Climate

Both oxygen and heat become elements of climate in the context of changing fire regimes. Changes in the composition of Earth's atmosphere provide a key constraint on wildfire prevalence. Atmospheric oxygen (O₂), in particular, can influence wildfire wherever fuel and ignition sources are present (Belcher et al., Reference Belcher, Yearsley, Hadden, McElwain and Rein2010). Oxygen levels today are 21%, but below 18.5% O₂, modeling and experiments indicate that fire activity would be greatly suppressed. Below 15% O₂, fire is unable to ignite or propagate (Belcher and McElwain, Reference Belcher and McElwain2008). In contrast, oxygen levels between 19-22% would rapidly enhance wildfire spread, while levels above 35% would cause wildfires to become so hot that they would consume most terrestrial vegetation (Scott, 2000; Lenton, 2001). Geological evidence suggests that wildfire activity first occurred during the Silurian (ca. 420 mya) and the Carboniferous (ca. 400 mya) periods (Scott and Jones, Reference Scott and Jones1994; Glasspool et al., Reference Glasspool, Edwards and Axe2004). Fires were also prevalent during the Permian, ca. 250–300 mya (Scott and Glasspool, Reference Scott and Glasspool2006; Yan et al., Reference Yan, Wan, He, Hou and Wang2016), when oxygen levels were higher than today.

Oxygen also affects fire locally through winds, which supply it and can make fires burn more rapidly. But when considering fire regimes, the influence of wind is through its effects on moisture and fire behavior rather than just oxygen, and thus winds become a function of changing climate. The absence of direct evidence for changing wind patterns over millennia, however, effectively places wind beyond the purview of paleofire studies. Large changes in atmospheric circulation that altered patterns of wind direction and speed would have affected fire regimes by influencing vegetation types and migration and impacting important fire-regime characteristics such as shifting typical fires from low to high-severity in a given region. Foehn winds, for example, play an important role in shaping fire regimes in mountainous areas, so as topography changes over long time scales their effects could become more important (Zumbrunnen et al., Reference Zumbrunnen, Bugmann, Conedera and Bürgi2009). Large shifts in atmospheric circulation during deglaciation may have had substantial impacts on dominant wind flows along ice sheet margins, for example, which in turn could have influenced fire-regime shifts by drying out vegetation or altering patterns of fire ignition and spread. Unusually high fire activity from 10,000-9,000 cal yr BP in the St. Lawrence region of North America could be attributable to a shift in atmospheric circulation (e.g., jet stream position or strength and associated pressure gradients and storm trajectories) as the Laurentide ice sheet was retreating (Carcaillet and Richard, Reference Carcaillet and Richard2000), for example, but only model experiments could disentangle such forces from other climate and vegetation shifts at that time. The importance of oxygen for a fire is thus analogous to the importance of climate for influencing fire regimes.

Heat represents one side of the original fire triangle, but in the context of fire regimes, heat can be considered in relation to both ignitions and changing climate conditions. The heat required to ignite fires has come primarily through lightning and volcanic activity for most of Earth's history, with lightning of primary interest during the Quaternary. Lightning strikes are abundant on Earth, occurring an estimated 44 ± 5 flashes per second on average, with about 78% of those located between 30°S and 30°N latitudes (Christian et al., Reference Christian, Blakeslee, Boccippio, Boeck, Buechler, Driscoll, Goodman, Hall, Koshak, Mach and Stewart2003). Lightning is particularly important at mid and high latitudes, and especially in boreal forests (Veraverbeke et al., Reference Veraverbeke, Rogers, Goulden, Jandt, Miller, Wiggins and Randerson2017). However, many wildfires today globally are ignited by humans, especially in fire-prone environments (Ganteaume et al., Reference Ganteaume, Camia, Jappiot, San-Miguel-Ayanz, Long-Fournel and Lampin2013; Guo et al., Reference Guo, Innes, Wang, Ma, Sun, Hu and Su2015; Syphard and Keeley, Reference Syphard and Keeley2015). The history of human fire use and how its importance has grown over time are contested (Bowman et al., Reference Bowman, Balch, Artaxo, Bond, Cochrane, D'antonio, DeFries, Johnston, Keeley, Krawchuk and Kull2011; Williams et al., Reference Williams, Mooney, Sisson and Marlon2015; Oswald et al., Reference Oswald, Foster, Shuman, Chilton, Doucette and Duranleau2020). Bowman et al. (Reference Bowman, Balch, Artaxo, Bond, Carlson, Cochrane, D'Antonio, DeFries, Doyle, Harrison and Johnston2009) suggest that human mastery of fire may have begun with the expansion of tropical savanna biomes 7–8 mya. Archaeological evidence suggests that early hominins moved into northern latitudes from Africa without the habitual use of fire, and that it was only after ~300,000 to 400,000 years ago that fire became a significant part of the hominin technological repertoire (Karkanas, Reference Karkanas, Shahack-Gross, Ayalon, Bar-Matthews, Barkai, Frumkin, Gopher and Stiner2007; Roebroeks and Villa, Reference Roebroeks and Villa2011). The subsequent human use and importance of fire has shifted in association with cultural and technological developments, first from domestic uses over hundreds of thousands of years, then expanding to foraging uses over tens of thousands of years, to agricultural fire during the Holocene, and finally to industrial uses during the past few centuries (Bowman et al., Reference Bowman, Balch, Artaxo, Bond, Carlson, Cochrane, D'Antonio, DeFries, Doyle, Harrison and Johnston2009). Disentangling the relative influence of lightning versus these human-started fires has been a primary focal area of paleofire research and is explored further below.

A focus on changing fire regimes requires consideration of climate processes affecting heat at broad spatial and long temporal scales. Changes in radiative forcing, for example, have important consequences for fire regimes over millennial scales (Daniau et al., Reference Daniau, Bartlein, Harrison, Prentice, Brewer, Friedlingstein, Harrison-Prentice, Inoue, Izumi, Marlon and Mooney2012), while shifting patterns of atmospheric circulation that determine land and ocean surface temperatures, precipitation, and the distribution of snow and ice (Bartlein, Reference Bartlein, Hostetler and Alder2014; Vanniere et al., Reference Vannière, Colombaroli, Chapron, Leroux, Tinner and Magny2008). Fischer et al. (Reference Fischer, Meissner, Mix, Abram, Austermann, Brovkin, Capron, Colombaroli, Daniau, Dyez, Felis, Finkelstein, Jaccard, McClymont, Rovere, Sutter, Wolff, Affolter, Bakker, Ballesteros-Cánovas, Barbante, Caley, Carlson, Churakova, Cortese, Cumming, Davis, Vernal, Emile-Geay, Fritz, Gierz, Gottschalk, Holloway, Joos, Kucera, Loutre, Lunt, Marcisz, Marlon, Martinez, Masson-Delmotte, Nehrbass-Ahles, Otto-Bliesner, Raible, Risebrobakken, Sánchez Goñi, Saleem Arrigo, Sarnthein, Stocker, Velasquez Alvárez, Tinner, Vogel, Wanner, Yan, Yu, Ziegler and Zhou2018) examine extensive paleoenvironmental data from the past 3.5 million years, including paleofire records, to identify the effects of warming on ecosystems analogous to what models predict for the coming century. They note that late-successional evergreen vegetation in Mediterranean areas was replaced by flammable, drought-adapted vegetation in response to climate warming of just 1-2°C in the past when drought thresholds are reached. Likewise, regional hydroclimatic fluctuations, can push tropical forests into fire-dominated savanna (Hirota et al., Reference Hirota, Holmgren, Van Nes and Scheffer2011), and grasslands into desert, as evidenced by the Green-Sahara—desert transition that occurred at the end of the African Humid Period (Kröpelin et al., Reference Kröpelin, Verschuren, Lézine, Eggermont, Cocquyt, Francus, Cazet, Fagot, Rumes, Russell and Darius2008). Such shifts are linked to temperature changes, such as those caused by variations in summer insolation that affect monsoons and other circulation patterns. Small changes in global mean temperature can therefore have enormous impacts on vegetation and fire regimes. The role of heat as a function of climate changes more broadly and their effects on fire are explored in more detail below.

