Pit characters determine drought-induced embolism resistance of leaf xylem across 18 Neotropical tree species

Abstract

Embolism spreading in xylem is an important component of plant drought resistance. Since embolism resistance has been shown to be mechanistically linked to pit membrane characters in stem xylem, we speculate that similar mechanisms account for leaf xylem. We conducted transmission electron microscopy to investigate pit membrane characters in leaf xylem across 18 Neotropical tree species. We also conducted gold perfusion and polar lipid detection experiments on three species covering the full range of leaf embolism resistance. We then related these observations to previously published data on embolism resistance of leaf xylem. We also incorporated previously published data on stem embolism resistance and stem xylem pit membranes to investigate the link between vulnerability segmentation (i.e. difference in embolism resistance) and leaf–stem anatomical variation. Maximum pit membrane thickness (T_pm,max) and the pit membrane thickness-to-diameter ratio (T_pm,max/D_pm) were predictive of leaf embolism resistance, especially when vestured pits were taken into account. Variation in T_pm,max/D_pm was the only trait predictive of vulnerability segmentation between leaves and stems. Gold particles of 5- and 10-nm infiltrated pit membranes in three species, while the entry of 50-nm particles was blocked. Moreover, polar lipids were associated with inner conduit walls and pits. Our results suggest that mechanisms related to embolism spreading are determined by T_pm, pore constrictions (i.e. the narrowest bottlenecks along pore pathways), and lipid surfactants, which are largely similar between leaf and stem xylem and between temperate and tropical trees. However, our mechanistic understanding of embolism propagation and the functional relevance of T_pm,max/D_pm remains elusive.

Introduction

Water transport in plants is essential for photosynthesis, productivity, and growth (Scoffoni et al., 2016). Water flows under negative pressure through the plant, from the soil to the atmosphere, through root, stem, and leaf xylem (Dixon and Joly, 1895). However, drought-induced embolism may interrupt water transport with increasing negative pressure due to water deficit. Embolism formation is, therefore, associated with hydraulic failure, which may negatively affect plant growth, especially in a context of increasing frequency of extreme drought events (Allen et al., 2015). Most studies over the last decades investigated embolism resistance of stem xylem, but recent technical advances allowed the acquisition of data on embolism resistance of leaf xylem to improve our understanding of whole-plant hydraulics and drought resistance strategies (Brodribb et al., 2016b; Scoffoni et al., 2017; Klepsch et al., 2018). In agreement with the ecological importance of embolism resistance of stem xylem (Martin-StPaul et al., 2017; Brodribb et al., 2020), recent work shed light on its anatomical determinants (Kaack et al., 2019, 2021; Levionnois et al., 2021a). Anatomical determinants of leaf embolism resistance, however, have been given less attention, despite the importance of leaf embolism resistance in shaping interspecific strategies in plant drought resistance (Blackman et al., 2019).

Inside the xylem, water flows through adjacent vessels across intervessel pit membranes. The pit membrane is composed of cellulose microfibril aggregates, which allow transport of xylem sap across vessels. Pit membranes also allow the entry of gas bubbles from an adjacent embolized vessel, which has been described as air-seeding (Tyree and Sperry, 1989; Mayr et al., 2014; Kaack et al., 2019; Avila et al., 2022). As a 3D porous medium, the pit membrane is crossed by numerous pore spaces, with a pore constriction as an effective bottleneck between each pore. The narrowest constriction in a pore pathway determines mass flow of water and gas within a single pore pathway, and the pore with the largest minimum pore constriction in all intervessel pit membranes between adjacent vessels governs embolism spreading and the minimum hydraulic resistance at the intervessel level (Zhang et al., 2020b; Kaack et al., 2021). A thick pit membrane (T_pm) has been shown to be associated with high embolism resistance in stem xylem, across different biomes and growth forms (Lens et al., 2011; Li et al., 2016; Dória et al., 2019; Trueba et al., 2019; Thonglim et al., 2020; Kaack et al., 2021; Levionnois et al., 2021a). Although embolism spreading appears not to rely on pore constriction size only, recent results strongly suggest that pit membrane thickness increases the likelihood of a narrow pore constriction inside pit membrane pathways due to an increasing number of pore constrictions with increasing pit membrane thickness (Zhang et al., 2020b; Kaack et al., 2021). Pit membrane thickness has been investigated in leaf xylem of a relatively small number of angiosperm species (Klepsch et al., 2018; Kotowska et al., 2020), and whether pit membrane thickness can predict leaf embolism resistance has not yet been tested for a large number of species. Also, leaf data on pit membrane pore size and pore constriction size remain relatively scarce.

Besides pit membranes, other bordered pit and conduit characters can influence xylem embolism resistance. A large pit membrane diameter has been shown to be associated with higher vulnerability to embolism in stems (Lens et al., 2011; Scholz et al., 2013; Levionnois et al., 2021a), maybe due to reduced resistance of pit membrane deflection during aspiration, which could favor embolism spreading (Choat et al., 2004; Tixier et al., 2014; Kaack et al., 2019). A recent study on the pit membrane diameter in petioles and stems of Magnolia grandiflora L. shows no difference between these two organs (Zhang et al., 2020a). Yet, the diversity of pit membrane diameter in leaf xylem remains poorly understood in comparison to stem xylem.

