Roads are amongst the most visible, direct anthropic activity impacting natural systems. It is estimated that road length will increase to 4.7 million km by 2050 in the world, an increase of 14%–23% compared to current estimates, and much of this increase will happen in countries with high biodiversity and important natural ecosystems (Meijer et al., 2018). Brazil, for instance, has more than 1.7 million km of roads responsible for 56% of national transport, and the fourth largest road network in the world (CNT, 2021). The country’s Southeast region has the largest road network, as well as the greatest number of vehicles (approximately 53,400,000) (IBGE, 2022). The nation’s road network is expected to increase by 10% by 2050 (Meijer et al., 2018).

Roads cause several negative impacts on the environment and biodiversity (for a review, see Bennett, 2017). The most obvious direct and immediate impact of roads on vertebrate biodiversity is through wild animal collision with vehicles. In Brazil, it is estimated that 475 million wild vertebrates are killed by vehicles every year (CBEE, 2015), although estimates vary substantially (e.g., for mammals, Pinto et al., 2022; González-Suárez et al., 2018) and may be underestimated (Slater, 2002; Teixeira et al., 2013) .

Another factor that significantly impacts wild animals is fire. Fires can occur naturally due to lightning strikes and it is an important process in certain ecosystems, such as in the Cerrado biome (Mistry, 1998; Fidelis and Pivello, 2011); however, the frequency and magnitude of fires caused by human activities are increasing worldwide, thereby negatively impacting native ecosystems (Keeley and Pausas, 2019; Kelly et al., 2020), and this can also be observed in Brazil (Pivello et al., 2021). Brazil is the country with the highest frequency of fires in South America (Li et al., 2020). Between 1985 and 2020, 19.6% of the Brazilian territory had been directly affected by fires at least once (Mapbiomas, 2021). Whether intentional or accidental, fires are responsible for the death of many wild animals every year. In 2020, for instance, an estimated 17 million animals died in the Pantanal biome due to fires (Tomas et al., 2021).

Considering the negative impacts that both fires and road networks can have on biodiversity, an aspect that has practically not been investigated is the synergistic effects of these factors on wildlife. Apart from the fact that roads are places where fires can start more frequently (Ricotta et al., 2018), an immediate effect of fires on wildlife might be the killing of animals due to collision with vehicles on roads running through or near areas affected by fires. Fires trigger a range of movements, both within, away from, and towards the burned (or burning) area (Nimmo et al., 2019). While small terrestrial animals may not be able to flee in response to fires, instead finding refuge underground or elsewhere locally (Ford et al., 1999; Engstrom, 2010) or dying in situ, larger animals and volant species are able to escape more easily by moving farther away from the fire site (Singer and Schullery, 1989; Engstrom, 2010; Silveira et al., 1999). However, in escaping from fires and moving more, animals may be more frequently killed while crossing roads, thereby leading to an increase in the number of roadkills during and soon after fires, particularly if fire severity is not high enough to cause large animal mortality (Jolly et al., 2022). Even when the fire is over, behavioral changes in animal movement may cause animals to move more frequently. For instance, bison and other grazers are attracted to burned areas (Shaw and Carter, 1990). Likewise, salamanders on burned areas move more, probably to find better habitats (O’Donnell et al., 2016), and some predators are attracted to burned areas due to increased detection of prey in the recently formed open areas (McGregor et al., 2015; Leahy et al., 2016; Doherty et al., 2022). In sum, while fire causes an immediate increase in species movement, roadkilling is expected to increase during and shortly after fires, at least when fires are not of high intensity. However, to the best of our knowledge, only four studies have been published reporting the effect of fires on roadkills (Coombs, 2015; Vujović et al., 2015; Bernardo, 2021; Cruz et al., 2021). Three of these found an increase in roadkilling of avian and mammal* species (Cruz et al., 2021; *marginally significant) and small reptiles (Coombs, 2015; Vujović et al., 2015) due to fires, while the fourth one found a decrease in roadkills except for some species of mammals (Bernardo, 2021).

In this study, we investigated whether there is a relationship between fires and the number of roadkilled wild vertebrates on a highway in Southeast Brazil. We hypothesized that there would be an immediate increase in roadkills during and after nearby fires.

There were 66 fires recorded between 2014 and 2021 along the analyzed highway stretch. The intensity of roadkills in the entire study area was 0.018 roadkills/km/day (8,318 roadkills in 2,557 days; Table 1). There was a significant interaction between before-after and treatment-control sections (F = 5.232; df = 1; p = 0.025), but no main effects of before-after (F = 0.771; df = 1; P = 0.383) or treatment-control (F = 0.612; df = 1; P = 0.437). The Wilcoxon rank test showed a significant increase in the number of roadkills after the fires in the (main) affected highway sections in comparison with before the fires (Z= −2.33; N = 66; P = 0.020), with an effect size of rc = 0.53. There were 17 roadkills recorded up to 7 days before the fires in the affected sections (Table 1), which corresponds to an intensity of 0.016 roadkills/km/day. The total number of roadkills in the 7 days after the fires was 42 (Table 1), which represents an intensity of 0.039 roadkills/km/day, 2.44 times or 144% higher than that before the fires and 2.17 times, or 117% higher than the roadkills over the entire highway stretch analyzed.

In addition, there was also an increase in the number of roadkills on the affected sections in comparison with paired, neighboring (control) sections after the fires (Z = −2.204; N = 66; P = 0.028). Taken together, these results show there was a significant increase in the number of roadkills after fires only in the main sections of the highway, that is, those closer to the fires or directly affected by them.

