•Severe drought may increase physiological stress on long-lived woody vegetation, occasionally leading to mortality of overstory trees. Little is known about the factors determining tree survival and subsequent recovery after drought.
•We used structural equation modeling to analyse the recovery of Scots pine (Pinus sylvestris) trees 4 yr after an extreme drought episode occurred in 2004–2005 in north-east Spain. Measured variables included the amount of green foliage, carbon reserves in the stem, mistletoe (Viscum album) infection, needle physiological performance and stem radial growth before, during and after the drought event.
•The amount of green leaves and the levels of carbon reserves were related to the impact of drought on radial growth, and mutually correlated. However, our most likely path model indicated that current depletion of carbon reserves was a result of reduced photosynthetic tissue. This relationship potentially constitutes a feedback limiting tree recovery. In addition, mistletoe infection reduced leaf nitrogen content, negatively affecting growth. Finally, successive surveys in 2009–2010 showed a direct association between carbon reserves depletion and drought-induced mortality.
•Severe drought events may induce long-term physiological disorders associated with canopy defoliation and depletion of carbon reserves, leading to prolonged recovery of surviving individuals and, eventually, to delayed tree death.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Episodes of drought-induced tree mortality are emerging as a global phenomenon and have already been reported from a variety of woodland and forest communities in many parts of the world (see review by Allen et al., 2010). Severe drought events may increase physiological stress on long-lived woody vegetation, occasionally leading to rapid defoliation and mortality of overstory trees (Bréda et al., 2006). In the Mediterranean basin, climate change effects interact with the effects of increasingly denser stands because of agricultural land abandonment (Poyatos et al., 2003), artificial afforestation (Martínez-García, 1999) and a decline in logging practices (Linares et al., 2009, 2010). In addition, rear-edge populations are likely to be particularly sensitive to the effects of increased aridity (Hampe & Petit, 2005). This pattern is well exemplified by Scots pine, a widely distributed tree species that has recently suffered episodes of drought-induced mortality in several of its southernmost populations (Martínez-Vilalta & Piñol, 2002; Hódar et al., 2003; Galiano et al., 2010).
Two main physiological mechanisms explaining the eventual drought-induced mortality of trees have been recently formalized by McDowell et al. (2008): hydraulic failure and carbon starvation. According to these authors, the relevance of each mechanism depends on the hydraulic properties of the species and on the intensity and length of the drought conditions. Anisohydric plants maintain relatively high stomatal conductance as soil water potential decreases and are likely to die through hydraulic failure when the drought is particularly intense (Tyree & Sperry, 1989; Davis et al., 2002; Brodribb & Cochard, 2009). By contrast, isohydric plants prevent desiccation by stricter stomatal control, causing a decrease of photosynthetic carbon uptake and plant starvation as a result of continued metabolic demand for carbohydrates and depletion of reserves (Parker & Patton, 1975; Martínez-Vilalta et al., 2002; Bréda et al., 2006). Isohydric plants are most likely to die during prolonged droughts of intermediate intensity. However, our current understanding of the prevalence of these mechanisms is still limited as they have proved difficult to demonstrate under natural conditions, and alternative mechanisms have been recently proposed (McDowell & Sevanto, 2010; Sala et al., 2010).
Mature trees store large amounts of mobile carbon pools that are mostly composed of nonstructural carbohydrates (NSC) and neutral lipids (triacylglycerols). The size of this pool may be considered an indicator of a plant’s ‘fuelling’ status with respect to carbon, because it should reflect any shortage (depletion) or surplus (accumulation) depending on the carbon source–sink balance (photosynthesis vs metabolism, growth and export; Mooney, 1972; Chapin et al., 1990; Körner, 2003). Thus, stored carbon accumulated in any organ or tissue during periods of carbon surplus can be later mobilized to sinks where carbon demand temporarily exceeds available carbon (Chapin et al., 1990). It is well known that this pool may become smaller when sources are removed, for example when trees are experimentally defoliated or pruned (Langström et al., 1990; Vanderklein & Reich, 1999; Kosola et al., 2001; Li et al., 2002) or may become larger when sinks are removed, for example when trees are experimentally debudded (Chapin & Wardlaw, 1988; Li et al., 2002). Drought normally impairs carbon assimilation through stomatal closure. However, drought also reduces carbon sinks, in particular, structural growth, so the relevant question is whether (or when) carbon sources are constrained more by drought than by carbon sinks. In fact, the carbon reserves stored in woody plants have been shown to remain stable, increase or decrease in response to drought, depending on the species and their ontogenetic stage (Sala et al., 2010). In any case, although there is some indirect evidence consistent with the carbon starvation hypothesis (Adams et al., 2009; Breshears et al., 2009), direct evidence is completely lacking to date (Sala, 2009).
