Evidence of local adaptation in litter flammability of a widespread fire‐adaptive pine

Fire is a strong ecological and evolutionary driver of plant species and terrestrial ecosystem dynamics. Numerous studies have shown that plant species vary widely in their flammability and associated traits that coincide with clear strategies of persistence in fire‐prone ecosystems. Yet, intraspecific variation and phenotypic plasticity in plant flammability has received substantially less attention. Some evidence has demonstrated high plasticity in flammability among populations in species adapted to low‐frequency, high severity fire regimes which suggests the capacity of local adaptation; however, intraspecific variability of species in high‐frequency, low severity fire regimes has not been conducted. Given that frequent fire‐adaptive species tend to produce fast‐flammable litter that is associated with survival following fire, we hypothesized that these species would have highly conserved leaf traits with low variation in litter flammability. We used an existing common garden study to test our hypothesis in Pinus rigida (pitch pine) with source populations that span most of its distribution in eastern North America. We examined both phenotypic and genotypic variation as well as phenotypic plasticity (relative distance plasticity index) in P. rigida leaf traits and litter flammability among populations grown in situ (population) and the same populations grown in a common garden or provenance study (provenance). Phenotypic and genotypic variation in fuel traits and litter flammability differed among source locations. Phenotypic plasticity in litter flammability among source locations were associated with fuel trait plasticity and biogeographic factors. Populations located in warmer and drier climates with presumably more frequent historical fire regimes were associated with greater plasticity in physical litter traits and their resulting flammability. Synthesis. Our study demonstrates the utility of common garden experiments to examine intraspecific variation and phenotypic plasticity in litter flammability. Our results indicate that litter flammability in Pinus rigida has the capacity for local adaptation. We hypothesize that intraspecific variation in plant flammability provides ecologically and evolutionary relevant information and that other frequent fire‐adaptive species will similarly demonstrate local adaptation in plant flammability.


