Extreme drought triggers parallel shifts in wood anatomical and physiological traits in upper treeline of the Mediterranean Andes

Abstract

Background

Treeline ecotones of Mediterranean ecoregions have been affected by the increasing intensity and severity of droughts. Even though the effect of droughts on forest dynamics has been widely documented, knowledge is relatively scarce of how extreme climate episodes affect the hydraulic structure and, therefore, the physiology of woody plants. The Mediterranean Andes have experienced an uninterrupted period of drought since 2010, including an extremely dry year in 2019 with approximately 80% rainfall deficit. Here, we investigated shifts in wood anatomical and physiological traits of Kageneckia angustifolia, an endemic treeline species, in response to this drought period.

Methods

We evaluated the xylem plasticity of three K. angustifolia populations across their natural distribution (31–35° SL) based on anatomical (vessel structure and distribution) and physiological (intrinsic water-use efficiency) variables in the tree rings. We focused on the period 2000–2020 that corresponds to before the megadrought (2000–2007), (ii) megadrought (2008–2018) and (iii) hyperdrought (2019–2020). The variables were annualized and analyzed by linear mixed-effects models.

Results

Our results provide insights to the anatomical and physiological mechanisms underlying the resilience of treeline forests to persistent droughts in central Chile. We found that the extreme drought in 2019–2020 triggered shifts in vessel size and frequency that increased hydraulic safety. These significant shifts in vessel traits occurred in parallel with a decrease in pit aperture area and an increase in water-use efficiency, further increasing the resilience of K. angustifolia to extreme drought stress.

Conclusions

Our results revealed coordinated shifts in vessel size and frequency and water-use efficiency in response to the megadrought, thereby reducing vulnerability to hydraulic failure. The apparent resilience of K. angustifolia to extreme droughts suggests that this adaptation to drought stress may increase its ability to tolerate novel climatic conditions of treeline environments of the Mediterranean Andes, although it is not clear whether these adaptations will be sufficient to persist in scenarios that predict intensification of climate stress. Finally, our results provide empirical evidence that integrating wood anatomical and physiological traits facilitates the understanding of resilience mechanisms that treeline forests develop in the face of increasing drought stress.

Introduction

Treeline ecotones in upper mountain ecosystems are affected by increasing intensity and severity of droughts and global warming, especially in Mediterranean ecoregions (Garreaud et al. 2020; Cook et al. 2022; González-Reyes et al. 2023). Even though the effect of droughts on montane ecosystems has been widely documented (Granda et al. 2018; Colangelo et al. 2021; Matskovsky et al. 2021; Gazol et al. 2022), knowledge is relatively scarce of how extreme climate episodes affect the hydraulic structure and, therefore, the physiology of woody plants. Therefore, it is relevant to understand how extreme drought phenomena affect the level of resilience of forests in contexts of climate crisis.

Due to the rapid climate changes that Mediterranean ecosystems are experiencing and their future predictions of intensification (Seager et al. 2019), upper treelines constitute a model system to evaluate resilience capacities through modifications of xylem anatomical traits and physiological processes (Pepin et al. 2015; McDowell et al. 2022). Drought-driven declines in ring growth are directly influenced by changes in cellular structure of wood, whose xylem structure captures inter-annual growth patterns (Fonti et al. 2010). The xylem anatomical functional traits provide deeper information than just the analysis of the variation in ring widths (Beeckman 2016; Gennaretti et al. 2022). The size and frequency of the vessels (Garcia-Gonzalez et al. 2016), as well as the structure of their interconnections (Mrad et al. 2018) are indicators of the hydraulic conductivity and carbon allocation capacities, both at the level of interannual and interspecific variations (Cuny et al. 2015; Zwieniecki and Secchi 2015). In parallel with temporal changes in vessel structure and distribution, intrinsic water-use efficiency (iWUE) relates atmospheric carbon uptake through photosynthesis to water loss through transpiration, reflecting physiological responses to water stress over time (Lavergne et al. 2020; Cherubini et al. 2021). Therefore, by integrating anatomical traits of wood linked to safety and hydraulic efficiency with physiological traits of the xylem (Pellizzari et al. 2016), it facilitates the understanding of the resilience mechanisms that treeline forests develop in the face of extreme climatic events, which is the primary purpose of this work (Martínez‐Vilalta et al. 2023).

