A unified framework for plant life-history strategies shaped by fire and herbivory

Introduction

Globally, fire and large mammal herbivores are two major consumers of aboveground plant biomass, particularly in tropical ecosystems, where they are important drivers of plant evolution and vegetation structure (Bond, 2005). Empirical evidence demonstrates that fire-adapted and herbivore-adapted plant communities in the same abiotic environments differ in species composition, structure, and plant functional traits (Collins & Barber, 1986; Anderson et al., 2007; Forrestel et al., 2015; Kruger et al., 2017). Indeed, fire and herbivory can be seen as ecological filters, where organisms exposed to these consumers must possess attributes enabling persistence and reproduction, or they will be lost from a community (Belsky, 1992; Cingolani et al., 2005). A substantial literature both in trophic and fire ecology exists, but these bodies of research have developed independently with different theoretical approaches. Fire is seen as a disturbance, whereas herbivory can be considered in terms of predator–prey dynamics; but neither of these theoretical frameworks is entirely satisfactory – see McNaughton (1983) for a discussion on this for herbivory, and Evans et al. (1989) for fire. There are several reasons why it would be beneficial to contrast these two ecological drivers as ‘consumers’ of vegetation (Bond & Keeley, 2005) in a common framework. First, individual plants are often exposed to both fire and herbivory over their lifetime. Second, the intensity and frequency of fire and herbivory depend to some extent on vegetation properties (Burkepile et al., 2013; Platt et al., 2016). Therefore, unlike a disturbance such as drought, there can be feedbacks between community composition and these consumers that can act either to promote or reduce their intensity. Finally, unlike most predator–prey relationships, it is possible for individual plants to survive, or even benefit from, a consumption event (Strauss & Agrawal, 1999; Gagnon et al., 2010).

Through the removal of aboveground biomass, both fire and herbivory can alter competitive interactions within communities by enabling tolerant plants to remain in environments where less-tolerant plants – but better competitors – would otherwise dominate (Collins & Barber, 1986; Cingolani et al., 2005). Intense consumption can alter the architecture of plants, where less intense consumption simply removes leaf material (Danell et al., 1994). Fire is episodic, and it is rare for an ecosystem to sustain more than one fire per year (usually every 2–5 years in tropical grasslands and savannas, and much less frequently elsewhere (Archibald et al., 2013)). Some insect herbivory is also episodic, but many other herbivores are always present, and it is possible to be exposed to repeated, frequent defoliation from herbivores within a day, week, and year (McNaughton, 1983). With fire, plants need to protect remaining, unconsumed living material from extreme heat, whereas with herbivory plants need to be able to withstand the physical action of tugging and breaking of the plant. Moreover, as fire is a physical process requiring energy, heat, and oxygen, and herbivory is a biological process requiring energy, water, and a range of other essential nutrients, these two consumers, while both consuming aboveground biomass, are not necessarily attracted to the same plant parts or plant traits.

Here, we contrast the approaches of fire and trophic ecology integrating these parallel fields of research to define a unified theoretical framework that enables predictions about community assembly and the viability of plant ecological strategies with varying regimes of fire and herbivory. Using grasses as an example, we identify plant functional traits associated with resistance and tolerance of fire and grazing, vs attraction and avoidance of fire and grazing. We use our proposed framework to assess the extent to which adaptations to fire and mammalian herbivory are compatible (i.e. result in the same plant functional types) or antagonistic (select for different plant life histories) and how this might have affected community assembly and, therefore, plant evolution. We discuss what this means for the structure and dynamics of ecological communities exposed to these consumers, and how these adaptations interact with other environmental drivers, such as aridity and cold temperatures. Finally, we consider whether and how the plant traits and life histories identified here relate to other theoretical frameworks in the broader plant trait and plant economics literature.

Contrasting current theoretical approaches in trophic vs fire ecology

In the trophic ecology literature, tolerance is generally defined as ‘the degree to which plant fitness is affected by herbivore damage relative to fitness in the undamaged state’ (Strauss & Agrawal, 1999). Resistance is a separate concept: ‘any plant trait that reduces the preference or performance of herbivores’. Other authors use different terms for the same concepts (Table 1), but it is generally agreed that these represent alternative life-history strategies; that is, plants with traits that make them unlikely to be eaten are not expected to have traits that confer tolerance (van der Meijden et al., 1988) – but see Núñez-Farfán et al. (2007) for further discussion.

