Subalpine dwarf shrubs differ in vulnerability to xylem cavitation: An innovative staining approach enables new insights

Dwarf shrubs are a key functional group of the high-elevation vegetation belt. Despite their ecological relevance and high sensitivity to environmental changes, the hydraulic adaptations and species-specific variations in drought tolerance within this growth form are mostly unknown. Here, we assess the xylem vulnerability to cavitation of important character species of the Alpine dwarf shrub heaths in Central Europe. Due to the high percentage of nonfunctional xylem areas in these species, vulnerability curves were gained by an innovative staining approach with safranin, validated by hydraulic and xylem anatomical measurements. The loss of 50% con-ducting xylem area appeared in the range of (cid:1) 1.78 to (cid:1) 2.91 MPa. Midday plant water potential during an intense summer drought remained above these critical thresholds but was correlated with cavitation resistance. No trade-off between hydraulic safety and efficiency was detected across analyzed species. We conclude that the characteristic clustered occurrence of species in the heterogeneous mountain terrain (mainly interpreted as a consequence of varying snow cover dynamics) is also reflected in species-specific adjustments in xylem hydraulics. The interspecific variety in vulnerability thresholds and plant water potentials during summer drought indicates different hydraulic strategies and adjustments in water relations between these co-occurring shrubs.


tion of important character species of the Alpine dwarf shrub heaths in Central
Europe. Due to the high percentage of nonfunctional xylem areas in these species, vulnerability curves were gained by an innovative staining approach with safranin, validated by hydraulic and xylem anatomical measurements. The loss of 50% conducting xylem area appeared in the range of À1.78 to À2.91 MPa. Midday plant water potential during an intense summer drought remained above these critical thresholds but was correlated with cavitation resistance. No trade-off between hydraulic safety and efficiency was detected across analyzed species. We conclude that the characteristic clustered occurrence of species in the heterogeneous mountain terrain (mainly interpreted as a consequence of varying snow cover dynamics) is also reflected in species-specific adjustments in xylem hydraulics. The interspecific variety in vulnerability thresholds and plant water potentials during summer drought indicates different hydraulic strategies and adjustments in water relations between these co-occurring shrubs.

