Introduction

The exceptional summer drought episode of 2018/2019 has exposed many European forests to extreme stress caused by the joint action of atmospheric and edaphic drought and heat (Obladen et al. 2021; Walthert et al. 2021; Frei et al. 2022). 2018 was the warmest year in Germany since the start of weather recording in 1881, with mean annual temperature (10.4 °C) being 2.2 °C above the long-term mean (1961–1990), and annual precipitation (435 mm) reaching only 58% of the long-term mean (747 mm). In correspondence, vapor pressure deficit reached record highs, thus enhancing atmospheric drought stress (Williams et al. 2013), while the deficit in the climatic water balance was the second largest ever recorded in the region (Schuldt et al. 2020). This drought event apparently was the most extreme in the last 2100 years in Central Europe (Büntgen et al. 2021).

In addition to climate warming, some Central European regions have experienced reductions in mid-summer precipitation in recent decades, which increases climatic drought in the physiologically most important season (Trnka et al. 2015; Bat-Enerel et al. 2022). As a consequence, widespread canopy dieback and tree mortality have recently been reported in various Central European regions (Schuldt et al. 2020; Senf et al. 2020; Braun et al. 2021; Thonfeld et al. 2022) with especially Norway spruce (Picea abies Karst.), but also European beech (Fagus sylvatica L.), Scots pine (Pinus sylvestris L.), Quercus species, and Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) being affected. The infestation of weakened trees with pest organisms often has accelerated stand decline. Climate models predict that such extreme hot droughts likely will increase in frequency and severity in Europe with advancing climate warming (Fischer and Schär 2008; Spinoni et al. 2018). This has raised concern among forest scientists and foresters about the suitability of Central Europe’s main timber species for silviculture in a future warmer climate.

Norway spruce, European beech, and Douglas-fir are the economically most important timber species of Central Europe, together with Scots pine and two oak species (Quercus petraea (Matt.) Liebl. and Q. robur L.) (Thünen Institute 2015). All three species were more or less affected by the hot droughts of the recent past, but their drought sensitivity and drought response strategies likely are dissimilar. F. sylvatica is the dominant tree species of Central Europe’s natural forests, which would cover about two thirds of Germany’s forest area in the absence of human influence (Bohn et al. 2003; Leuschner and Ellenberg 2017). Despite its competitiveness, F. sylvatica has been assessed as relatively sensitive to drought and heat (Gessler et al. 2004; Leuschner 2020; Schuldt et al. 2020; Walthert et al. 2021; Arend et al. 2022; Leuschner et al. 2023). P. abies is native to Central Europe’s higher mountains and is widespread in the boreal zone, but has been planted widely in lower montane and lowland regions outside of its natural range, since it is a fast-growing species whose wood is demanded for multiple purposes. Despite widespread damage by droughts, subsequent insect attack and windbreak in recent decades, it is still the economically most important timber species in Germany and other Central European countries, even though its percentage has decreased constantly in the past three decades (Thünen Institute 2015; Leuschner and Ellenberg 2017). Yet, numerous studies evidence the drought sensitivity of this species (Tumajer et al. 2017; Krejza et al. 2021). The North American conifer P. menziesii has been planted in Central Europe since the nineteenth century and covers about 2% of the recent forest area in Germany (Thünen Institute 2015) and 3% in France (Zeller et al. 2019). It is favored by European foresters as a possible replacement for P. abies in production forests due to its high productivity, excellent wood properties and assumed high drought resistance (Spiecker et al. 2019). From published work on subjective assessments of the drought resistance of the three species, a principal ranking of the species in the sequence P. abies < F. sylvatica < P. menziesii seems to emerge (Niinemets and Valladares 2006).

One of the traits determining a tree species’ drought vulnerability is the embolism resistance of its xylem. The P50- and P88-values, i.e., the xylem pressures at which 50% or 88% of hydraulic conductivity is lost, have widely been used for comparing tree species’ sensitivity to drought (Maherali et al. 2004; Choat et al. 2012; Lobo et al. 2017). It is thought that many woody species operate relatively close to their hydraulic safety margin (HSM), where minimum leaf water potential meets the critical value of embolism onset in the xylem. Yet, conifers tend to die at lower PLC (percentage loss of conductance) values than angiosperms (Urli et al. 2013), which may have led in many conifers to the evolution of larger HSMs than in angiosperm trees (Choat et al. 2012; Carnicer et al. 2013). Recorded P50 (and P88) values in mature trees of the three species here considered vary with hydrological site conditions, with more negative values often recorded at drier sites (Schuldt et al. 2016; Tomasella et al. 2017). According to the literature (Cochard 1992; Hacke and Sauter 1995; Sperry and Ikeda 1997; Piñol and Sala 2000; Maherali et al. 2004; Wortemann et al. 2011; Schuldt et al. 2016; Weithmann et al. 2022b), F. sylvatica shows the highest P50-values (ranging from − 2.8 to − 3.8 MPa), P. menziesii the lowest (from − 3.7 to − 5.5 MPa), and P. abies intermediate values (from − 3.5 to − 4.0 MPa). This would suggest the greatest embolism resistance in P. menziesii and the lowest in F. sylvatica, partly contradicting the sequence of drought resistances given for the three species in Niinemets and Valladares (2006). Less is known about the three species’ HSMs.

How plants regulate water potential variation determines their response to drought, notably the capacity to maintain carbon assimilation under water deficits. With respect to the ability to regulate leaf water potential (ΨLeaf), plants have frequently been categorized according to their degree of isohydry. While isohydric species maintain leaf water potential fairly constant, anisohydric species tolerate greater drops in ΨLeaf (Klein 2014). F. sylvatica has been characterized as markedly anisohydric (Leuschner et al. 2021), allowing for large diurnal and seasonal ΨLeaf variation, while P. abies seems to pursue a more isohydric strategy (Lyr et al. 1992; Leo et al. 2014; Zweifel et al. 2009). For P. menziesii, both isohydric and anisohydric regulation has been reported for different provenances and growth conditions (Anekonda et al. 2002; Warren et al. 2004; Link et al. 2014; Jansen 2017; Kerhoulas et al. 2020). A direct comparison of the three species with respect to traits determining drought vulnerability is lacking so far, especially for mature trees growing under similar environmental conditions. It, thus, remains unclear how the three species are to be ranked with respect to their drought vulnerability. Yet, a comparative assessment of the three species’ drought resistance is urgently needed, given their importance as timber species in Central Europe and the threat posed by global warming.

