Does Seasonality of Leaf Terpene Content and Moisture Content Trigger Leaf Flammability Seasonal Variation?

Given the importance of terpenes and fuel moisture content (FMC) on ammability, this work aims at checking how these parameters affect leaf ammability of different native and Wildland-Urban Interfaces species (Pinus halepensis, Cupressus sempervirens, Cupressocyparis leylandii, and Hesperocyparis arizonica) across seasons in the French Mediterranean region. We found that the highest terpene diversity and content seasonally varied according to the species, with diterpene content being lower in spring for C. leylandii, while monoterpene and diterpene content being higher in summer and winter, respectively, for P. halepensis. Flammability and FMC varied according to the season but the pattern differed among species. A signicant correlation between the latters was rarely observed and occurred in only one season, differing among species. The correlations between ammability and terpenes were mostly highlighted using single compounds, compared to subgroups, and they presented seasonal patterns varying among species. Checking the seasonal effect of groups of terpene compounds on ammability, there were seasonal differences in these groups according to the species and the variable tested. Mostly, these signicant compounds were not the most concentrated. The best ammability drivers of each model, mostly diterpenes, except for P. halepensis whose ammability was mostly drove by mono- and sesquiterpenes, changed among seasons according to the ammability variable considered. When a best driver remained the same in different seasons, its effect on ammability could be opposite. In contrast, FMC was generally not a signicant explanatory parameter of leaf ammability or did not improve the t of models.


Introduction
Most plants in the Mediterranean basin are known to emit volatile (the least complex molecules with lower weight such as mono-or sesquiterpenes) or semi-volatile (heavier molecules such as diterpenes) terpenes (Llusià and Peñuelas 2000). Those plants emitting terpenes may or may not have specialized structures in which large pools of these compounds can be stored (Staudt et al. 1993;Seufert et al. 1995;Loreto et al. 1996;Llusià and Peñuelas 1998). In addition to leaf moisture and physical characteristics (such as leaf thickness, leaf density, leaf surface, or surface-to-volume ratio) largely known to in uence leaf ammability ( in re activity and these compounds' dynamics due to global change factors (drought, warming and carbon dioxide emissions). The increase in temperature and decrease in water availability, on-going in a context of climate change, may produce an increase in ammability of species, especially in those that store terpenes.
Previous studies focusing on plant species commonly found in wildland-urban interfaces of SE France highlighted, for species containing terpenes, the composition and content of mono-, sesqui, and diterpenes as well as that of the single terpene compounds and their impact on ammability (Romero et al. 2019; Ganteaume et al. 2021). In any case, these studies did not assess the seasonal variation of such molecules and they also showed that species belonging to the same genus or to the same family differed in their terpene diversity and content highlighting these compounds as species-related traits.
The current work aimed at explaining the possible seasonal changes in four species' leaf ammability by the change in foliar hydration and terpene content throughout the year. In other terms, we are asking if the leaf ammability seasonal variation was triggered by a seasonal variation in terpene content and FMC and if the ammability drivers changed according to the season. It was also interesting to check if different species of the same genus (e.g. Cupressus) or family (e.g. Cupressacaea) presented the same pattern throughout the year, especially in terms of the seasonal variation in terpene content.

Materials And Methods
Species Studied and Sampling. The species studied in the current work, all conifers (one Pinacaea: Pinus halepensis Mill. 1768 and three Cupressacaea: Cupressus sempervirens L. 1753, Hesperocyparis arizonica Greene 1882 Bartel, (Adams et al. 2009), formerly Cupressus arizonica, and Cupressocyparis leylandii A.B. Jacks. and Dallim 1926) are common in the Wildland-Urban Interfaces (WUI) of the French Mediterranean region. P. halepensis is the only species native to SE France while H. arizonica comes from the southwestern USA and C. sempervirens forms natural forest stands, mostly in the eastern part of the Mediterranean basin (in some parts of Tunisia, Italy, and Greece, for instance). The latter species can present two distinct varieties: var. horizontalis characterized by a broad pyramidal crown and horizontally spreading branches (constituting the natural stands) and var. pyramidalis (or fastigata) characterized by a compact conical crown and small angles between branches and trunk which is the variety studied in the current work. C. leylandii is an intergenic hybrid of the yellow cedar (Callitropsis nootkatensis D. Don 1824) native to northwestern America (the North American continent) and of the Monterey cypress (Hesperocyparis macrocarpa Hartw. 1847), a species endemic of Monterey Bay in California (USA). All of these species can be involved in re propagation from wildland vegetation to nearby buildings, especially when they are used in ornamental hedges that provide a strong horizontal fuel continuity.