Recent studies also highlight the role of the chemical composition of sap, inner vessel walls, and pit membranes in embolism formation and spreading (Schenk et al., 2015, 2017, 2021). Polar lipids coating the pit membrane lower the surface tension of the air–water meniscus, reducing the pressure difference across the meniscus required for bubble formation (Yang et al., 2020). Thus, a low surface tension favors the formation of small bubbles, which can be stable under negative pressure by a surfactant coat (Schenk et al., 2017; Yang et al., 2020; Ingram et al., 2021; Schenk et al., 2021). While surfactant-coated nanobubbles have been visualized in xylem sap, it remains unclear whether or not, and how these are linked with embolism formation and gas solubility of xylem sap under conditions of changing pressure and temperature (Mercury et al., 2003; Schenk et al., 2016; Guan et al., 2021b). Polar lipids such as galactolipids and phospholipids have been observed in stem xylem in temperate, Mediterranean, and sub-tropical species (Schenk et al., 2018, 2021), but have not been studied in tropical rainforest species. Furthermore, whether or not such amphiphilic lipids are present in the leaf xylem remains an open question.

Leaf embolism resistance is also an important drought-resistance feature related to stem embolism resistance (Tyree and Ewers, 1991; Tyree and Zimmermann, 2002). The vulnerability segmentation hypothesis (hereafter called “vulnerability segmentation”) predicts that leaves are less resistant to embolism than stems. Positive segmentation (i.e. leaves being less resistant to the loss of conductance than stems) has been suggested in numerous studies (Chen et al., 2009; Johnson et al., 2011; Charrier et al., 2016; Hochberg et al., 2016; Johnson et al., 2016; Rodriguez-Dominguez et al., 2018; Skelton et al., 2018; Levionnois et al., 2020). There are also records of a lack of segmentation (leaves equally or more resistant to the loss of conductance than stems; Klepsch et al., 2018; Losso et al., 2019; Levionnois et al., 2020; Li et al., 2020; Smith-Martin et al., 2020). Important in this discussion is that most studies investigating vulnerability segmentation were based on measurements of whole-leaf conductance, integrating both xylem and outside-xylem pathways (Trifiló et al., 2016; Scoffoni et al., 2017). Differences between leaf and stem xylem in embolism resistance have only been investigated recently (Charrier et al., 2016; Hochberg et al., 2016; Skelton et al., 2017; Klepsch et al., 2018; Rodriguez-Dominguez et al., 2018; Skelton et al., 2018; Levionnois et al., 2020). The more segmented the hydraulic network of the leaf and stem xylem is, the longer is the plant desiccation time (Blackman et al., 2019; Levionnois et al., 2021b). If pit characters are the main determinants of xylem embolism resistance, vulnerability segmentation should be reflected in differences in pit characters between the leaf and the stem. One study attempted such investigation based on three species (Klepsch et al., 2018), but differences in pit characters between leaf and stem xylem are largely unknown.

Here, we aim to test if pit characters can predict (1) leaf xylem embolism resistance and (2) the possible difference in embolism resistance between leaves and stems for a phylogenetically diverse set of 18 tropical rainforest tree species. We expect the tropical species diversity and the broad phylogenetic diversity in this study to bring a broad diversity of anatomical structures. We measured pit characteristics based on transmission electron microscopy (TEM). We also conducted observations of polar lipids and applied a gold perfusion experiment to quantify pore constriction sizes in pit membranes of three species. We then related these anatomical observations to previously published data of xylem embolism resistance quantified as the water potential at 50% loss of conductivity (P_50,leaf) for the same set of species (Levionnois et al., 2020), and to P_50,stem and anatomical data to discuss the vulnerability segmentation issue (Ziegler et al., 2019; Levionnois et al., 2021a). We specifically addressed the following hypotheses: (1) Pit characters predict embolism resistance of leaf xylem, (2) Polar lipids occur in conduits of leaf xylem, and (3) Interspecific variation in vulnerability segmentation (i.e. the difference between P_50,leaf and P_50,stem) can be associated with and predicted by differences in pit characters between leaf and stem xylem.

Results

Leaf embolism resistance and anatomy

P_50,leaf (a complete list of traits with their abbreviations is presented in Table 1) ranged from −4.61 to −2.25 MPa across the species studied (Table 2). Interspecific variation in T_pm ranged from 147.5 to 429.8 nm, and T_pm,max was slightly larger, with values from 209 to 541 nm (Figure 1). D_pm values varied from 3.31 to 8.66 µm across the studied species (Figure 1). Vestured-pit species (Figure 1C) displayed significantly thinner pit membranes than nonvestured-pit, both based on T_pm (P < 0.05; t = 6.56; T_pm: 162.3 ± 9.8 and 293.5 ± 70.4 nm, respectively, for n = 5 and 13, respectively) and T_pm,max (P < 0.05; t = 6.76; T_pm,max: 227.0 ± 20.5 and 417.7 ± 96.2 nm, respectively, for n = 5 and 13, respectively). Xylem of the Fabaceae species Dicoryniaguianensis was found to have nonvestured pits, as could be expected based on its phylogenetic position within the Dialioideae (Jansen et al., 2001).

