Thirsty roots and hungry leaves: unravelling the roles of carbon and water dynamics in tree mortality



Featured paper: See also the Editorial by McDowell et al

Research into the mechanisms driving drought-related plant mortality has seen a focussed effort in recent years. Drought and water availability are pervasive factors influencing the distribution of forests and woodlands globally, particularly in water-limited environments, where evaporation exceeds rainfall and is therefore a major constraint on productivity. However, the lack of a predictive framework for forest mortality remains an important knowledge gap in ecosystem and biogeochemical models (Roxburgh et al., 2004). The paper by Hartmann et al. (pp. 340–349) in the Feature on drought-related mortality in this issue of New Phytologist provides some novel insights into the relative contributions of hydraulic failure and carbon (C) starvation in trees. In conjunction with recent activity focussed on unravelling the relative importance of the putative mechanisms of plant mortality, our goal should be to improve our understanding of C, water and nutrient dynamics in ecosystems and improve our capacity to predict conditions under which mortality is likely to occur.

‘What is the functional role of carbohydrates during drought?’

While much of the current research is focussed on the interaction of hydraulic failure and C starvation (McDowell et al., 2008), it is important to recognize that there are many physiological processes that might generate tree mortality. The evidence for hydraulic failure facilitating cell dehydration and eventual tree death is supported by a considerable body of research into the role of cavitation in limiting water transport. By contrast, tree mortality associated with systematic depletion and cessation of C metabolism remains difficult to separate from hydraulically mediated impacts on plant function. Despite this, evidence for drought-related declines in carbohydrate stores is beginning to emerge (Galiano et al., 2012). Hartmann et al. cleverly demonstrate potential experimental approaches that could be used to unravel the intricate interdependencies of C and water dynamics in plants. In their paper, Hartmann et al. designed experimental approaches aimed at disentangling the relative roles of hydraulic failure and C starvation during terminal drought in Norway spruce (Pinus abies) seedlings.

Using specifically designed small growth chambers, Hartmann et al. manipulated water availability and the supply of C by stripping CO2 from the atmosphere of the chambers. They grew Pinus abies at two atmospheric [CO2] (75 and 350 ppm), under high and low water availability. Low CO2 experiments are an ideal method for providing baseline data from which to assess the impacts of anthropogenic changes in the atmospheric composition (Gerhart & Ward, 2010). The very low atmospheric [CO2] treatments in this experiment were specifically designed to manipulate C supply and induce C starvation. The advantage of this approach was that it created direct C depletion effects that were independent of the indirect effects from an imposed drought. In this experiment, well-watered P. abies seedlings grown at very low [CO2] survived for c. 7 wk longer than droughted seedlings, regardless of whether the droughted seedlings were grown under ambient or depleted [CO2], leading to the conclusion that ‘thirst beats hunger’.

The changes in tissue carbohydrate concentrations associated with these treatments were particularly insightful for interpreting the relative contributions of hydraulic failure and C starvation associated with the observed mortality. Droughted seedlings grown in ambient [CO2] appeared to accumulate starch and sucrose, while droughted seedlings grown in depleted [CO2] exhibited depletion of starch and sucrose in some, but not all tissues. Importantly, the well-watered seedlings grown in the depleted [CO2] environment showed significant and almost complete depletion of carbohydrates in all tissues. This was one of the first studies to comprehensively demonstrate mortality as a function of C starvation, and the mechanisms that contributed to this mortality. However, uncertainty remains regarding details of how these depleted carbohydrate concentrations facilitated mortality and the sequence of processes that promoted cell death. For example, Hartmann et al. report that at death, in the well watered, low [CO2] seedlings, tissue relative water contents were below 10%, although this striking result is not really explored in detail. Low carbohydrate concentrations in roots may eventually inhibit fine root survival and water uptake, and potentially promote hydraulic limitations to survival.

On reflection, it may not be all that surprising that C starvation has been difficult to demonstrate during drought. The tight coordination of water and vapour transport in the soil-to-atmosphere continuum places physical constraints on the acquisition of C from the atmosphere. However, evaporation from plant surfaces can continue long after stomata have closed, and may indeed be exacerbated by higher temperatures associated with drought. The nature of the drought treatment imposed by Hartmann et al. meant that dehydration via hydraulic failure occurred before the complete depletion of carbohydrates. In this context, the findings of Hartmann et al. emphasizes the dominant role of hydraulic limitations for modelling drought mortality in forests.