In central Chile, treeline ecosystems are susceptible to droughts, with incidence in growth, reproduction and survival dynamics (Cavieres et al. 2021; Matskovsky et al. 2021; Tovar et al. 2022). This region, the only one in South America with Mediterranean forests, has been experiencing increasingly severe and persistent droughts. Since 2010, precipitation has consistently remained below historical averages (Garreaud et al. 2020), including years of extreme drought, such as 2019, where the precipitation deficit was approximately 80% (Alvarez-Garreton et al. 2021). Forest ecosystems throughout the Mediterranean region of central Chile have been strongly affected by these persistent droughts, with significant declines in growth (Venegas-González et al. 2023), weakening of plant–pollinator interactions (Arroyo et al. 2020), and more extensive areas of forests with symptoms of ‘browning’ (Miranda et al. 2020). However, the anatomical and physiological mechanisms that underpin tree responses to persistent and increasingly severe droughts in central Chile and other Mediterranean regions are still poorly understood.

In this study, we examined shifts in wood anatomical and physiological traits of Kageneckia angustifolia, an endemic tree species distributed across the treeline in the Mediterranean Andes, in response to persistent droughts for period 2010–2020. Combining a wide range of anatomical and physiological traits of K. angustifolia wood, the following hypotheses were evaluated: the hydraulic system of stems is modified by the incidence of persistent droughts, by forming a greater number of vessels but with a smaller diameter (e.g.,García-Cervigón et al. 2018; Castagneri et al. 2020) and reducing the inter-vessel pit area (e.g., Lens et al. 2011); and water use efficiency (iWUE) increases as the water stress gradient intensifies (e.g., Lévesque et al. 2014; Peñuelas et al. 2011). To explore these hypotheses, we: (i) built tree-ring chronologies to exactly determine the calendar dating of the ring widths and analyze the tree growth rate, (ii) built vessel chronologies for the period 2000–2020, and (iii) evaluated temporal variation in vessel features and iWUE patterns in response to droughts of increasing severity.

Materials and methods

Study site and climate

We selected three sites of K. angustifolia across the Mediterranean Andes of central Chile (31–35° SL), to cover its entire natural distribution. This species, despite having a wide distribution ~ 400 km latitude, is limited to small forest patches at the treelines of the Mediterranean Andes between 1500 and 2300 m a.s.l. (Piper et al. 2006). In each site, we took samples of a K. angustifolia population, which has the following features: (i) northern population, located in Tulahuen (TUL, − 30.99°S, − 70.64°W, 1900 m a.s.l.), with an average annual precipitation of 100 mm yr ^–1 and temperature of 16.5 °C; (ii) central population, located in Baños Morales (BM, − 33.8° SL, − 70.1° WL, 1800 m a.s.l.), with average rainfall of 350 mm yr^–1 and temperature of 14.6 °C; (iii) southern population, located in Teno (TEN, − 35.1° SL, − 70.5° WL, 1,600 m a.s.l.), has an average annual rainfall of 550 mm yr^–1 and a temperature of 12.8 °C (Fig. 1A). The wood anatomy of K. angustifolia populations demonstrates the high dendrochronological potential of the species, with semi-porous rings and increasing fiber wall thickness in latewood (Fig. 1B).

Fig. 1

A Map of three study sites located along gradients of temperature and precipitation in central Chile. Wood sections of K. angustifolia, B transversal (10x, bar = 500 µm), C tangential (40x, bar = 100 µm µm), and D radial (40x, bar = 100 µm). E Vessel element (2,200x, bar = 100 µm) showing the simple perforation plate and alternate inter-vessel pits

