Shoot-level flammability across the Dracophyllum (Ericaceae) phylogeny: evidence for flammability being an emergent property in a land with little fire
Summary
- Plant flammability varies across species, but the evolutionary basis for this variation is not well understood. Phylogenetic analysis of interspecific variation in flammability can provide insights into the evolution of plant flammability.
- We measured four components of flammability (ignitability, sustainability, combustibility and consumability) to assess the shoot-level flammability of 21 species of Dracophyllum (Ericaceae). Using a macroevolutionary approach, we explored phylogenetic patterns of variation in shoot-level flammability.
- Shoot-level flammability varied widely in Dracophyllum. Species in the subgenus Oreothamnus had higher flammability and smaller leaves than those in the subgenus Dracophyllum. Shoot flammability (ignitability, combustibility and consumability) and leaf length showed phylogenetic conservatism across genus Dracophyllum, but exhibited lability among some closely related species, such as D. menziesii and D. fiordense. Shoot flammability of Dracophyllum species was negatively correlated with leaf length and shoot moisture content, but had no relationship with the geographic distribution of Dracophyllum species.
- Shoot-level flammability varied widely in the genus Dracophyllum, but showed phylogenetic conservatism. The higher flammability of the subgenus Oreothamnus may be an incidental or emergent property as a result of the evolution of flammability-related traits, such as smaller leaves, which were selected for other functions and incidentally changed flammability.
Introduction
Terrestrial plants evolved around 420 Ma (million yr ago) (Wellman et al., 2003). These early plants provided fuel and increased oxygen in the atmosphere to support fire (Glasspool et al., 2004; Pausas & Keeley, 2009). Fire has influenced the evolution of plants, and plants have influenced fire behaviour (Bond et al., 2005; Keeley et al., 2011; Schwilk & Caprio, 2011). Plant flammability is a compound trait emerging from the chemical and physical characteristics of a plant (Schwilk, 2015; Pausas et al., 2017). Different plant species and individuals of the same species growing in different habitats vary in their flammability (Pausas et al., 2012; Murray et al., 2013; Wyse et al., 2016; Krix & Murray, 2018). Investigating the evolution of plant flammability can help us better understand the interaction between fire and plants, and allow us to better prepare for a warmer world, where fire risk may be higher in many regions (Doerr & Santín, 2016). However, the evolutionary mechanisms determining flammability, and whether flammability is selected for or has incidentally emerged, remain unclear (Mutch, 1970; Snyder, 1984; Bond & Midgley, 1995; Midgley, 2013; Bowman et al., 2014).
Macro- and microevolutionary approaches have been used to assess the evolution of plant flammability. The macroevolutionary approach uses a dated phylogeny to trace the evolution of flammability-related traits over extended temporal scales (Myr) (Pausas & Schwilk, 2012). The microevolutionary approach involves investigating variation in traits, such as branch shedding, within species or populations (Pausas, 2015). Several macroevolutionary studies have suggested that fire can be an important selective force on plant fire-related traits (Crisp et al., 2011; He et al., 2011, 2012; Pausas, 2015). For example, He et al. (2012) provided compelling evidence that fire has influenced the evolution of five fire-adaptive traits (bark thickness, serotiny, branch shedding, grass stage and resprouting capacity) in Pinus. Likewise, fire may have played a role in the origin of Banksia and the evolution of some traits, such as dead floret retention (He et al., 2011). However, previous macroevolutionary studies (He et al., 2011, 2012) have used qualitative traits (e.g. branch shedding/branch retention) rather than quantitative measures of flammability, which reduces the scope of available comparative phylogenetic analyses, such as estimation of phylogenetic signal. Phylogenetic signal is used to evaluate the correlation between species trait variation and phylogenetic relatedness, and has been used in a range of ecological and evolutionary research areas (Felsenstein, 1985; Pagel, 1999; Blomberg et al., 2003; Münkemüller et al., 2012). A strong phylogenetic signal indicates that closely related species have similar trait values, while trait similarity decreases with phylogenetic distance (Losos, 2008). Conversely, a weak phylogenetic signal suggests that a trait varies randomly across the phylogeny, implying that the trait is not passed down from ancestors (Kamilar & Muldoon, 2010). However, phylogenetic signal has rarely been used in studies of the evolution of plant flammability (but see Cui et al., 2020). Flammability can be quantified by four flammability variables: ignitability (how easily a plant ignites), sustainability (the length of time a plant sustains flames), combustibility (the intensity at which a plant burns), and consumability (the percentage of biomass consumed by fire) (Anderson, 1970; Martin et al., 1993). Estimating phylogenetic signal in these flammability variables will provide insights into the evolution of flammability. Furthermore, quantitative measures of flammability appropriately represent flammability as a continuous rather than a binary trait.