Figure 1

Transverse sections of leaf xylem from rainforest tree species based on TEM. A, Thin pit membrane in leaf xylem of the embolism vulnerable species L. poiteauii. B, T_pm in the embolism resistant species P. cochlearia. C, Vestured pit in E. grandiflora. D, Large pit membrane in an intervessel wall of the embolism vulnerable species C. schomburgkianus. Black bars (also white bars in (B)) indicate the scale.

Table 1

List of traits and abbreviations

Abbreviation	Trait and Definition	Measured Organ	Unit
P12,leaf, P50,leaf and P88,leaf	Water potential at 12%, 50%, and 88% leaf total embolism events	Leaf veins	MPa
P12,stem, P50,stem and P88,stem	Water potential at 12%, 50%, and 88% loss of stem hydraulic conductivity	Stem	MPa
Tpm	Mean pit membrane thickness	Midrib	nm
Tpm,max	Maximum pit membrane thickness	Midrib	nm
Dpm	Pit membrane diameter	Midrib	µm
Tpm,max/Dpm	Maximum pit membrane thickness :diameter ratio	Midrib

Table 2

List of the species studied and key leaf xylem traits

Species	Family	n	Vestured Pits	P50,leaf (MPa)	Tpm (nm)	Tpm,max (nm)	Dpm (µm)
Bocoa prouacensis Aubl.	Fabaceae	2	Yes	−4.61 ± 0.22	168	225	4.56
Chaetocarpus schomburgkianus (Kuntze)	Peraceae	3	No	−2.59 ± 1.02	371	514	8.66
Chrysophyllum sanguinolentum (Pierre) Baehni	Sapotaceae	2	No	−2.28 ± 1.71	430	533	5.05
Dicorynia guianensis Amshoff	Fabaceae	3	No	−3.48 ± 1.40	221	302	3.57
Eperua falcata Aubl.	Fabaceae	3	Yes	−3.68 ± 0.45	166	261	4.41
Eperua grandiflora (Aubl.) Benth.	Fabaceae	3	Yes	−4.40 ± 1.00	148	213	4.01
Eschweilera sagotiana Miers	Lecythidaceae	3	No	−2.25 ± 0.32	212	292	4.60
Goupia glabra Aubl.	Goupiaceae	3	No	−2.88 ± 0.47	269	323	8.58
Gustavia hexapetala (Aubl.) Sm.	Lecythidaceae	3	No	−4.45 ± 1.32	328	464	4.58
Lecythis poiteauii O. Berg	Lecythidaceae	2	No	−3.37 ± 2.29	188	326	3.86
Licania membranacea Sagot ex Laness.	Chrysobalanaceae	3	No	−3.35 ± 0.72	264	396	5.00
Manilkara bidentata (A. DC.) A. Chev.	Sapotaceae	3	No	−4.40 ± 1.64	293	492	4.48
Pradosia cochlearia (Lecomte) TD. Penn.	Sapotaceae	3	No	−4.51 ± 0.58	339	541	4.35
Protium opacum Swart	Burseraceae	3	No	−2.98 ± 1.16	364	526	7.77
Qualea rosea Aubl.	Vochysiaceae	2	Yes	−2.43 ± 0.05	158	209	6.45
Symphonia sp1	Clusiaceae	3	No	−2.51 ± 0.23	248	346	7.91
Tachigali melinonii (Harms) Zarucchi & Her.	Fabaceae	3	Yes	−3.86 ± 0.29	172	227	3.31
Virola michelii Heckel	Myristicaceae	3	No	−2.62 ± 0.90	289	375	6.19

Mean ± sd. There is no sd for T_pm, T_pm,max, and D_pm as these traits were measured on one sample per species. Trait abbreviations are shown in Table 1.

Based on multiple linear model analyses, an effect of vestured pits was found on the relationship between the T_pm,max-to-D_pm ratio and P_50,leaf (Figure 2A). Hence, species with relatively large T_pm in relation to their pit diameter showed high embolism resistance when the presence of vestures was taken into account (Table 3). The presence or absence of vestured pits did not significantly affect the relationships between T_pm and P_50,leaf, between T_pm,max and P_50,leaf, and between D_pm and P_50,leaf (Table 3). T_pm and T_pm,max were not significantly related to P_12,leaf, P_50,leaf and P_88,leaf, even if vestured-pit species were excluded from the analyses (Figure 2A and Table 4). However, we found a significant relationship between T_pm,max and P_50,leaf when vestured-pits were removed and a separate analysis was conducted at the individual level, that is, directly across tree individuals with both T_pm,max and P_50,leaf values, considering that there was only one measured tree per species for TEM measurements (R² = 0.663; Figure 2B;Supplemental Table S2). Therefore, at the individual level, T_pm were associated with high leaf embolism resistance. Increasing D_pm was associated with an increase in P_50,leaf and P_88,leaf (R² = 0.411 and 0.432, respectively), such that large pit membranes were associated with low leaf embolism resistance (Figure 2C and Table 4), whether or not bordered pits were vestured or nonvestured. Increasing the T_pm,max-to-D_pm ratio was associated with a decrease in P_12,leaf, P_50,leaf, and P_88,leaf, only when vestured-pit species were removed, such that relatively large T_pm in relation to their surface were associated with high leaf embolism resistance (R² = 0.321, 0.489, and 0.405, respectively; Figure 2D and Table 4). All analyses conducted at the individual tree level (i.e. conducting the analysis only on trees with both T_pm,max and P_50,leaf values, with only one measured tree per species for TEM measurement) are presented in Supplemental Tables S1 and S2.

