The way back: recovery of trees from drought and its implication for acclimation

Resilience is a key concept in ecology, describing the capacity of a system to resist a disturbance, recover from it and return to a stable state (Lloret et al., 2011; Hodgson et al., 2015; Ingrisch see Box 1). Whilst it is indisputable and a matter of intensive research (e.g. McDowell et al., 2008; Meinzer et al., 2009; Vicente-Serrano et al., 2013) that ‘resistance’, that is, the ability of individuals and ecosystems to persist and maintain their functioning during a disturbance, is a key aspect of resilience, the importance of ‘recovery’ after a disturbance event for resilience comes more and more into the focus of research (e.g. Brodribb Galiano et al., 2011; Arend et al., 2016; Hagedorn et al., 2016). Only recently, Ingrisch Spinoni et al., 2018), putting strong pressure on forests; global observations of drought-induced forest decline and tree mortality already show this pressure (Allen et al., 2010; Hartmann et al., 2018; Schuldt et al., 2020). Severe drought limits primary metabolic processes and tree physiological functions, and can strongly affect the whole ecosystem functioning and the ecosystem services that forests provide (Ciais et al., 2005; McDowell et al., 2008; Anderegg et al., 2016). Due to their long lifespan, trees must have evolved specific mechanisms to cope with recurrent droughts, that is, the ability to survive water limitation and recover from it. The recovery of trees from an incidence of drought may take days, weeks or even years, depending on the timing, duration and severity of the experienced stress, on tree ontogeny, height and social position, and taxonomic group (Hanson et al., 2001; Anderegg et al., 2015; Bennett et al., 2015; Grote et al., 2016). Ruehr et al. (2019) provided a conceptual framework of the relationship between stress intensity and recovery within one growing season. Whilst the effects of mild stress are fully and immediately reversible, higher stress levels are supposed to result in structural damage, causing recovery to be slow and dependent on regrowth of lost and damaged tissues. Moreover, a recent study of a global tree-ring database suggests that slow recovery is associated with higher mortality risk in conifers, whereas no such trend was found in woody angiosperms (DeSoto et al., 2020). It has also been shown that when drought events occur in close sequence (i.e. when a second event occurs before the system has fully recovered from the first) the overall drought impact and the total recovery time can increase (Mitchell et al., 2016). Duration of recovery also depends on the scale and complexity of the function examined. Whilst molecular functions and physiological processes might regain function within days to weeks (Bogeat-Triboulot et al., 2007; Hagedorn et al., 2016; Iovieno et al., 2016; Volkmann et al., 2016), recovery times of more than 1 year have been observed for more integrating processes such as gross primary productivity (Schwalm et al., 2017), tree growth (Huang et al., 2018), or water-use efficiency (Monserud Atkin Martin-StPaul et al., 2013; Felsmann et al., 2017). While such approaches are important to understand general mechanisms of acclimation, they do not fully reflect natural conditions, where stress (e.g. drought) events of a finite duration occur. It is not clear if such event-based stressors convey longer-term acclimation after the stress has ceased. Here, the subsequent recovery process that constitutes an important part of the tree’s overall drought response might be involved in acclimation to future stress periods. In general, it is unclear whether the extreme droughts that occurred in 2003 and 2018 in Central Europe induced long-term acclimation, and increased resilience of trees during subsequent stress periods. Such a process is referred to as ecological (stress) memory (Walter et al., 2011) and has been defined as a mechanism or different sets of mechanisms that let a plant react to direct environmental drivers not only determined by their genetic capacity but also by their experience of antecedent (stress) conditions (cf. Ogle et al., 2015; Gessler et al., 2017). The process through which the phenotype of an organism gets adjusted for an improved response to future stress, and which thus induces stress memory, has been named priming (Hilker Eilmann et al., 2006, 2009; Arend Fonti et al., 2013), which are thought to be less prone to hydraulic failure (Guet et al., 2015) even though also higher hydraulic conductance together with higher cavitation vulnerability upon drought has been observed (Guérin et al., 2020). As, however, the formation of smaller xylem conduits is negatively correlated with tree-ring width, growth limitations after a drought are not necessarily a sign of continuing vulnerability, but on the contrary would serve as a structural adjustment of the hydraulic system to increase drought tolerance. This would be in line with observations showing that slow-growing trees are less prone to hydraulic failure and mortality than fast-growing trees (Jansen et al., 2013; Gessler et al., 2018; Büntgen et al., 