Theory has been developed to predict when it would be beneficial for plants to invest resources in avoiding herbivory (Feeny, 1976; Coley et al., 1985; Herms & Mattson, 1992). The converse, that plants might require herbivory in order to be fit and that attracting herbivores could be advantageous, has also been debated (McNaughton, 1983, 1986; Janzen, 1984; Belsky, 1986), but the focus has been on consequences for productivity (overcompensation) rather than overall fitness (Belsky et al., 1993). De Mazancourt et al. (2001) demonstrated that plant–herbivore mutualisms were possible, but only in very restricted circumstances, and this has not been formulated in terms of individual-level selection for palatability per se.

In the fire ecology literature, by contrast, there has been more focus on the mechanism by which species persist in a fire-prone environment. Plants can resist fire (i.e. not be damaged by a fire event), they can avoid fire (i.e. not be burned in a fire event), or they can be burned in a fire event and regenerate, either from resprouting (individual-level persistence) or from seed (population-level persistence) (Whelan, 1995; Gignoux et al., 1997; Pausas & Lavorel, 2003). In the fire literature, there has also been discussion on the degree to which flammability, or the lack thereof, interacts with fire tolerance strategies: flammable plants have associated tolerance traits, and less flammable plants tend to show resistance (Schwilk & Ackerly, 2001). Moreover, flammability has been proposed to increase individual fitness through directing heat away from sensitive plant parts (Gagnon et al., 2010 – individual level) or creating a better environment for offspring by damaging other plants (Bond & Midgley, 1995 – population level) and could, therefore, be selected for, although there are those that argue that flammability is simply an emergent property of selection for other plant functions related to leaf economics (Midgley, 2013). See Pausas et al. (2017) for an extensive review of this topic.

Some herbivore ecologists are more aligned in their terminology to fire ecologists. Briske (1996), for example, defined ‘avoidance’ as the ability to reduce the probability of being eaten – resistance in Strauss & Agrawal's (1999) formulation – and then defined resistance as a concept which integrates both strategies to determine whether a plant can persist and reproduce in the face of a disturbance. Moreover, defence has sometimes been used as an umbrella term incorporating both tolerance and resistance, and elsewhere used synonymously with resistance (Stowe et al., 2000; Fornoni, 2011).

An integrated framework

We suggest that some of these confusions can be resolved by distinguishing between the ability to avoid a defoliation event all together; that is, be unpalatable or nonflammable, and the ability to ‘resist’ a defoliation event when it happens by protecting sensitive plant parts (Box 1). These two concepts are usually conflated into ‘plant defence strategies’ (e.g. Agrawal, 2011), but are quite different ecologically. The ability to ‘resist’ defoliation is a subset of a range of plant tolerance strategies to defoliation and says nothing about an individual plant's attractiveness to herbivores or fire (i.e. the probability of being eaten/burned). For example, Acacia species in African savannas with palatable leaves are also highly defended by spines to prevent too much biomass being removed (Charles-Dominique et al., 2017); that is, though they do not avoid herbivores, they resist them.

Once avoidance is clearly distinguished from resistance then it is easy to see plant life histories as existing on an axis from ‘avoidance’ to ‘attraction’ of fire or herbivory, and that where a plant sits on this axis affects the degree to which it is exposed to consumption, and therefore the strategies that it requires to persist in a community exposed to fire or herbivory (see Fig. 3).

We therefore propose that when considering fire/herbivory adaptations there are three distinct axes, associated with three distinct functions that act over three distinct time periods that need to be quantified for understanding plant life-history strategies (Fig. 1, Box 1). First is the ‘avoidance–attraction continuum’, which acts before a plant is defoliated and determines whether a defoliation event is likely to occur. Second is the ‘resistance’ continuum, which acts during the defoliation event and determines the amount and type of biomass that is damaged or lost by the event. Finally, the ‘tolerance/persistence’ continuum acts over the lifetime of the plant and beyond, integrating an individual plant's response to defoliation and whether a population and species can persist when exposed to a particular level/type of consumer.