Vegetation

Once ignited, fire requires fuel to sustain combustion. Vegetation serves as fuel for fires when it is abundant and dry; it can also prevent, reduce, or extinguish fire when it is sparse and/or wet. Vegetation characteristics thus shape the existence and behavior of fire, and changes in these characteristics can be both a cause and effect of changes in a fire regime over time. Evidence for such interactions between fire and vegetation are as old as the evidence for fire itself. Charcoal from the Late Devonian document burning when plants first began spreading beyond wetlands, with new plant types developing that could survive on dry land (Scott, 2000, 2001). Peaks of charcoal in marine sediments around 7–8 mya indicate that fire increased when grassland and savanna biomes expanded globally, suggesting the establishment of sustained fire-vegetation-climate feedback cycles in these environments (Keeley and Rundel, Reference Keeley and Rundel2005). In Australia, fire became progressively more important with the development of sclerophyll vegetation in particular (Martin, 2006; Lynch et al., Reference Lynch, Beringer, Kershaw, Marshall, Mooney, Tapper, Turney and Van Der Kaars2007). Evidence from the late Cenozoic shows increasing accumulations of microcharcoal associated with aridity and the expansion of steppe taxa in which fire easily propagates (Miao et al., Reference Miao, Fang, Song, Yan, Zhang, Meng, Li, Wu, Yang, Kang and Wang2016). Paleofire records from modern grasslands are sparse, and as the boundaries and ecological communities within biomes shift over time, even defining where and when grasslands existed in the past in order to analyze them is challenging (Leys et al., Reference Leys, Marlon, Umbanhowar and Vanniere2018). Nevertheless, the existence of fire-dependent ecosystems demonstrates that fire regimes are not only controlled by fuel biomass, but are themselves a major evolutionary force that affects vegetation structure and composition, and thus species interactions (Bond et al., 2005; Gill et al., Reference Gill, Williams, Jackson, Lininger and Robinson2009; Feurdean and Vasiliev, Reference Feurdean and Vasiliev2019).

Today, about 18 percent of all anthropogenic carbon dioxide emissions come from biomass burning over hundreds of millions of hectares, primarily in the woody and grassland savannas of Africa and Australia (Jacobson, Reference Jacobson2014). Yet, fire has a long history in most places on Earth, even in rainforests and on islands once thought to be fire-free (Burney, Reference Burney1987; Horn et al., Reference Horn, Orvis, Kennedy and Clark2000; Hope et al., Reference Hope, Stevenson, Southern, Clark and Anderson2009; Whitlock et al. Reference Whitlock, McWethy, Tepley, Veblen, Holz, McGlone, Perry, Wilmshurst and Wood2015). Remotely-sensed fire data and analyses show that fire in low-productivity ecosystems is limited by low biomass, while in high-productivity ecosystems, it is more sensitive to high temperatures (Pausas and Ribeiro, Reference Pausas and Ribeiro2013).

Vegetation characteristics, often associated with specific climates, can also inhibit fire, such as when fuels become too sparse to carry fire, or their structure inhibits fire spread. Analyses of pollen, charcoal, and phytolith assemblages from Ounjougou, Mali offer one example. In the early Holocene, expansive tropical grasslands and gallery forests with low biomass production supported low levels of fire activity. Increasing rainfall and greater seasonality led to higher biomass production, however, and increased charcoal production with higher inferred fire activity (Neumann et al., Reference Neumann, Fahmy, Lespez, Ballouche and Huysecom2009). Shifts in forest species composition, such as a reduction in conifers and an increase in broadleaf species, can also reduce fire activity by decreasing fuel flammability (Blarquez et al., Reference Blarquez, Ali, Girardin, Grondin, Fréchette, Bergeron and Hély2015; Feurdean et al., Reference Feurdean, Veski, Florescu, Vannière, Pfeiffer, O'Hara, Stivrins, Amon, Heinsalu, Vassiljev and Hickler2017). Even different conifer species with different structural forms can affect the continuity of fuels and thus their flammability. Increased burning in boreal forests are evident, for example, where Populus and Picea glauca-dominated forests are replaced by highly flammable Picea mariana forests (Hu et al., Reference Hu, Brubaker, Gavin, Higuera, Lynch, Rupp and Tinner2006; Higuera et al., Reference Higuera, Brubaker, Anderson, Hu and Brown2009). One unusually high-resolution record from Kettle Lake in the Great Plains of North America reveals a tight coupling between high fire activity during moist intervals when grass cover was extensive and fuel loads were high versus intervals with low fire activity when fuels loads decreased as a result of greater aridity (Brown et al., Reference Brown, Clark, Grimm, Donovan, Mueller, Hansen and Stefanova2005). Such examples highlight the interplay of vegetation abundance, flammability, effective moisture, and fire over time—complex dynamics that can drive substantial ecosystem change even in the absence of land-use changes or human fire use.

Human activities

The routine human use of fire for foraging and hunting has shaped many environments for tens of thousands of years (Pyne, Reference Pyne1997, Bowman et al., 2011). Paleofire data from large islands provide some of the strongest evidence linking human activities to widespread fires because fire was often extremely limited or absent prior to the arrival of humans. The establishment of cultivation at 5 ka is coincident with a large increase in fire in New Guinea, for example, and also with local evidence of anthropogenic fire use on the island (Haberle et al., Reference Haberle, Hope and Defretes1991; Haberle and David, Reference Haberle and David2004). Likewise, McWethy et al. (Reference McWethy, Whitlock, Wilmshurst, McGlone, Fromont, Li, Dieffenbacher-Krall, Hobbs, Fritz and Cook2010) have shown that a dramatic increase in burning occurred within 200 years of Maori arrival at all but the wettest sites in New Zealand.

Ample evidence of the growing influence of human use of fire for agriculture comes from China. The gradual expansion of intensive land use in China are recorded in black carbon mass sedimentation rates (Wang et al., Reference Wang, Xiao, Cui and Ding2013), and black carbon in a sediment core from Daihai Lake, Inner Mongolia, suggest the appearance of early agriculture and the expansion of human land use at ca 8000 and 3000 ka, respectively (Han et al., Reference Han, Marlon, Cao, Jin and An2012). In Turkey, sediment records of fire, vegetation, and geochemistry from the Bereket basin in the Taurus mountains during the past three millennia detail human-induced change in the area. In the Bereket surroundings, fires appear to have catalyzed a simplification of the vegetation structure and associated increases in soil erosion, pasture expansion, and intensive cultivation (Kaniewski et al., Reference Kaniewski, Paulissen, De Laet and Waelkens2008). Leunda et al. (Reference Leunda, Gil-Romera, Daniau, Benito and González-Sampériz2020) use paleofire data from the Central Pyrenees in Spain to show the transition from climate-driven fire regimes in the early-to-mid-Holocene to human-driven landscape change over the last 3700 years. Many similar examples exist throughout Europe. While there are almost no paleofire records from India, one short sediment core from a reservoir in northern Karnataka, indicates well-defined periods of burning that coincide with periods of open vegetation as defined by pollen analysis, and with periods of high settlement densities and intensive land use evidenced by archaeological data c. AD 1300-1600 (Morrison et al., Reference Morrison1994).

