Tree species flammability based on plant traits: A synthesis

Spatio-temporal feature attribution of European summer wildfires with Explainable Artificial Intelligence (XAI)

Wildfires are among the most destructive natural disasters globally. Understanding the drivers behind wildfires is a crucial aspect of preventing and managing them. Machine learning methods have gained popularity in wildfire modeling in recent years, but their algorithms are usually complex and challenging to interpret. In this study, we employed the SHapley Additive exPlanations (SHAP) value, an Explainable Artificial Intelligence method, to interpret the model and thus generate spatio-temporal feature attributions. Our research focuses on the forest, shrub and herbaceous vegetated areas of Europe during the summers from 2018 to 2022. Using burned areas, meteorology, vegetation, topography, and anthropogenic activity data, we established a wildfire occurrence model using random forest classification. The model was highly accurate, with an Area Under the Receiver Operating Characteristic Curve of 0.940. The SHAP results revealed six features that significantly influence wildfire occurrences: land surface temperature (LST), solar radiation (SR), Temperature Condition Index (TCI), Normalized Difference Vegetation Index (NDVI), precipitation (Prep), and soil moisture (SM). The tipping points for the positive or negative shifts in contributions are around 30 °C (LST), 2.20e7 J/m^2 (SR), 0.2 (TCI), 0.78 (NDVI), 2 mm/h (Prep), and 0.18 (SM). These predictors display strong spatial variability in their contribution levels. In Southern Europe, LST and SR emerge as the primary contributors to wildfires, making substantial impacts. Conversely, in regions at mid and high latitudes in Europe, NDVI, Prep, and SM assume a more prominent role in promoting wildfires, albeit with relatively smaller contributions. Furthermore, the disparities in SHAP values for TCI and SMCI across different years provide valuable insights into the effects of variation in regional meteorological conditions on wildfires. Our study provides a new approach to uncovering the spatio-temporal variations of feature contributions, which will help to better understand the mechanism of wildfire occurrence and enhance prevention and mitigation.

Vegetation-fire feedbacks increase subtropical wildfire risk in scrubland and reduce it in forests

Climate dictates wildfire activity around the world. But East and Southeast Asia are an apparent exception as fire-activity variation there is unrelated to climatic variables. In subtropical China, fire activity decreased by 80% between 2003 and 2020 amid increased fire risks globally. Here, we assessed the fire regime, vegetation structure, fuel flammability and their interactions across subtropical Hubei, China. We show that tree basal area (TBA) and fuel flammability explained 60% of fire-frequency variance. Fire frequency and fuel flammability, in turn, explained 90% of TBA variance. These results reveal a novel system of scrubland-forest stabilized by vegetation-fire feedbacks. Frequent fires promote the persistence of derelict scrubland through positive vegetation-fire feedbacks; in forest, vegetation-fire feedbacks are negative and suppress fire. Thus, we attribute the decrease in wildfire activity to reforestation programs that concurrently increase forest coverage and foster negative vegetation-fire feedbacks that suppress wildfire.

Measuring flammability of crops, pastures, fruit trees, and weeds: A novel tool to fight wildfires in agricultural landscapes

Fires on agricultural land account for 8-11 % of the total number of fires that occur globally. These fires burn through various crops, pastures, and native vegetation on farms, causing economic and environmental losses. Fire management on farms will be aided by understanding the flammability of plant species as this would allow the design of low-flammability agricultural landscapes, but flammability data on large numbers of agricultural species are lacking. Many crop and vegetable species are assumed to be low in flammability, but this has rarely been tested. Therefore, we examined the shoot and whole-plant flammability of 47 plant taxa commonly grown on farms in Canterbury, New Zealand, which included many globally common temperate agricultural crops. We demonstrated that most of the agricultural species were low to very low in flammability, with many of them (24 taxa; 51 %) not igniting in the experimental burning. Among different crop types, fruit crops and cereals had significantly higher flammability, while taxa categorized as vegetable crops, grazing herbs, pasture grasses, pasture legumes, and weeds were lower in flammability. We further showed that taxa with lower moisture content, higher retention of dead material and faster moisture loss rates were higher in flammability. The strong variation of flammability between the studied taxa suggests that the selection of suitable low flammability species and strategic redesign of agricultural landscapes with fire-retardant planting can be a useful tool to reduce fire hazards and impacts of wildfires in agricultural landscapes.

