A review of the effects of wildfire smoke on the health and behavior of wildlife

Climate change is intensifying global wildfire activity, and people and wildlife are increasingly exposed to hazardous air pollution during large-scale smoke events. Although wildfire smoke is considered a growing risk to public health, few studies have investigated the impacts of wildfire smoke on wildlife, particularly among species that are vulnerable to smoke inhalation. In this review, we synthesized research to date on how wildfire smoke affects the health and behavior of wildlife. After executing a systematic search using Web of Science, we found only 41 relevant studies. We synthesized findings from this literature and incorporated knowledge gained from fields outside wildlife science, specifically veterinary medicine and air pollution toxicology. Although studies that directly investigated effects of smoke on wildlife were few in number, they show that wildfire smoke contributes to adverse acute and chronic health outcomes in wildlife and influences animal behavior. Our review demonstrates that smoke inhalation can lead to carbon monoxide poisoning, respiratory distress, neurological impairment, respiratory and cardiovascular disease, oxidative stress, and immunosuppression in wildlife, including terrestrial and aquatic species, and these health effects can contribute to changes in movement and vocalization. Some species also use smoke as a cue to engage in fire-avoidance behaviors or to conserve energy. However, our review also highlights significant gaps in our understanding of the impacts of wildfire smoke on wildlife. Most notably, the lack of robust air pollution measurements in existing studies limits meta-analyses and hinders construction of dose-response relationships, thereby precluding predictions of health outcomes and behaviors under different air quality conditions, especially during extreme smoke events. We recommend that future studies leverage existing data sets, infrastructure, and tools to rapidly advance research on this important conservation topic and highlight the potential value of interdisciplinary collaborations between ecologists and atmospheric chemists.


Introduction
As climate change intensifies the frequency and severity of wildfires, communities around the world are increasingly vulnerable to smoke pollution (Jacob and Winner 2009). Increased wildfire activity has been linked to declines in average regional air quality and greater incidence of extreme air pollution episodes. For example, wildfires contributed to a recent increase in annual concentrations of fine particulate matter (PM 2.5 , particles smaller than 2.5 µm in aerodynamic diameter) in the United States (McClure and Jaffe 2018, Clay and Muller 2019), and smoke events in the Pacific Northwest, United States in 2018 and 2020 caused PM 2.5 to spike to concentrations well above the National Ambient Air Quality Standards (Washington State Academy of Sciences 2019, Liu et al 2021a). Wildfire smoke directly contributes to adverse respiratory and cardiovascular health outcomes and mortality in humans (Cascio 2018, Chen et al 2021; in fact, studies have shown that the chemical composition of PM 2.5 in wildfire smoke is more toxic than that of urban ambient PM 2.5 (Franzi et al 2011, Aguilera et al 2021).
Wildfire smoke also sickens non-human animals, as illustrated by numerous case studies in veterinary medicine that document morbidity and mortality in domestic animals exposed to smoke, including pets and livestock (Fitzgerald andFlood 2006, Marsh 2007). These case studies demonstrate that, like people, animals can suffer from carbon monoxide poisoning, thermal and chemical damage to lung tissue, and greater susceptibility to respiratory disease as a result of smoke inhalation (Wohlsein et al 2016). In fact, animal models, including mice, rats, rabbits, sheep, and monkeys, are often used to study the onset and progression of human disease following exposure to the toxic gases and aerosols found in smoke (David et al 2009). Although many animals in fire-prone habitats are able to detect and avoid wildfires, fires still pose direct threats to wildlife (Engstrom 2010, Nimmo et al 2021, including exposure to extreme heat and smoke. Yet, the impacts of wildfire smoke on the health and behavior of wildlife are largely unknown (Hovick et al 2017, Lee et al 2017, Erb et al 2018. This paucity of research on how wildfire smoke affects the health and behavior of wild animals hinders full consideration of the direct and indirect effects of wildfires when conducting risk assessments for wildlife and developing conservation plans. In addition, research on the impacts of wildfire smoke on wildlife is published in disparate journals spanning numerous disciplines (e.g. ecology, physiology, animal behavior, veterinary medicine, etc); as such, ecologists, wildlife managers, and other stakeholders may be challenged to identify relevant studies. To date, review papers have synthesized findings on firstorder effects of fire on animals, including injury, morbidity, and mortality (Engstrom 2010), considered behavioral responses of mammals to fire, specifically torpor , and discussed fire as an evolutionary force driving animal behavior and survival (Nimmo et al 2021), but none have focused specifically on the effects of smoke from wildfires on the health and behavior of wildlife.
Wildfires are an important type of natural disturbance (Turner 2010) in forests, grasslands, and deserts around the world, and many wildlife species benefit from resources available in post-fire landscapes (Smith 2000). However, just as people now grapple with health risks posed by routine smoke events, even in airsheds where smoke pollution was previously uncommon (Wilmot et al 2021), wildlife must also contend with greater-perhaps even novel-exposure to wildfire smoke with more intense wildfire activity. The magnitude of smoke events in the 21st century further underscores the urgent need to study the impacts of wildfire smoke on wildlife. Wildfire smoke persists in the atmosphere even after flames have subsided and can travel hundreds of miles, creating hazardous air quality conditions and degrading visibility across large geographic areas (figure 1). As a result, smoke from a single wildfire could impact the health and behavior of wildlife at a much larger spatial scale than the area burned. Direct effects of wildfire smoke on individuals could scale to influence the demography of wildlife populations, with cascading community-and ecosystemlevel impacts (figure 2).
In this review, we synthesized research to date on the effects of wildfire smoke on the health and behavior of wildlife. We focused specifically on the impacts of wildfire smoke rather than describing all immediate effects of wildfires on wildlife in order to more deeply investigate physiological and behavioral responses of wildlife to the large-scale smoke events that are becoming increasingly common around the world. Below, we (1) identify relevant literature on the effects of wildfire smoke on the health and behavior of wildlife, (2) highlight knowledge gaps, and (3) present opportunities for rapidly advancing research on this important topic, all of which should serve as a useful resource for guiding ecological studies and conservation actions.