As discussed by Strauss & Agrawal (1999), a plant's location on the avoidance–attraction continuum should strongly influence the type of plant resistance and tolerance traits that will be successful in a given environment; that is, not all portions of these three axes will be occupied, but by placing plants on these axes it is possible to identify all possible strategies for surviving and persisting in consumer-prone environments. This different approach (Fig. 1) integrates ideas from both fire ecology and trophic ecology, and it should be able to be applied in both contexts. It both resolves confusion over avoidance vs resistance herbivore defences and incorporates the idea that even the most tolerant plant needs to resist at some level – that is, some part of its growth form needs to be protected from damage for recovery to occur.

Moreover, once both fire and herbivore adaptations are viewed from this combined framework, it should be possible to assess the degree to which adaptations for each consumer are aligned, or whether they select for different types of organisms – for example, is it possible to evolve traits that allow a plant to resist both fire and herbivory, or are there trade-offs such that fire-adapted species are more susceptible to herbivory and vice versa?

Contrasting avoidance–attraction traits for fire and herbivory in grasses

The differences between fire and herbivory become very clear when one considers which traits are associated with palatable vs flammable grasses. Fire burns more easily through dry grasses, with a high energy content (high carbon (C) to nitrogen (N) ratio), because these are easier to ignite and sustain a fire (Simpson et al., 2016). Moreover, thin leaves arranged in an aerated canopy increase ignitability and fire spread rate (Schwilk, 2015). Low phosphorus (P) content and the presence of volatile oils (Scarff & Westoby, 2008; Ormeno et al., 2009) have also been shown to increase flammability.

By contrast, leaves with a high moisture content are preferred by grazers, as this minimizes dependence on external water sources (Jarman, 1973) and is associated with actively growing leaf material with higher crude protein levels (Murray & Brown, 1993). Indeed, grazers prefer forage with a low C : N, which is more digestible and also high P content, as these nutritional components are required as part of a balanced diet that optimally supports metabolic processes (Owen-Smith & Novellie, 1982). Large leaves, clustered together in the canopy, provide high biomass per bite and reduce foraging time, and are thus preferred (Stobbs, 1973).

Tannins, which deter herbivores (Cooper & Owen-Smith, 1985), also slow decomposition rates (Kraus et al., 2003), so will decrease palatability and increase flammability by making more dead fuel available for longer (Grootemaat et al., 2015). Sodium attracts grazers (McNaughton et al., 1997) and silica is thought to deter grazers (Massey et al., 2009), but these constituents have no known impact on fire spread.

Therefore, owing to the differences between fire, a physical process, and herbivores, which are biologically metabolizing their food, the traits associated with flammability are exactly opposite to those associated with palatability: plants that are very flammable are likely to be largely unpalatable, and vice versa (Fig. 2; Table 2).

Contrasting resistance and tolerance traits for fire and herbivory in grasses

Key to understanding the response functional traits for resisting and tolerating disturbances is assessment of what plant parts are being protected. Pausas & Lavorel (2003) suggest that a species can persist in an environment exposed to disturbance at an individual, a population, a community, or a landscape level, and propose that the expected traits for persisting at these levels would be very different. An example in fire-prone environments is the distinction between reseeders, which die in a fire and reproduce again from seed (population-level persistence), and resprouters, which persist at an individual level by resprouting from the base or stem after a fire (individual-level persistence) (Bond & Midgley, 2001; Pausas et al., 2004).

We suggest that this distinction can be taken further to assess what part of an individual is being protected. This can be leaf material, plant structure, aerial buds, flowering culms, basal buds, or root stocks. Resistance strategies thus range from preventing loss of photosynthesis to preventing death (Fig. 1), and traits associated with protecting leaf material and aerial meristems are likely to be different from those that protect roots or basal meristems.