Scots pine is a classical example of a species exhibiting isohydric stomatal control (Irvine et al., 1998). In a previous study of a Scots pine forest affected by a drought episode, occurred in 2004 and 2005, it was found that mortality and defoliation rates were associated with the local level of drought stress experienced at different altitudes. In addition, stand density, soil properties and mistletoe infection acted as long-term predisposing factors during the drought episode (Galiano et al., 2010). In the present study, we analysed the recovery of the previous Scots pine forest 4 yr after the drought event. The main objectives were to: (1) establish whether the amount of carbon reserves stored in stems were related to the recovery of trees, and, more specifically, to the amount of green leaves; (2) test alternative hypotheses relating the impact of drought on trees, the amount of green leaves and the amount of carbon reserves stored in stems (see the bullet points at the end of the paragraph); (3) estimate how the previous relationships may be affected by mistletoe infection and by the physiological performance of current needles; and (4) investigate how the amount of carbon reserves may determine the capacity of trees to create new photosynthetic tissue and the survival of trees 1 yr after the sampling. In relation to objective (2) we considered the following three hypotheses (cf. panels in Fig. 1):
•Hypothesis 1: the current amount of green leaves, reduced by the impact of the past drought episode, causes photosynthetic carbon uptake to decline and, consequently, depletion of carbon reserves stored in stems.
•Hypothesis 2: there is a depletion of carbon reserves determined by their use during the past drought episode, proportional to the intensity of drought stress experienced by each tree. This depletion remains several years after the drought episode and reduces the current capacity of trees to create new photosynthetic tissue.
•Hypothesis 3: both the amount of green leaves and the levels of carbon reserves are reduced by the direct impact of the past drought episode. These two variables are mutually constrained and, consequently, they are positively correlated.
Materials and Methods
The study was carried out in a Scots pine (Pinus sylvestris L.) forest located in the Central Pyrenees (Soriguera, Pallars Sobirà, 42°22′43′′ N, 1°6′29′′ E, c. 16 km2) with a stand density of 1071 individuals ha−1 and a basal area of 35.90 m2 ha−1. Scots pine forests in the area are mainly on northern slopes, are distributed at altitudes from 600 to 1500 m above sea level (asl), and most of them have traditionally been under important agro-pastoral use up until the early twentieth century. Although both natural and artificial afforestation have taken place in the area during the twentieth century, the current Scots pine population is at least 150 yr old, exhibits natural regeneration and is well within the natural distribution area of Scots pine (see Galiano et al., 2010). The Scots pine stands studied have not been managed since the 1980s (C. Fañanàs pers. comm., Catalan Forest Service). The shrub layer is predominantly occupied by Buxus sempervirens, Amelanchier ovalis and Lonicera xylosteum. Other species of trees, Quercus humilis, Quercus ilex and Betula pendula, occasionally appear in the understory, mostly at lower (Quercus species) and higher altitudes (B. pendula). Some younger Pinus nigra plantations interrupt the Scots pine forest. A well-established mistletoe (Viscum album L.) population, with individuals up to 30 yr old, infects most Scots pine trees in the area. Soils are calcareous, fairly rocky (44% weight of stones > 1 cm diameter) and belong predominantly to the clayey-loam texture class.
The climate of the region is characterized by an annual mean temperature of 9.6°C and an annual mean rainfall c. 643 mm (climate data for the period 1951–1999 from the Climatic Digital Atlas of Catalonia (CDAC) (Pons, 1996; Ninyerola et al., 2000), corresponding to the temperate oceanic sub-Mediterranean bioclimatic region (Worldwide Bioclimatic Classification System, 1996–2009). In years 2004 and 2005, the Iberian Peninsula experienced a severe drought episode preceded by several dry periods during the last decades (European Environment Agency (EEA), 2008; see the Supporting Information, Fig. S1). Damage on Scots pines became visually evident in summer 2005 (C. Fañanàs, pers. comm., Catalan Forest Service). This observation was supported by dendro-ecological analyses that showed lower growth and higher mortality in 2005 and subsequent years, although minor mortality episodes were also detected before 2005 (A. Heres, unpublished). At the time of sampling, average standing tree mortality was 14.3%, corresponding to 9.13% of total basal area, whereas defoliation (weighted by the basal area of each tree) was 45.18% (Galiano et al., 2010). Climate data showed the absence of new severe drought events after the drought episode occurred in 2004–2005 (Fig. S1).