| INTRODUCTION
Dwarf shrubs are an important functional group of high mountain vegetation, and several essential character species of subalpine and alpine plant communities belong to this specific growth form (Körner, 2003). Due to climate and land-use changes, an expansion of dwarf shrubs was reported for several mountain areas (e.g., Rixen et al., 2010;Tasser & Tappeiner, 2002), with multiple impacts on ecosystem structure and functioning (Maestre et al., 2016). In the European Alps, many sociophytical and ecophysiological aspects of dwarf shrub heaths were intensively studied due to a long tradition in environmental and botanical investigations (Grabherr, 1980;Tranquillini, 1976). The ecological relevance of this vegetation type and its high sensitivity to environmental changes were demonstrated (e.g., Körner, 2003;Larcher, 2003). However, important hydraulic traits of these plant communities, such as water regulation strategies, xylem water transport, xylem anatomical characteristics, and species-specific drought resistance, have rarely been investigated. A few studies dealt with stomata regulation and cuticular transpiration (Ganthaler & Mayr, 2015a, 2015bPisek & Winkler, 1953); root traits and belowground biomass (Larcher, 1977), in situ water potentials during summer (Anadon-Rosell et al., 2017;Körner & Mayr, 1981), and xylem vulnerability and efficiency (Ganthaler & Mayr, 2015a, 2015b, but they were mostly limited to single species. At high altitudes, plant life is commonly less constrained by moisture limitation than at low altitudes (Roe, 2005); however, local water shortages can periodically occur. These water shortages are especially prevalent in skeletal soils with thin humus layers and can amplify the effects of the overall extreme physical environment on mountain plants (Körner, 2003). These harsh environmental conditions include low temperature, persisting snow cover, intensive radiation and high wind load, and also apply to the characteristic dwarf shrub heaths in the Central Alps (Cernusca, 1976). Dominant species of these plant communities mainly belong to the family Ericaceae, which show clonal reproduction and form extensive networks of subterranean rhizomes.
However, the species differ in leaf habit (deciduous or evergreen), leaf characteristics (i.e., leaf area, cuticula thickness, stomata density), soil requirement (silicate or calcareous), rooting depth, and growth size (creepers to upright plants of 60 cm height; Larcher, 1977), and thus also very likely differ in their water relations.
Although individual species frequently coexist on Alpine sites, their distribution is strongly shaped by the topography, which was shown to be linked with snow cover height and duration during winter (Carrer et al., 2019;Neuner, 2014). There is evidence that individual dwarf shrub species suffer from drought stress if the snow melts earlier in the season (Rixen et al., 2010) and that varying snowpack dynamics affect the vegetation composition. In the specific case of subalpine dwarf shrub heaths, a strong snow persistence gradient was detected from Vaccinium myrtillus stands with on average 6-month snow cover to wind-exposed Kalmia procumbens carpets where snow persists only 1 month (Körner, 2003). Importantly, the topography of mountain terrains also leads to an uneven distribution of precipitation and water melt during the growing season, for instance, between ridges and depressions. It should also be noted that Alpine plants create their own microclimate based on the plant stature (such as cushions or dense carpets), with large effects on species-specific evaporative losses (Cernusca, 1976).
The relevance of summer drought and drought resistance in Alpine dwarf shrubs is largely unknown. Few studies dealt with Alpine shrubs, such as Rhododendron ferrugineum, Rhododendron hirsutum, and Juniperus communis (Beikircher & Mayr, 2008;Mayr et al., 2010;Unterholzner et al., 2020). The paucity of published data on both shrubs and dwarf shrubs is likely due to the scarce economic interest in shrubs compared to trees, concerning all habitat types, but also due to methodological challenges. The small dimensions of the twigs, the dense branching pattern, and a high proportion of nonfunctional xylem areas in dwarf shrubs (see more details below) make some hydraulic measurements (e.g., centrifuge technique; Cochard et al., 2013) unfeasible, and most other classical hydraulic methods (compare Sergent et al., 2020) are not easily applied or require specific adaptations.
The lower growth height of dwarf shrubs may imply different requirements for a safe and efficient water transport compared to trees (see, e.g., Tyree & Ewers, 1991, Patiño et al., 1995, Beikircher & Mayr, 2008, and hydraulic studies on this growth form may provide new insights into the physiology of plant water transport by presenting aberrations from widely analyzed model species. Moreover, dwarf shrubs are adapted to extraordinary climatic and environmental conditions at high elevation. Shallow and rocky soils coupled with strong exposure to wind and radiation on these sites reduce the soil water retention capacity and increase transpiration rates. Thus, the mosaic-like distribution of dwarf shrub species may also be related to the small-scale variation in water availability and in evaporative losses between ridges and depressions. Importantly, winter stress may require further morphological and anatomical adaptations on the whole plant level, which affect the water balance during both winter and summer and lead to specific xylem hydraulic adjustments, including the resistance to cavitation. The relevance of winter stress for plant hydraulics at higher elevation has been shown for several tree and shrub species (e.g., Mayr et al., , 2019. Preceding studies on Vaccinium species revealed a rather low resistance to cavitation compared to co-occurring trees (Ganthaler & Mayr, 2015a, 2015b) and a surprisingly high proportion of nonfunctional xylem areas in several dwarf shrub species (A. Ganthaler et al., unpublished). The latter can cause methodical problems, as nonfunctional conduits may not be permanently blocked, and pressures used in classical hydraulic measurements (e.g., Sperry method;Sperry et al., 1988) can reactivate nonfunctional conduits and lead to artificially increased hydraulic conductivities. Staining approaches, based on perfusion of stem segments with a dye and subsequent determination of the stained, therefore conductive, cross-sectional xylem area, have already been shown to be a reliable alternative method (Hietz et al., 2008, Nolf et al., 2016A. Ganthaler et al., unpublished) to quantify the percent loss of conductivity (PLC) with progressive dehydration. Although the approach is comparably time-consuming, it has further advantages as it enables insights into distribution patterns of cavitation events.
In the present study, we investigated (1) whether stem hydraulic vulnerability of dwarf shrubs species can be accurately determined by staining conductive xylem elements with safranin and (2) to what extent the vulnerability to drought-induced cavitation varies across co-occurring species in Alpine dwarf shrub communities. We hypothesized that (3) xylem vulnerability is related to the species' moisture indicator values and/or (4) to the experienced water potential (Ψ) during summer drought periods (Ψ drought ). shrub heath above the treeline on silicate soil (Grabherr, 1980;Larcher, 1977). It is characterized by large populations of A. uva-ursi, C. vulgaris, K. procumbens, V. gaultherioides, and V. myrtillus, co-occurring with R. ferrugineum shrubs. The second site in a calcareous area (Höttinger Bild, 980 m; 47 17 0 N; 11 22 0 E) was selected to include also the calcicole species E. carnea. A detailed characterization of analyzed species, including indicator values for moisture, light, temperature, continentality, and nutrients, is given in Table S1.
Measurements were performed between June and September 2016. For hydraulic analyses, whole plants, including major roots, were collected on the field sites and transported with their roots in water and covered with a dark plastic bag immediately to the laboratory, where they were rehydrated overnight. Measurements were made on healthy, several-year-old plants with a stem diameter between 2.5 and 4.5 mm.