Methods
In January 2021, we conducted two keyword searches using Web of Science (figure 3). Search terms included (1) 'wildfire * AND smoke * ' and (2) 'fire * AND smoke * ' . We performed a basic search and entered search terms into the topic field. We excluded 'news items' and 'meeting abstracts' as document types to focus on peer-reviewed literature, then further refined our search results to include only articles from categories relevant to our review (e.g. environmental science, ecology, biology, physiology, toxicology, health sciences, veterinary sciences, etc) A complete list of topic areas is provided in the appendix.
We reviewed the titles of articles in both sets of search results (n = 4314) (figure 3). We earmarked articles with titles that included any of the following for further review: (1) the name of a specific domestic animal, wildlife species, or taxa (e.g. mammals, birds); (2) a general reference to animals or wildlife; (3) an example of an animal behavior (e.g. migration); or (4) an example of a health effect (e.g. mortality). We did not further review papers with titles suggestive of inquiry into ecosystem-level impacts of wildfires or the effects of fire or smoke on vegetation. We also excluded titles with clear references to human demographic groups (e.g. children) or epidemiological study (e.g. hospitals, emergency rooms). Next, we reviewed abstracts of articles earmarked in the review of titles (n = 295) to assess their eligibility for a full-text review (figure 3). We assessed whether or not these papers presented research on the responses of animals to smoke from fires-regardless of the specific type of exposure investigated. Those that met these criteria were included in the list of papers that were read for this review (n = 72; figure 3).
We sorted papers into three categories: (1) experiments using animal models; (2) case studies from veterinary medicine; and (3) research on the effects of smoke on wildlife species (figure 3). For the last category, we considered 'wildlife' to include all non-domesticated fauna in terrestrial environments, including insects, as well as aquatic animals that breathe air, such as marine mammals. This allowed us to focus on the direct effects of smoke on wildlife, rather than explore all possible indirect effects that could arise from atmospheric deposition of airborne toxins found in wildfire smoke. At least one of the co-authors of this review read and documented findings from studies of animal models (n = 36) and case studies from veterinary medicine (n = 18). However, we focused our review on studies of the impacts of smoke on wildlife (n = 18). At least two co-authors of this review read and documented findings from each of these papers. In addition, we used Web of Science to conduct forwards searches, noting any citations of these studies that referenced all of the following in the title: (1) smoke, or more generally air pollution associated with fires; (2) animals, wildlife, or the name of a particular species or taxa; and (3) an animal behavior or health effect. We also conducted backwards searches using two different approaches: (1) we noted citations that may be relevant to our review while reading a paper, and (2) we conducted a backwards search of all works cited in the paper using the same criteria described for the forwards search. All additional studies on the effects of smoke on wildlife species identified in forwards and backwards searches were also reviewed (n = 16) (figure 3). In addition to noting major findings from these papers, we pulled out several key pieces of information to characterize the research and compare results across studies, including publication year, field of study, location, type of exposure, Figure 2. A conceptual model illustrating the various pathways by which wildfire smoke could influence wildlife populations and ecosystems. We hypothesize how effects of smoke on individual health and behavior could contribute to shifts in demographic rates, including growth and development, survival, and reproductive success. These parameters determine population growth, ultimately leading to community-and ecosystem-level impacts.
concentrations of air pollutants (if provided), and taxa and species of animals studied, as well as whether the animals were captive (i.e. kept in a laboratory or outdoor enclosure), or free-living (i.e. in the wild). Finally, we noted whether the ultimate goal of the study was to investigate effects of smoke in people or non-human animals. While preparing this manuscript, we learned of seven additional studies that considered the impacts of wildfire smoke on wildlife, which we also included in our review. Although we did not restrict our initial keyword searches in Web of Science by language, each subsequent step of our review was limited to text available in English. In addition, we were only able to review articles available through the University of Washington library system. All together, we reviewed 41 studies that considered the effects of wildfire smoke on wildlife (figure 3, table 1).

Results
We found that although research to date on the impacts of wildfire smoke on wildlife is limited, existing evidence suggests that smoke pollution has wideranging direct and indirect effects on both terrestrial and aquatic wildlife. Studies have linked smoke inhalation to acute and chronic health outcomes in animals and sought to characterize how smoke influences animal behavior. Whereas the designs of these studies are highly variable, two general approaches have emerged in the literature: (1) experiments in which animals were intentionally exposed to smoke or constituents of smoke in a controlled environment (i.e. 'controlled exposure') and (2) opportunistic monitoring of free-living animals or animals in captivity during wildfire smoke events (i.e. 'in situ exposure'). The studies we reviewed were conducted on five continents, including North America, Australia, Europe, Asia, and Africa (figure 4), and published between 1968 and 2021. They explored responses in a wide variety of taxa, including mammals, birds, reptiles, and insects, in both controlled and in situ settings (figure 5). A complete list of studies reviewed is provided in table 1. In sections 4 through 6, we detail findings from this literature while also incorporating knowledge gained from fields outside wildlife science, specifically veterinary medicine and air pollution toxicology. Finally, in section 7, we briefly summarize studies of indirect effects of smoke on wildlife and consider how a species' lifehistory strategy mediates its exposure to smoke pollution. Figure 3. A schematic illustration of methods used to identify literature included in our review. We show steps in chronological order, starting at the top and working down. Our initial keyword searches yielded 4314 titles, of which we found 295 were relevant based on a review of titles. After reviewing abstracts of these papers, we identified 72 that we thought merited a full-text review. We sorted these papers into three categories: experiments using animal models, case studies from veterinary medicine, and studies of wildlife species. We conducted backwards and forwards searches on papers specific to wildlife and included all studies of wildlife species in our review.