Leaves of the different species were sampled in Le Tholonet (southeastern France) where the climate is typically Mediterranean. Leaf collection was carried out in three seasons in 2016: winter, spring, and summer (i.e. January-February, April-May, and August-September) in order to grasp the variability in terms of weather conditions over a year that could affect terpene and moisture content. We chose to work only on fresh green leaves from plant canopy as a previous work on these species (Romero et al. 2019) showed that their terpene content did not differ from that of litter leaves (i.e. entire leaves undergoing the rst stage of decomposition on the ground).
For each species, a maximum of 25 g of mature green leaves was collected on ve different individual plants, with at least 4 m distance between plants. For each plant sampled, 6 g were used for the burning experiments, 5 g for FMC measurements, and 1 g for terpene analysis. Sampling was conducted at least 48 h following a precipitation event to avoid any impact of recent rain on FMC. Collected leaves were placed in plastic bags that were stored in a cool box for immediate transportation to the laboratory, minimising changes in water content. The samples were burned the same day they were collected on returning to the laboratory.
Terpene Identi cation and Quanti cation. After sampling, leaves were stored at -80°C to avoid any metabolic transformation. Terpene content was analyzed once for each species, using 500 mg of leaves collected from the ve different plants as presented in Romero et al. (2019). Terpenes were qualitatively and quantitatively analyzed using a gas chromatography coupled to a mass spectrometry (GC-MS, 7890B-Agilent Technologies®) as described in Romero et al. (2019). Terpene identi cation was achieved based on the molecule retention time (which was compared to that of the pure standard when available) as well as the molecule mass spectrum which was compared to available libraries (Adams 2007;Nist 2011). To complete this identi cation, experimental retention indexes were calculated for each molecule identi ed and compared to the theoretical retention indexes of these libraries. The terpene content - For each plant species, the contribution of terpenes was investigated both, at the subgroup (comprising monoterpenes, sesquiterpenes, and diterpenes), and the single compound (within each subgroup) levels.
Flammability Experiments. For each species, thirty samples of 1 g leaf samples were burned using a 500 W epiradiator composed of a 10 cm radiant disc according to the methodology presented in Romero et al. (2019). The variation in temperature was recorded every second during burnings using a thermocouple (chromel-alumel, k type, 30 μm diameter) positioned 1 cm above the disc center. As soon as the fuel was in contact with the epiradiator surface, time and temperature recordings were started. Five ammability variables were measured during the burning experiments: (i) time-to-ignition, (TTI, s), de ned as the time necessary for the fuel to ignite once laid on the radiant disc; (ii) ignition temperature (tTTI, °C), de ned as the temperature recorded when the ame appeared; (iii) aming duration (FD, s), time elapsed between the ame occurrence and its extinction; and (iv) the maximum temperature reached during the burning (Tmax, °C).
Just before the burning experiments, three samples of 5 g-leaf subsamples of each individual were ovendried for 48 h at 60°C in order to measure their moisture content (i.e. FMC, calculated on a dry mass basis) at the time of burning. Data Analyses. The statistical analyses were performed on each species' dataset taking into account the content of terpenes assessed at the subgroup, and single compound levels as well as FMC as explanatory factors of ammability. Although leaf thickness was an important driver of ammability (along with other structural leaf traits such as surface-to-volume ratio) and was taken into account in our previous work (Romero et al. 2019), it was not included in the present analyses as this parameter does not signi cantly change from one season to another. The different ammability variables (using a single mean value per individuals of each species) were used as dependent variables. All tests were performed using StatGraphics Centurion XVII -X64 software (StatPoint Technologies, Inc®).
First, we performed variance analyses (one-way ANOVA) to highlight the effect of season on FMC, terpene content and ammability for each species. In these analyses, the Kruskal-Wallis test was performed instead of the Fisher test because of the lower amount of data per season.
