All or nothing? The importance of considering partial dieback for tree responses to fire

Abstract Key message We introduce an approach to studying partial crown dieback that accounts for height profiles of bark thickness and stem vulnerability to fire, resulting in improved modeling of biomass loss. Abstract Fire mediates tree cover in savannas by causing topkill, typically represented as a binary process in which the whole stem either survives or dies, overlooking losses of foliage and branches from partial canopy dieback. To overcome this limitation, we introduce an approach that focuses on conditional probabilities of dieback of stems and branches, which we demonstrate with a Brazilian savanna tree. We quantified the probability of branch death as a function of bark thickness and height above ground, to parameterize a model of tree architecture for simulating aerial biomass losses under scenarios of differing fire intensity, maximum tree height, and investment in bark. The study population experienced a 43% loss of stem biomass when exposed to a prescribed fire, but the traditional all-or-nothing approach that ignores partial dieback accounts for only half of this loss. Simulations show that, in absolute terms, the traditional approach more substantially underestimates carbon losses in severe fires, but in relative terms, the underestimation is greater in mild fires. A benefit–cost analysis revealed that the observed investment in bark more closely matches the predicted optimal investment when we account for partial dieback. In scenarios of low fire intensity or taller tree stature, the model predicts lower investment in bark, compared to the default scenario. We introduce the concept of bark safety margin, which quantifies the relative protection afforded by bark in the main stem and branches. This study thus demonstrates the importance of considering partial stem dieback, in addition to offering a new approach for quantifying this dieback.