2019). Only recently, it has been shown that drought-induced growth reductions and wood anatomical adjustments come along with improved resistance to recurrent drought and less severe limitations of water and carbon balances (Tomasella et al., 2019). In that sense, negative growth legacies could be considered as positive acclimation responses with respect to future drought events, thus optimizing a tree's survival in the long-term. Such a strategy in the years after drought would reduce short-term performance to optimize long-term survival as suggested by Galiano et al. (2017). In their work with seedlings, these authors showed that after drought release, prioritized incorporation of new assimilates into storage pools occurred rather than investment in growth even though other constraints on growth such as reduced hydraulic functions cannot be ruled out. Moreover, after the end of a drought period, preferential transport of new assimilates to the roots for the re-establishment of root functioning and growth at the expense of the supply of aboveground tissues was observed (Hagedorn et al., 2016). Strategies of increasing carbohydrate storage (O’Brien et al., 2014) and prioritization of root compared to shoot growth (López et al., 2009) are known to increase drought resistance in seedlings and can, thus, be also seen as an acclimation induced by an antecedent and no longer present extreme event. In contrast to slow recovery and persisting growth depression, compensation of functioning after a drought (recovery trajectory with compensatory gain of function in Fig. 1), that is, the increase of a function above the initial pre-drought value, has also been observed but has received little attention in the stress-ecological studies. Still, the phenomenon of compensatory growth is well established in agriculture, and while mainly applied to the field of plant–herbivore interaction (Orcutt Anderegg et al., 2015; Huang et al., 2018). On the other hand, it has also be shown to be independent of changes in competition and is assumed to be an inherent physiological mechanism (Spieß et al., 2012; Arend et al., 2016; Trugman et al., 2018), as for example when the root and stem hydraulic system is overbuilt for the residual leaf area (Vanderklein Menezes-Silva et al., 2017), which might be responsible for longer-term responses. Molecular stress memory is assumed to be conveyed by sustained alterations in levels of key signaling metabolites or transcription factors or by DNA methylation and histone modification (Crisp et al., 2016). Only recently, two memory gene candidates have been identified, controlling in poplar trees physiological processes during drought stress, after recovery and during recurrent drought (Georgii et al., 2019). During stress recovery, RNA metabolism, posttranscriptional gene silencing, and RNA-directed DNA methylation might be able to reset epigenetic and transcriptional modifications thus altering the memory over time making the plants ‘forgetful’ (Crisp et al., 2016). Based on the earlier-mentioned considerations, we suggest that the current framework of recovery needs an amendment to include the acclimation potential of post-drought trajectories of functions. While we fully agree that the perturbation of a system is determined by the impact of an event and the recovery rate (Ingrisch one could test whether such compensation makes individuals more susceptible to new stress events during that overshooting period. On the ecosystem level, Mitchell et al. (2016) give an example of a second drought event that occurs before the system has fully recovered from a first event; the second event increased the drought impact on stand gross primary productivity, thus supporting the scenario in Fig. 1(c). However, we need more long-term, large-scale datasets covering different forest ecosystems and tree species to quantitatively explore whether slow recovery can convey acclimation or not. Individual trees of the same species in the same stand can have different strategies concerning hydraulic safety (Hentschel et al., 2014) and different provenances show different resilience towards drought events (Arend et al., 2011; Jansen et al., 2013). For (2) tree-ring series could be investigated for individual- or provenance-specific differences in recovery dynamics and screened for drought events that occur when part of the individuals or provenances have recovered while others have not yet, as shown in Fig. 1(b,c). These ideas suggest a change in perspective as we evaluate drought responses. We suggest that some deviations from a steady-state may in fact be acclimative, reflecting the use of environmental information from stress events to adopt more conservative (or less conservative) functional traits, and to retain those traits for some finite period. Although the thresholds for response and the timing of the responses are likely to vary, recognition of their ability to enhance or avoid risk provides a means of considering the dynamics of risk management following a stress event. All authors have contributed to the development of the conceptual framework and to the writing of the manuscript.