| Vulnerability curves
Saturated plants of A. uva-ursi, C. vulgaris, E. carnea, and K. procumbens (n = 14-21 per species) were dehydrated on the bench (Cochard et al., 2013;Sperry et al., 1988). At increasing levels of dehydration, whole plants were equilibrated in dark plastic bags and plant Ψ was determined on one side branch, up to 10 cm long, with a pressure chamber (model 1505D pressure chamber; PMS Instrument). One stem section per plant, 3-5 cm long and at least 3 cm distant from the basal end was excised F I G U R E 1 The study area close to Innsbruck (Tyrol) with typical subalpine dwarf shrub heaths and the growth and leaf habit of analyzed species under water. Therefore, samples were decorticated, and the ends were recut about 5 mm for several times (in total at least 1.5-2 cm) with a sharp wood carving knife to gradually release tension and remove microbubbles (Venturas et al., 2015). It was previously shown (Ganthaler & Mayr, 2015a, 2015b) that this sample preparation protocol causes no cutting or rehydration artifacts (see Trifil o et al., 2014;Wheeler et al., 2013) in Alpine dwarf shrubs, which are characterized by few mm long vessels.
Prepared samples were then connected to the safranin or water reservoir as described in the following sections.

| Staining conductive xylem with safranin
Samples were sealed in a hydraulic system (modified after Sperry et al., 1988) connected with a reservoir filled with 0.1% (wt/vol) filtered (0.22 μm) safranin. The pressure was set to 5 kPa, and samples remained connected until the outflow was deeply red stained. After a drying period of 30 min at room temperature, cross sections were made from the middle part of the stained stem sections with a slide microtome (Schlittenmikrotom G.S.L. 1, Schenkung Dapples) and analyzed with a light microscope (Olympus BX41; Olympus Austria) interfaced with a digital camera (ProgRes CT3, Jenoptik). Pictures were evaluated with the software ImageJ (ImageJ 1.45; public domain, National Institutes of Health) using the included tool "color thresholds" to determine the stained xylem area (A s ) and total xylem area (A t ).
As analyzed species have a high proportion of functionally inactive (permanently blocked or air-filled) xylem elements (A. Ganthaler et al., unpublished), these nonfunctional areas (A nf ) had to be excluded from the analysis in order to exactly determine the loss of conductive area with progressive dehydration. This included mainly the innermost yearrings for A. uva-ursi, K. procumbens, E. carnea, and the latewood of each year-ring in C. vulgaris, but also parenchyma rays (mainly E. carnea) and areas with clear wound reaction and deposit of secondary compounds (all species; see also Figures 2 and 3; see also Results section). All these xylem parts never stained, also in saturated samples, and did not contribute to conductivity. As the pattern and extent of these areas vary between species, they had to be excluded manually, based on the visible staining pattern and extensive experience with the determination of conductive xylem in saturated samples of these dwarf shrub species by staining and x-ray microtomography (microCT) scans (A. Ganthaler et al., unpublished). The percent loss of conductive area (PLA) was finally calculated as:

| Hydraulic and wood anatomical control measurements
To validate the staining method, a reference vulnerability curve was measured by hydraulic flow measurements for E. carnea, as in this species, nonfunctional xylem areas were shown to be permanently The initial hydraulic conductance (K i ) was measured at 5 kPa.
Samples were then flushed for 20 min at 70 kPa (Ganthaler & Mayr, 2015a) to remove embolism, and the hydraulic conductivity was measured again. Flushing was repeated until measurements showed no further increase in conductivity to obtain final hydraulic conductance (K f ). The PLC was then calculated as follows: Furthermore, xylem anatomical measurements were performed to assess the comparability of PLA and PLC and to validate vulnerability curves. Mean conduit diameter (d), mean hydraulic diameter (d h ; calculated according to Kolb & Sperry, 1999), and diameter size Dotted lines indicate the water potential at 12% and 88% PLC/PLA. Shaded areas represent the 95% bootstrapped confidence interval for fitted curves. No significant differences were observed distribution were determined for each species and separately for stained and unstained xylem areas on three samples per species with a PLA of about 50%. Measurements were performed according to Mayr (2015a, 2015b) on 641-985 conduits per species using the software ImageJ.

| Additional data
In order to get a comprehensive overview of plant vulnerability to xylem cavitation in the analyzed dwarf shrub communities and to improve the significance of correlation analyses with Ψ following a drought period (Ψ drought ) and specific hydraulic conductivity (k s ), a data set comprising both data gained in this and previous studies was used.
Additional data were extracted from Ganthaler and Mayr (2015b)

| Statistics
For vulnerability analyses, PLC/PLA was plotted versus the corresponding Ψ (data pooled per species) and a Weibull regression curve was fitted to each vulnerability curve (R package FIT-PLC, R i386 3.2.5; Duursma & Choat, 2017). Then Ψ at 12%, 50%, and 88% PLC (Ψ 12 , Ψ 50 , Ψ 88 ) was calculated. Differences between methods (hydraulic analyses versus staining conductive xylem, for E. carnea) and between F I G U R E 3 Determined conductive (red) and embolised (gray) xylem areas of stems of Erica carnea following progressive dehydration. Shown are seven representative measurements of the curve plotted in Figure 1a, including the original microscopic image (A) and determined xylem areas by image analysis (B). Note that nonfunctional xylem areas including central pith, parenchyma rays, and innermost inactive year rings were excluded from analysis (white areas) species were assessed using 95% confidence intervals obtained via bootstrap resampling (performed in R STUDIO). Student's t-tests and correlation analysis (Ψ 50 vs. Ψ drought and Ψ 50 vs. k s and k sc ) were performed using SPSS v.24.0 (SPSS Inc.) at a probability level of 5%.

| Vulnerability to xylem cavitation
Vulnerability curves of E. carnea obtained with the presented staining approach did not significantly differ from control measurements gained with the classical hydraulic method (Table 1, Figure 2).
Although the vulnerability curve obtained from staining conductive xylem was slightly steeper, the differences in Ψ 50 were negligible (0.14 MPa). Figure (Table S2) and no significant differences between conductive and embolized xylem areas ( Figure S1).

| Correlation with hydraulic conductivity
Within the analyzed group of species, no correlation between the vulnerability to xylem cavitation and the specific hydraulic conductivity (k s , related to the whole xylem cross-sectional area) or the corrected specific hydraulic conductivity (k sc , related to the functional xylem T A B L E 1 Vulnerability to xylem cavitation of analyzed dwarf shrub species obtained by hydraulic measurements (flowmeter) and staining conductive xylem (staining) in the present study and by Mayr (2015a, 2015b)  area) was detected (Figure 7). Notably, A. uva-ursi, the species with an exceptional high proportion of non-functional xylem areas, showed comparably high k sc .