Then, for each species and season, simple linear regression analyses (Fisher test) were performed to highlight any signi cant correlations (positive or negative) existing between leaf parameters (FMC, terpenes) and ammability. When FMC explained a signi cant proportion of the variability of the relationship with a given ammability variable, we used the residuals of the regression as a moisturecorrected measure of this variable. This corrected variable was then regressed against the terpene content in order to only highlight the effect of terpenes avoiding the bias of the above-mentioned factors (see Pausas et al. 2016). In the simple linear regression analyses, the adjusted R 2 value was used to account for the variation in ammability.
Finally, partial least square (PLS) regression analyses were performed to determine the relative importance of the different fuel characteristics ( rst using terpene content only, then adding FMC in the models) on each ammability variable. These analyses allowed highlighting a pool of signi cant terpene compounds, along (or not) with FMC, affecting ammability as well as those being the best drivers according to the season. The signi cance of components in the models was determined according to uncertainty tests carried out within a full cross-validation. The scaled regression coe cients of the PLS models provided information on the effect (positive or negative) of each parameter on ammability metrics and its relative weight in the tted model (absolute value) indicated the relative importance in predicting each ammability variable.

Results
Seasonal Variation of Terpenes and FMC. Analyses by GC-MS of the terpenes contained in leaves of the four species studied led to the identi cation of 55 different terpene compounds, Cupressocyparis leylandii presenting the highest terpene diversity (34 compounds identi ed: 11 monoterpenes, 12 diterpenes, and 11 sesquiterpenes) regardless of the season (Tab. 1). The season of the highest terpene diversity differed among species, summer for C. leylandii (displaying mostly diterpenes) as well as for Heterocyparis arizonica (displaying mostly sesquiterpenes), and winter and summer for Cupressus sempervirens (displaying mostly diterpenes and sesquiterpenes). For Pinus halepensis, the highest diversity was observed in the three seasons and concerned the monoterpenes. Overall, there was a low inter-individual variation of this number of compounds (<20%) regardless of the season (except for diterpenes for H. arizonica in winter and for P. halepensis in spring and summer).
Regarding the terpene subgroup, the monoterpene diversity did not signi cantly vary according to the season, except for C. sempervirens (KW= 10.29, p= 0.006) for which the number of molecules was the highest in winter and summer. Sesquiterpene diversity was signi cantly higher in summer for H. arizonica and in winter for C. sempervirens. Except for H. arizonica, the diterpene diversity seasonally varied and was signi cantly higher in summer for C. leylandii and C. sempervirens (KW= 14.0, p= 0.0009 and KW= 12.87, p= 0.0016, respectively) and in winter and spring for P. halepensis (KW= 10.96, p=0.0042). There was no seasonal variation in diversity of mono-and sesquiterpenes for C. leylandii as well as for P. halepensis, and of mono-and diterpenes for H. arizonica. For each species, the composition of the extracted compounds as well as the percentage of each family on the total amount of terpenes are presented in Suppl. Mat. 1.
Regarding the single terpene compounds, the content of the most concentrated compounds characterizing each species were the sesquiterpenes β-caryophyllene for P. halepensis and cadina-1(6),4diene <cis> for H. arizonica, the diterpene totarol for C. sempervirens, and the monoterpenes δ3-carene, αpinene, and β-pinene for C. leylandii. It is worth noting that some compounds were not found all year round. Indeed, for C. leyandii, two diterpene compounds (abietal-4 epi and cembrene A) were detected only in summer (and were among the main compounds, i.e. content ≥ 0.1 mg g -1 : 0.34 and 0.14 mg g -1 , respectively) as well as the two sesquiterpenes muurol-5-en-4-one and muurol-5-en-beta-ol <cis> (both minor compounds) for H. arizonica. For P. halepensis, the diterpene cembrene was not detected in summer (but was among the main compound in winter: 0.34 mg g -1 ). C. sempervirens presented the highest number of compounds (eight in total, one monoterpene, ve sesquiterpenes, and two diterpenes) missing over a year (mostly in spring) but all of them were minor compounds (Suppl. Mat. 1).
Regardless of the species, the annual total terpene content did not vary according to the season but the species studied presented different seasonal patterns for the different terpene subgroups (Fig. 1). For C. leylandii (Fig. 1a), only diterpene content signi cantly varied among seasons, the content being lower in spring than in the other seasons. Regardless of the season, monoterpenes presented the highest values in this species. For P. halepensis (Fig. 1b), mono-and diterpene content seasonally varied but presented different patterns, the monoterpene subgroup being more concentrated in summer while that of diterpenes showed higher values in winter. Regardless of the season, the sesquiterpene subgroup presented the highest content, mostly due to β-caryophyllene (on average, 45% of the total content). H. arizonica and C. sempervirens did not present any seasonal variation in their terpene content, regardless of the subgroup (Fig. 1c, d).
The results obtained at the terpene subgroup level were con rmed by the analyses at the single compound level. For C. leylandii, only the content of six diterpene compounds differed among seasons, all of them among the main compounds in one season or another (cembrene being a major compound in winter and summer, isophyllocladene, manool oxide and manool oxide 4 epi only in winter while abietal 4 epi and cembrene A were detected in summer only). For this species, the most concentrated compounds were found in winter and were diterpenes. For P. halepensis, only the content of the diterpene cembrene varied seasonally and this molecule was not identi ed in summer; the most concentrated compounds were found in summer and belonged to the subgroup of monoterpenes. As for the subgroup level, the content of each single compound did not varied across seasons for H. arizonica and C. sempervirens. Even if in these species, some compounds were not present all year round (therefore presenting marked seasonal differences), all of them were minor compounds (Fig. 2). One of the explanations of the very low number of signi cant results was the high variability of content among seasons for most compounds, regardless of the species (e.g. the monoterpene myrcene for P. halepensis, the diterpene totarol for C. sempervirens, or the monoterpene β-pinene for C. leylandii; Suppl. Mat. 1).
The moisture content of the leaf samples of the different conifer species ranged from 91.41 (± 13.72) % (P. halepensis) to 125.95 (± 4.03) % (H. arizonica). Except for H. arizonica that presented the highest values regardless of the season, FMC varied according to the season but the pattern differed among species, opposing C. leylandii (increasing trend from winter to summer) to C. sempervirens and P. halepensis (highest values in winter and lower foliar hydration in spring and summer) (Fig. 3).
Role of Terpenes and FMC on Species' Flammability among Seasons. According to the variable tested, ammability could present or not a seasonal variation and the species studied differed in their seasonal pattern (Fig. 4). Except for tTTI which did not show a clear seasonal trend, all of the other variables varied seasonally for C. leylandii, ammability being the lowest in winter (but it is worth noting the long TTI in summer as well). Only Tmax varied according to the season for H. arizonica (highest values in spring) and tTTI for C. sempervirens (highest values in summer) while tTTI and TTI differed among seasons for P. halepensis (highest ammability in spring). Surprisingly, a signi cant correlation between FMC and ammability was observed only in one season (which differed according to the species), except for C. leylandii for which the results were not signi cant throughout the year (Table 2). Only one ammability variable was signi cantly affected by FMC (which also differed according to the species): TTI for H. arizonica (in summer) and C. sempervirens (in winter) as well as Tmax for P. halepensis (in spring). These variables were therefore FMC-corrected when used in the univariate regression analyses linking ammability with terpenes. Regardless of the species, FMC decreased ammability.
The effect of the terpene subgroup content on ammability varied seasonally according to the species, except for H. arizonica (Table 3). Terpene content (diterpene) affected ammability (negatively, increasing TTI) only in summer for C. leylandii while, for C. sempervirens, this terpene subgroup enhanced ammability (increasing FD) in both spring and summer. For P. halepensis, the role of terpenes varied according to the season, diterpene and sesquiterpene content enhancing ammability (decreasing tTTI) in winter and summer, respectively, but presenting the opposite effect in spring. During this latter season, monoterpene content presented contrasting effects on ammability according to the variable considered (decreasing Tmax but also TTI).