Broadleaf tree phenology and springtime wildfire occurrence in boreal Canada

Although broadleaf tree species of the boreal biome have a lower flammability compared to conifers, there is a period following snow melt and prior to leaf flush (i.e., greenup), termed the "spring window" by fire managers, when these forests are relatively conducive to wildfire ignition and spread. The goal of this study was to characterize the duration, timing, and fire proneness of the spring window across boreal Canada and assess the link between these phenological variables and the incidence of springtime wildfires. We used remotely sensed snow cover and greenup data to identify the annual spring window for five boreal ecozones from 2001 to 2021 and then compared seasonality of wildfire starts (by cause) and fire-conducive weather in relation to this window, averaged over the 21-year period. We conducted a path analysis to concomitantly evaluate the influence of the spring window's duration, the timing of greenup, and fire-conducive weather on the annual number and the seasonality of spring wildfires. Results show that the characteristics of spring windows vary substantially from year to year and among geographic zones, with the interior west of Canada having the longest and most fire-conducive spread window and, accordingly, the greatest springtime wildfire activity. We also provide support for the belief that springtime weather generally promotes wind-driven, rather than drought-driven wildfires. The path analyses show idiosyncratic behavior among ecozones, but, in general, the seasonality of the wildfire season is mainly driven by the timing of the greenup, whereas the number of spring wildfires mostly responds to the duration of the spring window and the frequency of fire-conducive weather. The results of this study allows us to better understand and anticipate the biome-wide changes projected for the northern forests of North America.

Climatic conditions affect shoot flammability by influencing flammability-related functional traits in nonfire-prone habitats

Plant flammability is an important driver of wildfires, and flammability itself is determined by several plant functional traits. While many plant traits are influenced by climatic conditions, the interaction between climatic conditions and plant flammability has rarely been investigated. Here, we explored the relationships among climatic conditions, shoot-level flammability components, and flammability-related functional traits for 186 plant species from fire-prone and nonfire-prone habitats. For species originating from nonfire-prone habitats, those from warmer areas tended to have lower shoot moisture content and larger leaves, and had higher shoot flammability with higher ignitibility, combustibility, and sustainability. Plants in wetter areas tended to have lower shoot flammability with lower combustibility and sustainability due to higher shoot moisture contents. In fire-prone habitats, shoot flammability was not significantly related to any climatic factor. Our study suggests that for species originating in nonfire-prone habitats, climatic conditions have influenced plant flammability by shifting flammability-related functional traits, including leaf size and shoot moisture content. Climate does not predict shoot flammability in species from fire-prone habitats; here, fire regimes may have an important role in shaping plant flammability. Understanding these nuances in the determinants of plant flammability is important in an increasingly fire-prone world.

Leaf decomposition and flammability are largely decoupled across species in a tropical swamp forest despite sharing some predictive leaf functional traits

Decomposition and fire are major carbon pathways in many ecosystems, yet potential linkages between these processes are poorly understood. We test whether variability in decomposability and flammability across species are related to each other and to key plant functional traits in tropical swamp forests, where habitat degradation is elevating decomposition and fire regimes. Using senesced and fresh leaves of 22 swamp tree species in Singapore, we conducted an in situ decomposition experiment and a laboratory flammability experiment. We analysed 16 leaf physical and biochemical traits as predictors of decomposability and components of flammability: combustibility, ignitability and sustainability. Decomposability and flammability were largely decoupled across species, despite some shared predictive traits such as specific leaf area (SLA). Physical traits predicted that thicker leaves with a smaller SLA and volume decomposed faster, while various cation concentrations predicted flammability components, particularly ignitability. We show that flammability and decomposability of swamp forest leaves are decoupled because flammability is mostly driven by biochemical traits, while decomposition is driven by physical traits. Our approach identifies species that are slow to decompose and burn (e.g. Calophyllum tetrapterum and Xanthophyllum flavescens), which could be planted to mitigate carbon losses in tropical swamp reforestation.