Figure 2

Relations between pit membrane traits and drought-induced embolism resistance in leaf xylem. A, P_50,leaf according to maximum pit membrane thickness (T_pm,max) at the species level. B, P_50,leaf according to T_pm,max at the individual level, that is, the analysis was conducted only on tree individuals with both T_pm,max and P_50,leaf values, knowing that there was only one measured tree per species for TEM measurements. C, P_50,leaf according to pit membrane diameter (D_pm) at the species level. D, P_50,leaf according to the T_pm,max-to-D_pm ratio at the species level. Standard deviations are plotted around the mean for each species. Dashed lines are significant relationships for all species (vestured and nonvestured pit species). Solid lines are significant relationships when vestured-pit species were excluded. Pearson or Spearman correlation tests were used depending on normality. Full circles indicate nonvestured pit species, empty circles indicate vestured-pit species.

Table 3

Multiple models to test if the relationship between pit membrane traits and leaf embolism resistance (P_50,leaf) is affected by the presence or absence of vestured pits

Multiple Model	Explicative Variables	P-value	Adjusted R²
Tpm + Vestured		0.438	0.000
	Tpm	0.972
	Vestured	0.383
Tpm,max + Vestured		0.245	0.060
	Tpm,max	0.289
	Vestured	0.103
Dpm + Vestured		<0.05	0.284
	Dpm	<0.05
	Vestured	0.503
Tpm,max/Dpm + Vestured		<0.01	0.420
	Tpm,max/Dpm,max	<0.01
	Vestured	<0.01

Bold values represent significant relationships (P < 0.05). Abbreviations for all traits are presented in Table 1.

Table 4

Pearson and Spearman correlations between leaf embolism resistance and anatomical traits at the species level

			P12,leaf		P50,leaf		P88,leaf
Trait	Vesturesa	n	P-value	R²	P-value	R²	P-value	R²
Tpm		18	0.364	0.052	0.345	0.056	0.271	0.075
Tpm,max		18	0.570	0.021	0.810	0.004	0.673	0.011
Tpm	Excluded	13	0.337	−0.084	0.964	0.000	0.624	0.023
Tpm,max	Excluded	13	0.093	−0.235	0.302	−0.096	0.664	−0.018
Dpm		18	0.057	0.211	<0.01	0.411	<0.01	0.432
Dpm	Excluded	13	0.415	0.061	<0.05	0.406	<0.05	0.442
Tpm,max/Dpm		18	0.494	−0.030	0.100	−0.160	0.097	−0.163
Tpm,max/Dpm	Excluded	13	<0.05	−0.321	<0.01	−0.489	<0.05	−0.405

a

In the vestures column, excluded means that vestured-species were excluded from the analysis to exclude the potential effects of vestured-pits on the relationship between embolism resistance and the tested anatomical trait. Bold values represent significant relationships (P < 0.05). Abbreviations for all traits are presented in Table 1.

In the colloidal gold perfusion experiment of the three species representing low, intermediate, and high leaf embolism resistance, gold particles of 5 and 10 nm penetrated all pit membranes, but 50-nm particles were not seen to infiltrate any pit membrane (Figure 3). While gold particles of 50 nm were only found on the pit membrane surface, 20-nm gold particles rarely penetrated pit membranes of Goupiaglabra (Figure 3, C and D). With a maximum penetration depth of 37 nm, these 20-nm particles did not deeply penetrate into pit membranes. In Chaetocarpusschomburgkianus, the gold particles appeared to penetrate the central areas of the pit membranes (Figure 3, A and B), whereas they were well distributed throughout the entire membranes of Gustaviahexapetala (Figure 3, E and F). Gold particles of 5 and 10 nm in size were also found in aspirated pit membranes and showed no clear differences with nonaspirated ones.

Figure 3

Penetration of gold particles into intervessel pit membranes in transverse sections of leaf midribs based on TEM images. A and B, C. schomburgkianus. C and D, G. glabra. E and F, G. hexapetala (E and F). The perfusion contained an equal concentration of 5, 10, 20, and 50 nm colloidal gold particles. In all three species, 5 and 10 nm gold particles penetrated the pit membranes (B, D, and F). Arrows indicate gold particles of different sizes. The black vertical lines indicate the outlines of pit membranes. CML, compound middle lamella; PA, pit aperture; PB, pit border; PC, pit chamber; PC1, perfused pit chamber; PC2, pit chamber downstream of the perfusion; PM, pit membrane; SW, secondary cell wall; VL, vessel lumen.