Since both fires and roadkilling are processes responsible for a significant number of wild animal deaths worldwide, it is essential to investigate how these two processes can act together to increase the number of vertebrate deaths, particularly in megadiverse countries with high projected road construction, such as Brazil. In this context, this is one of a very few studies that has investigated the synergistic effect of fires and roadkilling of wild vertebrates. We found an increase of 144% in roadkills in highway sections affected by fires, which we consider an intermediate to high effect size (rc = 0.53) that should not be disregarded. We believe this pattern of roadkill increase due to fires is common, at least for fires of relatively small magnitude and duration, but it is underestimated due to the lack of studies.

The very few studies on the effect of fire on roadkilling have reported contrasting results depending on the taxonomic group, which is probably explained by the distinct species ability to escape fires as well as fire intensity. For instance, in the study of Cruz et al. (2021) carried out in Mira (Portugal), they found an increase of around 54% in roadkills after fires, with higher mortality of avian and mammal* species (*marginally significant). They argued that this was probably because most species of these taxa move more rapidly and farther away while looking for refuge and food during/after fires. Reptiles and amphibians, in turn, showed lower numbers of roadkills after the fires, probably because most of them died burned or asphyxiated before reaching roads (Cruz et al., 2021). In South Africa, in turn, 32 lizards (Chamaesaura anguina) were roadkilled after a fire, although the species is rarely found dead along roadways in other situations (Coombs, 2015). The author argued that this species moves relatively fast and was therefore able to escape the fire (Coombs, 2015), which apparently was not of high magnitude. In a study carried out in Montenegro, there was a considerable increase in deaths of Hermann's tortoises (Testudo hermanni), both burned and roadkilled, after a large fire nearby a road (Vujović et al., 2015). Finally, in a study with mammals on highway BR-163, Mato Grosso state, Brazil, there was an overall reduction in roadkills after fires, except for capybaras (Hydrochoerus hydrochaeris) and other semiaquatic mammals, which were subject to an increase in roadkills up to 6 months after the fires. The author argued that the fire was of high intensity and probably caused the death of most animals (Bernardo, 2021). As highlighted before, most of the fires recorded in this study were relatively small, being extinguished within a day, and all of them were 1 km or closer to the highway. However, we could not statistically verify how the number of roadkills varied per taxon.

Although the roadkill data used were not corrected for the bias generated by carcass persistence and searcher efficiency and taxonomic/body size groups, which are sources of error recognized in the literature (Santos et al., 2016; Teixeira et al., 2013), we do not believe our results are biased by these factors since the highway stretch was surveyed 24 h a day, 7 days a week by trained employees, including a biologist who helped during weekdays. Moreover, we do not think carcass persistence and searcher efficiency varied before and after fires, or in control and affected sections, therefore not affecting the analyses performed here.

The scientific research of road impacts on vertebrates rarely focuses on multifactor or synergistic impacts on wild vertebrates (Jaeger and Torres, 2021), although in some cases data can be obtained for such purposes. We believe this work can be used as a model to guide future research into the synergistic effect of fires and roadkills worldwide, and that such effect should be considered in environmental impact assessments (Jaeger and Torres, 2021), since roads not only cause roadkilling and increase the probability of human-induced fires (Ricotta et al., 2018), but also the fires themselves may increase roadkilling even more, as shown by this study.

In this study, we adopted an experimental design which allowed for temporal and spatial controls. The BACI/BACIP designs provide less biased results (Smith, 2002; Smokorowski and Randall, 2017; Christie et al., 2020) but has not been used to evaluate the impact of fire on roadkilling. Considering that data on fires can be obtained through satellite images and that long-term projects to monitor vertebrate roadkilling are performed routinely in many places around the world, it is relatively straightforward to investigate the effect of fires on roadkilling using this approach. Data on fires can be obtained retrospectively, as done in this study, thereby allowing such investigations. It is clear, however, that long-term data are usually needed to obtain adequate statistical power for analysis at lower taxonomic levels.

Future studies should disentangle how distinct species respond to fires and at which specific spatial and temporal scales such responses are significant. In the short term, some species may move to avoid being burnt during fire (Geluso and Bragg, 1986; Grafe et al., 2002; Pausas and Parr, 2018), while in the longer term researchers may need to track shifts in resources that emerge during post-fire vegetation succession. The studies of Cruz et al. (2021) and Bernardo (2021) found higher roadkills for some species several months after the fires had been extinguished, and Bernardo (2021) found distinct roadkill patterns depending on the scale analyzed. In addition, in Southeastern Brazil, Hilário et al. (2021) found that the number of rescued wild vertebrates involved in traffic collisions increased with the number of fires two months before in the studied region. Moreover, an additional variable that might be worth investigating is fire direction, that is, whether it spread towards the road or away from it, and fire magnitude. Fires of high magnitude that last for several days/weeks and burn large areas may simply kill most of the animals, decreasing roadkill numbers at least just after the fire.

Regarding roadkill mitigation strategies and considering the specificities of the highway stretch analyzed here, we recommend that traffic velocity should be reduced in areas recently affected by fires, in order to decrease roadkilling probability. As proposed by Coombs (2015), limiting traffic flow and/or vehicle speed on roads nearby areas affected by fires and/or installing corridors beneath roads can also be applied in some situations to facilitate migration of species to fire-free areas.

Finally, it is important to note that both fire susceptibility and roads have increased over the last decades in Brazil and worldwide (Meijer et al., 2018; Singh and Huang, 2022; CNT, 2021; Oliveira et al., 2022). We therefore anticipate that the synergistic effect of fires on roadkilling will increase worldwide in the next decades, which highlights once more the need to investigate the combined effect of such anthropic impacts on wild vertebrates. Recently, Nimmo et al. (2019) called attention to the need for integration between movement ecology and fire ecology. We propose including road ecology in such integration whenever roads are present.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper