A broad survey of hydraulic and mechanical safety in the xylem of conifers

1Evidence for drought-induced forest dieback has been reported worldwide. There is growing evidence that all forest types or climate zones are equally vulnerable to drought events, even in currently mesic environments. Although gymnosperms seem to be on average more resistant to cavitation (; S. Delzon et al., unpublished data) and have greater hydraulic safety margins per se than angiosperms, they are also not immune to drought-induced mortality.

2Resistance to cavitation is a crucial trait in trees to cope with drought stress in addition to, for example, rooting depth, internal water storage, and changes in biomass allocation or leaf anatomy. Indeed, substantial evidence of a direct causal link between drought resistance and cavitation resistance has been highlighted in both conifers and angiosperms. Global surveys of cavitation resistance in woody species have not surprisingly shown that species from xeric climates are more resistant to embolism than species from wet climates. Incorporation of phylogenetic information strengthened these adaptive inferences and suggests that cavitation resistance-related traits are under natural selection. Hence, there is convincing evidence that the geographical distribution of many tree species is determined by their ability to resist drought-induced embolism.

3Drought-induced embolism occurs in the xylem, in which water is transported under tension. While the exact mechanisms remain unknown, drought-induced embolism formation is thought to occur via air leakage from an embolized conduit (non-functional) to a functional conduit. When the pressure in the xylem is sufficiently negative, the rupture of an air–sap meniscus allows propagation of air bubbles through porous interconduit pit membranes. Cavitation resistance would therefore be influenced by the structure and function of bordered pits. Accordingly, variation in xylem anatomy, conduit characteristics, and bordered pits has been associated with cavitation resistance in both angiosperms and conifers.

4Cavitation in angiosperms occurs by air-seeding at the pit membrane level. Under well-hydrated conditions, the pit membrane separating two functional vessels is in a relaxed position (i.e. unaspirated) and sap flows through the pores of the membrane. When water stress occurs, the pressure difference between an embolized and functional vessel leads to the rupture of an air–sap meniscus located within the pit. Embolism formation in angiosperms seems to depend on the size of the largest pores in the pit membranes. In conifers, the intertracheid pits are morphologically characterized by a centrally located torus and a porous margo region. When the pressure difference in xylem increases, the deflection of the torus against the pit aperture seals off the embolized tracheid. This so-called ‘valve effect’ prevents the spread of air into the functional xylem and may depend on the torus diameter relative to pit aperture diameter.

5How does cavitation occur in torus–margo pits? Different mechanisms of air-seeding have been proposed to explain cavitation in conifers (for a review, see ). Recent studies demonstrated that as for angiosperms, cavitation occurs by rupture of an air–sap meniscus in the vicinity of the pit membrane, but the exact location of where the meniscus breaks is unknown. Two mechanisms are likely: (i) air bubbles pass through pores at the edge of the torus when the torus and the inner wall of the pit membrane are not perfectly sealed (seal capillary-seeding hypothesis;;; ); and (ii) the torus structure is not fully impermeable, meaning that air bubbles may pass through tiny pores (torus capillary-seeding hypothesis; ). According to recent studies, the first is more likely in most conifer families, while the second may be an additional mechanism in Pinaceae. Additional anatomical observations of bordered pits in a broader range of conifers are believed to provide further details about how embolism formation occurs in gymnosperms.

6Conifer tracheids are involved not only in water transport but also in mechanical support of the plant. The length of a tracheid, its diameter, and the thickness of its wall are factors contributing to mechanical strength. Based on measurements and theoretical estimations, cavitation resistance is correlated to the ‘thickness to span’ ratio of tracheids. Higher absolute resistance to cavitation is associated with lower negative sap pressure and requires stronger tracheids with a higher thickness to span ratio to resist to mechanical stress. An increase in the thickness to span ratio, probably due to reduced lumen diameter, thus also enhances the resistance to water stress. The physiological consequences of this trade-off between hydraulic safety and mechanical strength has received considerable attention. Therefore, conduit structure is potentially constrained by safety considerations.

7In this study, the linkage between xylem anatomy and resistance to drought-induced cavitation was assessed using a large database of 115 conifer species. By broadly sampling across conifer phylogeny and four major terrestrial biomes, the objectives were (i) to better understand how xylem anatomical properties associated with air-seeding influence the mechanical strength of the xylem; and (ii) to assess the evolutionary trends of xylem anatomy in the selected taxa. It was hypothesized (i) that the anatomy and functional properties of pit pairs strongly influence cavitation resistance in conifer species—more specifically it was believed that the valve effect (product of torus–aperture overlap and margo flexibility) plays a major role in cavitation resistance; and (ii) that increased cavitation resistance is associated with higher mechanical strength.