All samples of C. schomburgkianus, G. glabra, and G. hexapetala, which were selected for perfusion with the FM1–43 dye, showed positive signals for polar lipids on their intervessel walls, pits and inner vessel walls (Figure 4, B, D, F, and G). No signals were found in the control samples when excised at 488 nm (Figure 4, A, C, and E). In addition, G. glabra showed distinct FM1–43 signals at the edges of pit membranes (Figure 4D), while the outline of pit membranes was weakly visible only in the other two species (Figure 4, B and D).

Figure 4

Confocal laser scanning microscopy images of transverse sections of distal stem segments perfused with FM1–43 dye. A and B, C. schomburgkianus. C and D, G. glabra. E, F and G, G. hexapetala. The dye in yellow indicates the presence of amphiphilic, polar lipids at vessel walls and pits (B, D, F, and G). Nonperfused controls are provided in (A), (C), and (E). Lignin autofluorescence in blue at 405-nm excitation. Arrow head, intervessel pit; IW, intervessel wall.

In Figure 5, theoretical air-seeding pressure (i.e. the bubble point pressure difference required to push an air–water meniscus through a given pore size) was negatively and nonlinearly related to pore constriction size, such that for a given pore constriction size the air-seeding pressure was lower for the equilibrium surface tension of xylem sap (25 mN m⁻¹) in comparison to the surface tension of pure water (72 mN m⁻¹). The theoretical air-seeding pressure appeared to be close to the measured absolute values of P_12,leaf (2.5 ± 0.5, 2.2 ± 0.3, and 3.1 ± 0.4 MPa for C. schomburgkianus, G. glabra and G. hexapetala, respectively). There was also good agreement between the estimated air-seeding pressure and the estimated pore constriction sizes based on the smallest gold particles retained (10, 20, and 10 nm for C. schomburgkianus, G. glabra, and G. hexapetala, respectively), as long as the equilibrium surface tension of xylem sap lipids (25 mN m⁻¹) and a pore shape correction factor of 0.5 (Figure 5) was considered. Without this pore shape correction factor, which was required due to the highly irregular, noncircular geometry of pit membrane pores, there would be a large difference between the estimated and measured air-seeding pressure. Similarly, applying the Young–Laplace equation based on the high surface tension of pure water resulted in theoretical air-seeding pressures that were much higher than the equilibrium surface tension of xylem sap with surfactants.

Figure 5

Theoretical air-seeding pressure according to pore constriction size based on a modified Young–Laplace equation (see Materials and Methods for details) for different values of pore shape correction factor (ϰ = 0.5 or 1) and surface tension of xylem sap (γ = 25 mN m⁻¹ as equilibrium surface tension, and 72 mN m⁻¹ as surface tension of pure water). An increase in ϰ (dashed versus solid lines) or γ (orange versus blue lines) would imply unrealistic, very large pore constrictions to obtain air-seeding pressure in the range of the species studied. The plotted points correspond to absolute values of the measured P_12,leaf, which is considered to represent the water potential associated with the initial air entry causing incipient damage to the hydraulic system. The estimated size of pore constrictions is based on gold perfusion experiments. Cs: C. schomburgkianus; Gg: G. glabra; Gh: G. hexapetala. Standard errors are shown.

Leaf–stem variation in pit membrane characters and vulnerability segmentation

Values of T_pm,max ranged from 429 to 1,047 nm for stems (Figure 6A and Table 2; Levionnois et al., 2021a, 2021b), and were on average significantly larger in stem xylem than in leaf xylem across all species (Figure 6A). D_pm ranged from 3.30 to 8.35 µm for stems (Figure 6B and Table 2), and there was no significant difference in D_pm between leaf and stem xylems across all species (Figure 6B). The T_pm,max-to-D_pm ratio ranged from 32.4 to 124.5 nm µm⁻¹ for leaves, and from 64.1 to 309.1 nm µm⁻¹ for stem xylem across all species (Figure 6C and Table 2). There was a significant difference in the T_pm,max-to-D_pm ratio between leaf and stem xylem, such that pit membranes were relatively thicker but pit borders were smaller for stem xylem than leaf xylem across all species (Figure 6C).

Figure 6

Differences in pit membrane traits between leaf and stem xylem across species. A, difference between leaf and stem in the maximum pit membrane thickness (T_pm,max). B, difference between leaf and stem in the diameter of intervessel pit membranes (D_pm). C, difference between leaf and stem in the ratio between maximum pit membrane thickness to pit membrane diameter (T_pm,max-to-D_pm ratio). Gray box: stem xylem, white box: leaf xylem. The center line represents the median, leaf and right box limits represent lower and upper quartiles, respectively, whiskers represent 1.5-time the interquartile range and circles represent outliers. Comparison tests were conducted with Student’s, Welch’s, or Mann–Whitney–Wilcoxon’s test, depending on the parameters of the samples (effectives, normality of distribution, and variance).

