The Science of the Total Environment 262Ž. 2000 221᎐229

Climate change and forest fires

U M.D. Flannigana, , B.J. Stocksb, B.M. Wottonb

a Canadian Forest Ser¨ice, 5320-122nd St, Edmonton AB T6H 3S5, Canada bCanadian Forest Ser¨ice, P.O. Box 490, Sault Ste. Marie ON P6A 5M7, Canada

Received 22 November 1999; accepted 4 March 2000

Abstract

This paper addresses the impacts of change on forest fires and describes how this, in turn, will impact on the forests of the United States. In addition to reviewing existing studies on and forest fires we have used two transient general circulation modelsŽ. GCMs , namely the Hadley Centre and the Canadian GCMs, to = estimate fire season severity in the middle of the next century. Ratios of 2 CO2 seasonal severity ratingŽ. SSR over present day SSR were calculated for the means and maximums for North America. The results suggest that the SSR will increase by 10᎐50% over most of North America; although, there are regions of little change or where the SSR may decrease by the middle of the next century. Increased SSRs should translate into increased forest fire activity. Thus, forest fires could be viewed as an agent of change for US forests as the fire regime will respond rapidly to climate warming. This change in the fire regime has the potential to overshadow the direct effects of climate change on species distribution and migration. ᮊ 2000 Elsevier Science B.V. All rights reserved.

Keywords: Climate change; Forest fires; Forests; General circulation models

1. Introduction helps shape the landscape mosaic and influence biogeochemical cycles such as the carbon cycle. Fire is one of the dominant disturbances in the Forest structure and composition, now and in the forests of the United States. Forest fire is a pri- past, is influenced by the fire regimeŽ Heinsel- mary process that influences the vegetation com- man, 1973; Wright and Bailey, 1982. . The fire position and structure of any given location; fire regime has six components; fire frequency, size, intensity, seasonality, type and severity. The eco- logical importance of some of these component U Corresponding author. Tel.: q1-780-435-7338; fax: q1- parts of the fire regime has been put into per- 780-435-7359. spective by MalansonŽ. 1987 , Whelan Ž. 1995 . Fire E-mail address: mfl[email protected]Ž. M.D. Flannigan . frequency affects through interrupting