For grasses, resistance to fire requires protecting basal buds from heat, and this requires that they are well hidden in dense layers of leaf material (Daubenmire, 1968). This is achieved through intravaginal branching and retained leaf sheaths. Alternatively, erect culms and distal branching result in a flammable aerial leaf material that carries flames away from the base of the plant, achieving high fire resistance for basal buds (Gagnon et al., 2010) (Fig. 2; Table 2). By contrast, although resisting heavy grazing also requires protecting basal buds, the main risk is uprooting. We propose that a strong root system (root crown below the soil surface) combined with leaves and culms that break easily (low leaf tensile strength and weak nodes) can protect basal buds from grazing (Table 2).

All traits that allow plants to retain leaf material close to the soil surface (lateral (extravaginal) branching, prostrate culms, rooting at the nodes, basal leaf material)) could be considered leaf-level resistance traits against grazing (Fig. 2; Box 2). The lack of leaf abscission allows plants to retain dead leaf material, and has been demonstrated to protect new leaves from grazing (Mingo & Oesterheld, 2009), and some grasses (e.g. Pennisetum mezianum, Triodia basedowii) retain hard spikey culms above ground as a defensive structure to prevent loss of aerial leaf material (O'Reagain & Mentis, 1989; Drescher et al., 2006). This physical defence would also be termed ‘resistance’ in our framework as it protects palatable leaf material from being eaten.

Large bud banks and high photosynthetic rates enable rapid recovery post-fire and promote fire tolerance. These would also promote grazing tolerance; but to prevent death under a patchy, chronic disturbance like grazing, having large stored reserves is another key individual-level tolerance trait (Table 2).

Population-level persistence in fire-prone environments requires preventing seeds from being burned, and rapid germination and recruitment after fire. This is promoted by early seed set and seed release, smoke-stimulated germination, and seed dormancy (Pausas et al., 2017). Tall culms with wind-dispersed seeds promote long-distance dispersal that would enable landscape-level persistence (Boucher et al., 2017). In grazed environments, rapid clonal growth (through lateral spread and rooting at the nodes) promotes population-level persistence. Ectozoochory or edible inflorescences and endozoochory (Janzen, 1984) would be strategies for persisting at a landscape level in a grazed environment.

Clearly, resprouting and rapid growth after defoliation are shared individual-level tolerance traits for fire and grazing, but the resistance traits are often incompatible (Fig. 2; Table 2) – with lateral growth being a good way to hide from grazer mouthparts, and vertical growth being a good way to reduce heat at the soil surface.

Supporting Information Table S1 summarizes the available evidence linking each trait to the functions proposed here, and how to measure it.

Life-history strategies in consumer-controlled environments

The information presented so far supports the idea that avoidance and tolerance should be alternative life-history strategies (van der Meijden et al., 1988; Schwilk & Ackerly, 2001). It also aligns with recent evolutionary theory showing that ‘mixed strategies’ – involving particular combinations of traits associated with attraction vs resistance vs tolerance – could also be evolutionarily stable (Núñez-Farfán et al., 2007; Carmona & Fornoni, 2013). However, considering fire and herbivory together adds a layer of complexity, as plants with traits that enable avoidance of grazing automatically become more flammable, and vice versa.

When one considers fire and herbivory traits together in the context of the three axes: attraction–avoidance, resistance, and tolerance (Fig. 3) we expect that:

From this, four grass life-history strategies emerge (Fig. 3):

Assessing the range of growth forms that exist in tropical grasslands indicates that examples of these four life-history strategies can be found but that there are often multiple ways to achieve the same functional outcome (Table 3). For example, there are at least three different growth forms of grazing-resistant grasses (Hempson et al., 2015): mat-forming stoloniferous/rhizomatous grasses; cushion-forming grasses that have their culm bases below ground and are impossible to uproot; and cage-like, stemmy architectures that protect leaf material in the same way spiny trees do. Moreover, grasses can achieve fire resistance through spreading fire up and away from basal meristems (Gagnon et al., 2010) or by protecting meristems in dense basal tussocks (Trollope et al., 2002). The generalist tolerance strategy can be achieved through stored reserves and a physiology that enables continued regrowth despite substantial loss of photosynthetic tissue (McNaughton, 1983; Tiffin, 2000). However, there are many plants that tolerate both fire and herbivory through having flexible growth forms – growing laterally when exposed to herbivory, and growing vertically when ungrazed and burned (Hempson et al., 2015). This phenotypic plasticity represents a second generalist tolerator life history.