Conifer and angiosperm trees differ in their wood anatomy, with the latter developing more complex woody tissues with tracheae and higher parenchyma fraction in comparison to the simpler tracheid-dominated conifer wood. The tracheids of conifer xylem are usually smaller than angiosperm vessels and often more resistant to embolism formation (Tyree and Zimmermann 2002). There is some evidence that these differences in wood anatomy also relate to stomatal regulation stringency and patterns of ΨLeaf variation, as it appears that isohydric regulation is more frequent in temperate and boreal conifers, while angiosperm trees seem to display more often anisohydric regulation (Carnicer et al. 2013; Martinez-Sancho et al. 2017; Blackman et al. 2019). Yet, isohydric angiosperms and anisohydric conifers do also exist (Voelker et al. 2018; Leuschner et al. 2019). It has been postulated that many conifers lack the capacity to repair drought-induced embolism due to a low wood parenchyma fraction and, thus, lower non-structural carbohydrate levels in their stem wood, which could explain the need for greater HSM (Johnson et al. 2012; Carnicer et al. 2013).

Here, we present the results of a comparative study of hydraulic and leaf water status traits in F. sylvatica, P. abies, and P. menziesii trees under exposure to severe natural drought, which might help defining hydrological thresholds of these three important timber species. Our study in the exceptionally dry and hot summers of 2018 and 2019, enhanced by low precipitation in winter 2018/2019, is a welcome case study of the vulnerability of these three important Central European timber species to climatic extremes, as they are predicted to occur more frequently in future with climate warming (Fischer and Schär 2008; Meinke et al. 2010; Spinoni et al. 2018). We measured diurnal, seasonal, and inter-annual variation in stomatal conductance and leaf water potential, established branch xylem vulnerability curves (P12-, P50-, P88-values), and measured leaf tissue turgor loss point (πtlp), and leaf and branch morphological traits [leaf size, specific leaf area (SLA), sapwood-to-leaf area ratio (Huber value)]. This enabled us to explore the stringency of stomatal regulation in relation to leaf water potential variation and branch embolism resistance of the three species. The study was conducted in the upper sun crown of mature trees growing in close vicinity to each other that were accessed with a mobile skyjack. We tested the hypotheses that (1) the conifers P. abies and P. menziesii display a more isohydric regulation strategy, while F. sylvatica pursues an anisohydric regulation, tolerating larger ΨLeaf fluctuation, and (2) hydraulic safety margins decrease in the sequence P. menziesii > P. abies > F. sylvatica, since embolism resistance seems to decrease in this sequence, while the threat of encountering low water potential minima should increase. We define drought resistance in general as the ability of a plant to maintain fitness in the face of climatic and edaphic drought stress through mechanisms that help avoiding and/or tolerating tissue desiccation.

Materials and methods

Study site

The study was carried out in the Lüß Forest close to Unterlüß (Lüneburg Heath, northern Lower Saxony) in northern Germany (52°50′N, 10°19′E; elevation: 132–139 m a.s.l.). The region is characterized by a cool-temperate climate with a mean annual temperature of 9.0 °C and mean annual precipitation of 747 mm (DWD, German Weather Service, Offenbach, Germany, period 1981–2018). Soils are spodo-dystric Cambisols developed in sandy to loamy deposits of the penultimate (Saale) glaciation (Drenthe stadial) (6% clay, 15% silt, 79% sand) (Foltran et al., unpubl data). We chose stands of the three species F. sylvatica, Picea abies and P. menziesii in vicinity to each other (maximum distance: 2200 m) under comparable environmental conditions (Fig. S1). The physiological measurements were conducted in three mature neighboring trees per species that grew either in pure stands or mixtures with clear dominance of the respective species. The three trees each were located directly adjacent to each other and, thus, could be accessed with the skyjack within short time. All trees were of mature age (F. sylvatica and P. abies: 122 years, P. menziesii: 70 years), dominant individuals in the stands and in good health, with heights of 28–42 m and diameters at breast height (DBH) of 52–80 cm (Table 1). Tree height was measured in 2018 with a Vertex IV Hypsometer with T3 transponder (Haglöf, Längsele, Sweden) and DBH with a dendrometer band (UMS GmbH, München, Germany). The F. sylvatica stand had a mean DBH (all stems included) of 26.5 cm (range 7.2–59 cm), a basal area of 30.4 m2 ha−1, and a stem density of 320 ha−1. The P. menziesii stand had a mean DBH of 43.6 cm (range 11.8–64.3 cm), a basal area of 50.2 m2 ha−1, and a stem density of 264 ha−1. Finally, the P. abies stand had a mean DBH of 44.9 cm (range 12.8–66.4 cm), a basal area of ca. 30 m2 ha−1, and a stem density of 320 ha−1. The stands differ mainly with respect to the number of juvenile trees, which are higher in the F. sylvatica and P. menziesii stands. Hourly measured air temperature, air humidity, and precipitation data were obtained from a climate station set-up on an open field in close proximity to the study sites (52°49.830″N, 10°18.864″E). Vapor pressure deficit (VPD) was derived from air humidity and air temperature using the Teten’s formula. Volumetric soil water content was measured with TDR- and TensioMark-sensors at 5 cm, 20 cm, 50 cm, and 100 cm depth at a station located between the F. sylvatica and P. abies stand. The nearby P. menziesii stand stocked on soil of similar soil texture, and soil moisture conditions were, therefore, largely comparable. Additionally, daily precipitation data were available from a gauge installed on a nearby farm (52°49′42.5″N 10°14′55.9″E).