0048-9697r00r$ - see front matter ᮊ 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7Ž. 0 0 00524-6 222 M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 or terminating individuals’ life cycles. If fires re- fire severity on the landscape. The SSR is a final cur more or less regularly, selection pressure will component of the Canadian forest fire favor those organisms that better take advantage indexŽ. FWI system Ž Van Wagner, 1987 . . This of the recurrence at a given interval. Fire size SSR index is a seasonal mean of the daily esti- determines landscape patchiness and determines mate of the control difficulty of a potential fire the distance seed will have to travel for regenera- which, in turn, derives from an estimate of the tion. Fire intensity is equivalent to the amount of potential intensity of a fire. The FWI system is a energy released during a fire, and within the system of daily meteorological-based indexes de- confines of a single burn, can vary greatly depend- signed and used universally in Canada to estimate ing on fuel type and loading, topography, meteo- fire danger in a generalized fuel type. The success rological influences and characteristics of the pre- of this system in both Canada and beyond is due, vious , among others. The season of apart from the simplicity of its calculation proce- the year in which fire occurs is one of the de- dures, to the three fuel moisture models which terminants of the successional trajectories on form its core. The FWI system includes a litter which ecosystems embark after fire. Time of year moisture model, an upper organic moisture model may affect fire intensity through differences in and a deep organic or soil moisture model. The surface and crown fuel moisture contents. Seaso- different fuels represented by these models allow nal phenological state of the plants burned will a range of factors to enter into the final estima- determine the characteristics of the vegetative or tion of a particular day’s fire dangerŽ from rapidly seed reproductive response and have a pro- changing surface litter moisture which would nounced effect on the structure of post-fire eco- affect rates of spread of a fire, to the effect of systems and landscapes. Fire type refers to crown, long term drought on the difficulty of suppres- surface and ground fires, which are largely con- sion. . Though developed for a generalized fuel trolled by fire intensity and fuel characteristics type, the FWI system has been found to model Ž.structure, load and moisture . Fire type can vary fire potential in a broad range of fuel types from across the area of the burn, giving rise to a closed forest stands to grass quite wellŽ Van Wag- mosaic of post-fire plant communities that might ner, 1975. . Indeed fire management systems using be initiated by crowning, surface fires, intermit- the FWI system have been implemented in fuel tent crowning or a combination thereof. Fire types that differ significantly from the boreal, severity is a measure of fuel consumption, primar- such as Mexico and the state of Florida. ily the depth of burn in surface soil organic layers and is therefore another important controlling factor of post-fire structure and func- 2. Current impacts tion through direct impacts on underground plant root and reproductive tissues, soil seed bank, and In the United States, during the last 10 years forest floor microbial populations. These compo- Ž.1989᎐1998 an average of approximately 100 000 nent parts of fire regime with their intricate link- fires burn over an area of 3 300 000 acres an- age to forest ecosystem structure and function nually. There is a great deal of variability in the are, in turn, highly dependent on weather and fire statistics from year-to-year from a low in 1992 climateŽ Flannigan and Harrington, 1988; John- with area burned of just over 1 000 000 acres to a son, 1992; Swetnam, 1993. . The rapid response of high of over 6 500 000 acres in 1996. Typically, the fire regime to changes in climate is note- most of the area burned occurs in the westŽ in- worthy. Thus, if the expected changes in climate cluding Alaska.Ž and the south-east USDA Forest become a reality in the next century, an altered Service, 1998. . Most of the area burned occurs in fire regime could have the most immediate and May and June in the south-west and south-east significant impact on ecosystems. while July and August are the most active months In this paper we use an index called the sea- in terms of area burned in the north-west and sonal severity ratingŽ. SSR to examine changes in Alaska. Humans start most of the fires whether M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 223 accidentally or deliberately. From a suppression ignitions. This is important if land managers want perspective, however, lightning-ignited wildfires their activities to emulate natural disturbance over pose the most serious threat because detection the landscape; they should view the disturbance and response times can be significant, as lightning regime as dynamic and not staticŽ Bergeron et al., fires often start in remote areas that can be 1998. . Presently, the fire regime varies greatly difficult to reach. In terms of area burned statis- across the United States. In the south-east, tics, a relatively small percentage of the fires are loblolly pine Ž.Pinus taeda L. dominates as the responsible for the majority of the area burned. historical fire regime of frequent surface fires For example, 1% of all wildland fires in the Ž.2᎐15 years , which favored longleaf and slash western United States are responsible for 98% of pine Ž.Pinus elliotii Engelm. and has been re- the area burnedŽ. Strauss et al., 1989 . From a placed by a fire regime of infrequent r ᎐ historical perspective it appears that current lev- surface crown fires every 50 100 years primarily els of area burned are low compared to the last due to the influence of humans. In the pinyon- 500 yearsŽ. Leenhouts, 1998 . This amount of fire juniper forests of the south-west, grazing and fire activity significantly influences forest ecosystems. suppression has reduced the fire frequency and Schmoldt et al.Ž. 1999 provide a good overview of has allowed the expansion of juniperŽ Wright and the effects of fire disturbance on US ecosystems. Bailey, 1982. . In the west, frequent surface fires Ž The fire regime at any given location is the are common in the Ponderosa pine Pinus pon- . result of complex interactions between fuel, topo- derosa Dougl. communities. A complex fire graphy, ignitions and weather. Fuel type, struc- regime prevails over the wet northwestern forests ture, moisture and spatial continuity are impor- where surface fires occur every 50 years or so and ᎐ tant aspects in determining the fire regime. The stand replacing fires can range from 300 500 years or more. For interior Alaska, YarieŽ. 