8Anatomical observations and cavitation measurements were conducted on 60 conifer species. Additional data were retrieved from and (40 and 15 species, respectively) and completed by measuring novel characteristics. A total of 115 species, including seven families and 45 genera (see Table 1, and Supplementary Table S1 available at JXB online), were used to test the relationship between cavitation resistance and anatomical traits. All anatomical observations have been carried out on material that had previously been used for measuring cavitation resistance. These observations were limited to one individual for most species and, in cases where several samples were available, the sample that was closest to the average P 50 value; that is, the xylem pressure inducing 50% loss of hydraulic conductance, was selected. Although the intraspecific variability in pit anatomy can be considerable between organs (roots showed dramatic differences in conduit and pit anatomical properties compared with branches;;; ), here it was assumed that intraspecific variation was smaller than interspecific variation. Moreover, to avoid any additional intraorgan variability, only anatomical features in the xylem of young branches were measured.

9Four to five transverse sections were cut for each sample with a sliding microtome, stained with safranin (1%), and fixed on a microscope slide. Sections were observed with a light microscope (DM2500, Leica, Germany) at the University of Bordeaux. Five photos per section were taken with a digital camera (DFC290, Leica, Germany), and analysed with WinCell (Regent Inst., Canada).

10SEM observations were conducted at Ulm University with a Hitachi cold-field emission scanning electron microscope and with a benchtop scanning electron microscope at the University of Bordeaux (PhenomG2 pro; FEI, The Netherlands). For the Hitachi SEM, thin (1 μm) radial sections were cut in different parts of the stem, air-dried, coated with platinum using a sputter coater (Emitech Ltd; Ashford, UK) for 2min at 10 mA, and observed under 2kV. For the benchtop SEM, samples of 5–8mm long were cut with a razor blade in a radial direction. After drying for 24h in an oven at 70 °C, the samples were fixed on stubs, coated with gold using a sputter coater (108 Auto; Cressington, UK) for 40 s at 20 mA, and observed under 5kV.

11A transmission electron microscope was used to obtain ultrastructural details of pit membranes for 33 species. One stem per species was cut into 1mm3 blocks and dehydrated in an ethanol series. The ethanol was gradually replaced with LR White resin (London Resin Co., Reading, UK) over several days (for more details, see ). Then, transverse ultrathin (between 60nm and 90nm) sections were cut using a diamond knife and collected on 100 mesh copper grids. The ultrathin sections were manually stained with uranyl acetate and lead citrate. Observations were carried out with a JEOL JEM-1210 transmission electron microscope (Jeol, Tokyo, Japan) at 80kV accelerating voltage, and digital images were taken using a MegaView III camera (Soft Imaging System, Münster, Germany).

12All anatomical features related to tracheids and bordered pits (see Table 2) were based on earlywood, which is responsible for most of the hydraulic conductance. Based on light microscopy images (×200 magnification), the tracheid lumen diameter (D T; the simple average of the equivalent circle diameter) and the thickness of the double wall between neighbouring conduits (T W) were measured. Pit membrane diameter (D PM), aperture diameter (D PA), and torus diameter (D TO) were measured using SEM (Fig. 1). The distance between two pit borders (L PB; Fig. 1), the number of margo pores (N MP), the mean and maximum diameter of margo pores (D MP, D MPmax), and the mean diameter of pores in the torus (D TP) (Fig. 2) were measured using SEM images. Because the L PB parameter varies depending on the distribution of pits in tracheids, a constant distribution of pits along the entire length of a tracheid was assumed for these measurements. Average values for all anatomical features were determined based on a minimum of 20 measurements. All anatomical measurements were conducted using ImageJ freeware (W.S. Rasband, ImageJ, US National Institutes of Health, Bethesda, MD, USA, http://imagej.nih.gov/ij/, 1997–2012).

13The following functional properties of pit membranes and tracheids (Table 2) were estimated from anatomical measurements to investigate micro-morphological variation in relation to embolism formation.

14The margo flexibility [F=(D PM–D TO)/D TO], the torus overlap [O=(D TO–D PA)/D TO], and the valve effect (V EF=F×O) were estimated following. D PM is the pit membrane diameter, D PA the pit aperture diameter, and D TO is the torus diameter.