Acute and chronic health outcomes
Few studies have explicitly considered the impact of wildfire smoke on the health of wildlife (table 1); however, research from veterinary medicine and air pollution toxicology clearly demonstrates that smoke inhalation contributes to acute and chronic health outcomes in animals. Case studies detailing the symptoms, treatment, and recovery of pets and livestock following structural fires establish that animals are vulnerable to negative health outcomes from smoke inhalation (e.g. Drobatz et al 1999a, 1999b, Marsh 2007). In addition, there are numerous examples of laboratory experiments designed to investigate the effects of inhalation exposure to wildfire smoke in humans using animal models (e.g. Hargrove et al 2019, Martin et al 2020) including mice, rats, rabbits, and sheep. Although the objective of these studies is to characterize the underlying physiological mechanisms that contribute to respiratory and cardiovascular disease in humans, their findings allude to health effects we could observe in other mammalian species.
These experiments have incorporated in vivo, ex vivo, and/or in vitro approaches. Some studies have also used animal models to study possible treatment interventions to improve health outcomes in humans following smoke inhalation (e.g. Janssens et al 1994, Wang et al 1999, Wong et al 2004, Syrkina et al 2007, Hamahata et al 2008, Dunn et al 2018.
Research has largely focused on mammals, but all animals that breathe air-whether terrestrial or aquatic-are vulnerable to inhalation exposure to airborne toxins, including the reactive gases and aerosols that make up smoke (e.g. carbon monoxide (CO), hydrogen cyanide (HCN), and coarse and fine particulate matter (PM)). Many animals are susceptible to CO poisoning during smoke inhalation (Chaturvedi et al 1995, Fitzgerald and Flood 2006, Kent et al 2010, Ashbaugh et al 2012, Dörfelt et al 2014, Stern et al 2014, which can be fatal (Wohlsein et al 2016). CO binds to hemoglobin, a protein molecule containing iron that nearly all vertebrates (Ruud 1954) and many invertebrates depend on to carry oxygen through the bloodstream. This limits oxygen transport, resulting Table 1. Summary of studies of wildlife species included in our review. For each paper, we provided the in-text citation, broad taxonomic category and specific species studied, and type of exposure (i.e. in situ or controlled) investigated. We also indicated whether a study assessed impacts in captive or free-living animals. In addition, we included the location and continent where each study took place. If locations for studies of captive animals were not provided, we noted the location of the research institution of the first author. We also provided a description of smoke exposure. Finally, we indicated if a health outcome and/or a behavioral response was observed.  in low blood oxygen levels (i.e. hypoxemia) and insufficient supply of oxygen to tissues and organs (i.e. hypoxia) (Wohlsein et al 2016). Neurological symptoms of hypoxic brain damage could include confusion and stupor (Drobatz et al 1999a, 1999b, Mariani 2003, Kent et al 2010, Weiss et al 2011, Guillaumin and Hopper 2013. Hypoxia could also make animals more vulnerable to predation as they attempt to flee wildfires (Braithwaite and Estbergs 1987).
Smoke inhalation also causes both thermal and chemical damage to lung tissue in terrestrial and aquatic vertebrates (Fitzgerald and Flood 2006, Marsh 2007, Wohlsein et al 2016. As a result of this injury, fluid can accumulate in the lungs, a condition known as pulmonary edema (Bidani et al 1998, Jordaan et al 2020, which has been documented in pets and livestock exposed to smoke during structural fires (Verstappen and Dorrestein 2005, Fitzgerald and Flood 2006, Marsh 2007 (Mariani 2003, Fitzgerald andFlood 2006), wheezing (Kemper et al 1993), panting (i.e. polypnea) (Dörfelt et al 2014), coughing (Kemper et al 1993, Fitzgerald and Flood 2006, Dörfelt et al 2014, foaming at the nostrils (McPherson 1993, Wohlsein et al 2016, and rapid heart rate (i.e. tachycardia) (Dörfelt et al 2014), which are consistent with acute respiratory distress syndrome (Guillaumin and Hopper 2013). If untreated, smoke inhalation injury can quickly impair gas exchange, resulting in hypoxemia (Wohlsein et al 2016) and elevated levels of acid in the blood (i.e. acidosis) (Bidani et al 1998). For example, in a retrospective analysis of health records of captive bottlenose dolphins (Tursiops truncatus), researchers found that blood carbon dioxide (CO 2 ) levels were elevated in the month following a wildfire smoke event in 2003, possibly due to respiratory acidosis (Venn-Watson et al 2013). Airbreathing invertebrates might also be vulnerable to smoke inhalation. Tan et al (2018) investigated effects of smoke exposure in captive squinty bush brown butterflies (Bicyclus anynana) and found that particles accumulated in the entryway of spiracles-external openings in the exoskeleton that vent the insect respiratory system-but did not enter the trachea.
Wildfire smoke contributes to chronic respiratory and cardiovascular health outcomes in animals.
Smoke inhalation can jeopardize an animal's immune system, which is designed to protect the body from foreign matter, such as bacteria, viruses, and toxins. In mammals, smoke inhalation immediately triggers production of immune cells, including lymphocytes  . Distribution of studies that considered how wildfire smoke impacts the health and/or behavior of wildlife. Each column represents the number of papers we found on health effects or behavioral responses for a specific taxon, broken down by research approach (i.e. controlled exposure or in situ exposure). Some papers investigated both health effects and behavioral responses or considered multiple taxa and are therefore counted more than once. We did not find any papers on direct effects of wildfire smoke on amphibians.
toxic particles in wildfire smoke, they are unable to destroy them; this precludes macrophages from helping to prevent infection (Wohlsein et al 2016). Furthermore, toxins in smoke also destroy antioxidants, substances that neutralize free radicals-highly reactive, oxygen-containing compounds that damage tissue (Shalini et al 1994, Hamahata et al 2008, Wegesser et al 2010. Oxidative stress can ultimately contribute to compromised immune function by destroying macrophages or other types of immune cells (Franzi et al 2011, Williams et al 2013. Lung injury and a weakened immune response can leave animals more vulnerable to respiratory infection and illness, such as pneumonia (Attwood et al 1996a, 1996b, Marsh 2007, Simone-Freilicher 2008, Lange et al 2010, Guillaumin and Hopper 2013, Wohlsein et al 2016 or laryngotracheitis (Morris et al 1986). For example, captive bottlenose dolphins were three times more likely to have bacterial pneumonia at time of death after exposure to smoke during a wildfire in 2003 (Venn-Watson et al 2013). However, age also influenced the incidence of pneumonia, and after controlling for age, the effect of fires was no longer statistically significant (Venn-Watson et al 2013). A study of rhesus macaque monkeys (Macaca mulatta) housed in outdoor enclosures found that newborn monkeys exposed to wildfire smoke exhibited reduced lung capacity and weakened immune responses in adolescence compared to those born in a subsequent year with good air quality (Black et al 2017). Whereas this study was designed to investigate pediatric health outcomes in humans associated with wildfire smoke, these results suggest that wildlife could experience long-term, adverse health outcomes from a single smoke event. Smoke inhalation can also impair cardiovascular function in vertebrates (Kim et al 2014, Wohlsein et al 2016, Thompson et al 2018, Sharpe et al 2020 and repeated or prolonged exposure to smoke can lead to chronic heart disease (Thompson et al 2018, Martin et al 2020. Health outcomes associated with inhalation of wildfire smoke vary as a function of its toxicity. Research from air pollution toxicology demonstrates that toxicity of biomass smoke is dependent on its chemical and biological composition ( Smoke is also subject to chemical transformation during long-range transport (Jalava et al 2006). This suggests that the specific types of vegetation burned during wildfires, the stage and severity of the fires, and the distance smoke travels ultimately affect respiratory and cardiovascular health outcomes associated with smoke inhalation in wildlife.