1981 topography can influence the spread of fire found stand replacing fire frequencies of 26᎐113 through natural fire breaks such as lakes, rivers years depending on the vegetation type. and ridges; also, slope and orientation influence Fire does not influence the forest in isolation the fire spread. The frequency and timing of but rather interacts with other disturbances. After ignitions, whether natural or human-caused, can a fire, windthrow is more prevalent due to in- play a role in the fire regime. Weather is crucial creased exposure. StocksŽ. 1987 suggests that to the occurrence and growth of forest fires. spruce budworm wChoristoneura fumiferana Cle- Weather is important in terms of ignition agents mensŽ. Lepidoptera: Tortricidaex damaged forests where lightning is the key ignition agent for natur- may be more prone to fire and would burn more ally caused forest fires. Lightning is the result of intensely due to increased fire loads. Swetnam an electrical discharge from a thunderstorm, and BetancourtŽ. 1998 found regional synchrony which itself is a result of the appropriate meteo- in relationships between climate, fire and insect rological conditions, namely, atmospheric instabil- disturbances in the American south-west. McCul- ity, moisture and a lifting agent. The weather lough et al.Ž. 1998 have reviewed the interaction prior to ignition is important in determining the between fire and insects, which often affects suc- fuel moisture, which, in turn, will determine if cession, nutrient cycling and forest composition. ignition will occur and if the fire will grow. These These interactions between disturbances could be weather conditions that influence fuel moisture synergistic and may change rapidly as the climate include temperature, precipitation, wind speed changes. and atmospheric moistureŽ vapor pressure defi- Forests are of vital importance for the economy cient. . Fire growth is a function of a number of of many regions across the country. Fires can variables; but, if fuels are available and dry then affect the forest industry by competing for wood wind speed is the key factor. The fire regime at fibre especially in pulp and paper operations any location will probably change over time due where charcoal is difficult to remove from the to these interactions between weather, fuel and pulp and paper making process. Fire can also 224 M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 affect the recreational value of the landscape, individual fire is a result of the complex set of although as the public becomes educated, fire will interactions that include ignition agents, fuel con- be more acceptablew e.g. Yellowstone National ditions, topography and weather including tem- Park-BaskinŽ. 1999x . Finally, fire can pose a health perature, relative humidity, wind velocity and the concern as smoke from wild or prescribed fires amount and frequency of precipitation. Increas- can cause respiratory problems. ing temperature alone does not necessarily guar- Human activities greatly influence the fire antee greater fire disturbance. regime. Since European settlement, fire suppres- Price and RindŽ. 1994 modelled lightning fires sion has been practised across much of the United in the United States and then used the Goddard States. The principal goal for many land manage- Institute for Space Studies GCM to estimate the ment agencies was to exclude fire. It is only change in lightning fires and area burned in the = recently that the positive aspects of fire in pre- United States for a 2 CO2 scenario. Their re- serving our landscapes have been recognized. sults suggest a 44% increase in lightning-caused Other human activities such as land use change fires with an associated 78% increase in area = have influenced the fire regime. Clearing the burned for the 2 CO2 scenario. Tackle et al. forest for agriculture or urban development has Ž.1994 looked at the impacts on surface pressure reduced or fragmented forested areas. patterns on forest fires in the north-east United Not only does climate influence fire, but fire States and then used the Canadian GCM to pro- activity also can influence climate. Regionally, ject future impacts. Their results suggested an = large fires with their change in albedo and re- increased frequency of drying in the 2 CO2 moval of vegetation with the accompanying drop scenario but the results were not statistically sig- in evapotranspiration, can influence the energy nificant. Other studies also indicate significant budget and climate. Recent evidence suggests that increases in fire weather severity for Canada, local to regional changes in the land surface Alaska and RussiaŽ. Stocks et al., 1998 , which characteristics can significantly alter the climate should be accompanied by increases in the area of that regionŽ. Couzin, 1999 . burned. Results from Flannigan et al.Ž. 1998 show = a great deal of regional variation in the 2 CO2 fire weather for central and northern North 3. Impact of climate change on the fire regime America and Europe with some regions showing a significant increase in fire weather while other There are many general circulation models areas show a decrease in fire weather severity due Ž.GCMs that enable researchers to simulate the to increased precipitation amount and frequency. future climate. Although there are a number of Figs. 1 and 2 show the ratio of the mean shortcomings associated with GCMs, they provide seasonal severity ratingŽ. SSR for 2060 Ž approx. = ᎐ the best means available to estimate the impact 2 CO2 .Ž. over present day 1985 1994 using the of changes in the future climate on the fire regime Canadian GCM and the Hadley GCM, respec- at larger scales. Most models are in agreement in tively. The higher the SSR the greater likelihood predicting the greatest warming at high latitudes of increased fire activity given everything else and over land. In addition to temperature, other being equalŽ. fuels, ignitions, etc. . Note that the weather variables will be altered in the new cli- 1.0 contour indicates no change while areas above mate such as precipitation, wind, cloudiness, etc. 1.0 suggest an increase in the SSR whereas values The variability of extreme events may be altered below 1.0 denote areas where SSR decreases. as wellŽ Mearns et al., 1989; Solomon and Lee- There is a great deal of regional variation in the mans 1997. . Some studies suggest universal in- results. The Canadian GCM suggests over 30% creases in fire frequency with climatic warming increase in SSR for sections of the south-east, ŽOverpeck et al., 1990; Intergovernmental Panel midwest and Alaska whereas the Hadley GCM on Climate ChangeŽ.. IPCC 1996 . The universal- suggests increases near 20% for the north-east ity of these results is questionable because an and decreases in the Northern Plains. Figs. 3 and M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 225