| INTRODUC TI ON
Fire has been a major ecological and evolutionary driver that has long shaped plant species and terrestrial ecosystems world-wide (Bond & Keeley, 2005;Bowman et al., 2009). Not only do plants respond to fire, but plants also serve as the primary source of fuel that directly influences the ignition and combustion of fire (Rothermel, 1983).
Thus, many plant traits are interpreted to be adaptations to certain fire regimes that promote avoidance, resistance or recovery strategies following fire (Keeley et al., 2011;Keeley & Zedler, 1998). This process, referred to as 'niche construction', suggests that plants have traits selected by fire that, in turn, modify their environments and the environments of competing neighbours to promote fitness advantages (Schwilk, 2003). One such way species in fire-prone ecosystems can modify their environments is through their flammability, or a species' ability to ignite, combust and spread fire (Varner, Kane, Banwell, & Kreye, 2015).
While there has been considerable debate about the degree to which plant flammability is an evolutionary adaptation to fire, multiple lines of evidence continue to support this possibility. Early criticisms of the Mutch (1970) hypothesis, that fire was a selective force promoting ecosystem flammability, were dismissed for being 'group selectionist' (Snyder, 1984;Troumbis & Trabaud, 1989). Subsequent modelling and field experiments have demonstrated that enhanced flammability could have been selected as a suite of traits that confers fitness advantages at the individual plant level (e.g. Bond & Midgley, 1995;Kerr et al., 1999;Schwilk, 2003). Still, there is a persistent perspective that plant flammability may be environmentally determined rather than selected for by fire (Midgley, 2013). Isolating the causal mechanism of adaptation remains a challenge, but continued research examining possible evidence in support of or in refutation of existing ideas is needed to advance our understanding.
Three flammability strategies have been proposed that align with other known fire-adaptive traits and promote fitness benefits to plants in fire-prone environments (Pausas et al., 2017). In the first strategy, plant traits result in rapid low-intensity fires (i.e. 'fast flammable') often paired with thicker bark and self-pruning branches that limit fire-caused mortality and reduce competition from less fire-tolerant species. In the second strategy, species have traits that promote high-intensity, long duration fire (i.e. 'hot flammable') benefit from reduced competition and higher quality seedbed conditions that commonly occur following fire. These species tend to have other traits such as thin bark, retained branches, serotinous cones, resprouting or heat-triggered germination. Species with traits promoting poor flammability (i.e. 'non-flammable') are likely best suited to relatively fire-naïve ecosystems (i.e. avoiders) or by impeding fire (i.e. impeders sensu Kane et al., 2008). Fitness is enhanced in these cases through increased survival either by passively avoiding fire (e.g. some individuals survive fire in unburned refugia) or by actively reducing fire spread to limit damage (e.g. fire dampening litter). In these species, we would not expect to see other adaptations to fire.
Numerous studies have examined interspecific plant flammability (e.g. Kane et al., 2021;Kreye et al., 2018;Saura-Mas et al., 2010;Scarff & Westoby, 2006;Schwilk & Ackerly, 2001) with substantially fewer investigations into intraspecific plant flammability. Studies examining intraspecific variation of plant flammability have primarily focused on 'hot-flammable' obligate seeder species (mature plants that die but reproduce from seeds stored in the canopy or soil that germinate following fire) or 'non-flammable' species (Pausas et al., 2012;Romero & Ganteaume, 2020;Schwilk, 2003). For instance, Schwilk (2003) artificially enhanced flammability by adding dead branches in the obligate seeder shrub Ceanothus cuneatus that resulted in higher post-fire seed germination rates, demonstrating a clear fitness benefit of higher flammability. Another study, examining flammability within the shrub Ulex parviflorus, indicated that plants growing in areas with more frequent fire had enhanced flammability traits compared to areas with no recent fires (Pausas et al., 2012).
Thus, it is reasonable to expect that hot-flammable species would have high variation in flammability among populations due to higher fire intensity (i.e. stronger selective force) and faster turnover rates (via post-fire mortality of adult plants) that can contribute to greater incidences of local adaptation. We are unaware of any studies that have examined intraspecific variation in flammability among populations within 'fast-flammable' species. Since the primary fitness advantage of this strategy is survival during fire resulting in longer turnover rates, we might expect flammability traits in this strategy to be conserved with low intraspecific variability in flammability among populations.
Senesced leaves, or 'litter', often represent a substantial portion of the surface fuel layer and is one of the major drivers of fire behaviour in many woodland and forest ecosystems. Litter flammability varies widely among species reflecting particle and fuelbed trait differences (Burton et al., 2020;Popović et al., 2021;Varner, Kane, Kreye, & Engber, 2015). Physical traits such as long leaves and large for local adaptation. We hypothesize that intraspecific variation in plant flammability provides ecologically and evolutionary relevant information and that other frequent fire-adaptive species will similarly demonstrate local adaptation in plant flammability.

K E Y W O R D S
common garden, fire behaviour, fire ecology, fuels, intraspecific trait variation, Pinus rigida, plant traits, provenance surface area-to-volume ratios tend to result in more flammable litter (Engber & Varner, 2012). These traits also influence fuelbed attributes, such as bulk density and packing ratio that contribute to variation in litter flammability among species (Burton et al., 2020). Litter that has a high volatile content or high cellulose-to-lignin ratio also have higher flammability (Ormeño et al., 2009). While these characteristics are known to vary among species, intraspecific variation is less well-understood.
We chose to examine intraspecific variability in litter flammability of Pinus rigida (pitch pine) due to its wide ecological amplitude and well-known genotypic diversity. P. rigida is a mostly non-serotinous conifer that spans a wide range of climatic conditions in eastern North America extending from Georgia northwards to Maine, with outlying populations slightly farther north in southern Quebec and Ontario. Prior to colonization, much of this species' range experienced relatively frequent fire (<15-year fire return interval) with highly variable intensity depending on vegetation structure and substrate (Brose & Waldrop, 2006;Guyette et al., 2012;Stambaugh et al., 2015). Generally, P. rigida is capable of surviving surface fires because of its thick bark and rapidly burning litter (Little & Garrett, 1990;Varner, Kane, Banwell, & Kreye, 2015). This species also has the capacity to sprout basally or epicormically following fire. Interestingly, two known coastal populations located in the Pine Plains region of New Jersey and Long Island, New York are mostly serotinous with >90% of trees having closed cones (Givnish, 1981;Jordan et al., 2003). The serotiny level of populations generally declines rapidly away from these coastal populations (Ledig & Fryer, 1972). Past research has demonstrated high genetic diversity and high variability in growth rates among trees from a wide range of source populations (Guries & Ledig, 1982;Leland et al., 2016), indicating the potential for substantial variation in flammability among populations.
The primary objective of our study was to examine the intraspecific variability and phenotypic plasticity of litter flammability and associated traits in P. rigida across its distribution. Our hypothesis was that litter flammability would be highly conserved, or in other words, that litter flammability and associated traits would be consistent across a broad climatic and topographic gradient and that litter flammability from different source populations but grown in a common garden experiment (i.e. provenance study) would also be similar.
If this hypothesis is supported, our results would suggest that litter flammability in species adapted to low severity and high-frequency fire regimes may be highly conserved and less prone to variation due to local adaptation.

