Environmentally persistent free radicals are ubiquitous in wildfire charcoals and remain stable for years

Globally landscape fires produce about 256 Tg of pyrogenic carbon or charcoal each year. The role of charcoal as a source of environmentally persistent free radicals, which are precursors of potentially harmful reactive oxygen species, is poorly constrained. Here, we analyse 60 charcoal samples collected from 10 wildfires, that include crown as well as surface fires in forest, shrubland and grassland spanning different boreal, temperate, subtropical and tropical climate. Using electron spin resonance spectroscopy, we measure high concentrations of environmentally persistent free radicals in charcoal samples, much higher than those found in soils. Concentrations increased with degree of carbonization and woody fuels favoured higher concentrations. Moreover, environmentally persistent free radicals remained stable for an unexpectedly long time of at least 5 years. We suggest that wildfire charcoal is an important global source of environmentally persistent free radicals, and therefore potentially of harmful reactive oxygen species. Wildfire charcoal is an important source of environmentally persistent free radicals, a precursor to harmful reactive oxygen species, according to electron spin resonance spectroscopy analyses of natural pyrogenic carbon.

L andscape fires (i.e., wildfires, land management fires, and agricultural burns) affect 3-5 million km 2 annually around the world, burning~4% of Earth's vegetated land surface 1 . As a result of incomplete combustion, part of the biomass burnt during fire is transformed into charcoal (also known as pyrogenic carbon) 2 . The proportion of burnt biomass converted into charcoal is highly variable, and ranges from negligible up to 30%, depending on the characteristics of the vegetation and the fire itself 3 . Landscape fires have recently been estimated to produce about 256 Tg/year of charcoal globally, equivalent to 12% of the total carbon released by these fires 4 . Charcoal is more resistant to degradation than unburned biomass and can be preserved in soils, waters, and sediments for centuries to millennia 2 . About 200 Pg of charcoal is estimated to be stored globally in surface soils 5 and 480-1440 Pg in marine sediments 6 .
Wildfire charcoal is ubiquitous in many landscapes around the world and a comprehensive understanding of its effects on the climate system, ecosystem function, and environmental health is therefore fundamental. The impacts of charcoal on carbon and nutrient cycles have been widely investigated 7 , as well as its role as a source of pollutants such as polycyclic aromatic hydrocarbons 8 . There is, however, no information available to date concerning its potential as a source of environmentally persistent free radicals (EPFRs). In contrast to other free radicals with lifetimes in the range of microseconds to milliseconds, EPFRs can remain stable for prolonged periods of time (typically days to months) and are bound to particle surfaces. This makes them highly susceptible to redistribution within, and having impacts on, the natural environment [9][10][11] . EPFRs have recently been proposed as a new class of contaminant 12 because they can undergo a variety of reactions with water to form reactive oxygen species (ROS) 13,14 , which can have detrimental effects on ecosystem functions (see Fig. 1). EPFRs containing particles have thus been linked to adverse impacts on biological systems, including inhibition of plant germination 15,16 , reduced enzymatic activity 17 , cytotoxicity to single cells 18 , toxic effects on both bacteria 19 , and aquatic invertebrates 20,21 . Therefore, the presence of EPFRs in charcoal and the subsequent formation of ROS may lead to yet unexplored effects on ecosystem functions and human health. For example, in the context of post-fire erosion and transport of wildfire charcoal into water bodies 22 .
No measurements of EPFRs have been reported from wildfire charcoals to date, but high EPFR concentrations have previously been reported in human-derived pyrogenic products; for example, in particulate matter (PM) from fossil fuel combustion 23 and biochar 14 , a man-made product of pyrolysis with some similarities to charcoal but produced under industrial conditions 24,25 . There is currently no information available concerning the stability of charcoal-bound EPFR in natural environments but, under laboratory conditions, EPFRs in biochars were shown to remain stable for months 10,12 . Based on these observations, we hypothesized that EPFRs would also be present in wildfire charcoal, with EPFR concentration being affected by fire conditions, type of fuel, and time since fire. To test this hypothesis across a variety of charcoal types, we analysed the EPFR content of 60 charcoal samples collected in different regions of the world from ten wildfires (crown fires and surface fires) in a range of ecosystems (forest, shrubland, and grassland) and climatic regions (boreal, temperate, tropical, and subtropical). In addition, different sized charcoal particles were also analysed, as well as charcoal derived from different types of feedstocks (e.g., wood, leaf litter, grass). Furthermore, to assess the stability of charcoal EPFRs in natural environments, we also included samples collected at different times after fire (aged 0-5 years) (Tables 1 and S1). We found that all charcoals analysed via electron spin resonance (ESR) spectroscopy have high concentrations of EPFRs (10 18 to 10 19 spin/g; every spin/g corresponds to a free electron/ g), orders of magnitude higher than those previously reported for soils and airborne PM 12 . EPFR concentrations were related to charcoal carbonization degree, which in turn is determined largely by the type of fuel material (feedstock) and charring conditions. We also observe that EPFR concentrations in charcoal remained stable for an unexpectedly long time of at least 5 years.