Figs. 1᎐4. Ž.See legends on opposite page . 226 M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229

4 show the ratio of the maximum SSR for 2= The fire season will start earlier in the spring and r ᎐ CO2 present dayŽ. 1985 1994 using the Cana- extend longer into the autumn, yielding a longer dian Climate Model and Hadley model, respec- fire season. Wotton and FlanniganŽ. 1993 have tively. Maximum values were used as fire typically found that the fire season length in Canada on occurs during extreme weather conditions rather average will increase by 22% or 30 days in a = than mean conditions. The Canadian ModelŽ Fig. 2 CO2 climate. Vegetation type, amount and 3. shows SSR increases of approximately 30% for structure influence the fire regime characteristics; parts of Alaska and increases of less than 10% for thus any changes in vegetation due to changes in most of the western United States. The Hadley climate or fire regime would have a feedback on modelŽ. Fig. 4 shows most of the country with the fire regime. Human activities such as fire increases of less than 10% except most of Alaska management policies and effectiveness will cont- and a section from eastern Texas᎐Louisiana to inue to change. Other human activities such as eastern Kansas and southwestern Missouri. These conversion of forest lands to agriculture or urban figures indicate that changes in the climate can areas along with the fragmentation of the land- have dramatic changes in the fire regime. These scape will influence the fire regime. Also, re- increases in SSR should translate to correspond- search has suggested that the persistence of ing increases in fire activity, in particular area blocking ridges in the upper atmosphere will in- = burned. As a first guess we would expect in- crease in a 2 CO2 climateŽ. Lupo et al., 1997 . creases in area burned for the United States of This could have a significant impact on forest 25᎐50% by the middle of the 21st century with fires as these upper ridges are associated with dry most of the increases occurring in Alaska and the and warm conditions at the surface, which are southeastern United States. However, there may conducive to forest firesŽ. Skinner et al., 1999 . be regions in America like the Northern Plain These are all confounding effects that may dam- StatesŽ. Fig. 2 that may experience no change or pen or amplify the impact of a changing climate even reduced fire activity in the next century due on the fire regime. to increases in precipitation amount and fre- The important aspect of the impact of climate quency. Due to the coarse spatial resolution of change on forest fires with respect to the influ- the GCMŽ. approx. 400 km confidence in the ence on vegetation is that fire may be more results over complex terrain-like mountainous ar- important than the direct effects of climate change eas is low. In such areas, a regional climate model with respect to species distribution, migration, Ž.RCM should be used Ž Giorgi, 1990 . where the substitution and extinctionŽ Weber and Flanni- finer spatial resolutionŽ. approx. 40 km can resolve gan, 1997. . Fire can act as an agent of change to mountain ranges, etc. hasten the modification of the vegetation land- In addition to climatic influences on the fire scape into a new equilibrium with the climate if regime other factors such as ignition agents, species are able to migrate fast enough. This length of the fire season, vegetation characteris- might be true where the fire activity is expected tics and human activities such as fire management to increase in the next century and thus acceler- policies and landscape fragmentation may greatly ate changes in vegetation. In those areas of influence the fire regime in the next century. America that experience a reduced fire frequency Ignition probabilities may increase in a warmer due to the altered climate or human activities, the world due to increased cloud-to-ground lightning transition of vegetation types may be retarded. As discharges with warmingŽ. Price and Rind, 1994 . the climate warms, the favorable region for many