Table 1 Structural characteristics of the nine investigated trees in 2018.
Full size table

Stomatal conductance and leaf water potential measurements

All measurements were carried out on twigs and leaves of the fully sun-exposed upper crown. In case of F. sylvatica and P. abies, this were the uppermost twigs of the crown, in case of taller P. menziesii, we measured side branches in the uppermost third of the crown. Canopy access was achieved through a mobile skyjack (model DL30, DENKA-Lift A/S, Denmark), which allowed reaching 30 m height. Stomatal conductance (gS) and leaf water potential (ΨLeaf) measurements were carried out in diurnal measuring campaigns from 8 a.m. to 5 p.m. at time steps of about 1 h on various days between May and September 2018 (7–11 campaigns per species) and 2019 (4 campaigns per species). Midday leaf water potential (ΨMD) and midday stomatal conductance (gMD) measurements were taken between 12 a.m. and 4 p.m. on sunny or partly cloudy days, and daily minima were identified from those values separately for each year. The three lowest ΨLeaf values recorded in a summer were averaged and termed Ψmin. Stomatal conductance was measured with a Li-Cor 6400 gas exchange system (Li-Cor Inc., Lincoln, NE, USA) at ambient light, air humidity, and temperature conditions. The skyjack was placed in the center of the three studied trees of the same species, and each five leaves (or needle-bearing shoots) per tree on the same twig were measured every hour, rotating between the three trees each. We then averaged over the 15 measurements per hour and tree species. As the skyjack could not be moved so rapidly to accommodate all 9 trees of the three species on the same day, the three species were measured on different days, usually on the next days. Since all measuring days were sunny or partly overcast days, the atmospheric conditions on consecutive days were sufficiently comparable. Due to the hot and dry weather in the summers of 2018 and 2019, the majority of leaf conductance measurements were conducted under photon flux densities close to or above light saturation of photosynthesis of the species (c. 450–700 μmol m−2 s−1; Leuschner and Ellenberg 2017).

While F. sylvatica leaves were measured in the conventional leaf chamber (which was completely filled by the leaves), we used the conifer chamber for measurements on P. menziesii and P. abies short shoots. To relate photosynthesis to area, the cumulative needle surface area of all needles inside the chamber was determined with the software WinSeedle 2013 (Régent Instruments, Quebec City, QC, Canada). This was determined for each three shoots of a tree, and the area averaged. All leaves were marked at the beginning of the season to ensure that measurements always took place on the same leaves/shoots throughout the summer. In case of leaf damage or loss, it was replaced by another leaf on the same branch. We measured in both seasons (2018 and 2019) the same branches.

ΨLeaf was measured with a Scholander pressure chamber (1505D-EXP, PMS Instruments Company, Albany, OR, USA) on short shoots of F. sylvatica (typically bearing four leaves), P. abies, and P. menziesii (youngest shoots of ca. 10 cm length). We measured three leaves/shoots (one per tree) for ΨLeaf per hour and averaged the data. The shoots used for ΨLeaf measurement were taken from a branch in direct neighborhood to the one serving for stomatal conductance measurements. To characterize the three species along the isohydry–anisohydry continuum of water potential variation (Klein 2014), we chose the seasonal variability of ΨMD as a criterion, with more isohydric species showing less variability (Martinez-Vilalta and Garcia-Forner 2017).

Measurement of xylem hydraulic conductivity and vulnerability curves

During the second half of August in 2018 and 2019, each 6–8 sun-canopy branches per species of about 50 cm length and 10 mm in diameter were collected for hydraulic conductivity measurements and analysis of vulnerability curves. The branches were cut in the air and immediately wrapped in moist towels, which had been soaked in distilled water containing a sodium–silver-chloride complex (16 μg L−1 Ag and 8 mg L−1 NaCl; Micropur katadyn, Wallisellen, Switzerland) to restrict microbial activity. The samples were stored at 4 °C in the dark and used for hydraulic conductivity measurement and vulnerability curve establishment within 4 weeks. The twig segment used for conductivity measurement was marked in the field and all leaves distal to the segment were collected and stored in plastic bags at 4 °C. Additional branch samples (two per tree, i.e., six per species) were collected at the end of August 2018 in the sun canopies for establishing pressure–volume curves. The twigs were cut in the field and immediately transferred to polyethylene tubes containing demineralized water and covered with a plastic bag to prevent water loss, were re-cut under water in the laboratory and stored overnight in a cool and dark room (Koide et al. 2000; Prometheuswiki 2018).

Hydraulic conductivity (Kh, kg m MPa−1 s−1) was determined on 3–4 branches per tree by connecting the segment to the Xyl’em apparatus (Bronkhorst, Montigny-les-Cormeilles, France) and flushing demineralized, degassed and filtered (0.2 µm) water containing 10 mM KCl and 1 mM CaCl2 through the segment. We used different procedures for broadleaved F. sylvatica and the conifers. In case of F. sylvatica, branches were shortened under water to about 30 cm length, lateral branches were cut-off and the scars quickly sealed with instant glue (Loctite 431, Henkel, Düsseldorf, Germany) to prevent water leakage during measurement. The exact length of the segments and their diameters were measured twice at the basipetal and distal ends and four times along the segment. Subsequently, 1 cm of the bark at the basipetal end was removed and the twigs connected to the Xyl’em apparatus. After obtaining the actual hydraulic conductivity (Khact) under a small pressure difference of 6 kPa, samples were flushed at high pressure (120 kPa) for 10 min to remove potential emboli. Measurements at low pressure and flushing events were repeated until maximum hydraulic conductivity (Khmax) was reached. In case of the conifers, branches were shortened under water to a segment length of 5 cm that lacked lateral branches, and the bark was completely removed. Length and diameters were measured in the same way as for F. sylvatica. To avoid conduit sealing by resin released during flushing, samples were afterward stored under vacuum for at least 12 h, while kept in the flushing solution. Subsequently, Khmax was measured in the conifer samples at 6 kPa pressure. Hydraulic conductivity and flow rate were computed with the software XylWin 3.0 (Bronkhorst, Montigny-les-Cormeilles, France), which considers segment length. Specific hydraulic conductivity (KS, kg m−1 MPa−1 s−1) was computed by dividing Khmax by the maximal basipetal sapwood cross-sectional area (Acrossmax) without pith and bark (Hajek et al. 2014; Schuldt et al. 2016). To obtain Acrossmax without bark for the F. sylvatica twigs with bark, the following equation established by Schuldt et al. (2016) was used: Acrossmax = − 3.715 + (0.7698 * Acrossmax (bark)), with Acrossmax (bark) being the recorded Acrossmax including bark. Leaf-specific conductivity (KL, kg m−1 MPa−1 s−1) was calculated by dividing Khmax by the associated total leaf area of the branch segment.