Unless paleofire records are collected adjacent to archaeological sites, which is rare, is it difficult to know whether shifts in fire regimes are the result of changes in climate, vegetation, cultural practices, or some combination of two or more of these factors. Recent methodological advances are improving our ability to clearly identify the advent of human impacts on fire on millennial times scales, however. Research by Vachula et al. (Reference Vachula, Huang, Longo, Dee, Daniels and Russell2019) provide one example in a study that used organic geochemical biomarkers, including polycyclic aromatic hydrocarbons (PAHs) and sterols, to reconstruct fire activity in Beringia and human presence over the past 32,000 years. Understanding human use of fire in Beringia can shed light on the timing and pathway of human arrival to the Americas, which remains an important topic of debate in archaeology and anthropology. The debate centers on whether a “swift peopling” of the Americas from Asia occurred during the last Glacial termination via the Bering Land Bridge, or whether the eventual spread of populations from Beringia across the American continents were delayed by thousands of years.

Vachula et al. (Reference Vachula, Huang, Longo, Dee, Daniels and Russell2019) found that elevated levels of PAHs from 19-32kyr were coincident with coprostanol levels consistent with human presence. Coprostanol:stigmastanol ratios have been used to distinguish human and ruminant herbivore fecal inputs to lake sediments in Norway (D'Anjou et al., Reference D'Anjou, Bradley, Balascio and Finkelstein2012), and given the known coprolite sterol compositions of mammoth, which comprised about half the glacial animal biomass of northern Alaska (Mann et al., Reference Mann, Groves, Kunz, Reanier and Gaglioti2013), these data suggest increased burning during the glacial period in Beringia may have been due to human activities. Whether the fire and sterol data provide evidence of temporary or permanent residence in the area is unclear, but multiple spikes in the coprostanol:stigmastanol ratio during the interval from 32–19ka suggest that human populations may have inhabited Beringia much earlier than conventionally thought.

Matching fecal biomarkers with the timing of human presence and charcoal was used as evidence for human-lit fires in New Zealand at c. AD 1350, confirming the role of humans as the cause of dramatic deforestation at that time (Argiriadis et al., Reference Argiriadis, Battistel, McWethy, Vecchiato, Kirchgeorg, Kehrwald, Whitlock, Wilmshurst and Barbante2018). Other research by Brittingham et al. (Reference Brittingham, Hren, Hartman, Wilkinson, Mallol, Gasparyan and Adler2019) exploits the differences in the types of PAHs found in widely-dispersed wildfire emissions versus those found in particulate emissions of burned wood from Lusakert Cave in Armenia. The authors use their data to argue that Middle Paleolithic hominins (ca. 300,000 to 35,000 years ago) were able to control fire and utilize it regardless of the variability of fires in the environment, providing support for the intentional creation and control of fire by hominins independent of exploitation of wildfires in Eurasia. Such studies illustrate the potential for progress in advancing our understanding of how humans have shaped fire history and environmental change through time as our fire proxy repertoire expands.

Fire regimes track climate changes

Fire regimes not only track climate—they are finely tuned to them. On the broadest scales of Earth history, a warmer planet means a more fire-prone planet. About sixty-five million years ago, during the Paleocene-Eocene thermal maximum, global temperatures increased more than 5–8°C. The interval of warming was associated with a massive release of carbon to the atmosphere from volcanism, which lasted tens of thousands of years. The warming was associated with major increases in global biomass burning (Collinson et al., Reference Collinson, Steart, Scott, Glasspool and Hooker2007; Marynowski and Simoneit, Reference Marynowski and Simoneit2009). In contrast, the cold climate conditions that characterized the Last Glacial Maximum, when temperatures were about 6.0°C colder than today and 25% of the land surface was covered with ice, are associated with reduced biomass burning at the global scale (Power et al., Reference Power, Marlon, Ortiz, Bartlein, Harrison, Mayle, Ballouche, Bradshaw, Carcaillet, Cordova and Mooney2008).

Paleofire literature is replete with examples of how fire regimes have responded to the powerful, mostly gradual effects of the hierarchical controls that drive Earth's climate system (Bartlein et al., 2014). The paleoecological record from Cygnet Lake in Yellowstone National Park in the United States exemplifies the effects of gradual insolation changes on forest fires. Consistently surrounded by lodgepole pine forest for more than 11,000 years, the detailed environmental reconstruction shows how fire increased in direct response to increased summer warmth and dryness during the peak of northern mid-latitude summer insolation in the early Holocene (Millspaugh et al., Reference Millspaugh, Whitlock and Bartlein2000). As the Holocene progressed, summer insolation declined and fires gradually became less frequent.

Cygnet Lake's setting is unique, which made it possible to link fire directly to climate. The vegetation on the plateau where the lake sits reflects the influence of infertile rhyolite soils, which limit the establishment and growth of most other conifer species. This unique physical constraint limited the ability of vegetation around the lake to respond to climate changes (Whitlock, Reference Whitlock1993). In most places, vegetation change is also highly responsive to climate changes, interacting with fire regime characteristics including size, frequency, and timing, to catalyze additional change and ecosystem reorganization (Colombaroli et al., Reference Colombaroli, Vannière, Emmanuel, Magny and Tinner2008).

A 170,000-year marine sediment core off the coast of Namibia in southern Africa reveals the interaction of fire and fuels synchronized by orbital variations in insolation (Daniau et al., Reference Daniau, Goñi, Martinez, Urrego, Bout-Roumazeilles, Desprat and Marlon2013). Six fire cycles occurred during the long record that correspond both in timing and magnitude to the precessional forcing of north–south shifts in the Intertropical Convergence Zone. Fire increased with wetter and cooler climate conditions owing to a shift in rainfall amount and seasonality. The increased fire thus occurred in response to an increase in biomass produced by reliable rains and coupled with regular dry intervals that increased grass flammability, revealed through measurements of the elongation (length/width ratio) of microcharcoal particles.