When the same analyses were performed at the single compound level, taking into account both main and minor compounds, a higher number of signi cant correlations was highlighted between terpenes and ammability, the seasonal trend differing among species (Table 4). For H. arizonica, using the main compounds in the models, the correlations were not signi cant regardless of the ammability variable considered, as at the subgroup level. However, when the minor compounds were taken into account in the analyses, the correlations became signi cant in summer (only one compound affected ammability: the diterpene totarol, decreasing TTI) and in winter (four compounds, one sesquiterpene and three diterpenes differing from totarol, all enhancing ammability). For C. leylandii, signi cant relationships were highlighted in winter and summer but not in spring, regardless of the terpene subgroup. The main compounds involved were diterpenes and mostly mitigated ammability (all except manool oxide that presented the opposite effect on TTI and tTTI in winter). The minor compounds signi cantly affecting ammability, also in winter and summer, were diterpenes (still mostly mitigating ammability) and sesquiterpenes. The molecules differed between the two seasons but it is worth noting that one of the main compounds in winter, manool oxide, was found among the minor compounds in summer. For C. sempervirens, the signi cant molecules and their impact differed among seasons. They did not affect ammability in winter and only one compound (monoterpene δ3-carene) presented a signi cant effect (positive) on ammability in spring. Most signi cant correlations were highlighted in summer and involved the two main signi cant compounds, the diterpene totarol enhancing ammability in contrast to the monoterpene α-pinene. The signi cant minor compounds were all diterpenes and mitigated ammability except abietadiene. The role of the terpene compounds (both main and minor) on ammability was signi cant in the three seasons for P. halepensis, especially in winter and spring. The signi cant molecules (all sesquiterpenes or monoterpenes except one diterpene: manool oxide) and their impact differed among seasons (e.g. the monoterpene myrcene increasing ammability in winter in contrast to summer) except for the sesquiterpene germacrene (increasing FD in summer and winter) and the monoterpene α-pinene (mitigating ammability in winter and spring).
Seasonal Variation of Combined effects of Terpene Compounds and FMC on Flammability. Checking the possible seasonal patterns in the species studied, we found that there were seasonal differences in the groups of terpene compounds signi cantly explaining ammability and in their effects according to the species and the variable taken into account. The results were not signi cant in spring for C. leylandii (except for tTTI; Fig. 5), in winter, regardless of the variable considered for C. sempervirens (Fig. 6) and for H. arizonica (Fig. 7) regarding tTTI, and in summer regarding Tmax and tTTI for P. halepensis (Fig. 8).
The signi cant compounds highlighted in summer mostly enhanced FD for C. leylandii and H. arizonica in contrast to other molecules mitigating ammability (regarding Tmax and tTTI) for the latter species and for C. sempervirens (regarding Tmax). Those identi ed in winter presented the opposite effect on ammability for C. leylandii as well as a positive effect for H. arizonica and P. halepensis (only the monoterpene terpinolene regarding TTI). The compounds identi ed in spring enhanced ammability for C. sempervirens (FD and Tmax) and P. halepensis (tTTI ) while it was the opposite for C. sempervirens (TTI) and P. halepensis (FD and Tmax). Only three compounds were signi cant at the same time in winter and summer (e.g. the sesquiterpenes cedrol and β-elemene affecting Tmax with opposite effects) for C. leylandii and only one compound for H. arizonica (the diterpene nezukol correlated with TTI). Also three compounds were signi cant in spring and summer for C. sempervirens (e.g. the diterpene totarol correlated with FD), for P. halepensis (e.g. the monoterpene limonene correlated with Tmax), and for H. arizonica (e.g. the diterpene ferruginol correlated with Tmax, TTI, and tTTI). Two compounds were signi cant in winter and spring (e.g. the diterpene sempervirol correlated with FD, Tmax, and TTI) for H. arizonica and P. halepensis (e.g. the monoterpene terpinolene with TTI). The opposite effect could also occur between different variables (e.g. the monoterpene terpinen 4 ol with Tmax and Ttti for C. leylandii). Regarding this latter species, a compound signi cant in the three seasons was found (the sesquiterpene germacrene D presenting the opposite correlation with FD in winter-summer vs spring). Opposite correlations were also observed between different compounds of the same terpene subgroup accross different seasons (e.g. the diterpenes thunbergol vs totarol with Tmax for C. sempervirens) and sometimes within the same season (e.g. the monoterpenes α-pinene vs sabinene hydrate with FD in spring for H. arizonica).