The stem–leaf differences in T_pm, T_pm,max, and D_pm were not significantly related to the leaf–stem difference in P₁₂, P₅₀, and P₈₈ (Figure 7, A and B; Table 5). Increasing stem–leaf difference in the T_pm,max-to-D_pm ratio was associated with an increase in the leaf–stem difference in P₅₀ and P₈₈ (Figure 7C and Table 5). Therefore, a large difference in the T_pm,max-to-D_pm ratio (i.e. relatively thick and small pit membranes in stem xylem, but thin and large pit membranes in leaf xylem) was associated with a strong leaf–stem vulnerability segmentation, with leaf xylem being more vulnerable to embolism than stem xylem. For some species, small differences in the T_pm,max-to-D_pm ratio was associated with a negative leaf–stem vulnerability segmentation, with leaf xylem being more resistant to embolism than stem xylem.

Figure 7

Relations between the difference in pit membrane traits between stem and leaf xylem, and the difference in xylem embolism resistance between leaves and stems. A, Leaf–stem difference in P₅₀ according to stem–leaf difference in T_pm,max at the species level. B, Leaf–stem difference in P₅₀ according to stem–leaf difference in D_pm at the species level. C, Leaf–stem difference in P₅₀ according to stem–leaf difference in the T_pm,max-to-D_pm ratio at the species level. Standard deviations are plotted around the mean for each species. Dashed lines are significant relationships for all species (vestured- and nonvestured-pit species). Pearson or Spearman correlation tests were used depending on normality. Full circles indicate nonvestured pit species, empty circles indicate vestured-pit species.

Table 5

Pearson and Spearman correlations between leaf–stem vulnerability segmentation and anatomical traits at the species level

		P12,leaf – P12,stem		P50,leaf – P50,stem		P88,leaf – P88,stem
Trait	n	P-value	R²	P-value	R²	P-value	R²
stem Tpm – leaf Tpm	17	0.449	−0.038	0.593	0.020	0.374	0.053
stem Tpm,max – leaf Tpm,max	17	0.817	−0.004	0.290	0.074	0.529	0.030
stem Dpm – leaf Dpm	18	0.609	−0.017	0.070	−0.191	0.075	−0.184
stem Tpm,max/Dpm leaf Tpm,max/Dpm	17	0.619	0.017	<0.01	0.438	<0.05	0.282

Bold values represent significant relationships (P < 0.05). Abbreviations for all traits are presented in Table 1.

Discussion

Our results demonstrate that xylem resistance to drought-induced embolism is related to the structure of bordered pits in leaf xylem across a broad structural diversity of tropical rainforest tree species. Pit membrane thickness and diameter, in addition to the presence or absence of vestured pits, were predictive of leaf embolism resistance. This was also supported by pit trait variation explaining the interspecific variation in leaf–stem vulnerability segmentation (Levionnois et al., 2020). The gold perfusion experiment confirmed that embolism spreading is related to tiny pore constrictions (<50 nm) inside the pit membrane. Finally, the observation of polar lipids suggests that dynamic surface tension also applies to leaf xylem, which may play an important role in multiphase interactions between the gas, liquid, and solid phase, and therefore embolism formation and spreading in leaf xylem, similar to stem xylem.

Leaf embolism resistance

T_pm were associated with high leaf embolism resistance when pit morphology (vestured versus nonvestured) was taken into account (Figure 2B). This is in agreement with previous studies at the stem level across different biomes (Li et al., 2016; Trueba et al., 2019; Levionnois et al., 2021a) and plant forms (Dória et al., 2019; Thonglim et al., 2020), as well as at the intraspecific level (Thonglim et al., 2020). The T_pm–P₅₀ relationship in leaves of three species was left inconclusive based on Klepsch et al. (2018), while the 10 species studied by Guan et al. (2022) showed no significant relationship for leaf xylem. Here, the T_pm–P_50,leaf relationship was only found at the individual level (i.e. comparing measurements made on the same individuals only) and not at the species level (i.e. averaging P_50,leaf per species). This is quite logical, as for individuals for which we measured T_pm, P_50,leaf was measured on the same shoot, reducing potential sources of variation that may be encountered across trees of a given species (for instance, due to tree ontogeny and height, phenological stage, light exposure, life history, etc.). However, correlations of P_50,leaf with D_pm and the T_pm,max/D_pm ratio were lower at the individual tree level than the species level (Table 4; Supplemental Table S2). This may indicate that potential trait expression at the species level and realized trait expression at the individual level represent different approaches, and could each be valuable in different contexts (Poorter et al., 2018).

Recent findings based on gold perfusion experiments, TEM measurements, and pit membrane modeling support that pit membrane thickness increases the number of microfibril layers, increasing the likelihood of narrow pore constrictions in pit membranes, which finally increases the pressure difference across a bordered-pit required for air-entry by mass flow (Zhang et al., 2017, 2020b; Kaack et al., 2021). Our results support that this mechanism holds true for the leaf xylem across different species, and also for tropical rainforest tree species. We found the estimated size of pore constrictions to be similar across the species studied, well ˂50 nm, and down to ca. 5–10 nm (Figure 3). This is in agreement with former observations of temperate and Mediterranean species, and also demonstrates that estimated pore constriction sizes are similar between leaf and stem xylem (Choat et al., 2004; Zhang et al., 2017, 2020b). We found that measured values of P_12,leaf and estimated pore constriction sizes were close to the theoretical relationship between air-seeding pressure and pore constriction size based on a modified Young–Laplace equation (Figure 5). Indeed, P₁₂ represents the water potential associated with initial air-entry causing incipient damage to the hydraulic system (Meinzer et al., 2009). This finding supports pit membrane as a mesoporous medium (defined as having pores with a diameter between 5 and 50 nm), and a mechanistic linkage between pore constriction size and embolism propagation for leaf xylem. Moreover, G. glabra exhibited the thinnest pit membranes (among the species with no vestured pits) and had 20-nm particles penetrating the membrane, which is in line with the suggested relationship between pit membrane thickness and pore constriction size (Kaack et al., 2021).