For establishing vulnerability curves, the Cavitron technique was applied (Cochard et al. 2005), using a custom-made Cavitron rotor chamber (Delzon et al. 2010) attached to a commercially available centrifuge (Sorvall RC-5C; Thermo Fisher Scientific, Waltham, MA, USA). In F. sylvatica, an average maximum vessel length of 19.3 ± 2.6 cm has been reported (Lübbe et al. 2022), which makes this species suited for flow-centrifuge measurements with a 30 cm rotor. Three–four branches per species were cut under water to a length of 27.5 cm, and the bark removed at both ends for about 3 cm in case of F. sylvatica, or completely in case of the conifers, before cutting to avoid possible extrusion of resin. Conductivity measurements started with a negative pressure of − 0.83 MPa, and pressure was raised stepwise by 0.2–0.3 MPa until a percentage loss of conductance (PLC) of at least 90% was reached. PLC values were recorded with the software CaviSoft (Version 2.1, University of Bordeaux, France) and plotted against xylem pressure to generate vulnerability curves. To derive the xylem pressure causing 50% loss of conductance (P50), a sigmoidal function was fitted to the data points according to the equation given by Pammenter and Vander Willigen (1998), with PLC = 100/(1 + exp(s/25 * (Pi − P50))), in which s (% MPa−1) is the negative slope of the curve at the inflexion point and Pi the applied xylem pressure. The xylem pressures causing 12% (P12; embolism onset) and 88% (P88) loss of conductance were derived following Domec und Gartner (2001) with P12 = 2/(s/25) + P50 and P88 = − 2/(s/25) + P50.

Establishment of pressure–volume curves

Pressure–volume curves (PVC), i.e., the plot of water potential (− 1/Ψ in MPa) vs. 100 − total relative water content (RWC, %), were established in twigs sampled at the end of August 2018 with the over-pressurization technique (Tyree and Jarvis 1982; Koide et al. 2000) using several pressure chambers (1505D-EXP and M1000, PMS Instruments Company, OR, USA). Measurements took place on the day after sampling on two twigs per tree (six per species) that had been hydrated to near-maximum turgor overnight and that were measured in parallel, following the protocol given by Leuschner et al. (2019). Briefly, samples were once more re-cut before measurement, dried with tissue paper, weighed to 1 µg to determine initial fresh weight and immediately placed in the pressure chamber to measure the initial water potential. The pressure in the chamber was then increased to 0.4 MPa and kept at this level for 10 min. while the expressed sap was collected in a small vial filled with cotton wool. Subsequently, the pressure was released for 10 min to allow symplast and apoplast to equilibrate. This procedure was repeated in pressure steps of 0.3 MPa until 2.5 MPa were reached. To achieve a higher resolution in the linear part of the pressure–volume curve, the steps were reduced to 0.2 MPa between 2.5 and 3.3 MPa (final value). The vials were weighed to an accuracy of 100 µg before and after collecting the sap for calculating the amount of expressed sap. Finally, the leaves and shoots were dried at 70 °C for 48 h to determine dry weight. The following parameters were derived from the curves: osmotic potential at full turgor (π0) and at turgor loss point (πtlp), relative water content (symplastic water) at full hydration (RWC0), relative water content at turgor loss point (RWCtlp), and the bulk modulus of tissue elasticity (ε0) with ε0 being calculated with the standard major axes method (Sokal and Rohlf 1995).

Leaf morphological measurements

All leaves and needles on the measured twigs were scanned (V11 Epson Perfection, Epson, Nagano, Japan) and analyzed with the software WinFolia 2005 (Régent Instruments, Quebec City, QC, Canada) in case of leaves and WinSeedle 2013 (Régent Instruments, Quebec City, QC, Canada) in case of needles to obtain mean leaf/needle area and total leaf area on the twig (Amean and Atotal, cm2). To overcome the problem of comparing leaf and needle surface areas, we used projected surface area (silhouette area) for both types of foliage in subsequent calculations (e.g., for stomatal conductance), as has been done in earlier studies (e.g., Wallin et al. 1990; Stinziano et al. 2015). The option to calculate total surface area (upper and lower side) of the needles from the measured projected area by means of empirically determined factors (Keller and Wehrmann 1963; Perterer and Körner 1990; Goisser et al. 2016) was abandoned, as it would have inflated the foliage surface area of the conifers compared to F. sylvatica, and due to considerable variation in conversion factors, which vary between 2.38 (Schulze et al. 1977) and 3.2–3.3 (Perterer and Körner 1990; Goisser et al. 2016), probably due to the influence of local environmental conditions.

We further calculated Huber value (HV, 104 m2 m−2), i.e., the sapwood-to-leaf area ratio, by dividing sapwood cross-sectional area (see above) by the total leaf area distal to the cut. Leaves were oven-dried at 70 °C for 48–72 h and subsequently weighed to calculate specific leaf area (leaf surface to dry mass ration, SLA, in cm2 g−1).

Statistical analysis

All statistical analyses were carried out with the software package R (R Core Team 2013, version 4.0.5), except for the determination of P12/50/88-values, which were calculated with SAS 9.13 software (SAS Institute Inc., Cary, North Carolina, USA). Subsamples from a tree were averaged and the data subsequently tested for normal distribution with a Shapiro–Wilk normality test. The Wilcoxon signed-rank test was used to test for significant differences in the studied traits, comparing years and species based on means at the tree level (see Table 2). Linear models were fitted to the data after selection of the best-fitting model using the Akaike Information Criterion (AIC) and the model results depicted in the graphs. A significance level of p < 0.05 was used throughout the paper.

Table 2 Physiological and leaf morphological traits of the three species in the summers 2018 and 2019 (species-level means with SE in brackets)
Full size table

Results

Weather and soil moisture conditions in 2018 and 2019

While both summers 2018 and 2019 were exceptionally dry and hot, the two years differed in the timing of dry and moist periods. The summer 2018 was characterized by a very long period without larger rainfall events (no single event with > 13 mm d−1 between mid of April and September, rainfall total May–September: 94 mm, annual precipitation total: 435 mm). Soil water content varied in the whole profile to 1 m depth around 2 vol.% until October (maximum: 3%), and maximum air temperature reached 36.3 °C, minimum relative air humidity 22% and maximum VPD 4.3 kPa (Fig. S2 in the Supplement). Precipitation in summer was somewhat higher in 2019 (284 mm from May to September), but soil moisture reserves had not been recharged during the 2018/2019 winter, and soil water content consequently reached only 13 vol.% in mid-March 2019, to drop to 3 vol.% in early July. As in summer 2018, relative air humidity dropped to 21% in summer 2019, maximum air temperatures reached 37.3 °C and maximum VPD 4.7 kPa.

