Have plants evolved to self-immolate?

By definition fire prone ecosystems have highly combustible plants, leading to the hypothesis, first formally stated by Mutch in 1970, that community flammability is the product of natural selection of flammable traits. However, proving the "Mutch hypothesis" has presented an enormous challenge for fire ecologists given the difficulty in establishing cause and effect between landscape fire and flammable plant traits. Individual plant traits (such as leaf moisture content, retention of dead branches and foliage, oil rich foliage) are known to affect the flammability of plants but there is no evidence these characters evolved specifically to self-immolate, although some of these traits may have been secondarily modified to increase the propensity to burn. Demonstrating individual benefits from self-immolation is extraordinarily difficult, given the intersection of the physical environmental factors that control landscape fire (fuel production, dryness and ignitions) with community flammability properties that emerge from numerous traits of multiple species (canopy cover and litter bed bulk density). It is more parsimonious to conclude plants have evolved mechanisms to tolerate, but not promote, landscape fire.

Using a rainforest-flame forest mosaic to test the hypothesis that leaf and litter fuel flammability is under natural selection.

We used a mosaic of infrequently burnt temperate rainforest and adjacent, frequently burnt eucalypt forests in temperate eastern Australia to test whether: (1) there were differences in flammability of fresh and dried foliage amongst congeners from contrasting habitats, (2) habitat flammability was related to regeneration strategy, (3) litter fuels were more flammable in frequently burnt forests, (4) the severity of a recent fire influenced the flammability of litter (as this would suggest fire feedbacks), and (5) microclimate contributed to differences in fire hazard amongst habitats. Leaf-level comparisons were made among 11 congeneric pairs from rainforest and eucalypt forests. Leaf-level ignitability, combustibility and sustainability were not consistently higher for taxa from frequently burnt eucalypt forests, nor were they higher for species with fire-driven recruitment. The bulk density of litter-bed fuels strongly influenced flammability, but eucalypt forest litter was not less dense than rainforest litter. Ignitability, combustibility and flame sustainability of community surface fuels (litter) were compared using fuel arrays with standardized fuel mass and moisture content. Forests previously burned at high fire severity did not have consistently higher litter flammability than those burned at lower severity or long unburned. Thus, contrary to the Mutch hypothesis, there was no evidence of higher flammability of litter fuels or leaves from frequently burnt eucalypt forests compared with infrequently burnt rainforests. We suggest the manifest pyrogenicity of eucalypt forests is not due to natural selection for more flammable foliage, but better explained by differences in crown openness and associated microclimatic differences.