| DISCUSSION
The present data set demonstrates a range in vulnerability thresholds of about 1 MPa across studied dwarf shrubs, whereby vulnerabilities correlated with midday Ψ drought but not with k s or k sc . A direct comparison of the vulnerability curve based on the proportion of stained xylem with the curve gained by classical hydraulic measurements (performed for E. carnea; Figure 2) proved that the staining method revealed reliable results. The validity of the method for all analyzed species was underlined by the small species-specific differences in d and conduit size distribution, and by consistent d h in conductive and embolized xylem areas ( Figure S1; Table S2).
The latter gives evidence of equivalent conductive capacity in stained and unstained xylem areas and thus good agreement between PLA and PLC.
The approach thus seems like a valuable alternative or complementary method to measure xylem vulnerability (as previously shown for conifers; Mayr & Cochard, 2003, Hietz et al., 2008 also for angiosperm species with their complex anatomical wood structure. The slightly diverging curve  Water potential (MPa) F I G U R E 6 Vulnerability to xylem cavitation (water potential at 12% and 50% loss of conductivity/cond. area, Ѱ 12 and Ѱ 50 ) versus the water potential measured in situ after a drought period (Ѱ drought ). Shown are mean ± SE for each species (Au, Arctostaphylos uva-ursi; Cv, Calluna vulgaris; Ec, Erica carnea; Kp, Kalmia procumbens; Vg, Vaccinium gaultherioides; Vm, Vaccinium myrtillus) and the linear regression line. Areas below the 1:1 line are shed in yellow shape between the two methods may have different explanations: Firstly, the diffusion of dye can lead to partial staining of non-conductive xylem areas, and thus to an underestimation of embolized areas at moderate Ψ. Secondly, dye crosses and thereby stains larger vessels faster than smaller ones. This makes determination of the staining endpoint difficult and may cause an overestimation of embolized xylem areas at low Ψ. Though both aspects had a comparably low impact on the position of the vulnerability curve and no significant effect on the determined 50% threshold (Ѱ 50 ; Table 1). The third explanation that PLA does not exactly correspond to respective PLC as conductivities vary with conduit diameter and conduit frequency (McCulloh et al., 2010) can be excluded for analyzed species based on the xylem anatomical measurements (see above). Thus the news staining method enables reliable analyses of xylem vulnerability thresholds. However, the image analysis requires substantial experience and knowledge on the potential presence and distribution patterns of nonfunctional xylem areas in hydrated plants (compare Measurements with the staining approach on A. uva-ursi, E. carnea, C. vulgaris, and K. procumbens (Table 1 and Figure 4) revealed Ѱ 50 values comparable to thresholds of previously analyzed Vaccinium species (À1.97 to À2.70 MPa; Ganthaler & Mayr, 2015a, 2015b and close to co-occurring R. ferrugineum shrubs (À2.96 MPa; Mayr et al., 2010). The overall resistance to cavitation was rather low compared to co-occurring trees such as Pinus cembra (Ѱ 50 of À4.12 MPa) or Picea abies (Ѱ 50 of À3.52 MPa; . These findings strengthen the assumption that dwarf shrubs in Alpine environments in general exhibit a rather low cavitation resistance. On one hand, this characteristic may be related to their growth stature, as the reduced growth height leads to comparably low cumulative hydraulic resistance along the pathway. Thus, compared to trees, smaller Ψ gradients along the axis (Tyree & Ewers, 1991) and, consequently, less negative Ψ minima appear. It is not uncommon that plants with shorter stature exhibit low resistance to cavitation, often combined with low hydraulic efficiency (Gleason et al., 2016). Such a combination with lacking trade-off between xylem hydraulic efficiency and safety also applied to the analyzed dwarf shrubs (Figure 7). On the other hand, Alpine dwarf shrubs exhibit a large set of specific ecophysiological adaptations regarding nonxylem traits affecting their water relations. For instance, they are characterized by a 2-3 times higher belowground than aboveground biomass (Larcher, 1977). The extensive network of long and deep roots improves the water uptake and counterbalances the limited water retention capacitance of mountain soils and the modest annual precipitation on the study sites (about 800 mm; ZAMG, 2020), even for species growing on exposed, shallow-grounded places such as K. procumbens. Deeper rooting makes dwarf shrubs less dependent on the water content of upper soil layers and less vulnerable to drought conditions than, for example, herbaceous species (Anadon-Rosell et al., 2017). Dwarf shrubs are also characterized by a lower leaf area index and growth rate compared to trees (Körner, 2003), which may facilitate a water saving strategy.
This overall favorable hydraulic situation, as also demonstrated by only moderate Ψ even in a dry period (see Section 3), points to a rather unproblematic water regime in these plant communities, at least during summer. Similarly, Larcher (1977), Körner and Mayr (1981), and Anadon- Rosell et al. (2017) reported that Ψ of these species rarely drops below À1.5 to À1.8 MPa on sunny summer days, even after long periods without rain. Thus, plants do not reach critical Ψ for cavitation, despite the relatively high xylem vulnerability, but are operating with rather narrow hydraulic safety margins ($0.6-1.2 MPa) against critical levels of drought stress, comparable to several forest species across the world (compare Choat et al., 2012). Notably, A. uva-ursi and V. gaultherioides exhibited a significantly wider safety margin between Ψ 50 and Ψ drought compared to the other species. Within the analyzed species group, the vulnerability to drought-induced cavitation was correlated with Ψ drought (Figure 6), indicating that species-specific xylem characteristics are adjusted to (a) water availability and/or (b) plant water use strategies, which both affect plant Ψ drought .
We found no evidence for the former, as Ψ 50 was not related to the moisture ecological indicator values (Table S1) according to Landolt (2010). For instance, K. procumbens, the characteristic species on exposed ridges with a low indicator value for soil moisture, exhibited a rather high Ψ 50 , while V. gaultherioides, mainly growing in depressions and more humid sites, was significantly more cavitation resistant (compare Table 1).
In contrast, several relationships with plant traits affecting transpirational control and water losses became apparent. Several studies indicate that Alpine plants respond sensitively to the evaporative demand and species start closing their stomata with increasing vapor pressure deficit even before a relevant drop in plant Ψ (Bowman et al., 1995;Johnson & Caldwell, 1975). Moreover, according to gas exchange measurements of Larcher (1977) on four of the analyzed species, CO 2 uptake during dehydration is first limited in K. procumbens (at $18% relative water saturation deficiency) and C. vulgaris and V. myrtillus (at $27%), while V.   Pisek and Winkler (1953) showed that the cuticular transpiration of Alpine plants can vary substantially, for instance, between 5 mg h À1 g À1 fresh weight in A. uva-ursi to 24 mg h À1 g À1 fresh weight in V. myrtillus. It has to be considered that plant water losses are not only affected by leaf and stomata characteristics but are driven by the vapor pressure deficit, which in turn is strongly affected by the growth stature and overall aerodynamic canopy resistance (Körner, 2003). K. procumbens, for example, although it tolerates strong wind exposure on windswept ridges, forms low stature prostrate mats, which enable to maintain a high air humidity within the cushion (Cernusca, 1976). Without this protection, however, it develops quick and severe drought damage (Larcher, 2003), which is in accordance with the comparably high Ψ 50 and steep vulnerability curve found in our study.
No indication was found that deciduous dwarf shrub species operate closer to hydraulic limits compared to evergreen species (risking defoliation under stress), at least by analyzing the safety margin between Ψ 50 and Ψ drought (compare Figure 6). Vulnerability thresholds (Table 1)  with potential negative effects on ecosystems services and stability.