When the signi cant compounds highlighted in each model were compared to the main terpene compounds (content ≥ 0.01 mg g -1 ) identi ed for each species and season, it could be noted that signi cant compounds were not the most concentrated with a few exceptions. The following molecules mostly differed among species: the monoterpene α-pinene in spring and summer for H. arizonica, the monoterpene δ3-carene in winter for C. leylandii, the diterpene totarol as well as the monoterpene αpinene in spring and summer for C. sempervirens, and the monoterpene myrcene in summer as well as the sesquiterpene β-caryophyllene in spring for P. halepensis (Fig. 4 and Suppl. Mat. 1).
Regardless of the species, the best drivers of each model (terpene compound presenting the highest regression coe cient explaining most of the ammability variation) changed among seasons according to the ammability variable considered (Tab. 5). When the best driver was the same at different seasons and for a given ammability variable, its effect on this variable could change (for instance, the monoterpene terpinolene positively affected TTI in winter but negatively in spring) or not (for example, the diterpene sempervirol positively affected FD in winter and in spring). The best drivers were most often diterpene compounds (displaying mostly a negative effect in winter and summer for C. leylandii, a positive effect in spring but the opposite in summer for C. sempervirens, and a positive effect regardless of the season for H. arizonica), regardless of the season and the variable, except for P. halepensis. For this latter species, ammability was mostly driven by mono-(with a positive effect in spring in contrast to the other seasons) and sesquiterpene (mostly with a positive effect on ammability) compounds (Tab.

5).
The seasonal variation of the proportion of ammability (R 2 ) explained by terpene compounds (Tab. 6) was highlighted in the different models. When the results were signi cant within a given species, the season in which the coe cient of determination, R 2 , was the highest varied according to the ammability variable. The highest proportion of FD and TTI explained by terpenes occurred in the models run on the summer datasets except for H. arizonica and P. halepensis (spring datasets) while that of tTTI occurred in the models run on the spring dataset except for P. halepensis (winter dataset). The highest proportion of Tmax was found in the models run on the summer datasets for C. sempervirens and H. arizonica, on the winter dataset for C. leylandii, and on the spring dataset for P. halepensis.
Regarding the role of FMC in the explanation of ammability, in most cases, either FMC taken as parameter in the models or the results of the analyses were not signi cant, or the t did not improve when FMC was added to the models (Suppl. Mat. 2). The t improved when FMC was added to models in spring for C. leylandii (all results signi cant in contrast to the analyses taking into account only terpenes, except for tTTI for which the proportion of variation explained remaining the same) and for P. halepensis (regarding Tmax, FMC being the best driver). A better t also occurred in winter for H. arizonica and P. halepensis (for both species, regarding TTI only) as well as for C. sempervirens (regarding TTI and tTTI), in contrast to the previous results, and in summer (for FD and tTTI). For the former species, FMC was the only best driver of TTI in summer and in winter for the latter.

Discussion
Seasonality of Traits and Flammability. A seasonal variation in ammability for Mediterranean species has already been reported by several authors (e.g. Rodriguez Añón et al. 1995; Alessio et al. 2008b).
According to these previous studies, ammability -characterized by one (for C. sempervirens and H. arizonica) and up to three (for C. leylandii) signi cant variables -varied seasonally. A strong seasonality was found for TTI and Tmax (highest ammability in spring) only for C. leylandii, P. halepensis (for the former variable), and H. arizonica (for the latter) in contrast to C. sempervirens whose ammability was surprisingly higher in winter. Indeed, the driest seasons at the time of the sampling were spring and summer (Suppl.Mat.3).