Two syntheses of paleofire records from the GCD further demonstrate the sensitivity of fire regimes to temperature and precipitation over long time scales. In the first, 67 sites paleofire records from the Last Glacial Maximum showed that fire was consistently lower during the glacial period than during the Eemian and Holocene (Daniau et al., Reference Daniau, Harrison and Bartlein2010). Most records, particularly from the northern extra-tropics, show millennial-scale variability in fire regimes corresponding to the rapid climate changes associated with Dansgaard–Oeschger (D-O) cycles. Fire increased during D-O warming events and decreased during intervals of rapid cooling. A high-resolution wildfire record from Greenland ice cores shows similar results (Fischer et al., Reference Fischer, Schüpbach, Gfeller, Bigler, Röthlisberger, Erhardt, Stocker, Mulvaney and Wolff2015). The paleofire records indicate that strong climatic variability evident during the glacial period resulted in substantial changes in fire regimes even though biomass burning was less than today. A second synthesis using 679 paleofire records from the GCD show that the changes in fire regime over the past 21,000 years are predictable from changes in regional climates (Daniau et al., Reference Daniau, Bartlein, Harrison, Prentice, Brewer, Friedlingstein, Harrison-Prentice, Inoue, Izumi, Marlon and Mooney2012). Fire increases monotonically with changes in temperature and peaks at intermediate moisture levels, and temperature is the most important driver of changes in biomass burning over the past 21,000 years. Increasing fire with increasing temperature and peak fire activity at intermediate moisture levels is also evident in modern remotely-sensed area burned data (Daniau et al., Reference Daniau, Bartlein, Harrison, Prentice, Brewer, Friedlingstein, Harrison-Prentice, Inoue, Izumi, Marlon and Mooney2012; Bistinas et al., Reference Bistinas, Harrison, Prentice and Pereira2014).

In areas where fuels are limiting, wet/dry cycles are more closely associated with high fire activity than temperature, whether those cycles occur on a monthly basis or a millennial basis. Hao et al. (Reference Hao, Han, An and Burr2020) examined a 32,000-year sediment core from Qinghai Lake, the largest in China, and found that more wildfires occurred during the cold-dry late glacial period (32.8-11.7 ka), and fewer wildfires during the warm-humid Holocene. The study points to drought as the primary control of fire regimes on orbital time-scales in the region when lack of fuel biomass and continuity limited fire spread.

Not all climate changes, of course, are gradual. Evidence of the sensitivity of fire regimes to rapid shifts in temperature and precipitation is found even since the late glacial period. Fire across North America increased at both the beginning and end of the Younger Dryas Chronozone, for example (Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012). Fire regimes and vegetation in Europe also show a swift response to the 8.2 event (Davis and Stevensen, Reference Davis and Stevenson2007; Tinner et al., Reference Tinner, van Leeuwen, Colombaroli, Vescovi, van der Knaap, Henne, Pasta, D'Angelo and La Mantia2009). Regionally-specific rapid climate shifts are evidenced by paleofire records from central Europe (Finsinger and Tinner, Reference Finsinger and Tinner2007) and fire-prone communities of the Mediterranean (Linstädter and Zielhofer, Reference Linstädter and Zielhofer2010), from the tropical savannas of Africa (Colombaroli et al., Reference Colombaroli, van der Plas, Rucina and Verschuren2018), the shifting steppe-forest ecotone of southern South America (Iglesias et al., Reference Iglesias, Whitlock, Markgraf and Bianchi2014) and elsewhere. In the late Holocene, hundreds of paleofire studies show evidence of increased burning during the Medieval Climate Anomaly and decreased burning during the Little Ice Age, reflected in global and regional compilations of paleofire records (Marlon et al., Reference Marlon, Bartlein, Carcaillet, Gavin, Harrison, Higuera, Joos, Power and Prentice2008; Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012; Power et al., Reference Power, Mayle, Bartlein, Marlon, Anderson, Behling, Brown, Carcaillet, Colombaroli, Gavin and Hallett2013; Calder et al., Reference Calder, Stefanova and Shuman2019).

The strong evidence for climatic controls of fire regimes during Earth's geological past and throughout the Holocene is not mutually exclusive of strong human impacts on fire regimes in the more recent past. Debates over the relative importance of these two powerful (now interacting) forces center on when, at what scale, and to what degree each has shaped local and regional flora and fire. Paleofire records, particularly when examined in conjunction with pollen, climate, archaeological, and other data, have provided compelling evidence of cases with intense human impacts regardless of climate (e.g., McWethy et al., Reference McWethy, Whitlock, Wilmshurst, McGlone, Fromont, Li, Dieffenbacher-Krall, Hobbs, Fritz and Cook2010) as well as limited human impacts in comparison with those forced by climate (e.g., Oswald et al., Reference Oswald, Foster, Shuman, Chilton, Doucette and Duranleau2020). The fact that both vegetation and fire respond to climate, however, often makes it difficult to disentangle the two. More natural experiments, like Cygnet Lake, where the confounding effects of one or another of climate, vegetation and fire can be controlled, would help. To date, however, fossil charcoal and pollen-based records of intensive human impacts emerge at the global scale only during the industrial era (Marlon et al., Reference Marlon, Bartlein, Carcaillet, Gavin, Harrison, Higuera, Joos, Power and Prentice2008; Brugger et al., Reference Brugger, Gobet, Osmont, Behling, Fontana, Hooghiemstra, Morales-Molino, Sigl, Schwikowski and Tinner2019; Osmont et al., Reference Osmont, Sigl and Schwikowski2019).

“Slow” (interannual- to decadal-scale) socioecological processes are essential for predicting the future of wildfire and carbon emissions

While long-term climate dynamics set the stage for variations in wildfire on millennial and longer time scales, the interannual and decadal-scale coupling between fire, climate, vegetation, and human activities is more important for understanding our current challenges. Such intermediate-scale socioecological processes are what produce many of the landscape- to regional-scale changes in fire regimes that will unfold in the coming decades (Trouet et al., Reference Trouet, Taylor, Carleton and Skinner2009). Insight into the effects of El Nino Southern Oscillation (ENSO) on fires in the tropics can be gained through analysis of satellite data (van der Werf et al., Reference Van Der Werf, Randerson, Collatz, Giglio, Kasibhatla, Arellano, Olsen and Kasischke2004). But analyses of how global warming may additionally impact Australian bushfires, for example, would benefit greatly from historical and paleo data on fire activity and synoptic-scale atmosphere-ocean interactions beyond interannual scales. Disentangling the effects of anthropogenic climate change from this natural climate variability could reduce large uncertainties in expected fire occurrence in many parts of the world over the coming years and decades (Wittenberg, Reference Wittenberg2009; Eden et al., Reference Eden, Krikken and Drobyshev2020). Examination of feedbacks between fires and extreme climate conditions in Amazonia, for example, is needed to understand the likelihood that forests will reach “dieback” tipping points—knowledge essential for informing land management strategies (Brando et al., Reference Brando, Balch, Nepstad, Morton, Putz, Coe, Silvério, Macedo, Davidson, Nóbrega and Alencar2014). Likewise, identifying demographic trends, such as urbanization, rural abandonment, agricultural expansion, and development in the wildland-urban interface, in different regions can reveal how land-cover changes have and will continue to shape fuels and fire in the coming decades (Foster, Reference Foster2002; d'Amour et al., Reference D'Amour, Reitsma, Baiocchi, Barthel, Güneralp, Erb, Haberl, Creutzig and Seto2017). Development in the Wildland Urban Interface and migration away from coasts, as well as cultural changes that affect agriculture and land-use, have major implications not only for fuels and ignitions that affect fire regime trends, but also for the exposure of new populations to fire regardless of fire-regime shifts (Knorr et al., Reference Knorr, Arneth and Jiang2016). Sociodemographic changes can also influence fire attitudes and policy relating to hazard mitigation and fire suppression (e.g., Guyette et al., Reference Guyette, Muzika and Dey2002; Pausas and Fernández-Muñoz Reference Pausas and Fernández-Muñoz2012; Bühler et al., Reference Bühler, de Torres Curth and Garibaldi2013; Wu et al., Reference Wu, Knorr, Thonicke, Schurgers, Camia and Arneth2015).