VPD dependence of stomatal conductance

Stomatal conductance (gS) was relatively low in 2018 in all three species with most values ranging between 15 and 100 mmol m−2 s−1 only (Fig. 1). Higher gS values (up to 220 mmol m−2 s−1 in F. sylvatica and to 170 mmol m−2 s−1 in the conifers) were reached in all species in 2019, with the increase being most pronounced in F. sylvatica. gS was in both summers higher in F. sylvatica (68:9 ± 43.6 mmol m−2 s−1) than in the two conifers (P. abies: 32.3 ± 40.2 mmol m−2 s−1; P. menziesii: 24.0 ± 21.7 mmol m−2 s−1); P. abies tended to have a higher stomatal conductance than P. menziesii. The latter species maintained especially low stomatal conductance levels in mid and late summer 2018 (16.8 ± 10.2 mmol m−2 s−1 from mid-July–September), with only minor gS reduction upon a VPD increase. The dependence of gS on VPD was significant in all three species in 2018, but this relation diminished (or was weak) in 2019 in F. sylvatica and P. abies. The strongest gS decrease with VPD was recorded among the species in F. sylvatica in 2018 (p < 0.001, R2 = 0.45) and in P. menziesii in 2019 (p = 0.008 and R2 = 0.22), covering a VPD range of c. 0.4–2.3 kPa.

Fig. 1
figure 1

Dependence of leaf stomatal conductance (gS) on vapor pressure deficit of the air (VPD) in summer 2018 and 2019 for the three species (all measurements taken between 10 a.m. and 4 p.m. on various sunny or overcast days between June and September). Each point represents the mean of five replicate leaves/shoots measured per tree. R2 and p values are given for each curve per species per year. Only significant relationships are displayed with regression lines

Full size image

Relationships between g S and Ψ Leaf

The extent of seasonal variation in midday stomatal conductance (gMD) and midday leaf water potential (ΨMD) differed markedly between the species and also between the summers. In F. sylvatica and P. menziesii, the gMD − ΨMD relationship was positive and highly significant, while only a non-significant tendency appeared in P. abies (Fig. 2). Moreover, the gMD drop with a ΨMD decline was steeper in P. menziesii than in P. abies and F. sylvatica. F. sylvatica revealed in 2018 a characteristic anisohydric behavior with a marked ΨMD decline from about − 1.7 to − 2.7 MPa (Δ1 MPa) between June and September, even though midday stomatal conductance was reduced from about 70 to 20 mmol m−2 s−1. In contrast, P. abies exhibited in 2018 a typical isohydric leaf water status regulation pattern with maintenance of very low gMD values of 12.3 ± 11.1 mmol m−2 s−1 throughout summer, which resulted in an only minor ΨMD drop between June and September (by ca. 0.3 MPa; Fig. 3). P. menziesii also reached only very low gMD values of 12.5 ± 7.1 mmol m−2 s−1 between July and September, and showed minimal change in ΨMD (Δ 0.1 MPa), but it is likely that water potential has dropped from higher levels already in May or June. Patterns were different in summer 2019, when all three species showed a linear decline in leaf water potential from May to October. Daily ΨLeaf and gs minima started at higher levels in summer 2019 than in 2018, and the seasonal ΨMD drop was larger in P. abies in 2019 than in 2018 (Δ1.0 MPa vs. 0.3 MPa). gMD declined from early to late summer only in P. abies, but reached a late-summer peak in F. sylvatica and P. menziesii, surpassing 100 mmol m−2 s−1 (Fig. 3).

Fig. 2
figure 2

Relationship between midday leaf conductance (gMD) and midday leaf water potential (ΨMD) in 2018 (filled symbols) and 2019 (open symbols) in the three species. Each point gives a day’s leaf conductance at the time of daily leaf water potential minimum (ΨMD) (gMD values are means of each five replicate leaves measured per tree, while ΨMD values base on one measurement only). F. sylvatica: y = 113 − 27.9x, P. menziesii: y = 160 − 55.9x. Regression lines base on the pooled data of the two summers. Significant R2 and p values are given for each line per species

Full size image
Fig. 3
figure 3

Seasonal variation in daily stomatal conductance minimum (gMD) and leaf water potential minimum (ΨMD), according to measurements between 12 p.m. and 4 p.m. on several days from May until September 2018 and 2019. Each point shows one value per day per tree, whereas values of gS are means of 5 replicate leaves per tree. R2 and p values are given for each curve per species per year. Dots and lines in green: F. sylvatica, in blue: P. abies, in red: P. menziesii

Full size image

Diurnal ΨLeaf drops were in F. sylvatica usually associated with reductions in stomatal conductance (positive ΨLeaf − gS relationship), but negative relations were also recorded in the summers 2018 and 2019, demonstrating the complexity of this interaction (Fig. S3 in the Supplement). Both negative and positive ΨLeaf − gS relationships were also observed in P. abies and P. menziesii on the diurnal scale, but the gS range was generally smaller than in F. sylvatica. Positive relationships prevailed in the conifers especially in late summer (August and September). All three species showed a remarkable increase in maximum gS in 2019 compared to 2018 (F. sylvatica: from ~ 120 to ~ 215 mmol m−2 s−1; P. abies: from ca. 50 to ca. 160 mmol m−2 s−1; P. menziesii: from 47 to 125 mmol m−2 s−1, Fig. S3).

Pressure–volume curve parameters

Sun-crown foliage sampled during peak drought in late August 2018 showed for the three species remarkably similar leaf tissue osmotic potentials at full turgor (π0; range of species means: −2.05 to − 2.12 MPa) and at turgor loss (πtlp − 2.44 to − 2.51 MPa), relative water content at turgor loss point (RWCtlp; 0.88–0.90) and of the bulk tissue elastic modulus at maximum turgor (ɛmax; 15.04–21.43 MPa) (Fig. 4). Significant differences between species were only found for ɛmax, and a weak tendency toward a more negative πtlp in P. abies.

Fig. 4
figure 4

Osmotic potential at a full turgor (π0) and b at turgor loss point (πtlp), c relative water content at turgor loss point (RWC), and d bulk tissue modulus of elasticity at full turgor (ɛmax) (n = 6). Different capital letters indicate significant differences between species (p < 0.05)

Full size image

Xylem embolism resistance and stomatal closure

Although the P50-values were fairly similar among the three species, the vulnerability curves revealed a tendency toward earlier embolism onset (P12) in the conifers than in F. sylvatica in the 2018 samples (Fig. 5). However, after a shift toward less negative P12-values (and in F. sylvatica also: P50-values) in 2019, these species differences diminished. While the conifers showed only minor inter-annual change in embolism resistance, it appears that F. sylvatica was less resistant in 2019 than in 2018 (Table 2).