Table 3. There are theoretically four distinct life-history strategies associated with fire and grazing: these are exemplified in approximately eight different growth forms that are common in tropical grassland flora, each associated with particular combinations of plant traits (Box 2) that result in unique architectures.

Life-history strategy	Potential growth forms	Examples	Image	Key traits
(1) Fire resistor, grazer avoider	Aerial flammable tussocks – with vertical growth, and distal tillering that maintains flames above the soil surface (fire resistance largely achieved through being flammable)	Hyparrhenia filipendula, Schizachyrium sanguineum
	Basal resistor tussocks – with vertical growth and dense intravaginally branched culms that protect buds from fire (fire resistance largely achieved through protecting basal meristems)	Panicum natalense, Alloteropsis semialata, Aristida junciformis
(2) Grazer resistor, fire avoider	Mat-forming lawn grasses – with extravaginal branching, laterally growing stems and palatable high-density leaves (require grazing to avoid self-shading and prevent being outcompeted)	Stenotaphrum secundatum, Cynodon dactylon
	Cushion-forming grasses – maintain culms and leaf bases below the soil surface, leaving palatable leaf blades within graze height (require grazing to avoid self-shading and prevent being outcompeted)	Sporobolus nitens, Microchloa kunthii
	Stemmy, cage-like architecture that protects green leaves from being eaten	Pennisetum mezianum, Triodia basedowii
(3) Generalist tolerators	Compensators: tussock grasses that can resist uprooting and have stored reserves, and thus persist when lightly/briefly defoliated	Themeda triandra, Heteropogon contortus, Digitaria eriantha
	Growth-form switchers – these can grow laterally with stolons when grazed, but grow vertically in tall, fire-prone communities	Urochloa mosambicensis, Panicum coloratum
(4) Generalist avoiders	Sparsely branched tussocks with thin leaves, low productivity, and low bulk density	Aristida congesta, Eragrostis rigidior

• Artist: Daleen Roodt.

We have identified approximately eight growth forms (Table 3) that could be effective in consumer-controlled ecosystems), which are by no means a complete set; the universality of these growth forms needs to be tested with data from a wide range of grassy ecosystems. The growth form that dominates in a particular environment will depend on the degree of grazing or fire, as well as other environmental constraints plants are placed under (Coley et al., 1985). For example, in very mesic environments C is less limiting than N or P, so there could be selection for tall, stemmy, C-rich, architectures that promote height gain. These growth forms are also more flammable, so the fire resistor/grazer avoider strategy would be common. In arid environments, light competition is less severe, so the generalist avoider strategy might be able to persist, despite the reduced growth rates associated with small, sparse canopies. Very cold environments might not be conducive to extravaginal branching or distal tillering, which exposes buds, but a dense tussock growth form could confer tolerance to both fire and cold. The flexible growth-form switcher strategy is likely to be most effective in places where the consumer shifts from fire to herbivory over time, whereas the generalist tolerator (compensator) strategy is predicted to be effective when exposed to persistent but intermediate levels of either consumer. Nutrient-rich environments also probably enable the compensator strategy that requires high rates of regrowth and productivity. Moreover, seasonal aridity determines how effective fire avoidance can be. Only in places without seasonal aridity can plants maintain a high leaf moisture content all year; consequently, in temperate places, we expect fire avoidance and grazing tolerance to be the most common strategy.

Annual plants often represent ‘avoider’ life histories, or, alternatively, do not survive fire/herbivory and persist in these environments through effective recruitment from seed (see Table 2). Many of these predictions fit with what has been observed by existing global analyses of grazing traits (McIntyre et al., 1999; Díaz et al., 2007); however, here, fire is explicitly integrated into the same predictive framework.

The framework presented here differs from classic plant succession theory – often formalized for grasslands using the ‘Increaser/Decreaser’ strategy framework (Foran et al., 1978) – in a few key ways. ‘Increaser 2’ species are defined by Foran et al. (1978) as those that increase when landscapes are heavily grazed. Following the aforementioned framework, this could occur because they are unpalatable and avoided by grazers (Aristida congesta), or because they are palatable and resistant to grazing (Pennesitum clandestinum). From a land management point of view, it is essential to distinguish these two functional groups, because one is desirable to a cattle farmer and the other is not. This issue has been highlighted by Milchunas et al. (1988), but up to now there has been no formal way to distinguish these grazing functional groups.