Long-term fire-history data from both dendrochronological and paleoecological records, particularly when they overlap in space and time, can improve projections of changing fire activity and its effects on vegetation, climate, and populations beyond seasonal time frames. Indeed, without a better understanding of the key mechanisms driving these “slow” processes or even the ability to identify gaps in our awareness of which processes are most important in driving decadal-scale fire behavior, reducing uncertainty in estimates of future fire and associated carbon emissions will be exceedingly difficult (Kloster et al., Reference Kloster, Mahowald, Randerson and Lawrence2012). A focus on intermediate (interannual to centennial-scale) ecosystem processes requires the compilation and integration of diverse data sources (Fig. 2; after Gavin et al., Reference Gavin, Hallett, Hu, Lertzman, Prichard, Brown, Lynch, Bartlein and Peterson2007). In the western U.S. or southeastern Australia, some of these data are available and abundant if not stored together and in comparable formats (Mooney et al., Reference Mooney, Harrison, Bartlein, Daniau, Stevenson, Brownlie, Buckman, Cupper, Luly, Black and Colhoun2011; Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012; Williams et al., Reference Williams, Mooney, Sisson and Marlon2015), whereas in other areas, such as Africa and Siberia, fire as well as other relevant data (e.g., paleoclimate records) are sparse. In most cases, substantial data exist for one or two key components (e.g., fire or climate) but not all (e.g., fire, vegetation, human activities, climate). In northeastern North America, for instance, paleofire and archaeological data are relatively abundant but paleoclimate data are sparse (Oswald et al., Reference Oswald, Foster, Shuman, Chilton, Doucette and Duranleau2020; Marlon et al., Reference Marlon, Pederson, Nolan, Goring, Shuman, Robertson, Booth, Bartlein, Berke, Clifford and Cook2017). In parts of Asia, such as in Japan, rich historical documents complement archaeological and paleoecological evidence (Sasaki and Takahara, Reference Sasaki and Takahara2011). In much of Europe, historical, archaeological and paleoecological data are abundant and provide rich local evidence for long-term environmental change (e.g., Turner et al., Reference Turner, Roberts, Eastwood, Jenkins and Rosen2010; Roberts et al., Reference Roberts, Fyfe, Woodbridge, Gaillard, Davis, Kaplan, Marquer, Mazier, Nielsen, Sugita and Trondman2018).

Figure 2. Temporal distribution of record lengths and scales of variability captured by different fire proxies (after Gavin et al., 2007). Many proxy types can capture interannual to decadal scale variability to some extent, but such information is often pushing the limits of sediment proxies in terms of their data availability and accuracy. Nevertheless, many critical research questions lie at this intersection, making multiproxy data synthesis and integration increasingly imperative.

The integration of historical, dendrochronological, and lake-sediment-based paleofire data in the western US during the past 3000 years provides an example of how dramatically climate and human activities can shift fire regimes over several decades (Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012). Fire frequency and biomass burned here were both driven largely by temperature and drought until the 21^st century, when climate and fire became completely decoupled due to increasing human activities that reduced or fragmented fuels and excluded or suppressed fire. A “fire deficit” or gap thus ensued between the amount of fire expected given warmer climate conditions versus the amount of fire actually allowed to burn in the West. This “deficit” helps explain the current surge in fire activity in the West as fuels have accumulated and temperatures continue to warm, extending the fire season and drying out fuels for longer periods each year (Westerling et al., Reference Westerling, Hidalgo, Cayan and Swetnam2006; Williams et al., Reference Williams, Abatzoglou, Gershunov, Guzman-Moralez, Bishop, Balch and Lettenmaier2019).

Strong decadal to centennial trends in fire associated with changing human activities, land-use, and demographic patterns are a common feature of paleofire records around the world. Not only are clear increases in burning often evident at the establishment of cultivation as mentioned previously, but examples of subsequent shifts in the intensity and type of fire use are also abundant. Such data provide evidence that charcoal records can detect fine-scale changes in fire caused by human activities. Declines in fire and land-use intensity coupled with forest regeneration are documented, for example, by decreases in charcoal abundance and increases in maize and other pollen types associated with cultivation from sediment cores in Costa Rica in the late Holocene (Clement and Horn, Reference Clement and Horn2001). Also during the late Holocene in northwestern China, peaks in charcoal coincide with declines in arboreal pollen, while gradual increases in the abundance of pollen types indicating human activity, including Herba taraxaci and Radix asteris, are also observed (Zhang et al., Reference Zhang, Zhang, Kong, Yang, Li and Tarasov2015). In Europe, mid- to late-Holocene charcoal increases at Lago dell'Accesa in Tuscany, Italy, are synchronous with the development of settlements in the region, slash-and-burn agriculture, animal husbandry, and mineral exploitation. Human-caused burning in turn triggered significant changes in vegetational communities (e.g. temporal declines of Quercus ilex forests and expansion of shrublands and macchia) (Vannière et al., Reference Vannière, Colombaroli, Chapron, Leroux, Tinner and Magny2008). In east Africa, pollen, charcoal, and archeological data document the cumulative effects of climate and land-use change and how they have continuously shaped fire regimes over the past 6000 years (Marchant et al., Reference Marchant, Richer, Boles, Capitani, Courtney-Mustaphi, Lane, Prendergast, Stump, De Cort, Kaplan and Phelps2018).

Analysis of relatively slow processes only evident in long-term reconstructions can also yield insights into the importance of physical processes not considered in typical studies based solely on modern datasets. Calder et al. (Reference Calder, Stefanova and Shuman2019), for example, identified landscape patterns of climate-fire-vegetation interactions in subalpine ecosystems of Colorado involving “ribbon forests.” These high-elevation bands of forest became separated by meadows, evidenced by coincident increases in sagebrush (Artemisia) and other meadow taxa, and decreases in conifers, especially spruce (Picea). The vegetation changes occurred during an interval of warming about 1000 years ago, yet were also associated with decreased wildfire frequency as the high-elevation meadows served as novel fuel breaks.

The lack of integration of decadal to centennial trends in fire and fire emissions modeling is evident in simulations produced by dynamic global vegetation models (DGVMs) and in historical reconstructions of fire emissions based largely on paleoclimate and Earth system model output (van Marle et al., Reference Van Marle, Kloster, Magi, Marlon, Daniau, Field, Arneth, Forrest, Hantson, Kehrwald and Knorr2017). Despite integration of some paleoecological data, such simulations show essentially no multidecadal variability in global fire activity prior to the most recent decade or two. The results stand in remarkable contrast to most paleofire records, including those from the tropics and subtropics, where most burning occurs (e.g., Bush et al., Reference Bush, Hansen, Rodbell, Seltzer, Young, León, Abbott, Silman and Gosling2005; Colombaroli et al., Reference Colombaroli, Ssemmanda, Gelorini and Verschuren2014; Rucina et al., Reference Rucina, Muiruri, Kinyanjui, McGuiness and Marchant2009; Wick et al., Reference Wick, Lemcke and Sturm2003). Another example of the vital need to incorporate paleodata into modeling comes from Kelly et al. (Reference Kelly, Genet, McGuire and Hu2016), who demonstrate the importance of accurately estimating baseline levels of burning before trying to model boreal carbon emissions in the coming century. Informing an ecosystem model with paleofire data resulted in predictions showing that forests serve as a carbon source rather than a carbon sink, which they had initially indicated when not informed by paleofire data. Further work integrating diverse fire proxies, especially at decadal to centennial scales, is needed to improve fire modeling and prediction under rapidly changing climate conditions (Rabin et al., Reference Rabin, Melton, Lasslop, Bachelet, Forrest, Hantson, Li, Mangeon, Yue, Arora and Hickler2017).