Results and discussion
High concentrations of EPFRs in wildfire charcoals. All samples analysed contained notable EPFR concentrations (ranging from 2.46 × 10 18 to 1.49 × 10 19 spin/g, Fig. 2) that were orders of magnitude higher than those previously reported for soils and airborne PM (~10 16 to 10 17 spin/g) 12 . Concentrations in woody charcoal were generally at the upper end of the range of EPFR concentrations reported for woody biochars 10,26,27 .
A positive relationship has been reported between the concentration of transition metal oxides in biochar and EPFRs formation, with a charge transfer between the aromatic structures of biochars and transition metal oxides identified as an important mechanism in EPFRs formation 14,[28][29][30] . We found no such correlation in our data between transition metal content and EPFRs formation (Fig. S1, Electronic Supplementary Material), which may be due to the different formation conditions of charcoal and biochar, as discussed in the next section. Therefore, other factors such as the charring duration and fuel/feedstock properties also need to be considered to better understand EPFRs formation in wildfire charcoal.
EPFR abundance depends on carbonization. A positive relationship was found between the EPFR abundance in wildfire charcoal samples and their carbon content (Fig. 3a, Pearson's R = 0.83, p < 0.05), and a negative relationship with their molar O/C ratio (Fig. 3b, Pearson's R = −0.80, p < 0.05). These two parameters are indicators of the degree of carbonization, with the degree of carbonization increasing with higher carbon contents and lower molar O/C ratios 31 . This indicates that the degree of carbonization is a determining factor for the different EPFR concentrations observed in wildfire charcoal samples. The degree of carbonization is a function of both the properties of the original fuel material (i.e. feedstock), and formation conditions (i.e., temperature, duration, and oxygen availability) 32 .
Regarding type of original fuel material, carbonization during wildfires is generally higher for woody fuels than for fine materials such as leaves or plant litter, which usually burn more completely because of their larger surface area per unit volume 3,33 . In addition, the high lignin content of woody materials also leads to them experiencing less complete combustion during charring 34 . Furthermore, the pyrolysis of lignin is reported to result in a higher concentration of EPFRs than the pyrolysis of cellulose 10 . All these factors have probably contributed to the higher EPFR concentrations observed in woody charcoals than in fine-size charcoal for the range of samples measured in this study (Fig. 2). Both the degree of carbonization as well as the aromaticity of the original fuel material (e.g., linked to lignin content) can be linked to the size of aromatic clusters within the wildfire charcoal. These aromatic clusters are likely responsible for the stability of EPFRs 26 .
For most samples, produced under flaming conditions, the charring duration (i.e. the time above 300°C) 34 was relatively short, ranging between 80 and 680 s. For these samples, the trends described above were consistently observed when applying ESR spectroscopy to quantify EPFRs. This measurement also included the determination of g-factors. The g-factor is a characteristic value for the EPFR paramagnetic centre's electronic structure. It can thus be used to characterize the neighbouring chemical structure of the EPFR. All measured samples had a g-factor value of 2.0032 ± 0.0003, which indicates that the EPFRs were associated with oxygen-containing functional groups (n = 57, Table S2 and Fig. S2 in the Supplementary Material). Oxygen-centred EPFRs are stable in the air, whereas carbon-centred EPFRs are not 26 , which may explain why only oxygen-centred EPFRs were measured in all samples. It is possible that C-centred EPFRs indirectly formed O-centred EPFRs during their degradation. However, we have not identified any evidence that would confirm   Table 1.