New Zealand is an archetypal isolated oceanic ecosystem (McGlone et al., 2016). Most ecosystems in New Zealand experienced low fire frequencies before human arrival, primarily as a result of limited ignition sources (Perry et al., 2014; Kitzberger et al., 2016). Few of New Zealand’s indigenous woody species show adaptation to fire (Perry et al., 2014). The indigenous species with distinctive fire adaptations (e.g. serotiny in Leptospermum scoparium, resprouting in Discaria toumatou, Pteridium esculentum and Cordyline spp.) are closely related to eastern Australian species and have a history in New Zealand no earlier than the Pliocene (Mildenhall, 1980; Walsh & Coates, 1997; McGlone et al., 2005; Stephens et al., 2005; De Lange et al., 2010). As Keeley et al. (2011) emphasize, species are not adapted to fire but to fire regimes. Long and variable fire intervals during most of New Zealand’s ecological history (Perry et al., 2014), coupled with the loss of fire-adapted traits in some taxa (McGlone, 2006; Battersby et al., 2017), suggest that the evolution of New Zealand’s indigenous species was not influenced by fire (Lawes et al., 2014). Whether flammability is selected by fire or emerges incidentally has been widely debated (Mutch, 1970; Snyder, 1984; Bond & Midgley, 1995; Midgley, 2013; Bowman et al., 2014). However, most previous studies of the evolution of flammability have focused on species in fire-prone ecosystems (Pausas et al., 2012; Archibald et al., 2018). No study has used a macroevolutionary approach on a clade of species that evolved in the relative absence of fire, to evaluate the influence of factors other than fire on the evolution of flammability. Therefore, the New Zealand flora provides an opportunity to explore whether flammability is an incidental or emergent property (i.e. is not a specific fire adaptation; Mason et al., 2016).
In this study, we measured the shoot flammability of 21 Dracophyllum species (six in the subgenus Dracophyllum and 15 in the subgenus Orethamnus) from New Zealand. With reference to published Dracophyllum phylogenies (Venter, 2009; Wagstaff et al., 2010), we explore evolutionary patterns in shoot flammability across the genus. We also discuss the origin of flammability variation among the Dracophyllum genus and whether it was selected for or emerged incidentally.
Materials and Methods
Sample collection
The genus Dracophyllum Labili. (Ericaceae) contains 51 polymorphic species, divided into three subgenera: Dracophyllum, Oreothamnus, and Cordophyllum (Table 1) (Oliver, 1952; Venter, 2009). Dracophyllum reaches its greatest degree of species richness and morphological diversity in New Zealand with 35 species, ranging from low-growing cushion plants to small trees up to 14 m tall (Fig. 1) (Wagstaff et al., 2010). Of the 35 Dracophyllum species native to New Zealand, eight are restricted to the North Island, 21 occur only on the South Island, three can be found on both main islands, and three grow on nearby offshore islands (Venter, 2009). The high degree of polymorphism in this genus makes it a useful model for evolutionary research.