Field sampling methods
In August 2009, 42 Scots pine individuals were sampled. To distinguish between the effect of green needle amount and mistletoe parasitism on NSC reserves stored in stem sapwood, we selected trees with contrasted amounts of green needles combined with different mistletoe infection intensities. We refer to the amount of green leaves instead of defoliation because our measure includes drought-induced defoliation and any recovery occurred afterwards. Tree selection used the following criteria in order to minimize unwanted variation: diameter at breast height (DBH) from 15 to 40 cm; no signs of recent disturbance or management; distance between trees > 15 m; altitude from 900 to 1000 m asl; slopes from 10° to 40° and North aspect. For all individual trees, we visually estimated the percentage of green needles relative to the number in a healthy canopy of a similar sized tree in the study area. We minimized error by having the same person always observing the crown from the same position close to the stem and making the estimate separately for four different sections of the crown. Preliminary trials indicated that the average discrepancy between two trained, independent observers was < 10%. Furthermore, DBH and a detailed characterization of all mistletoe plants based on age were recorded for all individual trees. The mistletoe characterization included six categories (1, no bifurcations; 2, 1–2 nodes; 3, 3–5 nodes; 4, 6–10 nodes; 5, 11–20 nodes; 6, > 21 nodes), accounting for the dichasial branching pattern of annual growth (Zuber, 2004). A mistletoe index was calculated by adding the category of all mistletoe plants on each sampled tree. In addition, in July 2010, state (alive vs dead) and visual estimation of the percentage of green needles (%) was recorded again for all individual trees. We assessed the difference between the amount of green leaves in years 2010 and 2009. Larger positive differences imply greater recovery in the 1-yr-long time lapse.
Using a hand increment borer (5 mm diameter; Suunto, Vantaa, Finland), two stem cores were sampled to the pith from each tree at 1.35 m above the ground from the two sides perpendicular to the slope. One core was used to measure NSC reserves stored in stem sapwood (see the ‘Nonstructural carbohydrates’ section), and the other core was used to quantify growth (see the ‘Growth measurements’ section). We also collected 20–30 exposed and apparently healthy current-year needles from two midcanopy branches to analyse the carbon isotope composition and the nitrogen content (see the ‘Foliar carbon isotope and nitrogen content’ section).
Cores for quantifying growth were placed in wooden supports and taken to the laboratory for analysis. In the laboratory, all cores were air dried and sanded using progressively finer sandpaper until growth rings could be easily recognized. The last 30 ring widths, that is, years from 1980 to 2009, were measured to a precision of 0.01 mm using the windendro computer software (windendro 2004c, Régent Instruments Inc., Quebec, Canada). The COFECHA cross-dating program (Holmes, 1983) was used to help detect the occurrence of missing and false rings. Two cores did not cross-date well and were excluded from further analyses. Basal area increment (BAI) was used to characterize tree growth. The BAI was calculated from ring growth according to:
(R, the radius of the tree; t, year of tree ring formation). To characterize the intensity of drought stress experienced by trees during the drought episode, we computed the ratio between the mean annual growth from 1980 to 2003 and the mean annual growth from 2004 to 2007. Growth from 2004 to 2007 was considered to be potentially affected by drought, as drought effects have been shown to last for several years in trees, including Scots pines (Becker, 1989; Bréda et al., 2006). Larger values of this ratio indicate higher impact of drought on trees. We also computed the mean annual growth for years 2008 and 2009 to characterize the current growth rate.