In agreement with the idea that bordered pits play a role in mediating embolism spreading, species with large pit membranes (D_pm) were more vulnerable to embolism, with no effect of vestured pits (Figure 2C). This is in agreement with previous studies investigating the D_pm–P₅₀ relationship for stem xylem (Lens et al., 2011; Scholz et al., 2013; Levionnois et al., 2021a), although other studies did not confirm this relationship (Dória et al., 2018; Trueba et al., 2019). A recent study also supported this positive D_pm–P₅₀ relationship for M.grandiflora across three organs (stem, petiole, and peduncle; Zhang et al., 2020b). A mechanistic understanding of this relationship remains unclear. It can be speculated that the pit membrane diameter (or area) is related to its resistance to stretching and deflection, such that large pit membranes would more easily stretch when these experience a pressure difference across a bordered pit (Choat et al., 2008; Tixier et al., 2014). A stretched membrane could locally increase its pore constriction sizes due to re-arrangement of microfibrils, which may promote embolism spreading. However, an increase in the pore constriction size during pit membrane stretching may depend on pit membrane thickness and stiffness. Stretching of a T_pm could also compress microfibril layers, and then decrease its pore constriction sizes, as can be seen during pit membrane shrinkage by artificial dehydration or seasonal changes during the growing season (Zhang et al., 2020b; Sorek et al., 2021). Alternatively, large pit membranes could promote faster gas diffusion driven by a pressure difference across a pit membrane (Guan et al., 2021a) because cell walls are much thicker than pit membranes and possess pores <2 nm (Donaldson et al., 2015, 2018). Nevertheless, it is unknown how exactly dissolved or undissolved gas movement into sap-filled conduits may trigger the likelihood of embolism formation (Guan et al., 2021b; Schenk et al., 2021).

In agreement with a pit membrane stretching-based mechanism of embolism spreading, the T_pm,max/D_pm ratio was the best predictor of leaf embolism resistance in our study, as long as vestured-pit species were removed (Figure 2D). The T_pm,max/D_pm ratio was introduced in a recent study (Levionnois et al., 2021a), and was based on a simplification of the mechanical framework of Tixier et al. (2014), which investigated the mechanical behavior of a pit membrane subject to a strain. According to Levionnois et al. (2021a, 2021b), the T_pm,max/D_pm ratio was also the best predictor of embolism resistance for the stem xylem for the same set of species. The T_pm,max/D_pm ratio reflects an important feature of the pit membrane geometry and its stiffness, such that thick but small pit membranes would be relatively stiff, and then be more resistant to stretching and deformation when these experience a pressure difference across the pit border.

Although we lack detailed measurements of the mechanical properties of fresh (i.e. nondehydrated) angiosperm pit membranes, the presence or absence of vestures, which may support a pit membrane (Zweypfenning, 1978), adds further evidence for a functional link between embolism resistance and pit membrane stretching. We found an effect of pit morphology on the T_pm–P_50,leaf relationship, and also on the relationship between the T_pm,max/D_pm ratio and leaf embolism resistance (Table 3). Moreover, all vestured pit species displayed the thinnest pit membranes (Figures 1, C, 2, A and B), and did not exhibit a large T_pm,max/D_pm ratio (Figure 1D). However, this does not impede vestured pit species to exhibit high leaf embolism resistance. Eperuagrandiflora, for instance, which was the second most embolism-resistant species, showed relatively thin pit membranes (<200 nm; Table 2). An effect of pit morphology on the T_pm–P_50,leaf relationship, and thinner pit membranes for vestured pit species, was also found in stem xylem for the same set of species (Levionnois et al., 2021a). Due to their branching, which may fill up a large fraction of the pit border (Figure 1C), vestures have already been suggested to prevent pit membrane stretching and deflection caused by a pressure difference across a bordered pit (Zweypfenning, 1978; Jansen et al., 2003). Vestured pits have also been suggested to promote resistance to drought-induced embolism, due to their high occurrence in hot and dry environments (Jansen et al., 2004; Medeiros et al., 2019). If vestures, especially large and branched ones that occupy the entire pit border, are preventing pit membrane deflection, and aspiration (Zweypfenning, 1978; Jansen et al., 2003; Choat et al., 2004), it can be speculated that thin pit membranes provide reduced hydraulic resistance, but still sufficient safety to avoid embolism spreading due to pit membrane aspiration. However, further evidence linking pit morphology (vestured versus nonvestured), pit membrane thickness, and resistance to drought-induced embolism is needed for a larger number of species.