15The pressure difference between two adjacent tracheids required to break the air–water meniscus in the margo was calculated following the Young–Laplace equation:

16where τ (0.0728 N m–1 at 20 °C) is the surface tension of water, α is the contact angle between the microfibrils and the meniscus (assumed equal to 0°), and D MP is the average diameter of the margo pores.

17The pressure difference required to break the air–water meniscus in the torus when it is already sealed against the pit aperture was calculated as:

18where D TP is the mean diameter of a pore in the torus.

19The pressure difference required to aspirate the pit membrane onto the pit border was estimated according to:

20where N is the number of margo strands, which was assigned to the average value of 55 and 200 strands per pit. MOE is the modulus of elasticity of the strands (taken at 3.5 GPa; ) and e A is the margo spoke strain at aspiration (e A=0.03D PM/D PM–D TO).

21The pressure difference needed to break strands of the margo was estimated following:

22where J is the tension of a strand between the aspirated position and stretched position when the torus goes through the pit aperture and no longer covers the whole aperture [J=0.0147(e S–e A)MOE] and e S is the margo spoke strain in the stretched position [e S=(D TO–D PA+0.03D PM)/(D PM–D TO)].

23The hydraulic trade-off associated with cavitation resistance was quantified by calculating the pit aperture resistivity (R PA), margo resistivity (R M), and total pit resistance (R P) following and:

24where υ is the viscosity of the water (0.001 Pa.s at 20°C) and T PA is the thickness of a single pit border calculated from the double wall thickness (T PA=81% of T W; ).

25where N MP is the number of pores in the margo and f(h) is the function of h, the proportion of the margo occupied by pores [h=N MPπ(N MP/2)2/π(D PM/2)2]. The total pit resistance R P equals the sum of R PA and R M (R P =R PA+R M).

26The thickness to span ratio was also estimated as it reflects the tracheid contribution to conductance. It corresponds to T W/D T, where T W is the double wall thickness and D T the tracheid lumen diameter (the simple average of the equivalent circle diameter).

27The conduit implosion pressure was defined as the pressure difference across the tracheid wall required to cause bending stress to exceed the wall strength. It was estimated using methods described by:

28where ω is the strength of the wall material assumed to be 80MPa, and β is a coefficient taken as 0.25. The moment ratio, I H/I S, represents the ratio of the second moment of area of a wall with pit chamber (I H) to that of a solid wall with no pit chamber present (I S). showed that I H/I S does not change with air-seeding pressure, and is on average ~0.95 in conifers. The ligament efficiency [L E=1–D PA/(D PM+L PE)] quantifies the spatial distribution of the pit aperture in the wall.

29Cavitation resistance data were retrieved from; unpublished data). Data analyses were carried out for several cavitation traits, P 50 (xylem pressure inducing 50% loss of hydraulic conductance), S (slope of the vulnerability curves), P 12 (xylem air entry), and P 88 (xylem pressure at which 88% of conductivity is lost) (for more details, see ). As similar results were found for all cavitation parameters, it was decided to present the relationship between anatomical traits and P 50 for simplicity. However, xylem air entry pressure (P 12) was also used to test whether cavitation occurs before or after torus deflection.

30Cross-species correlations between xylem anatomical traits and cavitation resistance were tested with a Pearson correlation coefficient (r) and a Spearman correlation coefficient (s) for non-linear data. In addition, the species distribution in the four major biomes was retrieved from Delzon et al. (unpublished data), and variations of anatomical traits among biomes were assessed using a one-way analysis of variance (ANOVA). Data and statistical analyses were conducted using the SAS software (version 9.3 SAS Institute, Cary, NC, USA).

31Phylogeny can induce bias when testing for correlations in pairs of traits within a group of species. Because of shared evolutionary history between related species, the assumption of independence of classical statistical tests and correlations is disregarded. The phylogenetically independent contrast method (or PIC; ) is a common workaround for this issue: differences (or contrasts) in trait values are computed for each pair of species and each node of the phylogeny; these are statistically independent, and represent the evolutionary divergences in traits at each node (for more detailed information, see; ). Furthermore, the PIC method is used to test for correlated evolution between traits: a significant positive trend (i.e. PICs for trait A are positively correlated with PICs for trait B) means that for each node of the phylogeny a change in the value of trait A is associated with the evolution of trait B. Phylogenetically independent contrasts analyses (PICs; ) were run in R using the ‘ape’ package.

32This method requires knowledge of the phylogeny of the studied taxa with branch lengths. To this end, DNA sequences for three generally available genes (chloroplast genes rbcL and matK, and nuclear gene phyP) were retrieved from GenBank and aligned using PHLAWD (PHyLogeny Assembly With Databases;; http://code.google.com/p/phlawd). Alignements were visually checked and trimmed in MEGA 5.0. Maximum Likelihood (ML) phylogenetic analyses were run in RAxML (version 7.0.3; ). A separate GTR+CAT rate model was assigned to each partition, and each search was conducted 100 times and the best maximum likelihood (ML) tree was retained.

33Inclusion of phylogenetic information was useful to identify adaptive relationships between anatomical properties and cavitation resistance. Below both cross-species correlations between variables analysed with Pearson or Spearman correlation (r or s) coefficients and phylogenetically independent contrasts (PICs) are presented.

34Across species, cavitation resistance (P 50) was negatively correlated with pit aperture diameter (D PA; Table 3) showing that species with narrower pit apertures are more resistant to cavitation. However, these relationships were not supported by correlations using PICs (Table 3). No relationship was found between cavitation resistance and other anatomical traits such as tracheid lumen diameter (D T), torus diameter (D TO), or pit membrane diameter (D PM). The same results were found for correlations with P 12, P 88, and the slope of vulnerability curves (Supplementary Table S2 available at JXB online).

35P 50 was significantly correlated with functional properties of the bordered pit such as torus–aperture overlap (O) and valve effect (V EF; product of torus overlap and margo flexibility) (Table 3, Fig. 3). This relationship was also supported by the PIC analyses, suggesting a correlative evolution between cavitation resistance and pit membrane functional properties. Increasing valve effect (V EF) increased cavitation resistance. Variation in valve effect was mainly due to change in torus–aperture overlap as margo flexibility varied weakly among species. Following cross-species correlations, pit aperture diameter (D PA) contributed more to torus–aperture overlap than to torus diameter (D TO) (Fig. 3). However, the PIC analyses did not confirm this trend (Table 3).

36Based on interspecific analyses, rupture stretching pressure (P RS) was highly correlated to P 50 (Table 3). Margo strength increased with increasing cavitation resistance. However, P RS was always much lower than P 50 (P RS= –4.67 to –71.31MPa; P 50= –2.23 to –15.79MPa; Supplementary Fig. S1 available at JXB online), suggesting that cavitation took place before mechanical rupture of the margo strands. For all species, torus aspiration occurred at a relatively high xylem pressure (close to 0MPa, P TD= –0.03 to –0.33MPa; Supplementary Fig. S1 available at JXB online) as compared with the xylem air entry pressure (P 12= –0.91 to –12.66MPa; Supplementary Fig. S1 available at JXB online). Moreover, the pressure difference inducing margo capillary rupture (P MC) was lower (more negative) than the pressure difference required to deflect the torus towards the pit aperture (P TD) (P MC= –0.10 to –0.75MPa; Supplementary Fig. S1 available at JXB online). This means that torus deflection occurs before an air–water meniscus breaks through the margo of a pit membrane. Cavitation resistance was positively correlated with total pit resistivity (R P) and pit aperture resistivity (R PA; Table 3; Fig. 5A, C) but not with margo resistivity (R M; Table 3; Fig. 5B), suggesting that the margo pores are not involved in cavitation resistance in conifers. However, these trends were not confirmed by the PIC analysis (Table 3). Concerning the 70 species tested for the torus capillary-seeding hypothesis, only 16 species were found with punctured tori, mostly belonging to the Pinaceae family. The size of the pores varied from 12nm to 144nm. Only a weak but significant correlation was found between cavitation resistance and torus capillary-seeding pressure (P TC, s=0.52, P=0.046; Supplementary Fig. S2 available at JXB online). The air-seeding pressure based on the size of plasmodesmatal pores in tori was of the same order of magnitude as the P 50 values.