this hypothesis. Two samples were obtained from burnt logs after smouldering combustion instead of flaming combustion (Table S1). For these two samples, the charring duration was more than 7200 s (Table S1). As shown in Fig. 2, they report some of the lowest EPFR concentrations despite having the highest degrees of carbonization (i.e. highest C% and lowest O/C; Fig. 3). We hypothesize the prolonged exposure to high temperatures caused an extremely high degree of carbonization and loss of oxygen surface functional groups, which in turn decreased the EPFR concentrations. This is in good agreement with previous studies on biochars 35 and PM including soot 36 , which report reduced EPFR concentrations at very high degrees of carbonization. These results suggest that EPFR formation during both wildfires and biochar production can be expressed as a bell-shaped curve, with EPFR concentrations first increasing with the degree of carbonization (as most charcoal samples here), but subsequently decreasing following further carbonization, as reported for highly carbonized biochars 29,35,37 and the two smouldering charcoal samples here. It is worth highlighting that, of all samples examined in our study, these latter two charcoal samples are the most similar to biochar, in terms of the prolonged heating times compared to wildfire charcoal 24 .
In addition, to assess if the degree of carbonization depends on the particle size of the feedstock, we analysed woody charcoal particles in four different sizes classes (i.e., 0.5-1.0, 1.5-2.0, 2.5-3.0, and >5 cm) produced from the same original feedstock material (Pinus sylvestris) in the same fire (Swedish boreal forest fire: Fire ID 8 in Table 1). The EPFR concentrations tended to increase with increasing particle size, but the differences between the samples were small (p = 0.10, Fig. S3 in the Supplementary Material). Larger particles were very probably derived from larger pieces of wood, which in turn are likely to burn for longer periods of time and hence would be more highly carbonized, as indicated by a slight decrease in the molar O/C ratios from 0.19 ± 0.01 for pieces in the 0.5-1.0 cm size range to 0.16 ± 0.01 for pieces >5 cm.
EPFRs in wildfire charcoal remain stable for years. To explore the in situ stability of EPFRs in wildfire charcoal aged in natural environments, samples from the Canadian boreal forest fire (Fire ID 1, Table 1) were taken immediately after the fire, and also after 1 year, 3 years, and 5 years (Fig. 4 and Table S3). These samples showed an initial slight increase in EPFRs after the first year, possibly due to the formation of secondary radicals during oxidative aging as previously described for pyrolysis of tobacco 38 . Over the subsequent four sampling years, the EPFR concentrations decreased only slightly, with the most pronounced reduction occurring in the fifth year (Fig. 4). Throughout the 5 years, the EPFR concentrations remained of the order of 10 19 spin/g, indicating a high level of stability in the natural environment. This may partially be explained by the occurrence of a catalytic EPFR cycle causing continuous re-generation of EPFRs, similar to that previously described by Khachatryan et al. 39 , and/or the structural enclosure of EPFRs within the charcoal which would make them not highly reactive.
Environmental implications of EPFRs in wildfire charcoal. The EPFR concentrations in the wide range of wildfire charcoals examined here (2.46 × 10 18 to 1.49 × 10 19 spin/g) are orders of magnitude above those previously reported in soils and airborne Fig. 3 The degree of carbonization is related to measured concentrations of environmentally persistent free radicals. The abundance of environmentally persistent free radicals increased with increasing carbon content (a, Pearson's R = 0.83, p < 0.05) and decreased with increasing molar oxygen to carbon ratios (b, Pearson's R = −0.80, p < 0.05), both of which are proxies for the degree of carbonization. Black blocks (■) represent woody charcoal, grey triangles (▼) represent fine-size wildfire charcoal. Woody charcoal samples 6 and 7 (circled) were excluded from the correlation analysis because of the distinctive formation conditions for these two samples (smouldering for prolonged periods of time rather than flaming combustion, see Table S1). Error bars represent standard deviation. Fig. 4 Environmentally persistent free radical concentrations in woody charcoal over time since fire. Samples from Fire ID 1 (see Table 1), showing a high level of stability for EPFRs in a natural boreal conifer forest environment. Different lower-case letters at the top of the bars indicate statistical differences, with p < 0.05; all samples were analysed in triplicate. Error bars represent standard deviation. PM (~10 16 to 10 17 spin/g) 12 . Although the range of samples collected was limited to ten fires, the overall evidence of high EPFR concentrations in wildfire charcoals and their stability in the natural environment is strong. Since the most recent estimates put the global annual production of wildfire charcoal at 256 Tg 4 , wildfire charcoal may represent a substantial, as well as ubiquitous, global source of EPFRs. Charcoal is highly mobile in postfire landscapes and substantial quantities eventually reach the hydrological network, in either particulate or