Fig. 1. The ratio of the mean seasonal severity ratingŽ. SSR for 2060rpresent day Ž 1985᎐1994 . using the Canadian GCM. Fig. 2. The ratio of the mean seasonal severity ratingŽ. SSR for 2060rpresent day Ž 1985᎐1994 . using the Hadley GCM. Fig. 3. The ratio of the maximum seasonal severity ratingŽ. SSR for 2060rpresent day Ž 1985᎐1994 . using the Canadian GCM. Fig. 4. The ratio of the maximum seasonal severity ratingŽ. SSR for 2060rpresent day Ž 1985᎐1994 . using the Hadley GCM. M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 227 species shifts northward. This poleward migration sensitive processes, including decomposition and of the southern species would be enhanced by the nutrient cyclingŽ.Ž. Anderson, 1992 ; 2 northern presence of disturbed areas such as burns that forest ecosystems are typically nitrogen-limited would allow establishment of these migrating and can be expected to respond to ameliorated competitors. For example, increased fire fre- nutrient conditionsŽ. Van Cleve et al., 1986 ; and quency would hasten the change from forest to Ž.3 effects of altered decomposition rates on fire parkland to grassland. In areas of decreased fire regime via fire severity and changed organic layer activity, shade tolerant species and long-lived thickness. species would dominate the landscape and would The forests of the United States will respond be hard to displace thus retarding the poleward directly to changes in the climate gradually over migration of the southern species. Of course, it is time. However, the almost instantaneous re- possible that in some regions increases in other sponse of the fire regime to changes in climate disturbance regimes such as pests, diseases and has the potential to overshadow importance of windthrow could offset any decreases due to an direct effects of global warming on species dis- altered fire regime. Changes in climate and dis- tribution, migration, substitution and extinction. turbance regimes may lead to assemblages of Thus, fire is a catalyst for vegetation change. species that have never been encountered before Ž.Martin, 1993 . The various aspects of the fire regime can have significant influences on successional pathways as 4. Conclusions discussed above. Changes in climate and the fire regime will have an affect on carbon, nitrogen cycling and budgets. Disturbances such as fire The two GCMs in this study suggest increases could be critical factors in determining if US in SSR of 10᎐50% across much of the United forests are a carbon sink or source on a year-to- States by 2060Ž. Figs. 1᎐4 . The impact of climate year basis. The very close coupling of nitrogen change on the fire regime in the United States and carbon cycles within the plant and the ecosys- could have an almost immediate and significant tem as a whole makes them particularly suscepti- impact on ecosystems, due to likely increases in ble to modifications under global change as one area burned and fire intensityrseverity. Possible or the other cycle may be altered by elevated CO2 and climate change. Alteration in one of these manipulations of fuel type, load and arrangement two cycles can be expected to have immediate could be used to help protect local areas of high Ž. repercussions on the other because of the inter- value Pyne et al. 1996 . At the larger scale, action and feedback between the twoŽ Pastor and however, fuel management would not be feasible. Post, 1986; Reynolds et al., 1996. . For example, A fine balancing act is required by land managers the ability for increased carbon acquisition by to protect valuesŽ. people and resources from fire while at the same time allowing fire to resume plants in a higher CO2 atmosphere could be limited by available soil nitrogenŽ. N , which is in more of its natural role in ecosystem functioning turn controlled by decomposition rates. The main and maintenance. Fire may be becoming increas- avenue for interaction between carbonŽ. C and N ingly important in terms of determining whether cycles actually may be via decomposition and American forests are a carbon sink. Additionally, litter quality. The reciprocal linkage between in regions where there is potential for fires to ecosystem cyclesŽ. N and C and attributes Ž de- increase, the result may be competition between composition rates and litter quality. assumes the forest industry and fire for wood fiber. Con- added importance in northern and boreal forest tinued research is required to further explore the for several reasons:Ž. 1 greater tempera- relationships between the climate᎐fire association ture impacts are predicted for northern latitudes as well as the interaction between fire and other under climate change, affecting all temperature- disturbances and their impacts on vegetation. 228 M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229