Fig. 5
figure 5

Relationships between stomatal conductance (gS) and leaf water potential (ΨLeaf) (black dots, right y axis) (measured between 10 a.m. and 4 p.m. from June to September 2018 and 2019), and between percentage loss of hydraulic conductivity in branch xylem (PLC) and leaf water potential (ΨLeaf) (vulnerability curves at the species level; colored curves; left y axis) for the three species in 2018 and 2019. Each point represents the mean of 5 replicate leaves per tree. Shaded areas around VCs indicate 95% confidence interval of means. Dashed lines indicate P12- and P50-values, i.e., xylem pressures with 12% or 50% loss of conductivity, and the solid line shows the point of stomatal closure. Vulnerability curves and P12- and P50-values are means of the three measured trees per species and the each 3–4 replicate samples per tree. We assumed stomatal closure to have occurred at gS values equaling 5% of maximum gs recorded in the two years (gs95%). Stomatal closure occurring close to P12 hints at possible interactions between stomatal closure and the onset of embolism

Full size image

In summer 2018, stomatal closure (assumed for gs95%, i.e., 5% of maximum measured gs) occurred at leaf water potentials close to the P12-value in F. sylvatica (~ − 3.15 MPa) and P. menziesii (~ − 2.8 MPa), while P. abies closed its stomata well in advance of the onset of embolism (~ − 2.15 MPa). The P12-value shift to higher xylem water potentials and, thus, greater embolism vulnerability in F. sylvatica in 2019 resulted in a drop of the ΨLeaf value at stomatal closure below the species’ P12 threshold, even though F. sylvatica closed its stomata at higher leaf water potentials in 2019 (~ − 2.85 MPa) than in 2018 (~ − 3.20 MPa). In 2019, P. abies closed stomata later (~ − 2.5 MPa) than in 2018 (~ − 2.15 MPa), which, therefore, corresponded with the shifted onset of embolism. P. menziesii apparently closed its stomata also more sensitively in 2019 (at ~ − 2.65 MPa) than in 2018, maintaining the association of stomatal closure with the P12-value (Fig. 5).

In 2018, hydraulic safety margins (HSM) were largest in P. abies (1.42 MPa), intermediate in F. sylvatica (0.99 MPa) and smallest in P. menziesii (0.69 MPa) (Fig. 6). However, HSM changed from 2018 to 2019 due to shifts in P50 and observed Ψmin-values. While P. menziesii kept its HSM relatively constant with only a slight increase to 0.85 MPa, P. abies and F. sylvatica reduced their HSM to 1.13 MPa and 0.58 MPa, respectively, leaving F. sylvatica in 2019 with the smallest HSM of the three species (Table 2, Fig. 6).

Fig. 6
figure 6

Hydraulic safety margins (HSM) of F. sylvatica, P. abies and P. menziesii in the summers 2018 and 2019. Individual vulnerability curves for each tree are shown as percentage loss of conductivity (PLC, %) against xylem water potential (MPa) (mean of 3–4 samples each, colored lines) and at species level (black line). Shaded areas around VCs indicate 95% confidence interval of means. HSMs were calculated as lowest recorded ΨLeaf-values (Ψmin) minus P50-mean (average of all vulnerability curves of a species)

Full size image

Inter-annual variation in physiological and leaf morphological traits

The comparison of species-level leaf and xylem traits between the two years revealed high inter-annual variation in F. sylvatica, medium variation in P. abies, and low variation in P. menziesii (Table 2). All three species had a significantly lower xylem-specific hydraulic conductivity (KS) in 2019 than in 2018, while leaf area-specific conductivity (KL) increased toward 2019 in the conifers (significant) and in F. sylvatica (non-significant). While P. menziesii showed no inter-annual variation in all other investigated traits, P. abies and F. sylvatica reduced specific leaf area (SLA) and leaf size (A) (only F. sylvatica) but increased Huber value (HV) from 2018 to 2019. P12-, P50- and P88-values were significantly higher (less negative) in 2019 in F. sylvatica but not in the conifers. Hydraulic safety margins (HSM), as calculated with Ψmin, were significantly reduced in 2019 in F. sylvatica and P. abies but not in P. menziesii.

Discussion

Our results on seasonal ΨMD variation during the summers 2018 and 2019 support the conclusion of earlier studies (Backes and Leuschner 2000; Köcher et al. 2009; Leuschner et al. 2021; Walthert et al. 2021) that F. sylvatica pursues a strictly anisohydric strategy, supporting our first hypothesis. In our study, sun-leaf ΨMD varied by > 2 MPa (from c. − 0.7 to − 3.1 MPa) between June and September 2018, and by 2 MPa (− 0.8 to − 2.8 MPa) from May to September 2019. An even larger seasonal leaf water potential amplitude (> 2.5 MPa) has been recorded in Swiss F. sylvatica trees during the extreme 2018 drought (Walthert et al. 2021; Kahmen et al. 2022), when absolute ΨMD minima of − 3.3 MPa were reached in sun-exposed branches. From a literature review, it appears that ΨMD of adult F. sylvatica trees drops below − 2.7 MPa only during exceptional droughts as in 2018 (Leuschner 2020), while seasonal variation in normal years is usually smaller.

While P. abies revealed in 2018 a characteristic isohydric regulation with only minor seasonal drops in ΨMD from July to October (Δ ~ 0.3 MPa), it seems that the species’ isohydric strategy failed with the extension of the drought into 2019, and spruce apparently shifted in summer 2019 toward a more anisohydric behavior with a pronounced ΨMD decrease (Δ ~ 1.0 MPa) from May to September, similar to the potential drop observed in F. sylvatica. The isohydric behavior of P. abies is well documented (Lyr et al. 1992; Hartmann et al. 2013; Leo et al. 2014; Oberhuber et al. 2015) and meets the assumption that conifers often have a more stringent stomatal regulation with earlier stomatal closure upon water deficits than angiosperm trees (Zweifel et al. 2007; McDowell et al. 2008; Carnicer et al. 2013). However, ΨMD has been found to vary largely between sites and years across European P. abies populations, suggesting that the stringency of leaf water potential regulation may depend partly on local environmental conditions. We measured in P. abies ΨMD minima of − 2.35 and − 2.40 MPa in 2018 and 2019, which compares well with the minimum of − 2.50 MPa recorded by Lu et al. (1995) in the Vosges Mountains, France. In contrast, Schuldt et al. (2020) reported ΨLeaf minima up to − 4.20 MPa in this species in the very dry summer of 2018 in certain Central European stands, which later succumbed to death.