Progress has been made towards compiling and standardizing data within proxy types, promoted by the Past Global Changes Organization (PAGES) and its many initiatives around climate, land-cover, and fire, for example, but multi-proxy initiatives are more nascent. The increasingly availability of data and open-source tools make integration increasingly tractable, at least at regional scales, however. New data collection specifically targeting areas of interest remains critical, so that sites can be carefully selected to maximize the temporal and spatial areas overlap between multiple proxies that yield unique information about ecosystem responses to climate, disturbance, and other changes (Nolan, Reference Nolan2019). Such efforts are challenging given the different areas of expertise and traditions of different subdisciplines, but moving more quickly towards integrating diverse scientific approaches as well as proxies is essential to make progress on ecological challenges that don't fit neatly into disciplinary domains (Buma et al., Reference Buma, Harvey, Gavin, Kelly, Loboda, McNeil, Marlon, Meddens, Morris, Raffa and Shuman2019).

Current fire regime changes are unprecedented

Attributing individual weather events such as hurricanes and heat waves to anthropogenic climate change has been challenging. Attributing fire regime changes to global warming can be more even more difficult because of the history of land-use legacies and fire management policy that affect fuels. Yet evidence continues to build that increasing global temperatures are leading to extreme heat and drought that are in turn driving changes in patterns of fire activity (Partain et al., 2015; Abatzoglou and Williams, Reference Abatzoglou and Williams2016; Kirchmeier-Young et al., Reference Kirchmeier-Young, Gillett, Zwiers, Cannon and Anslow2019). Defining a specific event as “unprecedented,” however, remains contentious, especially in the context of an issue as politically-divisive as global warming in the U.S. Some have argued, for instance, that recent extreme weather events were not unusual because of their physical characteristics or causes, but rather because of their skyrocketing costs, which they attribute to population growth and development in vulnerable areas (Pielke and Landsea, Reference Pielke and Landsea1998; Crompton et al., Reference Crompton, Mcaneney, Chen, Pielke and Haynes2010). Such arguments, which attempted to temper concern about climate change and its effects on weather, fail to hold up, however, as intense flood events unfold in inland areas, water shortages accompany extreme drought in rural and urban areas alike, and unusually large and severe wildfires spread through wildlands, rural, suburban and urban areas regardless of community or ecosystem type, region, or even continent (Mutch et al., Reference Mutch, Rogers, Stephens and Gill2011; Burls et al., Reference Burls, Blamey, Cash, Swenson, al Fahad, Bopape, Straus and Reason2019; Boer et al., 2020).

Paleofire records provide a unique opportunity to assess the rapid emergence of seemingly novel and unusually severe wildfires over very long time periods and in diverse and widely-separated regions (Fig. 3). Individual paleofire records that capture charcoal from tens of kilometers or more around a site can provide long-term context for changes at local scales (Clark, Reference Clark1988; Tinner et al., Reference Tinner, Hofstetter, Zeugin, Conedera, Wohlgemuth, Zimmermann and Zweifel2006), while broad-scale syntheses allow comparisons across regions that reveal patterns shaped by regional and global processes (Marlon et al., Reference Marlon, Bartlein, Walsh, Harrison, Brown, Edwards, Higuera, Power, Anderson, Briles and Brunelle2009; Marlon et al., Reference Marlon, Bartlein, Daniau, Harrison, Maezumi, Power, Tinner and Vanniére2013). Here, examples of paleofire records that show “unprecedented” recent charcoal accumulation rates (influx, measured using diverse methods) are identified where unusually severe wildfires have occurred in the recent past: western North America, the Brazilian Atlantic Coast, and southeastern Australia (Fig. 4).

Figure 3. Locations of 12 paleoecological sites (blue) that show unusually high rates of charcoal accumulation during recent decades as compared with Holocene levels. All sites are from the Global Charcoal Database version 3 (Marlon et al., Reference Marlon, Kelly, Daniau, Vannière, Power, Bartlein, Higuera, Blarquez, Brewer, Brücher and Feurdean2016) are shown in gray. Selected sites in blue were selected because they have unprecedented rates of recent charcoal accumulation, but many other sites with similar patterns exist in the database.

Figure 4. Examples of 12 paleoecological sites showing unusually high rates of charcoal accumulation during recent centuries as compared with Holocene levels in North America (OK Lake, Manitoba; Andy Lake, Northwest Territories; Slough Creek Pond, USA; Foy Lake, USA), in Brazil (Lago do Pires; Morro de Itapeva; Serra Campos Gerais; and Serra do Araçatuba); and Australia (Kurnell Fen, Kings Tableland Swamp, Goochs Swamp, and Nursery Swamp).

Fires in western North America have been increasing for the past forty years. Abatzoglou and Williams (Reference Abatzoglou and Williams2016) find that over half (~55%) of the observed increases in fuel aridity across western US forests during this time are attributable to human-caused climate change. Despite a doubling of area burned since the early 1980s, however, some argue that current fire activity levels are only unusual in the context of the past few decades. Stephens et al. (Reference Stephens, Martin and Clinton2007), for instance, estimate that approximately 1.8 million ha burned annually in California alone prehistorically (pre-1800). They further estimate that prehistoric annual area burned in California was 88% of the total annual wildfire area in the entire U.S. from 1994–2004. The authors suggest that characterizing two million ha burning annually in California as extreme is a uniquely 20th or 21st century perspective, and that smoky skies must have been characteristic of the summer and fall in California during the prehistoric period.

Indeed, historical, dendrochronological, and paleofire evidence from the western U.S. are highly consistent with the characterization of 20^th century burning as extremely low compared with most of the past 3000 years (Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012). Thus, increases in area burned in and of themselves are not necessarily unprecedented. However, in the two centuries prior to Euro-American settlement of the West, burning was also extremely low due to cool climate conditions and limited drought during the Little Ice Age (Marlon et al., Reference Marlon, Bartlein, Gavin, Long, Anderson, Briles, Brown, Colombaroli, Hallett, Power and Scharf2012). The causes of pre- and post-settlement fire regime changes are completely different, however. The rates of change in burning and the spatiotemporal variability of changes are also unique in recent times, and not only due to the clustering of modern fires along road networks and coincidence of modern fire outbreaks with summer holidays (Bartlein et al., Reference Bartlein, Hostetler, Shafer, Holman and Solomon2008). Change in charcoal accumulation in lakes across North America during the Industrial Era are visibly anomalous in the past 1000 years, and in some cases the past 5000 or even 10,000 years (Delcourt et al., Reference Delcourt, Delcourt, Ison, Sharp and Gremillion1998; Toney and Anderson, Reference Toney and Anderson2006; Lynch et al., Reference Lynch, Calcote and Hotchkiss2006; Walsh et al., Reference Walsh, Pearl, Whitlock, Bartlein and Worona2010). Examples of this exist in Manitoba at Andy Lake, which had a fire in its watershed in 1994, and in the Northwest Territories at OK Lake, which had a fire in its watershed in 1980 (Lynch et al., Reference Lynch, Clark and Stocks2004). The large wildfires in Yellowstone National Park that burned in 1988, evident in the record from Slough Creek Pond (Whitlock et al., Reference Whitlock, Bartlein, Briles, Brunelle, Long and Marlon2008), and fire(s) that registered at Foy Lake in Montana with unknown dates (Power et al., Reference Power, Whitlock and Bartlein2011) appear unique at least at the local scale (Fig. 4).