Acknowledgements regime. In: Trabaud L, editor. The role of fire in ecological systems. The Hague: SPB Academic Publishing, 1987:49᎐63. Martin P. Vegetation responses and feedbacks to climate: a The HadCM2 data have been supplied by the review of models and processes. Clim Dynam 1993;8: Climate Impacts LINK ProjectŽ Department of 201᎐210. the Environment Contract EPG 1r1r16. on be- McCullough DG, Werner RA, Neumann D. Fire and insects half of the Hadley Centre and the UK Meteoro- in northern and boreal forest ecosystems of North Amer- ica. Ann Rev Entomol 1998;43:107᎐127. logical Office. The CCC GCM data have been Mearns LO, Schneider SH, Thompson SL, McDaniel LR. supplied by the Canadian Centre for Climate Climate variability statistics from general circulation mod- Modelling and Analysis of Environment Canada. els as applied to climate change analysis. In: Malanson GP, editor. Natural areas facing climate change. The Hague: SPB Academic Publishing, 1989:51᎐73. References Overpeck JT, Rind D, Goldberg R. Climate-induced changes in forest disturbance and vegetation. Nature 1990;343: Anderson JM. Response of soils to climate change. Adv Ecol 51᎐53. Res 1992;22:163᎐210. Pastor J, Post WM. Influence of climate, soil moisture, and Baskin Y. Yellowstone fires: a decade later ᎏ ecological succession on forest carbon and nitrogen cycles. Bio- lessons learned in the wake of the conflagration. Bioscience geochemistry 1986;2:3᎐27. ᎐ = 1999;49:93 97. Price C, Rind D. The impact of a 2 CO2 climate on light- Bergeron Y, Richard JH, Christopher C, Gauthier S, Flanni- ning-caused fires. J Clim 1994;7:1484᎐1494. gan M, Prairie YT. Variability in fire frequency and forest Pyne SJ, Andrews PL, Laven RD. Introduction to wildland composition in Canada’s southeastern boreal forest: a chal- fire. 2 New York, NY: John Wiley and Sons, 1996:769. lenge for sustainable forest management. Conserv Ecol Reynolds JF, Kemp PR, Acock B, Chen J-L, Moorhead DL. wx 1998; online 2Ž. 2 : Available from the Internet. URL: Progress, limitations, and challenges in modeling the ef- http:rrwww.consecol.orgrvol2riss2rart6 fects of elevated CO2 on plants and ecosystems. In: Mooney Couzin J. Landscape changes make regional climate run hot HA, editor. Carbon dioxide and terrestrial ecosystems. San and cold. Science 1999;283:317᎐318. Diego: Academic Press, Inc, 1996:347᎐380. Flannigan MD, Harrington JB. A study of the relation of Schmoldt DL, Peterson D., Keane RE et al. Assessing the meteorological variables to monthly provincial area burned effects of fire disturbance on ecosystems: a scientific agenda ᎐ by wildfire in Canada 1953 80. J Appl Meteorol 1988;27: for research and management. Gen Tech Rep GTR- PNW ᎐ 441 452. 455 US Dep Agric, For Serv, Pac Northwest Res Stn, 1999, Flannigan MD, Bergeron Y, Engelmark O, Wotton BM. Fu- 104 pp. ture wildfire in circumboreal forests in relation to global Skinner WR, Stocks J, Martell DL, Bonsal B, Shabbar A. The ᎐ warming. J Veg Sci 1998;9:469 476. association between circulation anomalies in the mid- Giorgi F. Simulation of regional climate using a limited area troposphere and area burned by wildland fire in Canada. model nested in a general circulation model. J Clim Theor Appl Clim 1999;63:89᎐105. 1990;3:941᎐963. Solomon AM, Leemans R. Boreal forest carbon stocks and Heinselman ML. Fire in the virgin forests of the boundary wood supply: past, present and future responses to chang- waters canoe area, Minnesota. Q Res 1973;3:329᎐382. ing climate, agriculture and species availability. Agric For Intergovernmental Panel on Climate ChangeŽ. IPCC . Climate Meteorol 1997;84:137᎐151. change 1995 impacts, adaptations and mitigation of climate Strauss D, Bednar L, Mees R. Do one percent of the forest change: scientific᎐technical analyses. Cambridge: Cam- bridge Univ Press, 1996:878. fires cause ninety-nine percent of the damage? For Sci ᎐ Johnson EA. Fire and vegetation dynamics: studies from the 1989;35:319 328. North American boreal forest. Cambridge: Cambridge Uni- Stocks BJ. Fire potential in the spruce-budworm damaged ᎐ versity Press, 1992:129. forests of Ontario. For Chron 1987;63:8 14. Leenhouts B. Assessment of biomass burning in the contermi- Stocks BJ, Fosberg MA, Lynham TJ et al. Climate change and nous United States. Conserv Ecol 1998;wx online 2:Ž. 1 Avail- forest fire potential in Russian and Canadian boreal forests. ᎐ able on the Internet URL:rrwww.consecol.orgrvol2r Clim Change 1998;38:1 13. iss1rart1 Swetnam TW. Fire history and climate change in giant se- Lupo AR, Oglesby RJ, Mokhov II. Climatological features of quoia groves. Science 1993;262:885᎐889. blocking anticyclones: a study of Northern Hemisphere Swetnam TW, Betancourt JL. Mesoscale disturbance and eco- CCM1 model blocking events in present-day and couble logical response to decadal climatic variability in the Amer- CO2 concentration atmosphere. Clim Dynam 1997; ican southwest. J Clim 1998;11:3128᎐3147. 13:181᎐195. Tackle ES, Bramer DJ, Heilman WE, Thompson MR. A Malanson GP. Diversity, stability, and resilience: effects of fire synoptic climatology for forest fires in the NE US and M.D. Flannigan et al. rThe Science of the Total En¨ironment 262() 2000 221᎐229 229