| Genus | Subgenera | Distribution |
|---|---|---|
| Dracophyllum (51 species) | Dracophyllum (21 spp.) | New Zealand (seven spp.), New Caledonia (eight spp.), mainland Australia (four spp.), Lord Howe Island (one sp.), Tasmania (one sp.) |
| Oreothamnus (29 spp.) | New Zealand (28 spp.), Tasmania (one sp.) | |
| Cordophyllum (one sp.) | New Caledonia (one sp.) |
All shoot samples from the 21 Dracophyllum species were collected during one summer/autumn season (November 2018 to April 2019) from public conservation lands in New Zealand under permit from the Department of Conservation. The collection sites were selected using information from Venter (2009), iNaturalist (https://inaturalist.nz/) and the Allan Herbarium (CHR). Healthy terminal shoots 70 cm in length were collected from healthy individuals, preserving branch architecture, and kept in separate sealed plastic bags to prevent moisture loss. For Dracophyllum species with branches shorter than 70 cm, such as D. densum, whole plants above the roots were collected. We sampled at least seven individuals of each species. Shoot samples were kept cool when collecting and then stored at 4–8°C as soon as possible. All shoot samples were burned within 1 wk of collection.
Data collection
We obtained the GPS coordinates of observations (until May 2019) for each Dracophyllum species from iNaturalist (https://inaturalist.nz/) (the observation locations that are obscured were excluded). For four species without accurate observation information on iNaturalist (the location of observations are obscured), observation records by Venter (2009) were used. The elevations of all observation records were estimated from the New Zealand national digital elevation model (25 m resolution, downloaded from https://lris.scinfo.org.nz/). Individual geographical references of each species were used to obtain climatic information (annual mean air temperature and annual mean precipitation) from the WorldClim database (30 s (c. 1 km2), https://www.worldclim.org/). The climate data were obtained by using the R package dismo (v.1.1-4) (Hijmans et al., 2017). To characterize the climatic and geographic conditions across each species distribution, the mean values for mean annual air temperature, mean annual precipitation, elevation, and latitude from across the recorded distribution for each species were calculated. The latitudinal range of observations was used as an indicator of the overall latitudinal range size for each species.
Flammability measurements
Shoot flammability was measured for each sample following the methods described by Jaureguiberry et al. (2011) and Wyse et al. (2016), using the same device as Wyse et al. (2016). The samples for flammability measurement were 70-cm-long shoot samples. For each species, at least seven samples were collected, each from a different individual plant. Before burning, all shoot samples were air-dried at room temperature for 24 h to match the sample moisture content to the ignition source (following Wyse et al., 2016, 2018). For the flammability measurements, samples were first placed on our device for preheating for 2 min at 150°C. Then, a blowtorch was turned on for 10 s to ignite the samples. Ignitability was represented by an ignition score (Padullés Cubino et al., 2018; Wyse et al., 2018). Ignitability was recorded first as time to ignition (between 0 and 10 s), which was then converted to an ignition score by subtracting the time to ignition from 10; for example, a sample that took 1 s (i.e. rapid ignition) to ignite had an ignition score of 9. Samples that did not ignite after 10 s were given a value of zero. The maximum temperature of flames during burning was measured using an infrared laser thermometer (Fluke 572; Fluke Corp., Everett, WA, USA) to represent combustibility. Samples that failed to ignite were given a value of 150°C, representing the grill temperature (Padullés Cubino et al., 2018; Wyse et al., 2018). Sustainability was measured as the period of time that a sample burned (i.e. had flaming combustion) after the blowtorch was turned off. Consumability was measured as the mean value of the percentage of burnt biomass after flaming combustion ceased, assessed by visual observation by at least two observers. Samples that did not sustain flaming combustion after the blowtorch was turned off were assigned scores of zero for sustainability and consumability.