The stem sapwood portion of one core per tree (visually determined) was separated for measuring the NSC reserves. The segments were wrapped in plastic straws and stored in a cooler over ice until sample processing in the laboratory on the same day. All sapwood samples were microwaved for 90 s to stop enzymatic activity, oven-dried for 72 h at 65°C and ground to fine powder. Nonstructural carbohydrates were defined as free sugars (glucose and fructose), low molecular weight sugars (free sugars and sucrose) plus starch, and were analysed following the procedures described by Hoch et al. (2002), with some minor modifications. Approximately 12–14 mg of sapwood powder was extracted with 1.6 ml distilled water at 100°C for 60 min. After centrifugation, an aliquot of the extract was used for the determination of low molecular weight sugars after enzymatic conversion of fructose and sucrose into glucose. Another aliquot was incubated with an amyloglucosidase from Aspergillus niger at 50°C overnight, to break down all NSC (starch included) to glucose. The concentration of free glucose was determined photometrically in a 96-well microplate reader (Sunrise Basic Tecan, Männedorf, Switzerland) after enzymatic conversion of glucose to gluconate-6-phosphate. Starch was calculated as NSC minus low molecular weight sugars. All NSC values are expressed as percent dry matter.
Foliar carbon isotope and nitrogen content
Current year needles were oven-dried for 72 h at 65°C and ground to fine powder. Ground samples were analysed for carbon stable isotope composition and nitrogen content at the Cornell Isotope Laboratory (COIL) at Cornell University, using a Thermo Delta V isotope ratio mass spectrometer (IRMS) interfaced to a NC2500 elemental analyser. The carbon stable isotope composition was expressed in delta notation: δ13C (‰) = (Rsample/Rstandard − 1) × 1000, where Rsample is the 13C : 12C ratio of the sample and Rstandard is the 13C : 12C ratio of the international Vienna Pee Dee Belemnite carbon standard. The accuracy of the δ13C measurements was 0.05 ‰. Nitrogen concentrations are expressed as per cent dry matter.
We used structural equation modeling (SEM) to analyse the complex relationships between the set of observed variables representing the impact of drought on trees, the current growth rate, the amount of green leaves and the carbon reserves stored in stems as well as how the previous relationships may be affected by mistletoe infection and by the physiological performance of current needles (leaf nitrogen (N) content and leaf δ13C; see Fig. 1). The SEM was performed by using amos 18 (Arbuckle, 2009). Some variables were transformed to achieve normality. Because some variables remained nonnormal after transformation, the generalized least squares (GLS) method was used to estimate the value of the unknown parameters of the model, as recommended in cases of lack of multivariate normality and small sample sizes (Iriondo et al., 2003). Model selection was performed using the Bentler’s comparative fit index (CFI), the Akaike’s information criterion (AIC) and the Bayesian information criterion (BIC). Some additional analyses were carried out with R version 2.11.1. (R Development Core Team, 2010) using the Mann–Whitney–Wilcoxon or Student’s t and Kruskal–Wallis tests to compare means between two or more than two groups, respectively.
The relationships displayed in the structural equation models reflect the main hypotheses of this study (cf. Introduction; Fig. 1). The ratio between the mean annual growth from 1980 to 2003 and the mean annual growth from 2004 to 2007 is an estimator of the intensity of drought stress experienced by trees during the drought episode. This ratio is expected to influence the current growth rate (mean growth in years 2008 and 2009). We expected that the amount of green leaves also explains the current growth rate. Mistletoe infection has been shown to be directly connected to current water use efficiency (leaf δ13C), stem sapwood NSC concentrations, leaf N content and current growth rate (Schulze & Ehleringer, 1984; Ehleringer et al., 1986; Ehleringer & Marshall, 1995; Escher et al., 2004). Leaf N content was directly connected to leaf δ13C and stem sapwood NSC concentrations on the basis that nitrogen concentration is strongly associated with Rubisco concentration in leaves and, consequently, with carboxylation efficiency (Field & Mooney, 1986). As leaf δ13C was used as an integrated measure of the ratio photosynthetic capacity to stomatal conductance (water-use efficiency; Farquhar et al., 1989), the direct relationship between leaf N content and NSC concentrations allowed us to distinguish the effects of leaf δ13C on carbon reserves via photosynthetic efficiency or via stomatal conductance. DBH was not introduced in the models because preliminary analyses showed no relationship between DBH and stem NSC concentrations, possibly because of the relatively narrow range of DBH considered in this study.