37Spearman correlations and PIC analyses showed a positive correlation between the thickness to span ratio (T W D T –1) and cavitation resistance (P 50; Fig. 4A; Table 3). Variation in T W D T –1 across conifers was mainly determined by changes in double cell-wall thickness (T W; s=0.60, PIC=0.57) rather than a change in tracheid lumen diameter (D T; Fig. 4B). Moreover, tracheid lumen diameter and wall thickness were positively correlated (s=0.65, PIC=0.52; Fig. 4C), meaning that wall thickness and tracheid lumen diameter varied in the same way when total tracheid size changed. Wall implosion pressure (P WI) was also positively correlated with cavitation resistance (Table 3; Fig. 4C). For most of the species, P WI was always more negative than P 50 (P WI= – 4.69 to – 32.14MPa), suggesting that conduit implosion does not occur before cavitation. Comparing the data plot in Fig. 4C with the 1:1 line, the difference between P WI and P 50 decreased from vulnerable species (low absolute P 50 values) to resistant species (high absolute P 50 values). The PIC analyses suggest that traits related to cavitation resistance and mechanical strength have evolved jointly.

38Significant differences in xylem anatomy and hydraulic properties were found between the four biomes. In general, torus–aperture overlap, valve effect, and thickness to span ratio were significantly higher for the Mediterranean biome than the other biomes (Table 5). Species from Mediterranean regions showed a 2-fold higher cavitation resistance (more negative P 50) than in the other biomes (Table. 5).

39Across the 115 species studied, bordered pit properties of earlywood tracheids (torus–aperture overlap and valve effect) are the best proxy to explain the variability of cavitation resistance: species resistant to cavitation have a high valve effect (corresponding to both torus–aperture overlap and membrane flexibility). This is consistent with the seal capillary-seeding hypothesis. Regarding the flexibility of the margo, previous studies suggested that this feature could play an important role in the process of cavitation. First, mentioned that margo flexibility could allow the torus to be pulled out through the pit aperture, so that species that are vulnerable to cavitation should have a more flexible margo. In contrast, reported that a high margo flexibility may facilitate the torus to move toward the pit border and improve the seal between the torus and the pit aperture. The present study showed that most of the valve effect efficiency is due to variation in the torus–aperture overlap, while the flexibility of the margo does not seem to play a substantial role. Furthermore, the results confirm that torus capillary-seeding may provide an additional air-seeding mechanism in Pinaceae. While this must be interpreted with caution because SEM observations do not allow pores that completely pass through the torus to be distinguished from those that are limited to the surface of the torus, showed that at least a few pores in each species with punctured tori completely pass through the torus. Punctured tori are also observed in some members of Cupressaceae and Cephalotaxaceae (see Table 4, and ). Plasmodesmata need to be associated with early developmental stages of the torus in combination with a lack of matrix removal from the torus by autolytic enzymes during cell hydrolysis. Thus, the taxonomic limitation of punctured tori to these families is assumed to reflect developmental differences in torus ontogeny. However, because of the lack of a distinct torus in Araucariaceae, no comment can be made on the mechanism of air-seeding for this family. Nevertheless, in terms of cavitation resistance, Araucariaceae are highly vulnerable to air-seeding (P 50= –2.02 to –3.3MPa). showed that the more cavitation resistant a species is the more pronounced is the torus–margo difference.

40In this study, two additional mechanisms of air-seeding were also investigated, and they were excluded in agreement with. The pressure difference needed to break the margo (P RS) was more negative than P 50, regardless of the number of margo strands (Supplementary Fig. S1 available at JXB online). Therefore, cavitation was not due to breaking of the margo strands but to capillary rupture of the air–sap meniscus. Furthermore, the results demonstrate that cavitation does not occur at the pores in the margo, but when the torus becomes aspirated against the pit border and seals the pit aperture.

41The present data indicated a strong trade-off between hydraulic and mechanical safety with a significant evolutionary association between increasing cavitation resistance and increasing thickness to span ratio of tracheids. As in previous studies, the present results show that species resistant to cavitation have thicker tracheid walls relative to lumen area. While ) and concluded that variation in the thickness to span ratio was determined by lumen diameter rather than cell wall thickness, the present study demonstrated the opposite: the wall thickness but not the lumen diameter appears to be responsible for the trade-off between hydraulic safety and mechanical strength. A mechanical constraint on xylem anatomy could explain this relationship. Higher cavitation resistance is associated with lower negative sap pressure, which requires tracheids with a higher thickness to span ratio to resist mechanical stresses. Cavitation resistance might therefore be indirectly linked to thickness to span ratio. From a functional perspective, thicker walls in relation to lumen area do not improve drought-related cavitation resistance as this phenomenon occurs at the bordered pit level. Taking into account that there is certainly a carbon cost limitation in building up tracheid walls, suggested that walls are close to their maximum thickness. Thereby, conifer trees can only compete for higher mechanical strength by narrowing their tracheid lumina, while levels of ray and axial parenchyma remain typically low. However, our study showed that as cell diameter increased, cell-wall thickness varied proportionally much more than lumen diameter. This suggests that there could be a minimum lumen diameter threshold that maintains a minimum level of hydraulic conductance.