Impacts on demography
Negative health outcomes associated with inhalation exposure to wildfire smoke could ultimately influence demographic rates in wildlife populations, including survival, growth, and reproductive success ( figure 2). ing a large-scale smoke event to increased mortality in dairy cows, specifically calves. Yet, we found only four studies that considered the effect of smoke on survival of wildlife species. A study of captive butterflies reared in smoky conditions found that caterpillars exposed to smoke exhibited a higher mortality rate than those in the control group, likely due to gas intoxication (Tan et al 2018). Gas intoxication was also hypothesized as a contributing factor to the death of lizards and snakes that did not survive prescribed burns in Tembe Elephant Park, South Africa (Jordaan et al 2020). Few specimens exhibited signs of burn injury, yet Jordaan et al found that 61% of specimens collected at the site of one fire exhibited pulmonary edema and noted particles accumulated in the lungs of two of these specimens. These findings suggest that reptiles that did not survive the fire died from asphyxiation, CO or HCN poisoning, or heat-induced cardiac arrest. Large mammals are also vulnerable to smoke inhalation- Singer et al (1989) reported that smoke inhalation injury or gas intoxication likely killed 246 elk (Cervus elaphus) that perished in the 1988 wildfires in Yellowstone National Park. Finally, Yang et al (2021) found that smoke from extensive wildfires in the Western United States contributed to a mass avian mortality event in 2020.
Wildfire smoke could also reduce growth rates and reproductive success. Tan et al (2018) reported that captive squinty bush browns exposed to smoke developed more slowly and weighed less as pupae. Cahill and Walker (2000) reported that the nesting success of Red-knobbed Hornbills (Aceros cassidix) declined at the Tangkoko Nature Reserve in Indonesia following extensive wildfires in 1997, possibly due to exposure to extreme heat and smoke. Although the Red-knobbed Hornbill example was the only study we found that considered the impacts of wildfire smoke on reproductive success in wild animals, a case study of domestic chickens exposed to smoke during a structural fire suggests that smoke inhalation could reduce egg production (Morris et al 1986). Previous research has linked other types of air pollution to reduced hatching success and lower clutch size in birds (e.g. Eeva and Lehikoinen 1995), which suggests that wildfire smoke could also impair avian reproductive success. In addition, PM-a major component of wildfire smoke-dirties bird feathers, which can render them less attractive to potential mates (Griggio et al 2011) and interfere with other colorbased signaling or camouflage.

Behavioral responses
Wildfire smoke can also trigger shifts in animal behavior, including movement and vocalization. Such behavioral changes could be due to underlying health effects (Erb et al 2018) or serve to limit exposure to airborne toxins (Singer et al 1989, Dickinson et al 2009, Liu et al 2021b. Some species rely on smoke as an early-warning signal that helps them to avoid wildfires (Engstrom 2010 (Robock 1991). Emerging evidence suggests that behavioral responses to wildfire smoke could ultimately influence the short-and long-term fitness of wildlife (Cheyne 2008, Erb et al 2018.

Effects of smoke on wildlife activity
Exposure to smoke can influence wildlife activity, including movement and vocalization. Case studies from veterinary medicine demonstrate that animals sometimes alter their behavior due to acute, adverse health effects associated with smoke inhalation; for example, pets and livestock exposed to smoke from structural fires can become agitated (Fitzgerald and  (Marsh 2007, Weiss et al 2011, Guillaumin and Hopper 2013. Researchers observed that pinecone lizards (Tiliqua rugosa) in captivity exhibited rapid tongue-flicking when exposed to smoke near their enclosure, a sign of agitation (Mendyk et al 2020). Animals in the wild could also alter their behavior in response to smoke pollution, possibly due to underlying health effects, as noted in one of the only studies to directly link wildfire smoke exposure to specific health outcomes for a wildlife species (Erb et al 2018). In this study, researchers documented the daily activity of male Bornean orangutans (Pongo pygmaeus wurmbii) before, during, and after an extensive wildfire smoke event in Indonesia. They also collected urine samples opportunistically to test for ketones, a marker of fat catabolism associated with energy expenditure. The researchers found that orangutans rested more both during and after the smoke event. Furthermore, after the smoke event, orangutans traveled shorter distances and increased their caloric intake, but expended more energy (i.e. increased fat catabolism). Despite conserving energy and eating more food, orangutans still burned more calories than they consumed after an extended period of smoke exposure, which suggests that smoke inhalation negatively affected their energy budgets. The researchers postulated this could have been due to stress or a heightened immune response (Erb et al 2018).
In addition to movement, smoke can also influence animal vocalization. For example, a study of singing behavior in Bornean white-bearded gibbons (Hylobates albibarbis) in Indonesia found that gibbons sang less when it was smoky-during months when wildfire smoke led to unhealthy air quality, both the number of days gibbons sang and the length of singing bout decreased (Cheyne 2008). Changes in vocalization during wildfire smoke events may ultimately influence entire soundscapes (Lee et al 2017). An analysis of audio recordings collected in Singapore during a haze event brought on by wildfires showed that wildlife acoustic activity, as measured by four acoustic indices, was negatively correlated with smoke pollution. Although the mechanisms driving this response were beyond the scope of the study, its authors hypothesized that several factors could have contributed to a decrease in acoustic activity, including reduced vocalization, a shift in ecological activity outside the recording period, or mortality due to direct effects of smoke exposure or reduced foraging success. Acoustic activity was suppressed for months following the smoke event, illustrating that smoke could have long-term impacts on species and communities (Lee et al 2017).

