Pushing back in time: the role of fire in plant evolution


He et al. in this issue of New Phytologist (pp. 184–196) have a paper titled ‘Banksia born to burn’ in which they used a dated molecular phylogeny of this iconic Australian genus to argue that we need to push back in time, by several tens of millions of years, the role of fire in plant evolution. Their study also notes several new traits they consider as fire adaptations, the phylogenetic position of these traits and the likely date of origin of these and other better known fire-traits.

At this stage it seems that ancient fire is uniquely Australian.

The evolutionary impact of fire on plant evolution is difficult to trace because the taphonomy process is against preserving evidence of fire in both space and time. First, bio-geographically, fires typically take place in semi-arid sites (at least seasonally) and at these sites, or seasons, the preservation of burnt plants is unlikely. Ironically, some of the best palaeo-evidence for fire is from soggy mires! These extremely mesic sites produce useful inertinite records in coal deposits, such as used by Glasspool & Scott (2010). Second, plants evolve to fire regimes, so the occasional fossil presence of burnt plants (such as of a single inertinite layer), may not be sufficient temporal evidence of a fire regime that is predictably frequent to initiate an evolutionary response by plants. Stronger evidence for the evolutionary role of fire (i.e. the occurrence of a fire regime), would be of the occurrence of dated plant parts or traits that are unequivocal fire adaptations. There are very few such plant traits which would fossilize.

One way around these taphonomic problems is the use of dated phylogenies of fire traits, such as by Simon et al. (2009) and Bytebier et al. (2011), who returned dates of between 1 and 15 million yr ago (Ma) for the evolution of fire adaptations. By contrast, He et al. have now argued that fire adaptations such as on canopy seed storage (serotiny), date back some 40 Ma earlier to the Palaeocene (60 Ma). They also suggest that other fire adaptations, such as the retention of flammable dead leaves and clonal resprouting, appear to have originated much later, in the Miocene (26–16 Ma). They acknowledge that their Palaeocene dates are surprising, because there is no fossil evidence of fire at this time in Australia. Global analyses suggest declining fire after the end of the Cretaceous and that fire only significantly picks up again c. 10 Ma (Bond & Scott, 2010). Because much of Australia is nutrient-poor, it is inherently fire-prone due to intense sclerophylly (hard leaves) and associated long leaf retention times and low levels of herbivory. Thus fire-adapted tall (> 20 m) eucalypt forests presently exist even in mesic areas (annual precipitation of > 1000 mm) in parts of Australia. The argument would be that in previously very mesic times, there may still have been some areas which could burn seasonally. This seems feasible.

Are He et al. correct in considering that serotiny is a fire adaptation? In serotinous plants, seeds are stored in closed cones in the canopy for several years. These cones typically only open after the plant has been burned and thus serotinous species only successfully recruit (i.e. seedlings grow to maturity) in the post-fire environment. Bradshaw et al. (2011) recently argued that plants from nutrient-poor environments tend to produce low numbers of seeds annually and that serotiny merely allows them to accumulate a seed-bank (i.e. that serotiny is an adaptation to nutrient poverty and only an exaptation to fire). The exaptation argument is not reasonable; there are many nonfire prone sclerophyllous forest species on nutrient-poor soils, none of which is serotinous. We thus agree with He et al. that serotiny is strictly associated with fire. The exaptation argument is further laid to rest by several points emerging from He et al. First, by focussing on Banksia hookeriana, they draw attention to the fact that many serotinous Banksia species have cones that will largely only open after their resin seal has melted during burning (pyrescence), rather than merely opening after having dried out after a fire or drought (xeriscence). They show that in the absence of fire, even if B. hookeriana plants die (such as they did in a drought), this does not lead to recruitment because the cones did not open. That He et al. date the origin of serotiny to the Paleocene implies that fire has regularly been occurring for an enormous amount of time, certainly long enough for several fire adaptations to have been selected for.

He et al. venture into the contentious topic of plant flammability, that is, the evolution of traits that influence whether a plant will burn in a fire and how this apparently suicidal behaviour could be selfishly beneficial (Bond & Midgley, 1995).