Testing for phylogenetic signal
A phylogeny for the 21 Dracophyllum species was obtained from a maximum parsimony tree (from Venter, 2009) for visualizing the flammability variation across the 21 Dracophyllum species. However, this phylogeny does not have branch length and cannot be used to calculate phylogenetic signal. A branch-length phylogeny was constructed with the chloroplast-encoded genes rbcL and matK using Mega7 software (Supporting Information Fig. S1) (Kumar et al., 2016). The chloroplast-encoded genes rbcL and matK were obtained for 14 Dracophyllum species from Wagstaff et al. (2010). We used the R packages picante (v.1.8) (Kembel et al., 2010) and phytools (v. 0.6-99) (Revell, 2012) to calculate the phylogenetic signal, Pagel's λ (Pagel, 1999), which is more appropriate than alternatives, such as Blomberg's K (Blomberg et al., 2003), for testing ecologically relevant traits (Molina-Venegas & Rodríguez, 2017). Pagel's λ varies continuously from zero to unity. A value of λ = 0 indicates no phylogenetic signal in the trait, that is, that the trait has evolved independently of phylogeny and thus close relatives are not more similar on average than distant relatives; λ = 1 indicates a strong phylogenetic signal, and that the trait has evolved according to the Brownian motion model of evolution. Intermediate values of λ indicate that although there is a phylogenetic signal in the trait, it has evolved according to a process other than pure Brownian motion (Kamilar & Cooper, 2013).
Statistical analysis
All statistical analyses were conducted with R 3.5.0 (R Core Team, 2018). Principal component analysis (PCA) of the four flammability components was performed to evaluate the shoot flammability for every species using the ‘princomp’ function in R. The value of the first PCA component was positively correlated with all flammability components, and was used as an aggregate index of shoot flammability (Wyse et al., 2016). All flammability components were compared across species using one-way ANOVA. The proportion of variation across/within populations was calculated by dividing the sum of squares across/within populations by the sum of squares total using ANOVA. Leaf length, shoot moisture and environmental conditions were compared with shoot flammability for each species by using generalized linear regression. Associations among the index of shoot flammability (PC1), leaf length and shoot moisture content were evaluated with partial correlation analyses (using the Pearson method), controlling for leaf length or shoot moisture content, using the R package ppcor (v.1.1) (Kim, 2015).
Results
Shoot flammability varies among Dracophyllum species
We collected 251 samples from 21 Dracophyllum species across the two main islands of New Zealand, at elevations ranging from 80 m to 1260 m above sea level (Table S1). These species range from low-growing sprawling shrubs (e.g. D. densum, D. kirkii) to small trees up to 14 m in height (D. elegantissimum) (Fig. 1). The proportion of variance in all shoot flammability traits other than burning time (sustainability) was higher between species (ignition score, 92.9%; maximum temperature, 51.4%; burning time, 44.0%; burnt biomass, 66.6%) than within species (ignition score, 7.1%; maximum temperature, 48.6%; burning time, 56.0%; burnt biomass, 33.4%). The four shoot flammability components were positively correlated (Table S2) and varied significantly across the Dracophyllum species (ANOVA, ignition score, F20,230 = 150.50, P < 0.001; maximum temperature, F20,230 = 12.15, P < 0.001; burning time, F20,230 = 9.03, P < 0.001; burnt biomass, F20,230 = 22.91, P < 0.001). Some species (e.g. D. sinclairii and D. trimorphus) on average ignited within 1 s (ignition score > 9), while others (e.g. D. fiordense and D. traversii) took longer than 8 s to ignite (ignition score < 2). Mean burnt biomass per species ranged from 5.0% for D. fiordense, to 90.0% in D. trimorphum. The mean maximum temperature of D. densum reached 771.5 ± 23.0°C (mean ± 1 SE), while the mean maximum temperature of D. fiordense was only 277.7 ± 61.7°C. Mean burning time varied from 6.7 ± 4.2 s (D. fiordense) to 157.6 ± 18.2 s (D. pronum). Dracophyllum fiordense was the least flammable of the 21 Dracophyllum species, with the lowest values for all four flammability components. Dracophyllum traversii and D. elegantissimum also showed low flammability, requiring more than 6 s to ignite and with burnt biomass < 15%. The two most flammable species were D. densum and D. pronum, with both sustaining a flame for more than 2 min.