Use of smoke as a cue 6.2.1. Fire avoidance behaviors
Wildlife across taxa, including insects, reptiles, and mammals, rely on smoke as a cue to engage in fire avoidance behaviors (Nimmo et al 2021). Insects may relocate after detecting smoke to evade fires. Researchers in Germany exposed European paper wasps (Polistes dominula) to biomass smoke and found that the insects increased their thorax temperature in response to this stressor (Höcherl and Tautz 2015). Many insects must warm up their thoraces before flying; as such, the results of this experiment suggest that smoke prompts a pre-flight warmup behavior in wasps that prepares them for a quick escape from nearby fire (Höcherl and Tautz 2015). Researchers exposed ants in Kenya to smoke generated by burning elephant dung and found that two of the four study species evacuated in response to smoke, relocating up to 1800 m (Sensenig et al 2017). Of the two ant species that responded to smoke, the subordinate ant species (Crematogaster nigriceps) evacuated twice as quickly as the dominant competitor (C. mimosae). These results suggest that subordinate species may be more willing to adopt a colonist strategy following disturbance and therefore are better equipped to escape and survive wildfires (Sensenig et al 2017). However, some insects, such as the Cape honeybee (Apis mellifera capensis), may not attempt to evade fire and instead use smoke as a cue to retreat to protective nest structures, (Tribe et al 2017). Regardless of their fire avoidance strategy, smoke could compromise the ability of insects to escape fires by impairing flight performance (Liu et al 2021b). Liu et al (2021b) found that the duration, distance, and speed flown by painted lady butterflies (Vanessa cardui L.) decreased following exposure to smoke, which could adversely impact other insect behaviors as well, such as foraging and migration (Liu et al 2021b).
Studies of captive lizards suggest that smoke can also trigger fire avoidance behaviors in reptiles (Mendyk et al 2020,Álvarez-Ruiz et al 2021. For example, captive Psammodromus lizards (Psammodromus algirus) exhibited a variety of escape behaviors when exposed to smoke, including running and scratching at their terrariums (Álvarez-Ruiz et al 2021). Furthermore, lizards were more likely to increase their activity in response to smoke if they were captured in habitats prone to wildfires, regardless of an individual's previous experience with fire. These results indicate that in areas that experience frequent fires, selective pressure drives greater sensitivity to smoke, increasing the ability of local populations to detect and evade fires (Álvarez-Ruiz et al 2021).
Smoke can also prompt mammals to arouse from torpor, enabling them to escape fires (Scesny 2006, Layne 2009, Stawski et al 2015, Nowack et al 2016, Doty et al 2018. However, not all torpid mammals flee in response to smoke, or react quickly enough to survive; responses to fire stimuli are likely to vary by species, sex, and individual (Layne 2009, Nowack et al 2016. In addition, lower ambient temperatures slow torpor arousal following smoke exposure, which suggests that torpid mammals are less able to evade fires on colder days (Layne 2009, Nowack et al 2016, Doty et al 2018. Furthermore, animals that detect smoke and arouse from torpor at cooler temperatures might not return to steady-state torpor, which increases their energy expenditure (Doty et al 2018).
To avoid fires, small animals might seek shelter underground or in rock crevices (Engstrom 2010); however, burrowing may not always protect animals from extreme heat and smoke. For example, Jordaan et al (2020) noted that fossorial species were wellrepresented in their samples of dead reptile specimens collected after prescribed burns in Tembe Elephant Park, South Africa. They hypothesized that cause of death was likely asphyxiation, gas intoxication, or heat-induced cardiac arrest, which suggests that even burrowing animals are susceptible to smoke inhalation during fires (Jordaan et al 2020).

Energy-saving behaviors
Some animals rely on smoke as an indicator of impending food scarcity, prompting them to engage in energy-saving strategies. Small mammals must maintain high metabolic rates, which is difficult after fires due to limited availability of food and water. Studies of captive small mammals show that smoke can increase use of torpor in some species, allowing animals to conserve energy and survive post-fire conditions (Geiser et al 2018). For example, exposure to smoke and a substrate of charcoal and ash increased duration of torpor in captive yellow-footed antechinuses (Antechinus flavipes) (Stawski et al 2017) and captive sugar gliders (Petaurus breviceps) (Nowack et al 2018). However, use of torpor after fires depends on food availability, and is likely to vary by species (Nowack et al 2018) and sex (Stawski et al 2017).

Resource availability
Pyrophilous insects (i.e. fire-associated species that benefit from resources available in post-fire landscapes) can use smoke as a cue to navigate toward fires, responding to thermal and/or olfactory signals (Schütz et al 1999, Klocke et al 2011,Álvarez et al 2015, Milberg et al 2015. Some are even known to swarm in smoke plumes, such as 'smoke flies' of the genera Microsania and Hormopeza, possibly to mate near burned trees where they deposit their eggs (Evans 1966, Snoddy and Tippins 1968, Sinclair and Cumming 2006. Schütz et al (1999) found that the antennae of fire bugs (Melanophila acuminata) respond to volatiles generated in the combustion of pine, suggesting that the smell of smoke helps some beetles detect and locate burned trees. Insects may also respond to visual cues of smoke plumes but results from experimental studies are ambiguous (Hinz et al 2018). An influx of aerial insects to burned habitats could enhance foraging opportunities for bats (Braun De Torrez et al 2018) and insectivorous birds; unlike small, quadrupedal mammals, bats might actually decrease their use of torpor after fires to take advantage of this increase in food availability (Geiser et al 2018). Raptors may also be attracted to smoke plumes, which could signal an opportunity to prey on insects and small mammals fleeing fire (Hovick et al 2017).
Whereas pyrophilous insects appear to rely on olfaction to locate burned areas, exposure to smoke can impair detection of other scents. Visscher et al (1995) found that the antennae of honey bees (Apis mellifera) exposed to smoke were less responsive to both a floral odor and alarm pheromones. This suggests that detection of smoke could have shortterm impacts on foraging and defensive behaviors in insects (Visscher et al 1995).