Of course, in nature drought is rarely the only agent implicated in mortality events. In Australia for example, nearly all documented mortality events were associated with heat waves and over half of these were associated with biotic agents. While it has been suggested that higher temperatures associated with droughts may result in more rapid depletion of carbohydrate reserves (Adams et al., 2009), heat waves during drought (when the capacity for evaporative cooling is significantly impaired due to stomatal closure), could be particularly important in generating plant mortality. Thus, mortality directly associated with tissue damage due to high temperatures may also be an important contributing factor. While much of the current research is focussed on determining key hydraulic and carbohydrate concentration thresholds, identifying critical temperature thresholds for many important physiological processes in trees has rarely been addressed. In a recent study of short-term leaf temperature responses to high temperatures, O'Sullivan et al. (2013) identified critical leaf temperatures (between 50 and 57°C) beyond which processes, such as respiration, sharply decline. More studies that examine temperature sensitivities for key physiological processes will be essential for predicting the vulnerability of plants and ecosystems to future extreme temperature events associated with predicted increased global temperatures (Peters et al., 2013). Clearly, protracted soil-water deficits can heighten tree exposure to multiple abiotic and biotic stressors and amplify mortality risk via significant changes in the energy and C balance of plants.

Interpreting variability in carbohydrate responses during drought

To the best of our knowledge, no studies have demonstrated complete exhaustion of carbohydrates during drought (Sala et al., 2012). In the Hartmann et al. study, tissue carbohydrate concentrations at mortality were higher in the droughted trees grown in depleted [CO2] than in the well-watered trees grown at depleted [CO2]. These observations raise the question: what is the functional role of carbohydrates during drought? Patterns of carbohydrate depletion in plants during experimental or natural drought studies vary considerably. While duration and intensity of the drought event will undoubtedly be an important component of this variation, so too will be the co-ordination of C supply from photosynthesis and C demand from growth respiration and osmotic regulation (McDowell, 2011). Based on earlier work of Hsiao, McDowell (2011) presented a conceptual summary of the coordination of growth, photosynthesis and respiration (including other C demands). This conceptual model predicted the trajectory of the non-structural carbohydrate pool during drought and identified stages of carbohydrate accumulation and depletion, which were related to the duration and intensity of the drought. The generality that carbohydrates accumulate during drought has been posited on evidence from experimental studies on predominately herbaceous species, where initial reductions in sink demand are sustained as drought progresses (Muller et al., 2011). The dynamics of drought in long-lived woody species may be different to that of herbaceous species as woody trees may experience extended periods of little or no C assimilation while demand for carbohydrates is maintained. Mitchell et al. (2013) examined drought mortality in tree species with contrasting growth and water-use strategies and found that the regulation of growth and water-use during drought played an important role in the regulation of carbohydrate pools during mortality. Thus, the fast-growing, profligate water-use strategies of Eucalyptus trees predisposed these species to rapid hydraulic failure, but overall non-structural carbohydrates remained relatively unchanged at mortality. By contrast, the slower growing Pinus radiata with very conservative water-use strategies, exhibited significant depletion in carbohydrate reserves at mortality. Thus, we need to better understand the degree to which growth and photosynthesis are indeed coupled or preferentially affected in the context of both the severity and duration of the drought stress event.

Towards understanding the ecological implications of hydraulic and carbohydrate traits

Hartmann et al. caution against extrapolation of their growth chamber results to natural ecosystems. Despite this caution, prediction of the impacts of drought on ecosystem processes is an important driver of ecophysiological studies. Based on their review of hydraulic vulnerability data from species around the globe, Choat et al. (2012) hypothesized that all biomes were ‘equally vulnerable to hydraulic failure regardless of their current rainfall environment’. A bold claim indeed! While it is evident that many species have evolved to match their hydraulic architecture to climatic and site conditions within narrow hydraulic safety limits, in isolation this provides limited predictive power in assessing community and ecosystem vulnerability to water deficits. In interpreting the findings of ecophysiological studies, it is important to place knowledge of the ecophysiological traits being measured within the context of existing ecological frameworks.