Sampling and processing of the ring-width chronologies

We sampled cores from the main stem and lateral branch discs closest to the base from 15 to 32 individuals of Kageneckia angustifolia at each study site (Table 1). The choice of collecting cores or branches of each tree was the following criteria: (i) when the main base was larger than the lateral branches, collecting with the increment borer was preferred, and (ii) when lateral branches were of a similar size to the main stem, we collected disks. In addition, we avoided selecting trees with damage, presence of pathogens, or reaction wood. Collections were made between February 2021 and April 2022. Wood samples were processed according to standard procedures in dendrochronology (Stokes and Smiley 1968). Due to the similarity of tree heights within and between the study sites (Table 1), it was not necessary to vary the sampling point, as recommended by Carrer et al. (2015). Tree ring width measurements (RW) were performed using a Velmex measuring system with a resolution of 0.001 mm. The quality of the cross-dating was checked with COFECHA software (Grissino-Mayer 2001; Holmes et al. 1986). After building the ring width chronologies, we focused on the last two decades (2000–2020) to evaluate the effects of the intense drought that occurred in this period on wood anatomy and physiology.

Table 1 Descriptive statistics for ring width chronologies and wood vessel traits across three populations of K. angustifolia (ring width and vessels features) in the Mediterranean Andes of central Chile

Drought periods

We divided each sample into three periods, which differ in drought intensity as identified by Garreaud et al. (2020) and Alvarez-Garreton et al. (2021): (i) reference period, before the megadrought (BMd: 2000–2007); (ii) the megadrought period (Md: 2010–2018), and (iii) the hyperdrought period (Hd: 2019–2020) (Fig. 2). We restricted the reference period (BMd) to 2000–2007, because Cook et al. (2022) showed a significant decline in soil moisture starting in 2008.

Fig. 2

Climatic variability from 1979 to 2020 in central Chile: A z values of maximum temperature during the austral summer (December to February); B z values of precipitation in winter–spring (May–November). In both climatic series, mean values (± SE) across the drought periods are showed. C Annual Palmer Drought Severity Index for the study region (30–35°S, 70.5–71.0°W) for the period 2000–2020, with the three drought periods highlighted (BMd before megadrought, Md megadrought, Hd hyperdrought). We compared the three periods using percentiles (Px) from the period 1901–2020. In addition, we show the spatial variation of the PDSI in exceptional years using data from the South American Drought Atlas SADA (Morales et al. 2020)

We further validated the selection of these periods by evaluating summer mean maximum temperature (Tmax), winter–spring total precipitation (Pp), and the Palmer Drought Severity Index (PDSI) (Additional file 1: Fig. S1). To evaluate temporal variation in Tmax and Pp, we used standardized data from four meteorological stations close to the three study sites, considering the period between 1979 and 2020. We fitted a generalized linear model (GLM) to evaluate the effect of the drought period on Tmax and Pp using the R function ‘glm’ (Bates et al. 2015). Comparisons among means for significant effects were made using Fisher’s LSD tests (p < 0.05). We found statistically significant trends for both variables (Fig. 2). PDSI was analyzed using data from the South American Drought Atlas SADA (Morales et al. 2020). Based on percentiles (Px) of PDSI for the period 1901–2020 for the study region (30.0–35.0° SL, 70.5–71.0° WL), we considered the baseline period of 2000–2007, because its drought severity was markedly lower than the two subsequent periods (Fig. 2C).

Wood anatomical traits

After building ring-width chronologies, we selected a subset of 45 samples (15 per site) to evaluate wood anatomical traits (in the case of northern site we included all trees sampled). We choose trees without defects or visible damage and without cross-dating problems. Wood samples from each individual were softened in a water/glycerin solution (4:1) and sectioned along the cross-sectional surface using a sliding microtome (Quintilhan et al. 2021). Sections were cleared with sodium hypochlorite (20%), dehydrated in an alcohol series (30–50%) and then stained with safranin (1%) (Santini Jr et al. 2019). We photographed samples with a digital camera coupled to an optical microscope (Digital Microscope Motic, software 2.0) and a magnification of 25x. We analyzed wood anatomical traits following the International Association of Wood Anatomists (IAWA) standards. Measurements were performed by the ImageJ program (version 1.53a for Windows). We focused on earlywood vessels, because they provide better information on intra-annual variations of hydraulic traits (Fonti et al. 2010). Vessels larger than 40 µm diameter were measured in an area of 1.35 mm². Vessel frequency per mm² (VF) and mean vessel diameter (VD), were measured for each ring. The vulnerability index as the risk of hydraulic failure due to vessel characteristics (VI) (Carlquist 1977) was calculated with the following equation:

$${\text{VI}}=\frac{{\text{VD}}}{{\text{VF}}}$$

(1)