future implications from GCM simulations. Int J Wildland Weber MG, Flannigan MD. Canadian boreal forest ecosystem Fire 1994;4:217᎐224. structure and function in a changing climate: impact on fire USDA Forest Service. 1991᎐1997 Wildland fire statistics. US regimes. Environ Rev 1997;5:145᎐166. Dep Agric, For Serv, Fire and Aviation Management, Whelan RJ. The ecology of fire. Cambridge: Cambridge Univ 1998:141. Press, 1995:346. Van Cleve K, Chapin FS III, Flanagan PW, Viereck LA, Wotton BM, Flannigan MD. Length of the fire season in a Dyrness CT. Forest ecosystems in the Alaskan taiga. Eco- changing climate. For Chron 1993;69:187᎐192. logical Studies 57. New York: Springer-Verlag, 1986:230. Wright HA, Bailey AW. : United States and Van Wagner CE. A comparison of the Canadian and Ameri- southern Canada. New York, NY: John Wiley and Sons, can forest fire danger rating systems. Can For Serv, Info 1982:501. Rep PS-X-59. 1975 Van Wagner CE. The development and structure of the Yarie J. Forest fire cycles and life tables: a case study from ᎐ Canadian forest fire weather index system. Can For Serv, interior Alaska. Can J For Res 1981;11:554 562. For Tech Rep 35. Ottawa, Ontario, 1987