At the landscape scale, changes in the disturbance regime introduce instabilities in forest age-class distributions (Bhatti et al., 2000) and eventually in the distribution of plant species. If changes in disturbances are caused or accompanied by changes in environmental conditions, the responses of the forest ecosystem can be very complex. Changes in disturbance regimes, spatially and over time, can be exacerbated or mitigated by human activities (e.g., by fire ignitions and fire suppression).
Large areas of mixed savanna-woodlands in dry tropical zones of Africa, South America, Australia, and large areas of tropical humid forests burn every year (WRI, 2000). These fires are part of the natural seasonal cycle of growth, decay, and combustion and are ignited by lightning strikes. However, humans have long played a significant role in modifying fire regimes by changing the season and frequency of burning and consequently vegetation composition and structure (Goldammer and Price, 1998). During the 1990s in tropical humid areas, major fires have occurred in the Brazilian Amazon, Mexico, and Indonesia (Kalimantan and Sumatra) and were particularly severe in 1997-1998 during an El Niño episode (FAO, 1997a; Nepstad et al., 1999). Fire also is a serious threat to native forests in many parts of tropical and nontropical developing countries. China, for example, suffered large losses in a single fire event in 1987, with 1 Mha (Anon., 1987) to 1.3 Mha (Cahoon et al., 1994) burned. Fire prevention and suppression capability depends on available infrastructure such as imagery, roads, machinery, and human capital. In general, developed countries are better able to manage fire in regions with roads; developing countries may lack one or all of the factors.
The Indonesian fires of 1997-1998 were associated with a significant, but not unique, drought over much of the region. Estimates of the area burned vary from 96,000 ha to more than 8 Mha (Harwell, 2000). Most of this area was not forest but scrub, grassland, and agricultural lands. Almost no fires occurred deep within undisturbed primary forest; most were associated with land-clearing for new settlements or plantations or with logging operations. The most persistent fires were seven clusters of fires along the edges of degraded peat-swamp forests in southern Sumatera and Kalimantan (Legg and Laumonier, 1999). The extent and persistence of these fires, and similar fires in Brazil, show the importance of interaction between climate and human actions in determining the structure and composition of tropical forests, land-use patterns, and carbon emissions.
In the boreal zone of Canada, there has been a marked increase in fire since about 1970, after a 5-decade decrease (Kurz et al., 1995). The area of boreal forest burned annually in western North America has doubled in the past 20 years (0.28% in the 1970s to 0.57% in the 1990s), in parallel with the observed warming trend in the region (Kasischke et al., 1999), despite much improved detection and suppression technology. Similar trends have been noted for Eurasian forests (Shvidenko and Nilsson, 1994, 1997; Kasischke et al., 1999). Whether these changes are the result of human-induced climate change or are a result of natural climatic variability is not certain (Clark et al., 1996; Flannigan et al., 1998). Changes in the disturbance regime over periods of decades result in changes in forest age-class distribution to younger versus old forests (Kurz and Apps, 1999). These changes will reduce the landscape-averaged biomass stock and dead organic matter pools, including soils (Bhatti et al., 2000). Hence, changes in NBP occur on scales of years to decades.
Fire frequency is expected to increase with human-induced climate change, especially where precipitation remains the same or is reduced (Stocks et al., 1998). A general but moderate increase in precipitation, together with increased productivity, also could favor generation of more flammable fine fuels. Miranda (1994) suggests an increase in risk, severity, and frequency of forest fires in Europe. Stocks et al. (1998) used four GCMs and found similar predictions of an earlier start to the fire season and significant increases in the area experiencing high to extreme fire danger in Canada and Russia. Some regions may experience little change or even decreases in fire frequency, where precipitation increases or temperature does not rise (as in eastern Canada, where regional cooling has led to decreased fire frequencyFlannigan et al., 1998). In most regions, there is likely to be an increased risk of forest fires, resulting in a change in vegetation structure that in turn exacerbates this risk (Cramer and Steffen, 1997).
During the past decade, forest fires in developed countries generally have become smaller, with the exception of the former Soviet Union (FAO, 1997a), Canada (Kurz et al., 1995; Kurz and Apps, 1999), and the United States (Sampson and DeCoster, 1998). Where observed, the slight decline in forest areas burned per year is believed to be in part a result of improved prevention, detection, and control of fires. However, many such protected forests have increasing fuel loads and an abundance of dead and dying trees that eventually will make them more susceptible to catastrophic fires (e.g., Sampson and DeCoster, 1998). Several authors suggest that climate change is likely to increase the number of days with severe burning conditions, prolong the fire season, and increase lightning activity, all of which lead to probable increases in fire frequency and areas burned (Price and Rind, 1994; Goldammer and Price, 1998; Stocks et al., 1998).
Other reports in this collection