499© The Ecological Society of Americawwwwww..ffrroonnttiieerrssiinneeccoollooggyy..oorrggForest fires have swept into public and policy aware-ness over the past several decades, with an increase inthe frequency of large fires in western North America(Westerling et al. 2006). At the same time, human settle-ments and other infrastructure are impinging on the wild-land interface at an unprecedented rate, resulting inannual US federal suppression costs exceeding $1 billionin 3 of the past 6 years (OIG 2006). These recent, largefires have been attributed to a variety of processes, fromlocal increases in fuel loads and shifts in stand composi-tion to regional and global changes in climate(Schoennagel et al. 2004). In some regions, such as thesouthwestern US, historical fire suppression has beenblamed for increases in stand density and fire severity(Allen et al. 2002). In contrast, climate change explainsrecent patterns in fire occurrence over broad areas ofwestern North America (Fauria and Johnson 2006;Trouet et al. 2006). The few forecasts that exist indicatethat the trend toward increased incidence of fires willcontinue (eg McKenzie et al. 2004; Flannigan et al. 2005).To place modern fire processes in a meaningful context,scientists and policy makers need a long-term view of firevariability. Fossil records describing past ecosystems(paleoecological records) can quantify the historicalrange of variability of fire occurrence. They can thereforeprovide an important reference for ecosystem-basedstrategies aimed at maintaining ecological processes,habitats, and species (eg Willis and Birks 2006; Figure 1).It should be made clear, however, that application ofpaleoecological data to forecasting the future is compli-cated by the fact that the future may not resemble anytime in the past (Jackson and Williams 2004), particu-larly with respect to the climatic and fuel controls of fire.While this may limit the potential for the past to serve asan analog for the future, there is an important need for amechanistic understanding of the processes that pro-PALEOECOLOGY PALEOECOLOGY PALEOECOLOGYForest fire and climate change in westernNorth America: insights from sedimentcharcoal recordsDDaanniieell GG GGaavviinn11**,, DDoouuggllaass JJ HHaalllleetttt22,, FFeenngg SShheenngg HHuu33,, KKeennnneetthh PP LLeerrttzzmmaann44,, SSuussaann JJ PPrriicchhaarrdd55,,KKeennddrriicckk JJ BBrroowwnn66,,77,, JJaassoonn AA LLyynncchh88,, PPaattrriicckk BBaarrttlleeiinn11,, aanndd DDaavviidd LL PPeetteerrssoonn99Millennial-scale records of forest fire provide important baseline information for ecosystem management, especiallyin regions with too few recent fires to describe the historical range of variability. Charcoal records from lake sedi-ments and soil profiles are well suited for reconstructing the incidence of past fire and its relationship to changingclimate and vegetation. We highlight several records from western North America and their relevance in recon-structing historical forest dynamics, fire–climate relationships, and feedbacks between vegetation and fire under cli-mate change. Climatic effects on fire regimes are evident in many regions, but comparisons of paleo-fire recordssometimes show a lack of synchrony, indicating that local factors substantially affect fire occurrence, even over longperiods. Furthermore, the specific impacts of vegetation change on fire regimes vary among regions with differentvegetation histories. By documenting the effects on fire patterns of major changes in climate and vegetation, paleo-fire records can be used to test the mechanistic models required for the prediction of future variations in fire.Front Ecol Environ 2007; 5(9): 499–506, doi:10.1890/060161IInn aa nnuuttsshheellll::• Paleoecological records from western North America revealthat forest fire frequency has varied continually over the pastseveral millennia• Fire histories are sometimes only marginally synchronousamong similar sites, complicating their direct connection toclimate change• These records also show how changing composition of forestsaffects their susceptibility to fire, sometimes overriding the roleof climate• Understanding large-scale controls of fire regimes is highly rel-evant to creating fire policy in the context of anticipated futurechanges in climate and fuels1Department of Geography, University of Oregon, Eugene, OR 97403*([email protected]); 2Department of Geography and the School ofEnvironmental Studies, Queen’s University, Kingston, ON, CanadaK7L 3N6; 3Department of Plant Biology, University of Illinois,Urbana, IL 61801; 4School of Resource and EnvironmentalManagement, Simon Fraser University, Burnaby, BC, Canada V5A1S6; 5College of Forest Resources, University of Washington, Seattle,WA 98195; 6Geological Survey of Denmark and Greenland,Copenhagen K, Denmark; 7Royal British Columbia Museum,Victoria, BC, Canada V8W 9W2; 8US Environmental ProtectionAgency, Clean Air Markets Division, Washington, DC 20460;9USDA Forest Service, Pacific Wildland Fire Sciences Lab, Seattle,WA 98103Forest fire and climate change DG Gavin et al.500wwwwww..ffrroonnttiieerrssiinneeccoollooggyy..oorrgg ©© The Ecological Society of Americaduced the variability observed in paleoecological records(Flessa et al. 2005).Paleo-records can provide information on the responseof fire and vegetation through periods of substantial cli-mate change and thus help to characterize the controls offire occurrence over long periods (Clark et al. 1996). Suchrecords offer spatial and temporal perspectives that are suit-able for multi-scaled analyses; these can be used to identifywhich processes regulating fire regimes are scale-invariant(ie operating at all times and within small or large areas)and which are scale-contingent (ie operating only at onetemporal and spatial scale; Figure 2). For example, will thelarge-scale fire–climate relationships of the past 30 years(eg Westerling et al. 2006) extrapolate to longer periodsthat encompass large-scale changes in climate and forestcomposition? Similarly, will these relationships scale downto smaller areas that vary greatly with respect to fuel loadsand probability of ignition? Answers to these questions areparticularly useful for placing modern fire–climate rela-tionships into a context that is relevant for local land man-agers. The perspectives from paleo-records are also espe-cially helpful for regions where fires recur at intervals