42The results confirm that hydraulic failure by implosion is unlikely in lignified tracheids of conifers. Conduit implosion of xylem has been observed in stems of a few Pinaceae, but only in tracheids with severe reduction of lignification in their secondary walls (; Donaldson, 2002). Indeed, some degree of lignification is required to allow normal xylem function and water conduction. The data show that the pressure needed to cause conduit implosion in lignified tracheids is related to cavitation resistance, but is for most species more negative than P 50. The minimum water potential measured in conifer species is generally less negative than P 50, suggesting that the conduit implosion pressure is unlikely to occur under field conditions. These results suggest that cavitation always occurs before xylem collapse. According to, safety factors for implosion are high compared with air-seeding. One interpretation of such a safety margin is that there has been strong selective pressure to avoid implosion. Even so, evidence of collapse has been observed in needles of Podocarpaceae, but localized in extra-xylary transfusion tracheids. Dysfunction by collapse seems more likely to occur in these cells than in the xylem because of their parenchymatous origin, irregular shape, large lumen, and high pit density.

43Only a few investigations were carried out on the relationship between conifer species distribution and xylem anatomy. Previous studies on hydraulic traits have provided evidence of considerable variations of cavitation resistance between species from contrasting environments, with individuals from xeric regions more resistant than those from mesic regions. and highlighted the existence of geographic variations in wood density with significantly denser wood in dryer habitats. The present study confirmed these trends and, in addition, it was hypothesized that xylem anatomical and, in particular, bordered pit properties would differ between the different biomes, in agreement with the variability of cavitation resistance. The distribution analyses strengthen the conclusion that species growing in arid environments such as the Mediterranean region present the following combination of distinct anatomical features: wide torus–aperture overlap, high valve effect, and large thickness to span ratio. Results from the variation in cavitation resistance in combination with the distribution of species showed a substantial ability of species from xeric habitats to resist drought-induced cavitation and support mechanical strength. However, it does not mean that conifers from Mediterranean habitats are immune to drought stress.

44Based on both cross-species correlations and PIC analyses, the wide taxonomic sample examined here enabled the demonstration that cavitation is most likely to occur by seal capillary-seeding via the overlap of the torus on the pit aperture, while torus capillary-seeding could provide an additional mechanism in Pinaceae. Testing the consequences of increasing cavitation resistance highlighted an indirect trade-off between hydraulic safety (P 50) and mechanical strength (thickness to span ratio) over a broad range of species. This study illustrates that the torus–aperture overlap and the thickness to span ratio represent the two most useful proxies to estimate cavitation resistance. It was also found that dysfunction by conduit implosion in xylem tracheids is unlikely as the theoretical implosion pressure is unrealistic for most species. Secondly, increased cavitation resistance did not come at a cost of decreased tracheid lumen diameter and should therefore have only a minor impact on hydraulic efficiency. Compared with angiosperms, conifers seem to be able to achieve greater cavitation resistance without considerably sacrificing hydraulic efficiency. This growth strategy could allow conifers to colonize seasonally arid habitats that are subject to freezing-induced embolism formation. The bordered pit anatomy of the xylem could slightly affect the ability of the species to resist drought-induced embolism and consequently conifer distribution. Although it is felt that the approach used here of studying few individuals per species is valid when covering a large number of species, further work on the intraspecific and intraindividual variability of conifers would be required to better understand hydraulic trade-offs and functional adaptations to different environments.

45Supplementary data are available at JXB online.

46Figure S1. Relationship between (A) cavitation resistance (P 50) and rupture stretching pressure, (B) xylem air entry pressure (P 12) and torus deflection pressure, and (C) margo capillary-seeding pressure and torus deflection pressure.

47Figure S2. Relationship between torus capillary-seeding pressure and cavitation resistance (P 50).

48Table S1. List of species studied with reference to their taxonomic family, origin, and average cavitation resistance values (P 50).

49Table S2. Pearson (r) and Spearman correlation (s) for relationship between anatomical/functional traits and cavitation traits (P 50, P 12, P 88, and slope) in conifers.