| Experimental design and litter collection
We based our experiment on an existing P. rigida provenance (i.e. common garden) study located within the Brendan T. Byrne State Forest (formerly the Lebanon State Forest; 39.923087 N, −74.431403 W) in Ocean County, New Jersey, USA (Ledig et al., 1976). The site has well-drained sandy soils common in the outer coastal plain of New Jersey. The provenance study was established in 1971 and planting followed a compact family (i.e. stands) design with seed sources from 29 different geographic locations with one to three 'families' randomly assigned to each of the three blocks.
We collected intact litter from 10 source populations (one family from each geographic location) that were planted within the provenance study site (hereafter 'provenance'). Source populations ranged from Georgia to Maine, USA (Figure 1), representing a large climatic and geographic gradient within the distribution of P. rigida (Table 1) were haphazardly collected within each site that corresponded to the seed source used for the provenance study. However, the exact genotypes of the trees sampled from the source populations remain unknown. We placed litter from each tree separately into labelled paper bags, allowed them to air-dry and shipped them to the HSU Wildland Fire Lab in Arcata, California for burning.
For each source population, we gathered geographic, topographic and climatic data for subsequent analyses. Geographic and topographic variables considered included latitude, longitude, elevation (m), distance to the coast (km) and distance to the provenance site (km) using a geographic information system and relevant data layers. Climatic variables considered in our study included annual precipitation (mm); minimum, mean and maximum temperatures (°C); Palmer Drought Severity Index (Palmer, 1965); and minimum and maximum vapour pressure deficit (kPa). We used 30-year (1981 to 2010) climate data from PRISM Climate Group (2020) for each of the 10 source locations.

| Fuel traits
Leaf trait measurements were taken from 20 randomly selected needles of each population and provenance litter collection.
Needle length was measured using electronic calipers to the nearest millimetre. Fuel moisture content (%; dry weight basis) was measured by taking ~5 g litter sample from each population and provenance sample. Each sample was then placed in a drying oven at 40°C for at least 24 h to estimate the moisture content under equilibrium conditions in the laboratory at the time of burning. Fuelbed depth measures are described below in the burning trials section.

F I G U R E 1
Locations of Pinus rigida (pitch pine) sites where litter collection occurred. Green shading denotes the current distributional range of pitch pine (Little & Garrett, 1990). The provenance site was located at the Lakehurst (LAHU), New Jersey, USA site. Location abbreviations are defined in Table 1. Numbered points refer to the original source locations and families (i.e. stands) that were used by Ledig et al. (1976) to establish the provenance study TA B L E 1 Source and collection locations and their associated climate, geography and topography information for the Pinus rigida (pitch pine) litter flammability study. The provenance study site was located within Lakehurst, New Jersey. NF, National Forest; NP, National Park

| Burning trials
We conducted litter flammability experiments for each source population and provenance separately using well-established methods (Fonda, 2001). Litter samples were burned at their equilibrium moisture content (Fosberg et al., 1970) to accentuate any potential differences in moisture gain or loss among samples.
However, fuel moisture content of samples did not significantly differ among source location or type, that is population or provenance (p > 0.08). Each 15 g litter fuelbed was evenly distributed