Potential hydraulic conductivity (K_p), i.e., potential stem conductivity/ratio of stem conductivity, c.f. Hagen–Poiseuille law (Sterck et al. 2008), was also calculated with the following equation:

$${K}_{{\text{p}}}=\pi \rho w/128\eta \times {\text{VF}}\times {{\text{DH}}}^{4}$$

(2)

where K_p is the potential specific stem conductivity (kg m MPa⁻¹ s⁻¹), $\eta$ is the viscosity of water at 20 °C (1.002 × 10^–3 Pa s at 20 °C), ρw is the density of water at 20 °C (998.2 kg m⁻³ at 20 °C), VF is the vessel frequency and D_h is the hydraulically weighted vessel diameter (µm). DH was calculated with the following equation:

$${D}_{{\text{h}}}=\frac{{\sum }_{n=1}^{N}{{d}^{5}}_{n}}{{\sum }_{n=1}^{N}{{d}^{4}}_{n}}$$

(3)

where d_n is the diameter of each vessel (Sperry et al. 1994). According to the Hagen–Poiseuille equation, D_h is linked to trade-off of hydraulic safety efficiency, which is negatively correlated with cavitation tension and, under current dry and hot environmental conditions, has high plasticity (Hacke and Sperry 2001).

Inter-vessel pits

We selected a subset of one wood core for each study site to evaluate variations in the inter-vessel pit size. We selected samples with visibly marked rings and with a high correlation with the master series. Each core was cut into two small blocks that included the BMd and Hd period. We focused the measurements on radial and tangential surfaces (Fig. 1B, C). Samples were dried and fixed on aluminum stubs using carbon adhesive tape. Later, samples were sputter-coated with a 20-nm-thick gold blade using a sputter coater Denton Desk V (Denton Vacuum, USA) and analyzed with scanning electron microscopy (SEM, JSM 6610-LV, JEOL, Tokyo, Japan). In total, we measured 60 randomly selected pits (Fig. 1E), which correspond to 20 for each study site and 10 inter-vessel pits per period and per study site. The diameter of the pits and their elliptical apertures were measured using a magnification of 2000–20,000. We divided diameters by two to obtain the radius, which we then used to calculate the pit area and its aperture area using the ellipse formula (π×L_i×W_i/4). Finally, we worked on the variables pit total area and pit aperture area (Additional file 1: Fig. S1).

Wood intrinsic water-use efficiency (iWUE)

We selected a subset of 15 cores (5 individuals per study site) to analyze the δ¹³C isotope composition for the period 2000–2020. We cut each core into three wood sections, each corresponding to the drought period considered in this study, e.g., BMd, Md, and Hd. To eliminate possible errors in the chemical process due to handling of the samples, we removed the upper surface of the wood using a sledge microtome (Gärtner and Nievergelt 2010). Samples were pooled by period and analyzed with an elemental analyzer–isotope ratio mass spectrometer (Flasha EA 112 coupled via Conflo IV to a Delta Plus XP mass spectrometer, Thermo Electron, Bremen, Germany). We used whole wood instead of cellulose, because previous studies on hardwood species showed that material from the sapwood is as useful as cellulose for studying environmental effects at a short-term scale (Weigt et al. 2015). Following international standards, we used Vienna Pee Dee Belemmnite. The carbon isotope ratio was calculated as

$${\updelta }^{13} {\text{C}}( \permil ) = \left[ {\left(^{13}{\text{C}}/^{12}{\text{C}}_{\text{sample}} \right)/\left(^ {13}{\text{C}}/^{12}{\text{C}}_{\text{standard}} \right) - 1} \right] \times 1000.$$

We then calculated the wood intrinsic water-use efficiency (iWUE, μmol CO₂ mol^–1 H₂O), carbon isotope discrimination (∆¹³C, ‰) and intercellular CO₂ concentration (C_i, ppm), using the ‘simple’ formulation (Lavergne et al. 2022). We used 25.5‰ as the apparent fractionation for wood tissue (Cernusak and Ubierna 2022). All estimates were calculated using the R package ‘isocalcR’ (Mathias and Hudiburg 2022) for each period and individual, accounting for elevation, mean annual temperature, and ambient CO₂ concentration. Mean annual temperature for each period was obtained from ERA5 re-analysis (Hersbach et al. 2020) for each study site.