A PCA was performed using the mean values of the four flammability components of each species to assess the overall shoot flammability across the 21 Dracophyllum species. The values of the first two axes of the PCA explained 78.6% and 12.3% of the variation, respectively (Fig. 2). The loadings of the four flammability components on the first axis were 0.491 (ignition score), 0.514 (maximum temperature), 0.448 (burning time) and 0.541 (burnt biomass). This index of shoot flammability (i.e. PC1) ranged from −4.50 to 2.45 and was positively correlated with all flammability components (Fig. 2). According to this index of shoot flammability, D. densum, D. pronum and D. marmoricola were the most flammable Dracophyllum species, while D. fiordense, D. traversii and D. elegantissimum were the least. The index of shoot flammability differed between the two subgenera, with species in subgenus Dracophyllum significantly less flammable than those in Oreothamnus (Fig. 2). Another PCA was performed using the flammability components of all individual samples and the results were similar (Fig. S2).
Shoot flammability across Dracophyllum shows phylogenetic conservatism
Although shoot-level flammability varied significantly across Dracophyllum species (ANOVA, PC1: F20,230 = 29.3, P < 0.001), integrating the flammability data and phylogenetic data showed that closely related species tended to have similar flammability (Fig. 3). We divided the 21 species into six identified clades (Fig. 3) based on their phylogenetic relatedness. Variation in flammability variables, except maximum temperature, was higher among clades (index of flammability, 65.9%; ignition score, 70.9%; burning time, 76.0%; maximum temperature, 44.5%; burnt biomass, 62.3%) than among species (index of flammability, 34.1%; ignition score, 29.1%; burning time, 24.0%; maximum temperature, 55.5%; burnt biomass, 37.7%). The phylogenetic signal across a subset of 14 of the 21 Dracophyllum species showed that flammability components, except burning time, were highly phylogenetically conserved, and the index of shoot flammability showed a significant phylogenetic signal (Table 2). Although shoot flammability showed phylogenetic conservatism, it exhibits obvious lability between some closely related species; for example, D. menzisii and D. fiordense are closely related, but their shoot flammability differs significantly (ANOVA, P < 0.001 for all flammability components; Fig. 3).
| Traits | Pagel's λ | |
|---|---|---|
| λ | P | |
| Ignition score | 0.994 | < 0.001 |
| Maximum temperature | 0.923 | 0.031 |
| Burning time | 0.344 | 0.092 |
| Burnt biomass | 0.910 | 0.027 |
| Principal component 1 | 0.914 | 0.010 |
| Shoot moisture content | 0.260 | 0.241 |
| Leaf length | 0.987 | < 0.001 |
- Bold denotes statistical significance (P < 0.05).
Shoot flammability of Dracophyllum species decreases with leaf size and shoot moisture content
Leaf size varied across Dracophyllum species and discriminated the subgenera with Dracophyllum having longer leaves than Oreothamnus. Shoot flammability of Dracophyllum species was negatively associated with leaf length (R2 = 0.525 and P < 0.001 for PC1; Fig. 4); that is, species with longer leaves were less flammable. Ignition score and burnt biomass were negatively related to leaf length, while maximum temperature and burning time were less strongly related (Table 3). Shoot moisture content was significantly negatively correlated with shoot flammability (R2 = 0.536 and P < 0.001 for PC1; Fig. 4; Table 3). Partial correlation analysis showed that shoot flammability (PC1) is significantly correlated with shoot moisture content (r = −0.53, P = 0.016) and leaf length (r = −0.55, P = 0.013), after controlling for leaf length and shoot moisture content, respectively. The phylogenetic signal of leaf length and shoot moisture content showed that leaf length was highly phylogenetically conserved, while shoot moisture content was not phylogenetically conserved (Table 2).
| Leaf length | Shoot moisture content | |
|---|---|---|
| Principal component 1 | R2 = 0.525, P < 0.001 | R2 = 0.536, P < 0.001 |
| Ignition score | R2 = 0.647, P < 0.001 | R2 = 0.393, P = 0.002 |
| Maximum temperature | R2 = 0.209, P = 0.037 | R2 = 0.446, P < 0.001 |
| Burning time | R2 = 0.283, P = 0.013 | R2 = 0.277, P = 0.014 |
| Burnt biomass | R2 = 0.590, P < 0.001 | R2 = 0.575, P < 0.001 |
- Bold denotes significance (P < 0.05).