FMC also presented seasonal variations (except for H. arizonica) with differences among species. This result agreed with Pelizzaro et al. (2007) who showed that different Mediterranean species did not show the same seasonal pattern of FMC throughout the year and, in some of them, moisture content did not vary. Usually, for tree species, there is no pronounced decrease in FMC at the driest season, as found Viegas et al. (2001) for P. halepensis, the trees presenting deeper root systems than most Mediterranean shrubs (Kummerow 1981;Correia et al. 1992; Alessio et al. 2004). However, this result was in accordance with what we found for H. arizonica but not for P. halepensis. In some cases, FMC showed seasonal patterns that differed from those of ammability (e.g. for C. sempervirens). C. sempervirens and P. halepensis presented lower moisture content in summer and spring, the driest seasons in 2016 (Suppl. Mat. 3), but the former was more ammable in winter according to tTTI only (the results for the other variables were not signi cant). This means that some species were more ammable during the most humid season (winter), as C. leylandii. This species presented the lowest FMC in winter in contrast to the other species despite the fact that they were all sampled from the same location and at the same time; therefore, they were subjected to the same weather conditions throughout the year. One explanation for this latter species' peculiar pattern would be that the plants sampled could have been watered during summer (as located not far from housing  Alessio et al. 2008b), the seasonal pattern of ammability is usually in accordance with the progressive increase in temperatures and water demand. This would also agree with the results obtained for P. halepensis. Indeed, in the current study, the seasons with the lowest leaf hydration regarding this species were spring and summer (to a lower extent), corresponding to the highest ammability. However, the effect of FMC (negative) on ammability was signi cant only in one season, which differed according to the species (higher FMC in summer for H. arizonica and in winter for C. sempervirens, affecting only TTI, as well as in spring affecting only Tmax for P. halepensis). We did not nd, as Pelizzarro et al. (2007), that most species showed a maximum of moisture content and an increase in TTI in spring; but, as these authors highlighted, some of our species (H. arizonica, C. sempervirens) did not present any seasonal variation of TTI, or of FMC (for H. arizonica, only).
Terpene content and diversity also presented seasonal variations with patterns that could differ from those of ammability according to the species. Indeed, total terpene diversity was higher in summer for C. leylandii (due to higher diversity in diterpenes) and H. arizonica (due to higher diversity in sesquiterpenes) while P. halepensis and C. sempervirens presented the same pattern as ammability (high diversity and ammability in spring and winter respectively). The seasonal variation of terpene content also differed according to the level of terpene identi cation (i.e. subgroup and single compound) but was signi cant only in two species. Indeed, diterpene content was higher in winter and summer for C. leylandii and in winter for P. halepensis for which monoterpene content was higher in summer as well, which agreed with Alessio et al. (2008a). In contrast to our results regarding this latter species, Lluisiá and Peñuelas (2000) highlighted a drop in terpene content in spring, regardless of the compound.
Seasonal variations in terpene concentrations (and emissions) have been reported in several previous studies and have con rmed that the season of peak concentrations seemed to be species-speci c: in spring for Scots pine and some junipers (Adams, 1970), in summer for Douglas r and other coniferous species (Wagner et al. 1990; Gambliel and Cates 1995; Zou and Cates 1995), in autumn-winter but with larger terpene concentrations also observed at the end of summer, for P. halepensis ( Lluisá and Peñuelas 2000). In the current study, when a variation of the terpene subgroup content was highlighted, it was due to the seasonal variation of some terpene single compounds. However, most terpene single compounds did not present a signi cant seasonal variation (none for C. sempervirens and H. arizonica, only one for P. halepensis, and six for C. leylandii), for which only trends could be highlighted mostly due to the high inter-individual variation of their content (high standard deviation; Suppl. Mat. 1). Often, their seasonal trends could differ among species (e.g. monoterpene myrcene presented the highest content in spring for C. leylandii while it was in summer for P. halepensis). Previous studies also showed a spatial variation in terpene content resulting from differences in environmental conditions, such as drought stress or soil leylandii's and C. sempervirens' ammability being mainly affected by diterpenes (e.g. negative effect in summer for the former and positive effect in spring and summer for the latter) while P.halepensis was more affected by the other subgroups. Moreover, a same subgroup could present antagonist effects on ammability from one season to another (negative effect of diterpenes in winter becoming positive in spring). Regarding the single compound level, the seasonal trend also differed among species. The effect was positive mostly in winter for H. arizonica (two diterpenes) while it was negative mostly in winter (six diterpenes) for C. leylandii and mostly in summer (three diterpenes, one monoterpene) for C. sempervirens. Variable effects due to mono-and sesquiterpenes in the three seasons (between two and ve compounds involved) occurred for P halepensis. This underlines the importance of searching for terpene effects at a more re ned level than subgroup content, also using the minor compounds in the analyses (given some correlations became signi cant). Our results also showed that the effect of one single compound could differ among seasons, depending on the ammability variable considered, con rming the result at the subgroup level (the monoterpene myrcene presenting a positive effect on Tmax in winter and on TTI in summer for P. halepensis). This result stresses the importance of taking into account several components of ammability, as already highlighted by Ganteaume et al. (2021). It is worth noting that, in contrast to the results of Alessio et al. (2008), species with higher content in terpene volatiles were not the most ammable. Indeed, P. halepensis, the most ammable species in the current work, did not present the highest terpene content (as opposed to C. leylandii). This could be due to the interaction between FMC and terpene content. Checking the possible seasonal patterns in the species studied, we found that there were seasonal differences in the groups of terpene compounds signi cantly explaining ammability and their effects (positive or negative) contrasted according to the species and the variable taken into account. Only a few compounds were signi cant at the same time in two seasons, mostly for H. arizonica (only one in the three seasons, the sesquiterpene germacrene D for P. halepensis). Except for H. arizonica, these signi cant compounds could present an opposite effect on ammability among seasons according to the variable tested. These opposite effects could also occur between different variables and between different compounds of the same terpene subgroup in different seasons and sometimes in the same season, regardless of the species. Previous studies already showed that monoterpene content, for instance, presented the opposite effect on ammability according to the molecule considered (Owens et al. 1998). Except for a few signi cant compounds, which mostly differed among species and seasons, they were not the most concentrated molecules.