Statistical analysis

To examine variation in wood anatomical and physiological characteristics, we fitted linear mixed-effects models with drought period (BMd, Md and Hd), DBH (as a proxy for tree size), and the two-way interactions between them as fixed effects, and individual nested within sites as the random effect using the R-package ‘lme4’ (Bates et al. 2015) in software R (R Core Team 2020). We used the annualized variables to avoid problems with unbalanced data by period (Venegas‐González et al. 2020), which was obtained by dividing the data by 8, 9 and 3 years for BMd, Md and Hd, respectively. To meet normality assumptions, ring width, K_p, and vulnerability index were log-transformed prior to analysis; we also log-transformed DBH to linearize its relationships with response variables. We assessed model assumptions using the ‘check_model’ function in the R package “performance” (Lüdecke et al. 2021), which provides visual evaluations of the normality of model residuals and random effects. Because the full model for wood iWUE and C_i was rank deficient, we removed the interaction with study site and period. Comparison among means for significant effects were evaluated using Fisher’s LSD tests (P < 0.05).

Results

Tree-ring and vessel trait chronologies

The wood anatomy of K. angustifolia populations demonstrates the high dendrochronological potential of the species, with semi-porous rings and increasing fiber wall thickness in latewood (Additional file 1: Fig. S2). From the 77 trees sampled, 67 were successfully cross-dated (87%, r = ∼0.55 and RBar = ∼0.31) and were used to build the ring width chronologies for each population (Table 1). The northern and southern populations are comprised of adult trees with an average age of 53 years with a maximum age of 74 years, while the central population is younger (average age of 22 years). Annual growth, i.e., average tree-ring width, is similar across all populations, with mean values between 1.40 and 2.06 mm yr⁻¹.

We built chronologies of wood anatomical traits covering three periods of increasing drought severity across the range of Kageneckia angustifolia in central Chile (2000–2020, Fig. 3). In total, we analyzed 56,637 vessels of K. angustifolia trees for the period 2000–2020 (Table 1). We found smaller vessels, in terms of mean vessel area and hydraulic diameter, in the center and southern populations. On the other hand, the central and northern sites have the highest annual rates of radial growth, but lower vessel frequency, in contrast to what we observed in the southern population. The highest value of potential hydraulic conductivity (K_p) was found in the southern population, which was 104% and 50% higher than that of the northern and central populations, respectively. Consequently, we found the highest values of the vulnerability index in the northern population (Table 1).

Fig. 3

Annual variation of A tree ring width (RW, mm yr⁻¹), B hydraulic diameter (D_h, µm) and C vessel frequency (VF, ind mm⁻²) for the period 2000–2020. The black line is the mean and the shaded area is the standard error. Red vertical lines represent the start of the megadrought in 2010 and the extreme drought in 2019

Tree responses to drought intensity

We found that wood anatomical and physiological traits varied significantly in response to drought severity (Additional file 1: Table S1). In general, most of the total variation of each wood trait was explained by random effects, i.e., variation within study sites and individuals (conditional R²: 0.58–0.88; marginal R²: 0.04–0.39). All wood traits exhibited a statistically significant response to drought intensity. Furthermore, we found that hydraulic diameter (D_h), potential hydraulic conductivity (K_p), and ring width (RW) increased with increasing tree size (DBH) (Additional file 1: Fig. S3). In addition, the interaction between drought period and DBH was statistically significant for vessel frequency (VF) and the vulnerability index (VI) (Additional file 1: Fig. S4).