Ecological and evolutionary implications

When ordering grass communities across a ‘consumer’ gradient from frequent fire to intense grazing one expects turnover in the functional types that persist and dominate (Fig. 4). As the generalist strategies are only effective at intermediate levels of fire and herbivory, the prediction would be that there is higher functional diversity in these environments, and that fewer strategies should exist in extreme fire and extreme grazing situations. For the same reason, we would also expect generalist species to have larger range sizes. Fig. 4 predicts that high grazing can potentially result in two distinct ecosystem states: grazing lawns (McNaughton, 1984) or systems where only generalist avoiders can persist. There is ample field evidence for these two different grazing endpoints (Mack & Thompson, 1982; Milchunas et al., 1988), and understanding the conditions that result in each state is an important management issue.

The axis from fire to grazing (Fig. 4) can occur across regional gradients, where fire-prone mesic ecosystems transition into grazer-dominated ecosystems at lower rainfalls (Bond, 2005; Archibald & Hempson, 2016). However, it is also possible for feedbacks between grass communities and their consumers to maintain either fire-adapted or grazer-adapted grasslands within a single landscape (Hempson et al., 2019). Here, the genetic pool of grasses is similar, and differentiation occurs through filtering of grass functional types and through feedbacks to consumer regimes. Because grazing promotes low-statured grasses with leaf traits that deter fire, and fire promotes tall-statured grasses with leaf traits that deter herbivores (Fig. 2; Table 2), these habitat types should be quite distinct in terms of species composition. If the grasses in grazing lawns are also phylogenetically distinct from those in surrounding tall grass landscapes, this would be strong evidence for different evolutionary origins of fire- vs herbivore-adapted ecological strategies. Alternatively, grasses from all lineages have the capacity to evolve fire and herbivore tolerance, in which case these communities would not show strong phylogenetic patterns.

Field data indicate that geographically distant grasslands show similar functional and phylogenetic responses to changes in fire regime, but divergent responses to changes in grazer regime (Forrestel et al., 2014, 2015). This implies that grazer adaptations vary more between regions than fire adaptations do, and are dispersed more widely in the phylogeny, a hypothesis that can now be tested using the traits and life-history strategies identified here. Different suites of grazers, with different feeding ecologies, evolved independently across the globe (Owen-Smith, 2013), whereas fire regimes are an outcome of climate and fuel properties and will converge in conditions where these are similar (Archibald et al., 2013). Demonstrated links between fire and grasses of just one lineage, the Andropogoneae, support this (Ripley et al., 2015; Simpson et al., 2016).

Moreover, plant height is a key plant functional trait (Westoby, 1998; Díaz et al., 2016), and selection for a vertical (fire resistor) vs lateral (grazer resistor) growth will have consequences for light competition as well as other aspects of plant life history. For wind-pollinated plants, like grasses, height affects gene flow and dispersal distance (Rodger et al., 2018), with consequences for plant range size and rates of speciation (Boucher et al., 2017). We would, therefore, expect that grazer-adapted grasses should have smaller ranges overall, although this would depend on dispersal mode.

It is important to recognize that fire is not actually a very strong filter for grasses: the absence of wood and the concentration of meristems at/below the soil surface mean that most fires burn material that is already dead and ready to be discarded – slow creeping back-fires (Trollope et al., 2002) are the exception here and might be the selective force behind leaf-sheath retention and dense basal tussocks. Individual-level resistance to fire, therefore, becomes less important than community-level processes after the fire (Pausas & Lavorel, 2003). Grass species with high growth rates and rapid height gain are effective competitors for space and light in the high-resource environment after a fire, and these tall grasses are more flammable (high rates of biomass accumulation and connected fuels). Thus, tall grasses competing for light reinforce a fire feedback to increase flammability (D'Antonio & Vitousek, 1992; Rossiter et al., 2003). High-fire environments, therefore, exclude other herbaceous growth forms more through competition and shading than through the frequency of fire, whereas it is the fire itself that excludes many woody growth forms from these ecosystems.