Regardless of the species, the best drivers of each model mostly changed among seasons according to the ammability variable considered. When the best driver was the same accross different seasons, the effect on the variable could change in some cases. These compounds were mostly diterpenes (with variable effects according to the species and the season), except for P. halepensis mostly driven by monoterpenes (variable effect in spring) and sesquiterpenes (positive effect on ammability) compounds. For a given species, the seasonal variation of the proportion of ammability (R 2 ) explained by terpene compounds varied according to the ammability variable (in spring and summer for H. arizonica and C. sempervirens, winter and spring for P. halepensis, and in the three seasons for C. leylandii. Despite the fact that some studies showed that ammability strictly depends on leaf water availability for a given species, the moisture content and the plant structure interact and indirectly in uence the re behaviour also by the terpene content and that, in some cases, there was an interaction between both leaf parameters. In the current study, we also highlighted that each species presented its own seasonal pattern regarding the different parameters. Regarding the role of FMC in the explanation of ammability, most of the time, FMC was not signi cant or did not improve the t of the models (most changes occurred in spring for C. leylandii, when the terpene content was the lowest, as well as in winter and summer for C. sempervirens, but at a lower extent). Other fuel parameters could play a more important role in the explanation of ammability in the seasons for which the ts were the poorest.

Conclusions
The aim of this work as to assess what part of terpene content and FMC played on the seasonal variation of ammability of different conifer species found in the WUI of SE France (Pinus halepensis, Cupressus sempervirens, Cupressocyparis leylandii, and Hesperocyparis arizonica) and if the ammability drivers varied seasonally. The season of the highest terpene diversity (diterpene composition for C. leylandii as well as P. halepensis, and sesquiterpene composition for H. arizonica) and the terpene subgroup content differed among species, presenting different seasonal patterns (diterpene content lower in spring for C. leylandii, while monoterpene and diterpene content higher in summer and in winter, respectively for P. halepensis). FMC varied according to the season but the pattern differed among species and could also differ from that of ammability according to the variable tested.
Signi cant correlations between FMC and ammability were rarely observed (solely in one season, for only one ammability variable, differing among species) in contrast to those with terpene content (however, more frequent at the single compound level than for the terpene subgroup), the seasonal trend differing among species. Checking the seasonal variation of groups of terpene compounds on ammability, we found that there were seasonal differences in these groups according to the species and the variable tested, the molecules often displaying opposite effects (among ammability variables or seasons). As a whole, these signi cant compounds were not the most concentrated, except for a few molecules. The best drivers of each model, mostly diterpenes, except for P. halepensis (mono-and sesquiterpenes), changed among seasons according to the ammability variable considered. When the best driver was the same at different seasons, for a given ammability variable, its effect on this variable could change. In contrast, mostly, FMC was not a signi cant parameter or did not improve the t of models.
Even if Pinus halepensis is the most ammable species, the ammability of the others studied species could also be enhanced by some terpene single compounds, especially in spring and summer (while it is in winter and summer for P. halepensis), meaning that these species can act on the re risk not just in summer, as it is commonly assumed in the Mediterranean regions, and this has to be taken into consideration in the forest and wildland-urban interfaces management in this area.