dissolved form 40,41 . When in contact with water, be it in rivers, lakes, oceans, or soils, EPFRs can form various types of ROS such as, for example, hydroxyl radicals (•OH) 13,39 . In laboratory-produced biochars, most of the ROS-forming reactions take place within the first few hours of water contact 10 . However, our results indicate that EPFR in wildfire charcoals can remain highly stable in natural environments over far longer periods of time. This suggests these longlived EPFR may not be very reactive and/or undergo to date unexplored catalytic cycles similar to previously described mechanisms described by Khachatryan et al. 39 , including charge transfer reactions with transition metal oxides from the surrounding environment and/or the mineral fraction within the wildfire charcoals. In both cases, they may still produce some quantities of ROS when in contact with water. Whether EPFRs in wildfire charcoal do indeed pose an environmental risk remains unclear, and the rate of ROS production from EPFRs in field aged wildfire charcoal will be an important area for future research. Even at low concentrations, EPFR-derived ROS in environmental systems may influence ecosystem functions including organic matter transformation 42 , plant germination 15,16 , aquatic algal growth 43 , and enzymatic activity 17 , as well as having adverse effects on bacteria 19 and invertebrates 20,21 . Wildfires are a natural, necessary and frequent disturbance in many of the world's ecosystems 44 . However, fire frequency and severity are increasing in many regions due to human activity and a warming planet 4,45 . For example, the annual area burned in California has increased by over 400% over the last five decades and ecosystems that are less prone to wildfires, such as tropical rainforests or arctic tundra, are now suffering extensive fires 45,46 . In such regions, the increasing EPFR quantities, and possible ROS production, associated with higher wildfire charcoal production may be of concern and thus the potential environmental effects require further investigation.

Methods
Wildfire charcoal sample collection. Samples were collected from areas burnt by either actual or experimental wildfires, between 1 day and a few months after the fire. The maximum temperatures (T max ) and charring times for the experimental fires were recorded using K-type thermocouples (≤1 mm diameter) that had been attached to the surfaces of the woody samples or were placed within the litter/grass fuel layer, and connected to data loggers (Lascar, Easylog). The loggers were buried in the adjacent soil to protect them from thermal exposure and recorded the temperature at the interface between the samples and the surface-flames at 1-10 s intervals during the fire (depending on the fire). Additional samples were collected at annual intervals following the Canadian boreal forest fire, for up to 5 years after the fire. Details concerning the sampling locations, the characteristics of the fires, and the ecosystems are provided in Table S1. Samples were either air-dried or dried in an oven at low temperature (<45°C, >48 h), and then stored in airtight containers in the dark prior to analysis. Each sample was crushed with a ceramic mortar, sieved to <250 µm, and then dried overnight in a laboratory oven at 80°C before taking any measurements.
Physical and chemical characterization of wildfire charcoals. Total C, H, and N concentrations were determined using an elemental analyser (Elementar Var-ioMacro) and O concentrations then calculated by mass balance: O% = 100 − (C + H + N + ash). The mineral ash content was determined by weighing samples before and after heating at 750°C for 6 h in a muffle furnace.
ESR spectroscopy. Solid dry samples were analysed at room temperature using an X-band Bruker Elexsys-II E500 ESR spectrometer (Bruker Biospin GmbH), with a modulation frequency of 100 kHz and a microwave frequency of 9.8 GHz. The sweep width was 100 G, the sweep time 20 s, the modulation amplitude 1.00 G, the centre field 3508 G, the microwave power 0.16 mW, and the resolution along the X axes was 1024 points. ESR spectra were simulated and the area under the curve determined by double integration of the spectrum. A reference-free quantitation of the number of spins was performed, as has been described elsewhere 47 . The unit spin/g can be understood as "free radical/g".
The g-factor was estimated using Bruker Win ESR Acquisition software. Measurements were performed using 100 µL capillaries and a sample weight of 2-10 mg. All measurements were replicated using three subsamples and the results normalized to sample weight.
Data analysis. For the data analysis and production of figures, we used OriginPro 2018 software. Correlation analyses were performed using linear regressions and reporting Pearson's R. The two smouldering-produced charcoal samples 6 and 7 were excluded from our analysis of the relationships between carbon content, O/C ratios, and EPFR abundance because they strongly differed in their material properties compared to other charcoal samples. To compare multiple sample groups, the Tukey's post-hoc test was used for pairwise comparison following single factor ANOVA in OriginPro 2018.

Data availability
The data generated during the current study are available at https://doi.org/10.5281/ zenodo.4564056.