With increasing drought severity, radial growth (RW) and the vulnerability index (VI) declined significantly during the Hd period (Fig. 4A–E). In contrast, vessel frequency (VF) increased in response to drought intensity (Fig. 4B). Hydraulic diameter (D_h) and K_p increased during the megadrought period (Md), but decreased during the hyperdrought period (Hd), reaching values similar to those before the start of the megadrought period (BMd) (Fig. 4C, E). On other hand, iWUE increased with drought severity and was highest in the Hd period (Fig. 4F). We found that carbon isotope discrimination (∆¹³C) also declined significantly on both drought periods, while C_i increased during the Hd period, but not significantly (Additional file 1: Table S1, Fig. S5).

Fig. 4

Estimated values of wood anatomical and physiological traits of K. angustifolia populations across drought periods (before megadrought—BMd: 2000–2007, megadrought period—Md: 2010–2018, hyperdrought period—Hd: 2019–2020. A Ring width (RW, mm yr⁻¹); B vessel frequency (VF, ind mm⁻²); C hydraulic diameter (DH, µm); D vulnerability index of the xylem to embolism (VI); E potential hydraulic conductivity (K_p, kg m MPa⁻¹ s⁻¹); F wood intrinsic water-use efficiency (iWUE; μmol CO₂ mol⁻¹ H₂O). Differences between periods for each site were evaluated using Fisher's LSD test (p < 0.05)

We only found significant differences between drought periods for inter-vessel pit aperture area, but not inter-vessel pit total area (Fig. 5, Additional file 1: Fig. S6). These differences are attributed to the reduction of the pits with respect to their length (19%), width (20.2%), and area (37.1%) during the Hd period.

Fig. 5

Boxplots of inter-vessel pit area and inter-vessel pit aperture area of studied K. angustifolia trees between the periods: A before the megadrought (BMd: 2000–2007), and B the hyperdrought period (Hd: 2019–2020). **Indicate significant differences between period with p < 0.01

Discussion

In this study, we explored the anatomical and physiological mechanisms that underpin the response of a treeline forest species of K. angustifolia to increasing drought intensity. Our results revealed coordinated shifts in vessel size and frequency and water-use efficiency to enhance the hydraulic safety. The apparent resilience of K. angustifolia to extreme droughts suggests that this adaptation to drought stress may increase its ability to tolerate novel climatic conditions at the treeline environments of the Mediterranean Andes.

Vessel traits in response to drought intensity

Consistent with previous studies (Barichivich et al. 2009), we found that the ring width decreased in response to the historic drought of 2019 (Fig. 4). This is likely due to the fact that during the formation of tree rings favorable climatic conditions allow for the formation of larger diameter vessels, which improves water transport capacity and leads to an increase in tree growth (Garcia-Gonzalez et al. 2016; Nola et al. 2020; Gennaretti et al. 2022). Our analysis identified evidence of shifts in vessel traits in response to drought severity that are consistent with greater hydraulic safety, underlying the decrease in ring width. Under harsh climate conditions, e.g., low precipitation and high temperatures, vessel frequency is expected to increase and vessel diameter to decrease, thereby increasing hydraulic safety as these changes reduce the risk of cavitation (Sperry et al. 2006; Choat et al. 2012; Gea-Izquierdo et al. 2012; Gleason et al. 2016). Particularly after the extreme drought in 2019–2020, we observed an increase in vessel frequency and a decrease in hydraulic diameter, both aspects linked to increasing hydraulic safety (Fig. 3). These changes in vessel traits likely reduce the vulnerability of K. angustifolia to cavitation as evidenced by the decrease in the vulnerability index, which captures vessel diameter to frequency ratio (Fig. 3D). However, we also found that during the hyperdrougth period (Hd) there was a decrease in potential hydraulic conductivity, indicating a reduction in lumen area, which likely decreases stem (Bryukhanova and Fonti 2013). Previous studies focused on conifers have shown lower theoretical hydraulic conductivity, suggesting a long-lasting process of hydraulic deterioration in their water transport system, that could be linked to crown dieback and a decline in tree vigor (Pellizzari et al. 2016; Puchi et al. 2021). Consequently, it is necessary to continue monitoring these populations. Furthermore, we found that the decrease in D_h and K_p observed between the Md and Hd periods is not due to ontogeny, and rather, could indicate a pre-adaptive response of the species to this historical climatic event.