For P. menziesii, evidence of both isohydric (Bansal et al. 2015; Voelker et al. 2018) and more anisohydric behavior (Phillips et al. 2002; Warren et al. 2003; Jansen 2017) has been reported. Our results show a ΨMD decrease by ~ 0.6 MPa in summer 2019 and hint at a larger ΨMD drop also in early summer 2018 (for which we have no data), suggesting a more anisohydric regulation, even though less pronounced than in F. sylvatica. The few available water potential measurements on mature P. menziesii trees show only moderate seasonal reductions in ΨMD to around − 2.2 to − 2.6 MPa (Brix and Mitchell 1986; Andrews et al. 2012) or ~ − 2.1 MPa (Running 1976) in North American stands differing in annual precipitation, suggesting an intermediate position on the iso/anisohydric axis.

It appears that tree species can exhibit considerable intraspecific variation in the degree of isohydry and, therefore, typically cannot be characterized as strictly anisohydric or isohydric (Cocozza et al. 2016). It seems likely that most tree species with a somewhat broader hydrological and climatic niche are exhibiting a certain degree of plasticity in their stomatal regulation strategy, allowing for shifts along the isohydry–anisohydry continuum in dependence on water availability, temperature, development stage and other cues. This suggests adopting a more dynamic picture of the stomatal regulation strategy of trees than is conventionally done.

All three species showed a relatively small stomatal conductance in the two summers relative to literature data, reflecting the exceptional drought and heat in 2018 and 2019. gS values of F. sylvatica sun-crown leaves remained < 150 mmol m−2 s−1 in 2018 and < 200 mmol m−2 s−1 in 2019, which is well below the range of reported daily maximum gS values of up to 300 mmol m−2 s−1 for this species (Leuschner 2020; Leuschner et al. 2021). Similarly, P. abies and P. menziesii reached only relatively low maximum gS values in 2018 and 2019 at our site (in most cases < 180 and < 130 mmol m−2 s−1, respectively). In the exceptionally dry 2018 summer, P. abies showed gS maxima of only 68 mmol m−2 s−1 and maintained very low midday stomatal conductance (gMD) values throughout the summer (12.3 ± 11.1 mmol m−2 s−1). Since our measurements began in 2018 in late June, it is possible that the conifers started the growing season with somewhat higher stomatal conductance, but reduced gs rapidly at the onset of summer drought.

The relationship between gS and ΨLeaf is often described by a positive sigmoid function (Klein 2014), when potential decreases lead to, or coincide with, a gS reduction. Positive relations should be more clearly visible upon isohydric regulation, while strict anisohydric behavior may lead to negative relations, in which higher transpiration causes ΨLeaf to decrease, as has been reported for F. sylvatica (Leuschner et al. 2021). Our data show for the dry summers 2018 and 2019 in all three species not only positive, but also negative, diurnal gS − ΨLeaf relationships, hinting at complex interactions between turgor- and/or ABA-regulated feedback and VPD-driven feedforward regulation of gS (McAdam et al. 2016; Franks et al. 2017). A dominant positive gS − ΨLeaf relationship appeared, when day-to-day changes during a drying cycle were considered, as expressed by the positive gMD − ΨMD relation. Over a few days to weeks, increasing drought intensity reduced both gS and ΨLeaf in F. sylvatica as well as in P. menziesii (no relation appeared in P. abies).

Our stomatal conductance data produced no evidence for the assumption that F. sylvatica down-regulates gS less sensitively in response to increasing VPD than P. abies, on the contrary. This questions a close association between isohydric behavior and stringent stomatal regulation (Martinez-Vilalta and Garcia-Forner 2017; Leuschner et al. 2021).

While the global data set of Choat et al. (2012) and other published data suggest that P. menziesii and P. abies should possess a more embolism-resistant branch xylem than F. sylvatica, this is not reflected in our data. The three species showed in both study years fairly similar P50-values in the range of − 3.3 to − 3.8 MPa. The differences in P50 between F. sylvatica and P. menziesii were significant in 2018, despite considerable between-tree variation in the vulnerability curves of P. menziesii especially in 2019. Especially for P. menziesii, other studies have reported far more negative P50-values than observed here (e.g., McCulloh et al. 2014: <  − 5 MPa). Such contrasting outcomes of species comparisons are likely explained by considerable inter-population variation in the P50-values of P. menziesii (Dalla-Salda et al. 2011, 2014; Eilmann et al. 2013), reflecting the wide climatic niche occupied by this species and high genetic variation. The relatively high P50-values of P. menziesii in our study might partly be a consequence of the relatively deep soils, which can buffer drought effects. In apparent contrast, inter-population variation in P50 of F. sylvatica has been found to be remarkably low, at least across climatic gradients in Central Europe (Wortemann et al. 2011; Schuldt et al. 2016; Weithmann et al. 2022b).

In our study, F. sylvatica and P. menziesii operated with relatively similar hydraulic safety margins (HSMs) of 0.6–1.0 MPa (Ψmin − P50), while the HSM of P. abies was larger (1.1 − 1.4 MPa). The similarity in HSMs of F. sylvatica and P. menziesii is surprising, given the assumed lower capacity of conifers for embolism repair (Johnson et al. 2012; Carnicer et al. 2013). Yet, after the 2018 drought, Arend et al. (2022) did not observe embolism repair in F. sylvatica trees suffering from impaired xylem function. In any case, it appears that adult F. sylvatica trees operate during mild to moderate drought with a HSM of at least 0.1 to 0.8 MPa in their sun-crown branches (Leuschner 2020 and references therein), while Dietrich et al. (2019) give an even higher value of 1.5 MPa. For P. menziesii, higher HSMs than observed here have also been reported (3 MPa, McCulloh et al. 2014). Our HSM data derived from the Ψmin − P50 difference in two extremely dry summers suggest a species ranking in the sequence P. abies > F. sylvatica ≈ P. menziesii, which changes to F. sylvatica > P. abies > P. menziesii, when the Ψmin − P88 difference is used in case of F. sylvatica. This contradicts the widespread assumption that conifers in general operate with wider branch xylem HSMs than angiosperm trees (Choat et al. 2012; Carnicer et al. 2013), disproving our second hypothesis. Due to the wide range of reported P50 and ΨMD values, it appears that hydraulic safety margins are year- and site-dependent. Studies covering a wide range of different environmental conditions would be needed to draw a more general conclusion about species differences in hydraulic safety.