We therefore expect:

Discussion

We have demonstrated here that making sense of the ecological strategies found in consumer-controlled environments requires integrating understanding across disciplines. In particular, in grasslands, it is only possible to develop clear predictions around community assembly and evolution when the drivers of fire and large mammalian herbivory are considered together, because adaptations for one type of consumer can affect how susceptible plants are to other consumers. We therefore advocate that datasets on fire and grazing traits be collated and the expected relationships tested.

Key to integrating fire ecology and trophic ecology was developing a theoretical framework related to attraction/avoidance, resistance, and tolerance of consumption (Fig. 1). This framework blends aspects of previous conceptual models, but it makes a strong distinction between resistance traits (that act while a plant is being consumed) and attraction/avoidance traits (that determine whether plants are likely to be consumed in the first place). Moreover, it also forces trophic ecologists to raise questions about the degree to which palatability can be selected for – something fire ecologists have been grappling with for years.

Once fire and herbivore adaptations can be placed on common axes (Fig. 3), it is possible to assess the degree to which they are correlated or antagonistic. In this example, we show clearly that fire and herbivore adaptations in grasses are often contradictory: that flammable grasses are not palatable, and that grazing-resistant grasses are not necessarily fire resistant (Fig. 2; Table 2). This then results in expectations about the dominance of different grass life-history strategies across environmental gradients, and also the degree to which fire and herbivory can shape ecosystems and act to reinforce particular consumer regimes through altering species composition.

It would be good to ask similar questions regarding fire and herbivory adaptations in trees and nongrass herbaceous species, and to expand this thinking to other consumers, such as insect herbivores or birds (Lee et al., 2010). Testing this framework in these different contexts will demonstrate its universality.

The tolerance strategies associated with different positions on the avoidance–attraction continuum have already been identified for trees (Fig. 3). For example, in pines, there is correlated evolution in traits associated with attracting and tolerating vs avoiding and resisting fire (Schwilk & Ackerly, 2001; He et al., 2012). Moreover, within savanna trees, it has been shown that growth forms associated with resisting herbivory can make trees less resistant to fire (Archibald & Bond, 2003) and that this can result in sorting of savanna tree communities in space (Charles-Dominique et al., 2015; Osborne et al., 2018) and over time (Staver et al., 2007) as the consumer changes. Thus, one of the major predictions of this framework appears to hold true for woody species.

It has always been difficult to fit life-history strategies associated with herbivory and fire into classic ecological theory (Bond & Midgley, 2001; Pausas & Keeley, 2014). The ruderals in Grime's (1977) competitors–stress tolerators–ruderals scheme are predicted to occur in highly disturbed environments, but this strategy represents only population-level persistence and cannot encompass the full spectrum of strategies described here (Figs 1, 3). However, as already discussed, it should be possible to make predictions about what combination of avoidance, resistance, and tolerance traits are more likely for plants in different environments, and these need to be reconciled with the strategies identified for C capture, water and nutrient use efficiency, and reproduction in the same environments. For example, avoiders, which have defensive chemicals in their leaves, have been shown to characterize Grimes's ‘stress tolerator’ strategy because protecting leaf and aboveground biomass is important when this biomass is hard won (Coley et al., 1985).

Many of the traits we discuss here in the context of fire and herbivory are also considered important in the broader trait and plant economics literature, such as C : N ratio and plant height (Westoby, 1998; Wright et al., 2004; Díaz et al., 2016). However, classic leaf trait data are not sufficient to fully quantify consumer-driven life histories, and there is a need for further quantification and empirical testing of these consumer-related traits. To enable this, in Table S1 we describe sampling protocols for the fire and grazer traits mentioned in Table 2 and contrast them with similar traits in woody plants. We include traits associated with population- and landscape-level persistence, although we were not able to elaborate on these here.

Author contributions

SA, GPH, and CL jointly conceived the manuscript and developed the ideas. SA wrote the manuscript with contributions from GPH and CL. GPH perfected the figures.