Effects on wildlife habitat
Several studies have sought to quantify potential exposure to smoke for species that use specific habitats, although they did not evaluate the impact of smoke on animal health or behavior (e.g. O'Brien et al 2006, Bova et al 2011, Thompson and Purcell 2016. O'Brien et al (2006) measured air quality in a hole that could be used as a nesting cavity by Cuban parrots (Amazona leucocephala) during a prescribed burn. They found that as flames passed the cavity entrance, smoke accumulated inside for about 20 min, and CO 2 concentrations sharply increased to 2092 parts per million (ppm). O'Brien et al (2006) described these conditions as 'benign,' but pointed to the lack of research to date on inhalation exposure to air pollution in birds. Thompson and Purcell (2016) took a similar approach to assess the vulnerability of fishers (Pekania pennanti) to smoke during prescribed burns, measuring the concentration of CO in tree cavities that were previously used or could be used as den sites. They found that whereas levels of CO during burns might not be harmful to adult fishers, they are hazardous to developing fetuses and newborns. Dickinson et al (2010) used air quality data collected during prescribed burns to determine if smoke exposure endangered Indiana bats (Myotis sodalis). They determined that CO concentrations during lowintensity prescribed burns were unlikely to be dangerous but suggested that bats that roost in foliage or under bark could be more vulnerable to gas intoxication during fires than bats that roost in cavities or crevices where concentrations of poisonous gases are lower (Dickinson et al 2010). Dickinson et al (2009) found that CO concentrations during a prescribed burn did not exceed the threshold at which incapacitation of bats would likely occur; however, they noted that bats that roost closer to the ground would be more at risk of exposure to elevated concentrations of toxic gases. Cave-roosting bats in particular could be in danger of smoke inhalation because caves could fill with smoke before bats have a chance to escape (Dickinson et al 2009. These findings illustrate that exposure to air pollution during wildfires varies widely, depending on the specific habitats used by wildlife. Although we primarily focused our review on the direct effects of wildfire smoke on the health and behavior of wildlife, it is worth considering how smoke pollution indirectly affects wildlife by driving short-term changes in habitat. Smoke limits visibility (Haider et al 2019) and cools air temperatures (Robock 1991)-changes in the physical environment that could influence the health and behavior of wild animals. While vegetative succession following wildfires generates habitat for a wide variety of fauna (Smith 2000, Jones and Tingley 2021, Stillman et al 2021, smoke from wildfires also has immediate impacts on plant growth. For example, wildfire smoke triggers seed germination in plants that grow in fire-prone habitats (Van Staden et al 2000). Smoke can also positively or negatively influence plant productivity, depending on the extent to which aerosols absorb or scatter sunlight, as well as ambient concentrations of co-pollutants that damage plants (Hemes et al  2020). Furthermore, pollutants in smoke can deposit on soils or vegetation, which can indirectly affect wildlife (Phaneuf et al 1995). Plants can absorb toxins in smoke that, if consumed, could compromise the health of herbivorous animals (Tan et al 2018). Wildfire smoke also affects aquatic habitats (Jaafar and Loh 2014). Smoke limits how far light penetrates underwater, which can influence the vertical distribution of microorganisms (Urmy et al 2016) or primary productivity of coral reefs (Risk et al 2003). Atmospheric deposition of aerosols in smoke can also degrade water quality (Phaneuf et al 1995, Earl and Blinn 2003, Corbin 2012, which can in turn alter the composition of macroinvertebrate communities (Earl and Blinn 2003) and negatively affect the health of fish or other water-breathing animals (Gresswell 1999, Gonino et al 2019).