| Phenotypic plasticity
To quantify phenotypic plasticity of fuel and litter flammability traits for each location, we calculated the relative distance plasticity index (RDPI). This index represents an ecological and evolutionary relevant measure of environmentally driven trait variation that is well-correlated with most other measures of plasticity and is more amenable to statistical analyses than other measures (Valladares et al., 2006). RDPI ranges from 0 (no plasticity) to 1 (maximum plasticity) and was calculated for each location as where d ij→i ′ j ′ is defined as the absolute difference between trait values for all possible pairs of populations and provenance samples of a given source location. Each of these differences were then relativized by dividing by the corresponding sum of the two values being compared x i � j � + x ij , representing the maximum possible plasticity between the two samples. n is the total number of distances calculated per source location.

| Litter flammability
P. rigida litter flammability differed by source type and location ( Figure 3). Based on PERMANOVA, litter flammability differed by source type (F 1,171 = 4.78, p = 0.008) and the interaction of source type and location (F 9,171 = 1.92, p = 0.016), but did not differ by location alone (F 9,171 = 1.01, p = 0.45). However, when maximum flame height was examined alone using an ANOVA, only source type was

| Phenotypic plasticity
Phenotypic plasticity, measured as RDPI, in litter flammability differed among source locations and was associated with F I G U R E 2 Pinus rigida (pitch pine) litter traits by source location for population and provenance source types. Locations are arranged from north to south and the abbreviations are defined in Table 1. Capitalized letters above the boxplot denote significant differences among source locations of population samples, while lower case letters denote differences among source locations of provenance samples. An asterisk denotes significant differences between provenance and population samples within a location

| DISCUSS ION
We present the first study to examine distribution-wide, intraspecific variation in litter traits and litter flammability of a fire-adaptive tree species. Our hypothesis that intraspecific variation in litter flammability of P. rigida, a 'fast-flammable' species, would be highly The provenance site used in our study coincidentally occurred within the centre of the known serotinous populations of P. rigida (Givnish, 1981). Past research suggests that serotinous pine species tend to have lower litter flammability than non-serotinous, fastflammable pine species (Stevens et al., 2020). It is possible that the relatively low productivity soils of the provenance site promoted conditions that resulted in shorter needles and more compact fuelbeds across source locations with concomitantly lower flammability. It would have been interesting to examine if cooler and wetter populations that were grown in the warmer and drier portion of the range (e.g. CHIL, CHAT) would have demonstrated plasticity that resulted in longer needles and higher flammability. However, a provenance site of P. rigida in this portion of range was not available.
Phenotypic variation was greater than genotypic variation in some fuel traits and flammability measures of P. rigida. We suspect that this difference may in part be due to ontogenetic differences between population and provenance types, as has been found to affect flammability in other pine species (Romero & Ganteaume, 2020 (Ghalambor et al., 2007). Our results suggest that phenotypic plasticity of litter flammability in P. rigida is locally derived and that this species may have the capacity to adapt to potential climate change-induced alterations to fire regimes. Thus, it is possible that fast-flammable species may in part modify their environment by producing longer needles that form less compact fuelbeds that can facilitate lower intensity and shorter duration surface fires.
The topic of intraspecific trait variation in plants has received increasing attention over the past decade and the understanding of its importance to ecological and evolutionary processes is developing (Westerband et al., 2021). Our findings contribute to the advancement of this topic by demonstrating local or regional variability in litter flammability of a frequent fire-adaptive species.
Taken together with previous findings (Pausas et al., 2012;Romero & Ganteaume, 2020), we propose that the degree of intraspecific variability differs among fire-adaptive strategies in consistent ways that reflect different selective mechanisms of fire.

ACK N OWLED G EM ENTS
The authors dedicate this paper to the late F. Thomas Ledig, whose pitch pine provenance trials inspired this work and provided the opportunity to conduct it. They also acknowledge John Hom for his contribution of remapping, marking and inventorying the provenance plantation which made this paper possible. They thank Neil

CO N FLI C T O F I NTE R E S T
The authors do not have any conflict of interests to declare.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/1365-2745.13857.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available at https://doi.org/10.6084/m9.figsh are.19143083 .