In contrast to the remarkable similarity in branch hydraulic properties and also in the water potential at turgor loss point among the species, the conifers and F. sylvatica differed in the inter-annual variability of embolism resistance. We found only for F. sylvatica a marked shift of P12, P50 and P88 from 2018 to 2019 by 0.3 − 0.6 MPa toward less negative potentials, while only little inter-annual change in branch hydraulic properties was recorded for the conifers. One possible explanation for decreased embolism resistance after the extreme 2018 drought is cavitation fatigue, likely caused by embolism-induced damage to conduit walls and pit membranes (Hacke et al. 2001; Christensen-Dalsgaard and Tyree 2013; Hillabrand et al. 2016), which has been observed in angiosperms, but seems to be rare in conifers including P. abies (Feng et al. 2021). Cavitation fatigue as a possible explanation for drought legacy effects has not yet been shown for F. sylvatica, but our results seem to indicate its existence.

P. abies maintained in both years a larger HSM than the other species, partly due to its stringent stomatal regulation. It closed its stomata in 2018 already 0.5 MPa above the P12 threshold, which agrees with results of other studies reporting a water-saving, cavitation-avoiding strategy in P. abies (Anfodillo et al. 1998; Lu et al. 1995; Sellin 2000). In our and some other studies, ΨMD did not drop below − 2.5 MPa (e.g., Lu et al. 1995; Cochard 1992), thus avoiding critical xylem tensions even in a drought as severe as in 2018. However, P. abies may well suffer catastrophic hydraulic failure, as was observed in summer 2018 in regions such as Switzerland, where cavitation triggered P. abies dieback in subsequent weeks (Schuldt et al. 2020; Arend et al. 2021). However, ψmin reached − 4.2 MPa at this site, greatly surpassing the corresponding P12-value (− 3.08 MPa), which contrasts with our results from the same year. Apparently, P. abies is drought-sensitive especially on shallow soils (Schmidt-Vogt 1977–1989; Modrzyński 2007; Lévesque et al. 2013), where the bulk of fine root biomass is restricted to the uppermost 20 or 30 cm of the profile. The deep sandy soils at our site, which allow somewhat deeper root penetration (Pietig et al., unpubl. data), might well explain the species’ low vulnerability in this stand. Thus, P. abies may well be vulnerable to hydraulic failure and related dieback, despite its more isohydric behavior. This conclusion matches observations in a throughfall exclusion experiment in a mixed F. sylvatica-P. abies stand, where P. abies was more vulnerable to drought than F. sylvatica (Pretzsch et al. 2020; Grams et al. 2021).

Another important result is that HSMs were always positive in P. abies and P. menziesii, despite extreme drought in both summers, supporting the notion that stomatal closure normally prevents xylem pressure from traversing the steep section of the vulnerability curve (Meinzer et al. 2009). Yet, the negative Ψmin − P12 difference in F. sylvatica in summer 2019 indicates that emboli likely had formed at this instance in this species. Clearly, our results inform only about the species’ response to the exceptional 2018/2019 drought, but do not reflect the behavior in climatically average years, which is better understood from the findings of other published studies.

Even though P. menziesii exhibited in 2018 an isohydric leaf water status regulation with the lowest gS of all species, keeping stomata nearly closed throughout summer, the species operated with the smallest HSM close to its hydraulic limit. From the end of May onwards, the lower sun crown was hit by mealybug infestation (probably Douglas-fir mealybug, Puto profusus), and the needles discolored from mid of July onwards. This might have weakened the trees’ defense due to impaired carbon assimilation and/or hydraulic failure. However, the P. menziesii trees did not suffer from needle shedding or crown dieback in the subsequent months and years. P. abies and F. sylvatica, in contrast, were neither affected by widespread pathogen attack nor pre-senescent leaf/needle fall. However, F. sylvatica responded to the 2018 drought with the largest shift in leaf morphological and branch hydraulic traits, greatly decreasing leaf area and SLA and increasing Huber value in 2019, while reducing HSM. It is well established that F. sylvatica exhibits high plasticity in leaf morphological and/or hydraulic traits, which may increase drought acclimation (Schuldt et al. 2016; Weithmann et al. 2022a, b). In contrast, inter-annual trait variation was smaller in the conifers, probably in part due to the evergreen habitat. Constancy in morphological and anatomical traits was particularly striking in P. menziesii.

Conclusions

Our study during the exceptional 2018/2019 drought confirms the principally more anisohydric behavior of F. sylvatica and the more isohydric strategy of P. abies and P. menziesii. Nevertheless, our results contradict widely held assumptions of plant hydraulics, notably (1) the linkage of anisohydry to lose stomatal regulation (which is not the case in F. sylvatica), and (2) the existence of larger HSMs in conifers than in angiosperms (which is not true for P. menziesii at this site). Comparison with literature data indicates considerable spatial variation in the species’ hydraulic safety margins, which can relate to edaphic and climatic gradients and genetic differences between populations, and may be due to the fact that our data reveal HSM minima, as they are encountered during extreme drought. Extrapolating our results from a sandy site to other edaphic conditions should be done only with great caution, as especially root system structure can differ between sites and populations. Rooting patterns and root physiology likely have a large and insufficiently understood influence on tree hydraulics and canopy water relations.

Our species comparison demonstrates that the ‘conventional’ analysis of branch xylem hydraulic safety, HSMs and stomatal regulation strategies is not sufficient for assessing the species’ drought vulnerability at this sandy site. Moreover, we were not able to confirm the numerous literature reports about the high drought susceptibility of P. abies, nor the often-assumed higher drought resistance of P. menziesii (e.g., Vitali et al. 2017; Thomas et al. 2022). We conclude that a more comprehensive vulnerability assessment of these tree species has to consider additional traits such as rooting depth, shoot and root desiccation tolerance, the drought and heat sensitivity of photosynthesis and radial growth, evidence of drought-induced carbohydrate depletion, and the recovery potential after damage, among others.

Author contribution statement

CL conceived the specific research idea, while CA conceived the interdisciplinary research group, CL and CA received the funding, KS conducted the field work and analyzed the data, MF collected part of the field data, CL and BS supervised the field work, and KS and CL wrote the paper. All authors commented on the text and approved it.