CO2 -induced biochemical changes in leaf volatiles decreased fire-intensity in the run-up to the Triassic-Jurassic boundary

The Triassic-Jurassic boundary marks the third largest mass extinction event in the Phanerozoic, characterized by a rise in CO₂ -concentrations from c. 600 ppm to c. 2100-2400 ppm, coupled with a c. 3.0-4.0°C temperature rise. This is hypothesized to have induced major floral turnover, altering vegetation structure, composition and leaf morphology, which in turn are hypothesized to have driven changes in wildfire. However, the effects of elevated CO₂ on fuel properties, such as chemical composition of leaves, are also important in influencing fire behaviour, but yet have not been considered. We test this by selecting three Triassic analogue species grown experimentally in different atmospheric compositions, and analyse variations in leaf chemistry, and leaf level flammability. These data were used to inform a fire behaviour model. We find that all three species tested showed a reduction in their volatile component, leading to lower flammability. Accounting for these variations in a model, our results suggest that leaf intrinsic flammability has a measurable impact on modelled fire behaviour. If scaled up to ecosystem level, periods of elevated CO₂ may therefore be capable of inducing both biochemical and morphological changes in fuel properties, and thus may be capable of influencing fire behaviour.

Drivers of Flammability of Eucalyptus globulus Labill Leaves: Terpenes, Essential Oils, and Moisture Content

Mediterranean climate regions have become more vulnerable to fire due to the extreme weather conditions and numerous Eucalyptus globulus plantation areas. The aim of this study is to analyze the fire hazard related to E. globulus in a forest fire scenario, based on the contrast of thermochemical parameters and their relationship with chemical properties, considering the predominant exotic forest species (E. globulus, Pinus radiata, Acacia dealbata, and Acacia melanoxylon) present in the Valparaiso region, Chile. The results revealed that although all of the studied species were highly flammable, E. globulus was extremely flammable, as its leaves contain high concentrations of essential oils, monoterpenes, and sesquiterpenes, which can generate a flammable atmosphere due to their low flashpoint and the strong negative influence shown between the essential oils, volatile terpenes, and limonene concentration. Moreover, the heat of combustion of E. globulus was positively correlated with its high essential oil contents. Finally, all of the studied species had low flashpoints and high heating values; therefore, they are predisposed to ignite in the presence of a heat source, releasing high amounts of energy during combustion, which contributes to the risk of the formation and spread of canopy fires among these tree formations.

Increasing radiant heat flux affects leaf flammability patterns in plant species of eastern Australian fire-prone woodlands

Leaf flammability is a functional trait that can vary widely among plant species. At present, however, the effects that increasing radiant heat flux have on variation in leaf flammability among species are not well understood. Yet, such effects could have important implications for wildfire models that take into account species' differences in flammability. We examined how five leaf flammability attributes spanning ignitibility (times to incandescence and flaming), sustainability (incandescence and flame durations) and combustibility (proportion of leaves entering flaming combustion) responded to increasing radiant heat fluxes (29.6 to 96.6 kWm^-2 ) in 10 species of fire-prone woodlands. As radiant heat flux increased, times to incandescence and flaming became significantly faster and proportions of leaves entering flaming combustion became significantly higher. In contrast, incandescence duration became significantly shorter at high radiant heat flux. Differences among species in these flammability attributes decreased with increasing radiant heat flux, with species becoming significantly more similar to each other. Differences among species in flame duration, however, were not significantly affected by increasing radiant heat flux, with leaf flaming durations in each species remaining relatively fixed across the radiant heat flux gradient. Our findings show that leaf flammability is significantly affected by increasing radiant heat flux. We suggest that of the flammability attributes assessed in our study, flame duration is the most informative to include in wildfire models which explicitly consider species' flammability, given that differences among species in flame duration are maintained across a radiant heat flux gradient.

A Predictive Model of Leaf Flammability Using Leaf Traits and Radiant Heat Flux for Plants of Fire-Prone Dry Sclerophyll Forest

The differential flammability of individual plant species in landscape-scale fire behaviour is an important consideration, but one that is often overlooked. This is in part due to a relative dearth in the availability of plant flammability data. Here, we present a highly accurate predictive model of the likelihood of plant leaves entering flaming combustion as a function of leaf mass per area (LMA), leaf area (LA) and radiant heat flux using species of fire-prone dry sclerophyll forests of south-eastern Australia. We validated the performance of the model on two separate datasets, and on plant species not included in the model building process. Our model gives accurate predictions (75–84%) of leaf flaming with potential application in the next generation of fire behaviour models. Given the global wealth of species’ data for LMA and LA, in stark contrast to leaf flammability data, our model has the potential to improve understanding of forest flammability in the absence of leaf flammability information.