| CONCLUSIONS
The comparably low xylem resistance to cavitation in combination with several root and leaf functional traits optimized for water acquisition and saving indicate that the hydraulic strategy of Alpine dwarf shrubs is oriented toward avoidance of low Ψ rather than withstanding low Ψ. Although several physiological traits require further analysis (e.g., rooting depth, stomata regulation, and recovery from cavitation-induced embolism) this strategy may be critical under changing climatic conditions. Pronounced interspecific differences in xylem vulnerability are probably coordinated with occurring xylem tensions in the field, which in turn are strongly influenced by growth stature, leaf characteristics, and stomata regulation. Remarkably, the high variety of ecophysiological adaptations to the environment reported for these co-occurring species also applies to their water use strategy and coordination of individual hydraulic traits.

ACKNOWLEDGMENTS
The study was supported by an L'Oreal Austria fellowship "For Women in Science" to Andrea Ganthaler and the Austrian Science Fund (FWF) project P29896 and P32203, and was conducted in the frame of the research area "Mountain regions" of the University of Innsbruck. We are grateful to anonymous reviewers for constructive comments that helped to improve the manuscript.

Stefan Mayr and Andrea Ganthaler designed the experiments, Andrea
Ganthaler performed the measurements, analyzed the data, and wrote the manuscript, Stefan Mayr reviewed and complemented the manuscript.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as all new created data is already contained within this article.