Stress events rarely occur in isolation and the capacity of an ecosystem to absorb these stress events provides insights into ecosystem resistance. Conversely, ecosystem processes are rarely static, and disturbance regimes play an important role in maintaining community dynamics (e.g. the role of fire in maintaining savanna structure). Variation in hydraulic thresholds among species such as the water potential at stomatal closure, minimum seasonal water potential, and the water potential at 50% loss of conductance (i.e. traits related to the plants hydraulic vulnerability), are indicative of the capacity of that species to withstand drought stress, or the resistance of that species to water deficits (Fig. 1). By contrast, mechanisms that govern the rate of recovery of these systems to repeated disturbances are measures of the resilience of these ecosystems (Fig. 1). The capacity of these species to recover from drought stress will be governed to some extent by the carbohydrate reserves remaining when the drought is alleviated (Sala et al., 2012). Thus, tissue carbohydrate concentrations may be an important resilience trait. The ecological concepts of resistance and resilience are thus conceptually similar to species traits of drought tolerance (e.g. isohydry/anisohydry). Species with a high resistance to water deficits, (e.g. isohydric species) may have low resilience to these stressors, indicated by low capacity for recovery (Ogasa et al., 2013). Alternatively, species with low resistance to water deficits (e.g. anisohydric species) may have a higher capacity for recovery from the water stress events, as measured by limited carbohydrate depletion during drought or the propensity for coppicing or epicormic recovery. Thus, when considering the susceptibility of ecosystems to future droughts or other extreme events, it is important to assess traits that quantify the vulnerability of a species to the stress (e.g. resistance traits) and to assess traits that underpin the capacity of species to recover from these events. Furthermore, we need to be careful of developing false dichotomies (hydraulic failure vs C starvation, isohydric vs anisohydric, resistant vs resilient). Evidence for functional convergence (Meinzer, 2003) suggests that plant strategies occupy a response continuum to resource constraints and that individual species occupy different regions of these continuums.

Figure 1.

A simplified conceptual diagram of plant responses to drought. The response curve (green line) represents changes in plant function in response to a single drought event. The magnitude of the response is determined by (A) resistance traits, whereas the time for plant function to return to pre-drought conditions is determined by (B) resilience traits. Some examples of such resistance or resilience traits are presented.

Improving predictions of drought mortality, concluding remarks

While models that incorporate known hydraulic thresholds can predict hydraulic failure in trees (e.g. Battaglia et al., 2004), and we can use these as a basis for predicting drought-induced mortality, our poor understanding of C dynamics under well-watered and droughted conditions limits our capacity to integrate carbohydrate dynamics into forest growth models even though there is evidence for the role of carbohydrates in recovery (Sala et al., 2012). The key insights gleaned from this paper offer a framework for building this capacity. The approach of Hartmann et al. can be used, in conjunction with techniques such as stable isotopes, to address questions such as:

  • The role of active and passive storage of carbohydrates in trees, including an understanding of the limits to storage and depletion of these reserves (Sala et al., 2012).
  • The limits to carbohydrate transport and phloem function under stress.

These types of questions require elegant experiments that define the nature of complex interactions in physiological processes. Studies that examine the hydraulic and C dynamics in an integrated manner, but can also tease apart the key mechanisms, will be particularly important for improving models of forest mortality. Furthermore, we must translate this mechanistic knowledge into a broader ecological context that accommodates different species strategies and environmental conditions so that we may assess the impacts of future disturbances on ecosystem processes.


In writing this commentary, the authors acknowledge the useful insights and discussions of all of their colleagues during recent workshops on forest mortality hosted by the Australian Centre for Ecological Analysis and Synthesis (, a facility of the Australian Government's Terrestrial Ecosystems Research Network, a research infrastructure facility established under the National Collaborative Research Infrastructure Strategy and Education Infrastructure Fund – Super Science Initiative – through the Department of Industry, Innovation, Science, Research and Tertiary Education.