The observed shift towards a greater efficiency of the hydraulic safety in response to drought severity could be due to the fact that species' range is restricted to high elevations, consistent with the ‘pre-adaptation hypothesis’ (Peñuelas et al. 2001; De Kort et al. 2020), where tree growth rates are typically lower than species that occur at lower elevations (Körner and Hoch 2023). Even under normal environmental conditions, tree species of Mediterranean mountain forests of central Chile have exhibited slow radial growth rates, which are lower than 1.5 mm yr⁻¹ (Venegas-González et al. 2018; Matskovsky et al. 2021). Moreover, K. angustifolia, as a semi-ring-porous species, produces earlywood with wider vessels (Fig. 1B), which is associated with greater xylem plasticity, which in turn facilitates greater hydraulic efficiency (García-González et al. 2016; Rodríguez-Ramírez et al. 2022).

Further supporting the shift towards greater hydraulic safety, our results indicate a decrease in pit aperture area with increasing drought intensity. The trade-off between safe and efficient xylem transport is related to protection against blockage from cavitation and embolism, and low flow resistance for the vascular tissue (von Arx et al. 2016), which inferred directly changes in pit membranes (Wheeler et al. 2005; Mrad et al. 2018). Smaller elliptical pit apertures have been found to be positively correlated with enhanced resistance to cavitation (Schmitz et al. 2007; Lens et al. 2011; Zhang et al. 2017). Wheeler et al. (2005) found an inverse relationship between the pit surface area per vessel and water potential, inducing tradeoffs between xylem safety and hydraulic efficiency for eleven tree species of the Rosaceae family, the same family as that of K. angustifolia (Hacke et al. 2006; Choat et al. 2008; Pittermann et al. 2010; Lens et al. 2011). Together with the plastic responses of vessel traits to increased drought intensity, the decrease in pit aperture area supports the idea that K. angustifolia is resilient to water stress, which supports the pre-adaptation hypothesis, i.e., that species or populations that evolve in stressful environments have a high capacity to adapt to climate change (Li et al. 2018; Xu et al. 2020).

iWUE in response to drought intensity

We found that K. angustifolia increased wood iWUE in response to increasing drought intensity. This response supports the idea that K. angustifolia exhibits a drought avoidance strategy (Martínez-Ferri et al. 2000), by closing its stomata to reduce transpiration (Saurer and Voelker 2022). In line with our results, declines in tree growth and increases in iWUE in other high-elevation Mediterranean forests have been observed in recent decades in response to global change (Andreu‐Hayles et al. 2011; Peñuelas et al. 2011; Granda et al. 2014; Wu et al. 2015). In contrast, forests that have experienced drought-induced mortality have shown an increase in Δ¹³C and a decrease in iWUE, suggesting higher stomatal conductance rates (Linares and Camarero 2012; Hentschel et al. 2014; López et al. 2021). Therefore, our results suggest that K. angustifolia has largely avoided drought induced mortality via a reduction in stomatal conductance and shift in vessel traits towards greater hydraulic safety. As the drought in our study region has continued since 2020, it remains unclear the extent to which K. angustifolia will be able to continue tolerating water stress.

Conclusions

Our study provides insights to the anatomical and physiological mechanisms underlying the resilience of treeline forests to persistent droughts in central Chile. We found that the extreme drought in 2019–2020 triggered shifts in vessel size and frequency and frequency that increased hydraulic safety, thereby reducing vulnerability to hydraulic failure. These shifts in vessel traits occurred in parallel with a decrease in pit aperture area and an increase in water-use efficiency, further increasing the resilience of K. angustifolia to drought stress. Therefore, our results suggest a high capacity of K. angustifolia to tolerate droughts, even though it is not clear whether these adaptations will be sufficient to scenarios of greater intensification of climate stress in the future. Given the extent of the impacts of the megadrought on tree growth and productivity of Mediterranean forests in central Chile, it is crucial to extend the analysis of drought responses to other tree species in these ecosystems, which will allow us to test if other high-elevation species also exhibit a greater capacity to tolerate droughts. Furthermore, examining non-structural carbohydrate (NSC) dynamics during the megadrought may further elucidate the capacity of tree species in these ecosystems to endure persistent droughts.