Prior to the 1800s, variations in burning in the Western U.S. resulted from gradual changes in both lightning and human-caused ignitions. Today the additional influences of suppression as well as anthropogenically-induced climate changes that are extending the fire season, producing more extreme high temperatures, and shifting hydroclimatic patterns are becoming evident (Westerling et al., Reference Westerling, Hidalgo, Cayan and Swetnam2006; Abatzoglou and Williams, Reference Abatzoglou and Williams2016). The combined effects of climate change with those in land-cover, development, aggressive fire suppression, the spread of invasive species, and other factors have fundamentally altered fire activity in recent decades not only in terms of area burned, but also in terms of timing, severity, location, and other characteristics (Westerling et al., Reference Westerling, Hidalgo, Cayan and Swetnam2006; Lentile et al., Reference Lentile, Morgan, Hudak, Bobbitt, Lewis, Smith and Robichaud2007, Bartlein et al., Reference Bartlein, Hostetler, Shafer, Holman and Solomon2008; Fusco et al., Reference Fusco, Finn, Balch, Nagy and Bradley2019). The paleofire record captures the anomalous activity of the past two centuries in their long-term context, and as new cores are collected and analyzed, fire activity at regional scales during the most recent decades will be better contextualized as well. In Alaska, the anomalous nature of regional fire regimes is clear. Fire frequency and biomass burning is increasing now more rapidly than at any time in the past 10,000 years based on 14 paleofire records from the Yukon Flats, and this is an area that has been largely unaffected by land-use changes (Kelly et al., Reference Kelly, Chipman, Higuera, Stefanova, Brubaker and Hu2013). Thus, while it is accurate to say that burning was more widespread in the West prior to the 1800s than it is today, it is also accurate to say that at least some of the recent fires in the West are unprecedented in the paleofire record.

Recent burning along the Brazilian Atlantic Coast provides additional examples of unprecedented change with no comparison in the paleofire record (Fig. 4). Deforestation fires affect the livelihoods and health of nearby communities in myriad ways and have tremendous consequences for biodiversity in tropical rainforests and the global carbon cycle (Brando et al., Reference Brando, Soares-Filho, Rodrigues, Assunção, Morton, Tuchschneider, Fernandes, Macedo, Oliveira and Coe2020). Concerns that a tipping point may exist where rapid forest destruction becomes assured even without further human impacts increase during extreme wildfire seasons such as what occurred in 2019 when the rate of deforestation rose to its highest level in 11 years (Elassar, Reference Elassar2019). In a synthesis of charcoal accumulation rates in different Amazonian biomes, Cordeiro et al. (Reference Cordeiro, Turcq, Moreira, Rodrigues, Simões Filho, Martins, Santos, Barbosa, da Conceição, de Carvalho Rodrigues and Evangelista2014) found that fire associated with recent intense land-use changes had no precedent during the Holocene history of the Amazon. Similar examples can be found at many sites in Brazil, including Lago do Pires, Morro de Itapeva, Serra Campos Gerais, and Serra do Araçatuba (Behling, Reference Behling1995; Behling, Reference Behling1997a; Behling, Reference Behling1997b, Behling, Reference Behling2007), but others exist through South America, for example in Columbia (Berrío, et al., Reference Berrío, Hooghiemstra, Marchant and Rangel2002), Mexico (Caballero et al., Reference Caballero, Vázquez, Lozano-García, Rodríguez, Sosa-Nájera, Ruiz-Fernández and Ortega2006), and Chile (Haberle and Bennett, Reference Haberle and Bennett2004). While the anomalous changes are not always evident in the original publications of these paleofire records, reanalyses that convert concentration values to accumulation rates, which account for changes in sedimentation at each site, reveal the nature of recent increases in burning more clearly (Fig. 4, based on reanalyzed data from the GCDv3, Marlon et al., 2016). Accurately dating the most recent (uppermost) samples in sediment cores is very difficult, and so obtaining precise dates for recent abrupt changes in charcoal accumulations typically requires validation with independent evidence of burning. Nevertheless, examples such as those described above are not particularly rare; many records from most parts of the world show rapid and dramatic increases (often followed by equally dramatic decreases) in charcoal abundance in recent decades.

A final set of examples showing the unprecedented nature of recent fire-regime changes in their paleoecological context come from southeastern Australia, from lakes not far from where the bushfires of 2019 occurred (Fig. 3). Just south of Sydney, for example, paleofire data from a coastal wetland on the Kurnell peninsula provide a history of fire extending back more than 9000 years. Low levels of fire activity are recorded early in the record but charcoal steadily increases throughout the Holocene. Only two samples span the most recent century, with the earlier one showing extremely high fire activity relative to previous levels, and the more recent sample having almost no charcoal. About 150km to the west of Sydney in the Blue Mountains National Park, several other records show unprecedented fire activity in the most recent decades or century. At Goochs Crater, fire activity was relatively high prior to 9000 yr BP and after 6,000 years BP, when effective moisture was generally low (Black and Mooney, Reference Black and Mooney2006, Black et al., Reference Black, Mooney and Attenbrow2008). A mid-Holocene interval of low charcoal accumulation is consistent with an estimated decrease in summer temperature in south-eastern Australia of about 2°C between 9000 and 8000 years BP (Goede et al., Reference Goede, Mc Dermott, Hawkesworth, Webb and Finlayson1996), which would have shortened the fire season. Burning in the recent historic past at this site clearly occurred at a level without precedent in the previous 14,200 years. Similar findings come from Kings Tableland Swamp, (Chalson, Reference Chalson1991, Black, Reference Black2001), also in the Sydney Basin, and Nursery Swamp in the Australian Capital Territory (Rogers and Hope, Reference Rogers and Hope2006; Mooney et al., Reference Mooney, Harrison, Bartlein, Daniau, Stevenson, Brownlie, Buckman, Cupper, Luly, Black and Colhoun2011). Fire activity at Kings Tableland Swamp was minimal for the past 18,000 years, with evidence of small increases in fire ca. 16,000 and 3,000 years ago. The past few decades, however, have seen an exponential increase in the amount of charcoal recorded at the site, clearly unprecedented in this case in the past 18,000 years.