Trees with < 50% of green leaves 4 yr after the drought episode (2009) contained 34% less carbon reserves stored in stems than trees with > 50% of green leaves (Mann–Whitney–Wilcoxon W =118, P <0.05; Fig. 2a). In 2010, 47% of trees had increased the amount of green leaves with respect to 2009, whereas 28% and 24% had maintained and decreased the amount of green leaves, respectively. Similarly, trees with < 50% of green leaves in 2009 presented 70% lower growth rates in years 2008 and 2009 with respect to those trees with > 50% of green leaves (Student’s t = −4.321, P <0.001; Fig. 2b). Interestingly, amounts of carbon reserves analysed in 2009 were positively associated to the recovery of green leaves 1 yr later (2010), as shown by significant differences between trees that increased and those that decreased the amounts of leaves (Kruskal–Wallis χ2 = 7.681, P <0.01; see Fig. 3a). It is also remarkable that the four trees that died between 2009 and 2010 had much lower stem NSC concentrations in 2009 than those that survived (Mann–Whitney-Wilcoxon W =3, P <0.001; Fig. 3b). Sapwood area of trees was not related to NSC concentrations (r =0.065, P =0.685) and, thus, differences in NSC concentrations between defoliated/undefoliated trees and between dead/living trees were not caused by differences in sapwood area.
All SEM models present the following significant relationships (Fig. 4, Table S1): the current growth rate (mean growth in years 2008 and 2009) was negatively influenced by the direct impact of the past drought and the reduction of photosynthetic tissue, and by the indirect impact of mistletoe infection through the reduction of leaf N content; leaf N content and leaf δ13C were positively related (Fig. S2). Despite this positive relationship, the fact that the current water-use efficiency (estimated from leaf δ13C) was negatively related to NSC concentrations, and there was no relationship between leaf N content and NSC, suggests an important role of stomatal closure in determining stem carbon reserves.
In addition to the previous relationships, the SEM model that most adequately describes our data was model (a), consistent with hypothesis H1 (χ2 = 7.381, P =0.390, CFI = 0.985, AIC = 49.381, BIC = 85.872; Fig. 4a), explaining 45% of the total variability of NSC concentrations. According to this model, the current amount of green leaves was reduced by the impact of the past drought episode and determined the current amount of carbon reserves. Thus, the impact of the drought on trees has an indirect effect on NSC concentrations, mediated by the amount of green leaves (Table S1a). However, the SEM model corresponding to hypothesis 3 also provided an acceptable fit (P >0.05, CFI > 0.9, ΔAIC ≈ 1 and ΔBIC ≈ 3 compared with model (a); see Fig. 4c). Thus, we cannot discard the possibility that carbon reserves were also constrained to some extent by the direct impact of the past drought episode and that their relationship with the amount of green leaves may be reciprocal. Model (b), consistent with hypothesis 2, was clearly worse than any of the two other models (CFI < 0.9, ΔAIC ≈ ΔBIC ≈ 3 units; see Fig. 4b).
Scots pine trees in the studied area are still recovering from a drought episode that occurred in 2004 and 2005. Carbon reserves stored in stems are related to the recovery of pine trees and, particularly, to the amount of green leaves. Trees that failed to recover > 50% of green leaves contain very low levels of carbon reserves. Indeed, the entire Scots pine forest studied contained very low amounts of carbon reserves compared with other healthy Scots pine forests in southern Europe (Hoch et al., 2003).
Determinants of Scots pine drought recovery
Our SEM results showed that the current amount of green leaves was reduced by the impact of the past drought episode and support the notion that depletion of carbon reserves occurs primarily as a result of reduced photosynthetic carbon uptake, caused both by reductions in leaf area and by stomatal closure. Water scarcity during drought reduces stomatal conductance, leaf area, radial growth and bud production (Tyree et al., 1993; Bréda et al., 2006; Pichler & Oberhuber, 2007). The impairment of buds may limit the capacity of trees to create new photosynthetic tissue and twigs during subsequent years following the drought episode (Power, 1994; Stribley & Ashmore, 2002). However, the reduction of leaf area is a well known mechanism of acclimatization to water shortage in Scots pine (Martínez-Vilalta et al., 2009), and has been shown to induce growth reductions by limiting carbon assimilation in other pine species (e.g. Borghetti et al., 1998). It should be noted that our design does not allow separating the immediate impact of drought on leaf area (via needle shedding) from the delayed impact arising from, for example, the effects of drought on subsequent bud production.
Our results could not exclude the possibility that levels of carbon reserves were also constrained to some extent by the direct impact of the past drought episode. In agreement to the carbon source–sink balance theory (Mooney, 1972; Chapin et al., 1990; Körner, 2003), the depletion of carbon reserves in our case suggests that carbon assimilation (source activities), mainly impaired by leaf-shedding and stomatal closure, was more constrained than carbon sink activities as a result of the drought episode. The reciprocal relationship between the amount green needles and carbon reserves may constitute a feed-back that limits the recovery of trees and explains the delayed effects of drought on tree growth and survival, which have been reported elsewhere (e.g. Manion, 1991; Bréda et al., 2006; Bigler et al., 2007).