Shoot flammability of Dracophyllum species has no relationship with their distribution
The latitudinal range of Dracophyllum species had no relationship with their shoot flammability (Table S3). Of the 21 Dracophyllum species, four species occurred only in the North Island and 13 species only in the South Island (Table S1). Shoot-level flammability of Dracophyllum species did not differ between the two main islands of New Zealand (Fig. 5). Geographic and climatic conditions (latitude, elevation, mean annual air temperature and mean annual rainfall, Table S1) were not correlated with shoot-level flammability (Fig. 6; Table S3).
Discussion
Plant flammability can vary widely across species (Engber & Varner, 2012; Fuentes-Ramirez et al., 2016; Simpson et al., 2016; Wyse et al., 2016; Padullés Cubino et al., 2018). However, how the components of flammability vary between closely related species and whether such species have similar flammability, especially at the shoot level, has rarely been reported (Engber & Varner, 2012; Cornwell et al., 2015). Based on quantitative measures of shoot flammability from many species of the highly polymorphic genus Dracophyllum, we demonstrated that flammability can vary widely at the genus level. For example, some Dracophyllum species, such as D. trimorphum, had on average 90% of their biomass consumed by fire in our device, while someindividuals of some other species (e.g. D. fiordense) could not be ignited (Table S1). The existence of high flammability among Dracophyllum species confirms that high flammability species can occur in communities that rarely experience fire (Bowman et al., 2014; Calitz et al., 2015; Wyse et al., 2016; Cui et al., 2020). Despite the wide variation in the genus, shoot flammability was generally more similar in close relatives than distant relatives (Fig. 3; Table 2). Thus, although flammability is a compound trait affected by many functional traits, it retains phylogenetically conserved patterns. Phylogenetic analysis of the shoot flammability of 194 vascular plant species (from 72 families) showed a statistically significant phylogenetic signal in shoot flammability across the Trachaeophyta (Cui et al., 2020), indicating that flammability was phylogenetically conserved at higher taxonomic levels.
The phylogenetic component of shoot flammability is consistent with the notion that flammability, as an emergent trait, may be selected for. However, a more likely explanation for the patterns of flammability seen in New Zealand Dracophyllum is that flammability is an incidental or emergent property, and it is environmentally determined (Snyder, 1984; Midgley, 2013). The variation in flammability between the closely related species D. menzisii and D. fiordense provided evidence for flammability being affected by the environment. Dracophyllum menzisii, which mainly grows on mountain slopes, has never been recorded in forest communities and is the only true grassland species in the genus, while D. fiordense usually occurs in high rainfall areas and receives additional moisture from mist (Venter, 2009). The difference in flammability between these two closely related species (Fig. 3) may be a result of adaptation to different environments, indicating that flammability is environmentally determined in this case.
The subgenus Oreothamnus diverged from the subgenus Dracophyllum in the Pleistocene approximately 1–2 Ma becoming relatively more flammable (Gibbard & Van Kolfschoten, 2004; Wagstaff et al., 2010). Fire in most regions of New Zealand is believed to have been infrequent before human settlement, although charcoal is found in New Zealand sediments of all ages (Ogden et al., 1998; Perry et al., 2014). This low fire frequency suggests that fire was not a selective force during the divergence of low/high flammability subgenera in Dracophyllum. A more likely selective force than fire regime during this period may have been the shifts in climate associated with glacial/interglacial periods. The repeated climatic changes and glaciation during the Pleistocene are believed to have shaped the New Zealand flora (Wardle, 1988; Winkworth et al., 2005; Heenan & McGlone, 2013; Millar et al., 2017). The cold and dry climate of glacial periods may have influenced the origin and evolution of Oreothamnus species, and selected for certain traits, such as smaller leaves, that facilitated frost and drought tolerance (Lusk et al., 2016; Reichgelt et al., 2017), and that incidentally increased flammability. Consequently, we conclude that high shoot-level flammability in Oreothamnus is an incidental or emergent property associated with leaf form as selected through the glacial cycles of the Pleistocene. This result indicates that flammability could be an incidental property, at least in ecosystems with little fire.