Discussion
We found that the available literature clearly demonstrates that wildfire smoke has direct and indirect effects on wildlife, including terrestrial and aquatic species (figure 2, table 1). Smoke inhalation contributes to adverse acute and chronic health outcomes in animals (Venn-Watson et al 2013, Black et al 2017, including CO poisoning, respiratory distress, neurological impairment, respiratory and cardiovascular disease, oxidative stress, and immunosuppression. These health effects could contribute to changes in wildlife activity, including movement (Erb et al 2018) and vocalization (Cheyne 2008). Animal behavior could also be influenced by changes in the physical environment that co-occur with smoke pollution, such as reduced sunlight or cooler air and water temperatures. Finally, many species that depend on fire-prone habitats have evolved to use smoke as a cue to engage in fire avoidance (Nimmo et al 2021) or energy-conserving behaviors  or perceive smoke as a signal of resource availability (Schütz et al 1999). Both the immediate, direct effects of wildfire smoke on the health and behavior of animals and the long-term impacts of smoke on wildlife habitat could ultimately influence the demography of wildlife populations (figure 2).
However, our review also demonstrates that a limited number of studies have investigated-or even considered-the impacts of wildfire smoke on wildlife (table 1). For decades, naturalists have observed how wildlife respond to smoke from wildfires (e.g. Komarek 1969, Braithwaite andEstbergs 1987) and noted the vulnerability of animals exposed to smoke during wildfires or prescribed burns (e.g. Geluso et al 1986). Yet, we found few peer-reviewed studies that directly investigated health outcomes or behavioral responses in wildlife associated with inhalation or detection of biomass smoke. After conducting a comprehensive search, we only identified 41 relevant studies, several of which did not explicitly test for an effect of smoke on animals and, instead, only considered how the presence of smoke could explain the responses observed (e.g. Cahill and Walker 2000, Jordaan et al 2020). Furthermore, research to date is unequally distributed across taxa (figure 5) and world regions (figure 4), with most studies conducted on mammals (39%) or insects (29%) in North America (37%), followed by Europe (17%) and Australia (15%) (figure 4). Our keyword searches were conducted in English, which could have influenced the geographic distribution of the studies we reviewed.
Researchers have used a variety of methods to investigate the impacts of smoke on wildlife, which makes it challenging to compare findings across existing studies. Monitoring animals before, during, and after wildfires or prescribed burns (e.g. Dickinson et al 2009, Jordaan et al 2020 allows researchers to study how free-living animals respond to the onset and progression of a smoke event and enables direct inference about the impacts of biomass smoke on wildlife. However, such studies are difficult to plan, tend to be logistically complicated, and can jeopardize the health and safety of the research team (Erb et al 2018). Alternatively, researchers have studied how animals respond to smoke generated in a controlled environment, such as a laboratory or outdoor enclosure (e.g. Nowack et al 2018, Tan et al 2018. This approach may be easier to implement because it does not require coordination with a fire management team or planning fieldwork around unpredictable wildfires. Controlled conditions also allow researchers to investigate specific health outcomes and behaviors in animals that would be difficult to assess in the wild. However, despite attempts to simulate biomass smoke that is representative of what animals would be exposed to during a wildfire smoke event in their natural habitat (e.g. Layne 2009), controlled exposure experiments cannot reproduce the exact air quality and visibility conditions animals are likely to encounter in the wild. In addition, controlled studies are often limited to smaller species that are relatively easy to capture (e.g. insects, small mammals), and captive animals could exhibit behavioral changes during experiments that arise from confinement and should not be attributed to air pollution exposure (Sterner 1993a(Sterner , 1993b. In sum, the experience of smoke exposure for animals is likely to be vastly different between in situ and controlled studies, which makes it difficult to compare their findings. Another limiting factor in connecting findings from existing research is the lack of robust air pollution measurements during field studies and experiments. Primary components of wildfire smoke include water vapor, CO 2 , CO, PM, volatile organic compounds, nitrogen oxides, and hazardous air pollutants, such as acrolein, benzene, and formaldehyde (De Vos et al 2009). However, the exact biological (Kobziar and Thompson 2020)  . Without measuring the concentrations of reactive gases and aerosols animals are exposed to, it is impossible to construct dose-response relationships for specific health outcomes (Jaafar and Loh 2014, Sanderfoot and Holloway 2017). Furthermore, the composition of smoke could affect the visual and olfactory cues that elicit behavioral responses in wildlife. For example, Komarek observed that the behavior of Carolina grasshoppers (Dissoteria carolina) varied depending on smoke conditions-when exposed to dense, white smoke, the grasshoppers ceased all activity, yet when exposed to black smoke, grasshoppers exhibited fire avoidance behaviors. To facilitate comparisons and meta-analyses of findings across studies, it is critical that future investigations move beyond qualitative descriptions of smoke and actually quantify exposure by measuring concentrations of specific gases and aerosols (Engstrom 2010, Sanderfoot andHolloway 2017).
More research is needed to identify which taxa and species are most threatened by wildfire smoke and determine how their vulnerability is influenced by physiology, behavior, and life-history strategy. It is well-established that birds are more sensitive to air pollution than other taxa (Brown et al 1997) and therefore more likely to be susceptible than other animals to direct health effects associated with smoke inhalation. Cetaceans, like birds, exchange most of the air in their lungs with each breath, which might put them at greater risk than other mammals of experiencing adverse health outcomes during smoke events (Venn-Watson et al 2013). Animal behavior and habitat use within and across species can also influence smoke exposure, thereby mediating risks. For example, bats that roost at higher heights are more protected from toxic gases during prescribed burns, and bats in torpor are less exposed to airborne toxins than they would be if they were active (Dickinson et al 2009). Furthermore, overlap between the timing of smoke pollution episodes and life-history events likely contributes to species-specific vulnerability to wildfire smoke. For instance, birds attending to chicks (Cahill and Walker 2000) or bats caring for pups (Dickinson et al 2009) are likely more threatened by heat and smoke during fires than adults not tending to offspring, and fossorial reptiles are in greater danger when they come to the surface to feed or seek a mate (Jordaan et al 2020). Comparing species distributions with spatial and temporal trends in air pollution could help wildlife managers determine if smoke should be considered alongside other threats, such as habitat degradation, when developing wildlife conservation plans.
Animals have evolved alongside wildfires for thousands of years, but megafires driven by climate change are generating novel disturbance stressors, such as large-scale smoke events, that could exert selective pressure on wildlife (Nimmo et al 2021). The fire regimes species are adapted to are changing, and the traits that allow them to co-exist with fire and smoke may not be sufficient in the age of megafires (Nimmo et al 2021). For example, typical fire avoidance behaviors might not be sufficient to protect wildlife from injury or morbidity during more severe, fast-moving fires (Engstrom 2010, Nimmo et al 2021, and even animals that are not in the direct path of fires can still be exposed to dangerous levels of wildfire smoke (Erb et al 2018) (figure 1). As climate change intensifies smoke pollution, more animals are at risk of acute and chronic health outcomes associated with smoke inhalation, which could lower survival and reproductive success (figure 2). Over time, animals may adapt behavioral responses to detect hazardous air quality and limit their exposure to toxic gases and aerosols; however it is also possible that during large-scale smoke events, even well-adapted species may not find any refuge. Fire-adapted species might respond to visual and olfactory cues during large-scale smoke events even when fires are far away, which could have cascading impacts on wildlife communities. Pyrophilous species that rely on smoke as a cue to navigate toward burned areas may become disoriented during large-scale smoke events that occur hundreds of miles from fires, which could lead to reduced fitness and increased vulnerability to predation. Other species that use visual and olfactory cues from smoke to initiate fire-avoidance behaviors may do so at the expense of unnecessary energy expenditure when a fire is not an immediate threat (Dickinson et al 2009). Animals that exhibit escape behaviors when it is smoky could also be more vulnerable to predation; natural history observations suggest that raptors hunt insects and small mammals at the edge of fires (Braithwaite and Estbergs 1987) and may be attracted to smoke plumes as a signal of prey availability (Hovick et al 2017). Additionally, prey species often use scent cues to detect and avoid predators (Blumstein et al 2002); large-scale smoke events may mask olfactory signals and affect the ability of prey to detect predators, further increasing their predation risk. Shifts in predator-prey interactions during smoke events could ultimately influence wildlife populations and community dynamics (figure 2).
Earlier and more prolonged wildfire seasons might pose novel threats to species that now encounter wildfire smoke during a critical stage of their life cycle, such as reproduction or migration. For example, the breeding phenology of songbirds may increasingly overlap with the smoke season, which could adversely impact songbirds in a reproductive state. Individuals that breed earlier, thereby avoiding reproductive activities during peak smoke season, could have higher reproductive success. This could lead to the evolution of traits, such as more synchronous or asynchronous breeding (Iwasa and Levin 1995), depending on the risks and benefits associated with the timing of breeding in relation to the threats posed by smoke events. Similarly, climate change is thought to be driving earlier breeding periods in many songbirds (Hällfors et al 2020), a trend that could be reinforced as smoke pollution worsens air quality during the summer months. Although wildfire smoke could function as an ecological disturbance that forces some species to adapt their life-history strategies, it is unlikely that all species threatened by smoke pollution will be able to adapt their phenology to match changing environmental conditions (Both and Visser 2001). More research is needed to assess how the frequency and timing of massive smoke events affects species adaptations to fire across different fire regimes.
We did not find any studies that explicitly linked wildfire smoke to demographic rates in wildlife populations (figure 2); however, emerging evidence suggests that the impact of large-scale smoke events on survival of wildlife species could be substantial. Yang et al (2021) found that air quality contributed to the spatial distribution of bird deaths in a mass avian mortality event in the Western United States in late summer 2020 (Yang et al 2021). This was not the first study to suggest that air pollution has negative demographic consequences for bird populations-a recent study also found that reductions in ozone (O 3 ) pollution in the United States prevented the loss of more than one billion birds (Liang et al 2020). Although O 3 is not a component of wildfire smoke, concentrations of O 3 can be higher on smoky days (Brey and Fischer 2016). Smoke inhalation has also been implicated in the death of insects (Tan et al 2018), reptiles (Jordaan et al 2020), and mammals (Singer et al 1989). Taken together, these findings emphasize the need to consider if and how wildfire smoke affects demographic rates in wildlife populations (figure 2).
Shifts in animal behavior during wildfire smoke events might ultimately affect the probability of observing wildlife, which has important implications for wildlife research and monitoring. For example, animals that use smoke as cue to engage in fireavoidance (e.g. burrowing) or energy-conserving behaviors (e.g. torpor) could be more difficult to observe during wildfire smoke events (Geiser et al 2018). Sanderfoot and Gardner (2021) investigated how wildfire smoke affected detection of 71 common bird species in Washington, United States and found that particle pollution during the wildfire season influenced the probability of observing 37% of study species-as PM 2.5 increased, 16 species were less likely to be observed and 10 species were more likely to be observed. These results suggest that species-specific behavioral responses to wildfire smoke ultimately influence researchers' ability to detect wildlife. Failing to account for how smoke affects observations of wildlife could bias inference about wildlife activity and population demographics (Sanderfoot and Gardner 2021).
To develop effective policy for wildlife conservation, we must rapidly expand our understanding of the effects of wildfire smoke on wildlife. We believe that ecologists and wildlife managers are wellpositioned to tackle this challenge by leveraging preexisting resources and infrastructure to address critical knowledge gaps. For example, camera traps, GPS tags, and acoustic recorders are often deployed in fireprone areas as part of long-term monitoring projects, many of which are likely to overlap with the wildfire season (figure 6). Data collected by these instruments could be paired with long-term air quality monitoring data to investigate how wildfire smoke drives shifts in observations of wildlife (e.g. Lee et al 2017) or explore specific behavioral responses to smoke pollution, such as movement and vocalization. This equipment could also be deployed to monitor wildlife before, during, and after prescribed burns. Studies of marked individuals pre-and post-fire could also provide insight into the direct effects of fires on demographic rates (Engstrom 2010). In addition, retrospective analyses of health records of captive animals housed in outdoor enclosures at zoos and aquariums could be used to assess how sudden, extreme smoke events influence the health of wildlife across a wide variety of taxa (Venn-Watson et al 2013, Black et al 2017. Finally, data from existing large-scale databases, such as the North American Breeding Bird Survey, eBird, eMammal, iNaturalist, Movebank, and Map of Life, could be used in correlative studies to link smoke exposure to observations of wildlife. To facilitate comparison of future studies, we recommend that researchers at minimum (1) identify the primary type of vegetation burned during prescribed burns or wildfires, or alternatively the substrate burned to generate smoke in controlled experiments and (2) incorporate measurements of PM 2.5 during exposure. PM 2.5 is often the focus of epidemiological investigations into the impacts of wildfire smoke on public health (McClure and Jaffe 2018, ; as such, there is a multitude of resources available to characterize particle pollution during smoke events or controlled experiments, including data from ground-based air pollution sensors, air quality models, and satellite instruments (Diao et al 2019). Data from groundbased air quality monitors are considered the 'gold standard' for estimating exposure to air pollution (Diao et al 2019) and are often available to the public-for example, the U.S. Environmental Protection Agency provides data on air pollution across the United States, Puerto Rico, and the U.S. Virgin Islands on the web at www.epa.gov/outdoor-airquality-data. If data from ground-based monitors is not available at relevant spatial and temporal scales, atmospheric scientists might rely on statistical interpolation or Land-Use Regression (LUR) models to build PM 2.5 exposure estimates (Jerrett et al 2005, Zou et al 2009. Alternatively, output from chemical transport models (CTMs) can be used in retrospective analyses and forecasting (Zou et al 2009). CTMs simulate air pollution by modeling transformation and transport of emissions (Jerrett et al 2005); examples of CTMs include the Community Model for Air Quality (CMAQ) and the Weather Research and Forecasting Model-Chemistry (WRF-Chem). Satellite data are also increasingly used to build PM 2.5 exposure estimates, although measurements from instruments on polar-orbiting satellites are only available once or twice a day (West et al 2016, Diao et al 2019. Some of these approaches could be readily implemented with minimal training (Diao et al 2019), but others require technical knowledge. Regardless, careful consideration of the location and behavior of the target population is essential in determining exposure to specific pollutants. We recommend that ecologists studying the impacts of wildfire smoke on wildlife collaborate with atmospheric scientists to build PM 2.5 exposure estimates using the best available tools.