Despite the fact that several studies have reported that mistletoes extract carbohydrates from their hosts (Watson, 2001; Escher et al., 2004), we did not observe any effect of mistletoe infection on the carbon reserves stored in stems. Instead, mistletoe absorption of nutrients noticeably reduced leaf nitrogen content, probably causing a reduction in the photosynthetic rate of pine needles (Meinzer et al., 2004). Mistletoe infection, thus, may reduce growth by limiting carbon assimilation rather than by reducing long-term carbon reserves or by changing the water balance of the tree. Other studies have also documented growth reductions in hosts caused by mistletoe infection (Dobbertin, 2005; Rigling et al., 2010). Although our sampling design (cf. ‘Field sampling methods’ section) did not allow us to test the well-known relationship between mistletoe infection and needle loss (Dobbertin & Rigling, 2006; Galiano et al., 2010; Rigling et al., 2010), it is likely that mistletoe infection had also an indirect impact on tree growth, mediated by the amount of green leaves.
Implications for the mechanisms of drought-induced mortality
Repeated surveys in 2009 and 2010 allowed us to detect four trees that died in this 1-yr-long time-lapse. These trees contained extremely low carbon reserves in 2009. Consistent with this, surviving trees that exhibited less favorable canopy recovery from 2009 to 2010 also had lower of carbon reserves in 2009. These results show for the first time a direct association between carbon reserve depletion and drought-induced mortality in trees, and are consistent with the carbon starvation hypothesis (cf. McDowell et al., 2008). Previous evidence in favor of the carbon starvation hypothesis (Adams et al., 2009; Breshears et al., 2009) was indirect and consistent with alternative interpretations (Leuzinger et al., 2009; Sala, 2009). Adams et al. (2009), for example, found that higher temperatures in a period of protracted water stress increase respiration rates and accelerate mortality. Our study contributes to the current debate on the role of carbon reserves depletion on tree mortality (McDowell & Sevanto, 2010; Sala et al., 2010) by explicitly measuring carbon reserves stored in stems and showing that an isohydric tree such as Scots pine may actually deplete their carbon reserves and starve to death.
Although stomatal closure is normally described as the foremost mechanism that limits carbon uptake in the context of the carbon starvation hypothesis (Breshears et al., 2009), in our case the reduction in photosynthetic area owing to drought-induced leaf-shedding seems to be at least as important, as already suggested by McDowell et al. (2008). In any case, our observations on the mechanism of mortality remain somewhat preliminary, as our study was not designed to identify the mechanisms of drought-induced mortality in trees, and we cannot exclude that other mechanisms, different from the carbon starvation hypothesis, might have been in operation. More research is needed to establish the causal relationships linking the dynamics of carbohydrate reserves and mortality, and to determine whether our results are generalizable to other species or situations.
In conclusion, our study illustrates how a severe drought episode can produce a progressive loss of forest resilience by depleting the ability of surviving plants to grow and survive future stressful events (Lloret et al., 2004), despite the release in competition that normally accompanies episodes of drought-induced mortality (Martínez-Vilalta & Piñol, 2002; Bigler et al., 2006). Our results add to the evidence provided by earlier studies showing that long-term physiological disorders induced by a drought event occasionally lead to prolonged recovery phases of the surviving trees and/or eventual death (Bréda et al., 2006), and points to carbohydrate reserves as a key determinant of both tree survival and recovery.
We thank the Catalan Forest Service, and especially Carles Fañanàs Aguilera, for facilitating our field work and for their generous comments. We appreciate help from the undergraduate students (Josep Barba and Mar Unzeta) that were involved in this study. We are also indebted to the Anna Sala’s plant physiology laboratory (The University of Montana, Missoula, USA) for teaching the analysis technique for carbohydrates stored in wood, and to the Department of Genetics (The Autonomous University of Barcelona, Spain) for allowing us to use their laboratories to perform the analyses. This study was supported by the Spanish Ministry of Education and Sciences via competitive projects CGL2006-01293, CGL2007-60120, CSD2008-0004 and CGL2009-08101, and by the Government of Catalonia via AGAUR grant 2009 SGR 247. LG was supported by an FPI scholarship from the Spanish Ministry of Education and Sciences.