The common factor associated with these examples of unusual yet seemingly unrelated fire events in distant places is human activities, especially intensive, industrialized agriculture and land clearance. The most recent syntheses showing many nearby sites with high recent charcoal accumulation rates, perhaps most evident in Alaska (Kelly et al., Reference Kelly, Chipman, Higuera, Stefanova, Brubaker and Hu2013), is now additionally reflecting the impacts of human activities on global climate, and particularly the increases in global temperatures that extend fire seasons and cause an increase in the number of very hot, dry days. Globally, satellite data show that the increase in temperatures has lengthened the fire season between 1979 and 2013 (Jolly et al., Reference Jolly, Cochrane, Freeborn, Holden, Brown, Williamson and Bowman2015). In addition to more frequent extremely hot days, nighttime temperatures are rising (Donat et al., Reference Donat and Alexander2012), which may lead to increases in fire duration and ultimately fire season length. New data syntheses and analyses will be needed to examine the mechanisms behind the most recent changes in fire activity (Andela et al., Reference Andela, Morton, Giglio, Paugam, Chen, Hantson, Van Der Werf and Randerson2019).

Fire scientists must help the public understand the connections between global warming and wildfire

Climate scientists have been sounding the alarm about the risks of a warming planet and its effects on wildfires and many other impacts for decades. The societal response continues to fall far short of what is needed to slow global carbon emissions. For many individuals, climate change impacts have been perceived as distant in space (e.g., affecting those in other countries) as well as in time (e.g. affecting future generations). At least in the U.S., it is only over the past few years that a clear upward trend in public concern has begun to emerge (Ballew et al., Reference Ballew, Leiserowitz, Roser-Renouf, Rosenthal, Kotcher, Marlon, Lyon, Goldberg and Maibach2019). In 2019, registered U.S. voters said that out of 29 issues, ranging from healthcare to international trade, global warming was the 17th most important issue in determining their vote in the 2020 presidential election. A large partisan divide was evident, however, with global warming ranking 29th out of 29 voting issues among conservative Republicans, and 23rd among liberal and moderate Republicans. Among moderate and conservative Democrats, however, it was the eighth highest priority voting issue, up eight places from a year prior. Among liberal Democrats, it was ranked third, with environmental protection ahead of it—ranking second as a voting priority. Overall, global warming as a national policy priority for the president and Congress has increased dramatically in the U.S. among Democrats, and somewhat for Independents, while remaining relatively unchanged among Republicans. Thus, for the first time in American electoral history, climate change has become a top tier voting issue for a major political party (Leiserowitz et al., Reference Leiserowitz, Maibach, Rosenthal, Kotcher, Bergquist, Gustafson and Ballew2019).

Because climate change is a highly complex scientific issue, scientists have necessarily played (and continue to play) a critical role in communicating with and educating the public about its causes, consequences, and solutions. Knowledge about these complexities is extremely low among the public (Leiserowitz et al., Reference Leiserowitz, Smith and Marlon2010), and majorities trust climate scientists more than any other group (e.g. other scientists, television weather reporters, family and friends, or your primary care doctor) for information about this issue (Leiserowitz et al., Reference Leiserowitz, Maibach, Roser-Renouf, Feinberg and Rosenthal2015). Wildfires burning in unusual places, at unusual times, lasting longer or burning hotter than they otherwise would because of extreme heat and drought caused by global warming provide an opportunity for climate and fire scientists to approach the topic when it is already in the news. Scientists can use these extreme weather events made worse by global warming as “teachable moments”—not for those directly experiencing the impacts who need to be focus on preparedness, safety, and self-protection—but rather for those safe from the direct impacts but who are paying attention to these events and are seeking to understand why what we are seeing is unusual. It is exactly at these moments when many are most open to learning what we know from paleoecology about the true range of variability in Earth systems.

Surveys of the American public indicate that many are already making a connection a between global warming and uncontrollable wildfires (Marlon and Cheskis, Reference Marlon and Cheskis2017). Yet, it is only in states that experience wildfires that majorities believe global warming is making them worse. In California and Colorado, for example 69% and 66% of the public, respectively, believe that global warming is increasing the severity of wildfires. In Texas, 61% think that global warming is making wildfires more severe. But fewer people make this connection in states where wildfires are less common. In Ohio, 36% think wildfires are worse now due to global warming. Such differences in perception indicate that people often do not come to understand global warming's consequences until they or their local communities experience them directly, and even then, two individuals who experience the same fire can come away with remarkably different interpretations about the cause and appropriate responses (Howe et al., Reference Howe, Marlon, Mildenberger and Shield2019). More efforts are needed by scientists and many others to help the public understand the seriousness of climate change, the need to act quickly, and critically, the value of acting—even if many impacts are already underway and unavoidable at this stage. Reducing the rate of warming, reducing the magnitude of change, can still have enormous benefits to current and future generations, to ecosystems, and to other species. Articulating the strong and varied connections between wildfires, human activities, and global warming presents a unique communication opportunity for paleoecologists.

Maintaining scientific integrity when discussing the relationships between wildfires and climate change is essential. While the attribution of worsening wildfires to climate change is complex, informing the public about the connections between the two does not have to be (Doerr and Santin, Reference Doerr and Santín2016). The public does not need to and in most cases does not want to know about the nuances of fire regimes and their causal relationships to specific components of the climate system. In the vast majority of cases, the public primarily needs to know that managed fires are a critical component of healthy ecosystems, that aggressive suppression of all fires is extremely counterproductive, and that the impacts of unconstrained fossil fuel burning are extremely serious and will continue to worsen if little is done to lower carbon emissions. The growing body of climate attribution studies now provide ample scientific evidence that human activities are increasing the impacts of most extreme weather events (e.g., Trenberth et al., Reference Trenberth, Fasullo and Shepherd2015; Herring et al., Reference Herring, Hoell, Hoerling, Kossin, Schreck and Stott2016; Wang et al., Reference Wang, Zhao, Yoon, Klotzbach and Gillies2018). New statistical learning and climate model simulations now show that even when given a single day of globally observed temperature and moisture, the fingerprint of human-caused climate change is detectable (Sippel et al., Reference Sippel, Meinshausen, Fischer, Székely and Knutti2020). The effect of increasing carbon pollution from anthropogenic activities on global climate is thus now affecting all weather events. And yet, much can still be done to reduce future damages. Effective solutions are at hand—from the individual to the international level—for reducing global warming and for improving fire management (Schoennagel et al., Reference Schoennagel, Balch, Brenkert-Smith, Dennison, Harvey, Krawchuk, Morgan, Moritz, Rasker, Turner and Whitlock2017; Moritz et al., Reference Moritz, Batllori, Bradstock, Gill, Handmer, Hessburg, Leonard, McCaffrey, Odion, Schoennagel and Syphard2014).

The increasing pace of climate change and its impacts are beginning to drive concern among the public (Howe et al., Reference Howe, Marlon, Mildenberger and Shield2019). Wildfires are not the only events gaining attention, but they may be one of the most powerful, visual examples of a rapidly changing climate. From the standpoint of human perception, sea level rise is too slow, too incremental, and limited to the coasts to engage most people. Drought is also a slow process and is ill-suited to the highly visual media of television and the internet. Wildfire, in contrast, has arresting imagery and stories, and unlike hurricanes or flooding, it is intuitively and viscerally consistent with the idea of “global warming.” While care in any discussion of wildfire is critical to avoid describing all fire as destructive, the concrete images and strong affective associations of unusually large and severe events that are destructive is vital for helping people understand the meaning of a warmer world for our own health, our families, and our communities (Weber, Reference Weber2006). Even when confronted with unmistakable evidence that climate is changing and fire is responding, people still find it difficult to understand, because it is indeed “unprecedented.” Yet, by showing them that fire has always been sensitive to climate changes, that reticence to believe it's happening now might be eroded.