 weeks from year to year (Goulden et al., 1996a,b). The NEP of boreal forests is even more sensitive to the onset of the spring thaw, which can vary from year to year by 2 to 3 weeks and may result in variation in annual carbon sequestration of 1.0 t C ha-1 yr-1 (Havranek and Tranquillini, 1995; Bergh et al., 1998; Goulden et al., 1998; Lindroth et al., 1998; Bergh and Linder, 1999; Chen et al., 1999).
Other reports in this collection