They make the novel observation that some banksias retain dead leaves. This is probably the clearest flammability trait of all. The low water content of dead leaves means this trait will definitely impact local fire intensity, it presumably incurs a cost to the host plant (some reduction of nutrient cycling, some mechanical support) and there is no obvious secondary or exaptation benefit of retaining leaves. Dead leaf retention by Banksia may be matched in other fire-prone ecosystems. For example, we have now observed it in Protea amplexicaulis, a short serotinous reseeder in the Cape Proteaceae that retains dead leaves for up to 6 yr, a trait not previously noted in a monograph on Protea (Rourke, 1980).

But can dead leaf retention be explained? They argue that the benefit is that during fire, the incineration of dead leaves (as opposed to mere singeing, or killing, of live leaves) will cause highly localized release of mineral nutrients to the soil. This could later benefit the seedlings of flammable burnt mother plants. It is not obvious that fire-released nutrients from flammable plants would have limited dispersal, nor whether seedlings of serotinous plants would also have limited and congruent dispersal. Seeds of serotinous species often have dispersal structures (such as wings in Banksia). Besides wind dispersal they are also moved by water and accumulate in groups in micro-weirs (Lamont et al., 1993). Thus the selfish fertilization hypothesis of He et al. seems unlikely and needs further information on nutrient benefits and seed dispersal. He et al. also note that many Banksia species retain flammable florets on serotinous cones. Whilst the retention of florets may increase fire temperatures around cones and thus facilitate opening of resin sealed serotinous cones after fire, it fails to address why this pyrescence evolved in the first place. For example, pyrescence does not occur in Cape Proteaceae.

Nevertheless, their phylogenetic analysis has several implications for the understanding of high flammability (given that sclerophylly in seasonal climates is already a likely exaptation for flammability). First, the Banksia phylogeny shows reseeding evolved very early on and that most flammable plants are reseeders. This indicates that high flammability is not a suicidal behaviour, because reseeders would die in fire anyway; they die in even the mildest of fires. Second, He et al. show that high flammability traits evolved much later than did serotiny. The occurrence of serotiny already implies the selective agent of a fire regime. That flammability evolved after serotiny therefore suggests that high flammability traits do not influence whether fire would occur or not in a community, but rather how it influences local intensity. Thus highly flammable plants would not influence community fire frequencies. But just why plants evolve to burn intensively, still remains speculative.

Are the dates provided by He et al. correct? Molecular dating has its own internal issues, such as whether there is reasonable resolution and calibration, and their dating will no doubt be critically evaluated. These issues notwithstanding, their results gain support from a recent paper by Crisp et al. (2011). Also working on iconic Australian fire taxa, Crisp et al. (2011) argue that deeply buried epicormic resprouting strands in the stems of Australian Myrtaceae (eucalypts or gum trees) are fire adaptations and that these too have a 60 Ma origin. Eucalypt trees are singularly impressive in their rapid epicormic resprouting response after massively intense canopy fires and therefore it is highly likely that these epicormic strands are fire adaptations. The striking agreement between the dates of these two papers on two unrelated taxa is thus suggestive of an ancient fire regime in this continent. Serotinous taxa imply fire-prone communities not just fire-prone species!

At this stage it seems that ancient fire is uniquely Australian. Molecular dating of African Proteaceae suggest they are a recent dispersal event; Valente et al. (2010) have a crown age of 17.7–25.2 Ma for Cape Protea. Serotiny also seems to us to be a basal trait in Protea. Bytebier et al. (2011) have a date on fire stimulated flowering of Cape orchids of 13–15 Ma, again suggesting fire is more recent in Africa. Simon et al. (2009) argued that South American cerrado has a 10 Ma origin, but that most of the diversification would have been within the last 4 Ma. A draw back of the dated phylogeny approach to date the onset of fire regimes is that it is dependent on how old the study taxon is. In this sense nonangiosperms, such as conifers, may be useful to push fire back even further. For example, Widdringtonia in South Africa has fire-adapted taxa that are serotinous and resprout.

Glasspool & Scott (2010) showed that fossil evidence for the occurrence of fires (percentage of inertinite in coal) was high throughout the Cretaceous. Bond & Scott (2010) argued that this may have opened up ecosystems and facilitated the expansion of weedy angiosperms. This ‘fire in the Cretaceous hypothesis’ cannot be tested on Banksia because the taxon is too recent. Glasspool & Scott (2010) also noted that inertinite declines through the Palaeogene and Neogene in the Northern Hemisphere. The results of He et al. and Crisp et al. (2011) suggest that, by contrast, fire was still burning regurlarly during this period, at some reasonable scale, in some parts of the Southern Hemisphere.