Linking functional traits to flammability can facilitate the prediction of flammability across species (Alam et al., 2020). Different plant traits may influence different aspects of flammability, and traits important for crown fire behavior will differ from those important for surface fires (Schwilk & Caprio, 2011). Many studies exploring the influence of traits on flammability have considered the flammability of small plant components, mostly leaves and small twigs (Alessio et al., 2008; De Lillis et al., 2009; Engber & Varner, 2012; Murray et al., 2013; Grootemaat et al., 2015; Pausas et al., 2016; Simpson et al., 2016). The flammability of these small components may not adequately represent the flammability of an entire plant. Moisture content is generally accepted to be a strong determinant of fuel flammability (Dimitrakopoulos & Papaioannou, 2001; Ganteaume et al., 2010; Grootemaat et al., 2015). Leaf size affects litter flammability (Schwilk & Caprio, 2011; Cornwell et al., 2015), but how leaf size influences canopy fires has not been adequately studied. That shoot-level flammability of Dracophyllum species was negatively correlated with leaf size and shoot moisture content suggests that shoot moisture content and leaf size can help predict the shoot flammability of unmeasured Dracophyllum species. However, whether leaf size is negatively correlated with shoot flammability at higher taxonomic levels or across species in other genera is unclear. For example, Padullés Cubino et al. (2018) found leaf length to be positively related to shoot flammability in species from tussock grasslands, while Alam et al. (2020) found no correlation between leaf length and shoot flammability in an analysis of 43 tree and shrub species.
We have demonstrated that shoot-level flammability of 21 Dracophyllum species varied widely and was negatively correlated with leaf size and shoot moisture content. Shoot-level flammability showed phylogenetic conservatism across the Dracophyllum phylogeny, but also occasional lability between some closely related species, perhaps as a result of differing habitats. Subgenus Oreothamnus, which arose (1–2 Ma) in the Pleistocene and may have evolved in the absence of fire, exhibited high flammability, suggesting that the climate of the Pleistocene may have favoured and selected for characteristics, such as smaller leaves, that were suited to other functions (e.g. drought and frost tolerance) and incidentally increased shoot-level flammability. Our study has provided evidence that, at least in relatively fire-free environments, high flammability could be an incidental or emergent property that comprises traits that arose in response to selective forces independent of fire. However, other studies have suggested that flammability can evolve in fire-prone habitats (Pausas et al., 2012; Moreira et al., 2014; Cui et al., 2020), emphasizing the importance of considering fire regimes when examining the evolution of plant flammability.
Acknowledgements
We wish to thank the staff at the Allan Herbarium (CHR) for access to their collection of Dracophyllum species and site information on where to collect Dracophyllum species. We also want to thank the staff of the Department of Conservation of New Zealand, Myles Mackintosh, Dongyu Cao and Eva van den Berg for their help with the sample collection. The collection of samples was authorized by the Department of Conservation New Zealand under collection authorization 65543-FLO.
Author contributions
This project was developed and designed by XC, TJC and AMP. Plant samples and flammability data were collected by XC, MAA and KM. XC conducted the statistical analyses with advice from AMP, TJC, GLWP and SVW. XC built the phylogenies with published data. XC, TJC and AMP led the writing of the manuscript, with input from all co-authors. All authors gave final approval for publication. Thanks to Tim Brodribb and three anonymous reviewers for their helpful comments on the text.