Conclusion
The frequency and severity of large-scale smoke events are increasing as climate change intensifies global wildfire activity (Westerling et al 2011, Abatzloglou andWilliams 2016), posing new risks to wildlife (Nimmo et al 2021). Despite substantial research linking wildfire smoke to adverse health outcomes in humans, few studies have investigated the physiological and behavioral responses to wildfire smoke in animals (figure 4, table 1) (Erb et al 2018. However, research to date suggests that smoke inhalation contributes to negative acute and chronic health outcomes in a diversity of air-breathing animals, including mammals, birds, reptiles, and insects (figure 2, table 1). Detection of smoke triggers fire-avoidance and/or energyconserving behaviors in some wildlife species, and some species use smoke as a cue to navigate toward fires to take advantage of resources available in burned habitats (figure 2). However, even species that are adapted to fire-prone habitats are at risk of health outcomes linked to smoke inhalation, and it is unclear how they will cope with more extreme smoke pollution episodes. To inform the study and conservation of wildlife in a rapidly warming world, it is imperative that we expand our knowledge of wildfire smoke impacts on wildlife. Bridging the divide between the disciplines of ecology and atmospheric science will be essential in meeting this goal. We strongly recommend that scientists and managers build interdisciplinary partnerships and leverage existing data sets, infrastructure, and tools to quickly and efficiently address knowledge gaps and tackle research questions of global importance.

Data availability statement
No new data were created or analysed in this study.

Acknowledgments
We thank Tom Gillman for his help creating the conceptual model in figure 2, and Trent Roussin for his assistance reviewing titles from results of our initial keyword search. We thank two anonymous reviewers and the editor for helpful feedback on earlier drafts. We also acknowledge the use of imagery from the NASA Worldview application (https://worldview.earthdata.nasa.gov/), part of the NASA Earth Observing System Data and Information System (EOSDIS) (figure 1). This work was supported by the National Science Foundation Graduate Research Fellowship Program

Conflict of interest
We have no conflicts of interest to declare.

Appendix
We refined results from our initial keyword search in Web of Science to include only articles from categories deemed relevant to our literature review. We considered the top 100 Web of Science categories for each keyword search. Articles from the following categories were included from each keyword search. fire * AND