We found positive signals for polar lipid surfactants on the intervessel walls, pits, and inner vessel walls in the leaf xylem of the three species investigated—C. schomburgkianus, G. glabra, and G. hexapetala—which displayed low, middle, and high leaf embolism resistance, respectively (Figure 4). These findings are in agreement with previous studies (Schenk et al., 2017, 2018; Guan et al., 2021b) conducted on stem xylem for temperate, Mediterranean, and subtropical species. Thus, our results expand this observation to leaf xylem and to tropical rainforest trees, adding further evidence to the hypothesis that polar and amphiphilic xylem lipids are a universal feature of angiosperms (Schenk et al., 2021). Neutral lipids, however, such as triacylglycerol, were recently found to form incrustations or coatings on pit membranes either due to artificial drying of wood samples, or as a natural process in nonfunctional, embolized conduits in living plants (Yamagishi et al., 2021). These coatings of neutral lipids differ strongly from the amphiphilic lipids observed here, which mainly include phospholipids and galactolipids originating from the cytoplasm of vessel elements before cell death (Scott et al., 1960; Esau et al., 1966; Jansen and Schenk, 2021).

Lipid surfactants are assumed to coat nanobubbles in the sap and to make inner conduit walls and pit borders more hydrophilic, favoring the formation of small bubbles, which can be stable under negative pressure, but reducing the likelihood of surface bubbles on inner conduit walls (Schenk et al., 2017, 2021; Yang et al., 2020; Ingram et al., 2021). This is in agreement with the finding that across species, P_12,leaf is close to the estimated air-seeding pressure based on the Young–Laplace equation when taking into account the equilibrium surface tension of xylem sap lipids based on a dynamic surface tension concept (Yang et al., 2020). Conversely, surface tension of pure water would require a much higher, unrealistic, air-seeding pressure for the estimated pore constrictions (between 10 and 50 nm; Choat et al., 2004; Zhang et al., 2017, 2020b). How exactly the concentration and chemical composition of polar lipids is linked to nanobubble formation and stability requires further research (Guan et al., 2021b).

Leaf–stem vulnerability segmentation

The range of T_pm and T_pm,max was larger for stems than for leaves across the species studied (Table 2). This is in agreement with a recent study across the same species showing that the range of P_50,stem was larger than that of P_50,leaf, and that P_50,stem values contribute considerably to variation in leaf–stem vulnerability segmentation (Levionnois et al., 2020).

As hypothesized, our study shows that interspecific variation in leaf–stem vulnerability segmentation is anatomically determined. We found a significant relationship between the stem–leaf difference in the T_pm,max/D_pm ratio and vulnerability segmentation, with a strong segmentation being associated with large differences in the T_pm,max/D_pm ratio (Figure 7C). Moreover, we found a strong difference in the T_pm,max/D_pm ratio between leaf and stem xylem, with leaf xylem exhibiting relatively thin but large pit membranes (Figure 6C). This result further strengthens the putative functional importance of the T_pm,max/D_pm ratio, potentially via a pit membrane stretching mechanism. The stem–leaf difference in the T_pm,max/D_pm ratio was first related to the stem–leaf difference in T_pm, as we found thinner pit membranes in leaf than in stem xylem (Figure 6A). This suggests that T_pm may be involved in the leaf–stem vulnerability segmentation, based on the functional importance of pore constrictions, even if we did not capture a relationship between the stem–leaf difference in T_pm and vulnerability segmentation (Figure 7A). Guan et al. (2022) showed that leaf xylem was equally or more resistant to stem xylem for 10 temperate tree species, with larger pit borders (D_pm) in stem xylem than leaf xylem, but no significant difference in T_pm and the T_pm,max/D_pm ratio between leaf and stem xylem. In Klepsch et al. (2018), one species exhibited a positive vulnerability segmentation (P_50,stem < P_50,leaf; Laurusnobilis), with no difference in T_pm, but with larger pit membranes (D_pm) in leaf xylem than stem xylem, leading to a higher value of the T_pm,max/D_pm ratio in the stem, which is partly in agreement with our study. The other species exhibited a negative vulnerability segmentation (P_50,stem > P_50,leaf; Betulapendula), with thinner pit membranes found in the stem, in agreement with the cross-organ variation of T_pm determining vulnerability segmentation (Klepsch et al., 2018).

Leaf–stem vulnerability segmentation, when positive (P_50,stem < P_50,leaf), has been shown to correlate with enhanced plant drought resistance (Blackman et al., 2019; Levionnois et al., 2021b). When leaf xylem first embolizes during a period of water stress, water will no longer flow to leaves, and the total plant transpiration is significantly reduced. In this way, the desiccation time required to deplete soil and plant water pools could be enhanced. Vulnerability segmentation may be an important mechanism of plant drought resistance, despite being not pervasive across species (Klepsch et al., 2018; Losso et al., 2019; Levionnois et al., 2020; Li et al., 2020; Smith-Martin et al., 2020). In this study, we show that the leaf–stem variation in pit characters may be an important component of plant drought resistance. However, leaf–stem vulnerability segmentation may also be determined by other mechanisms. For instance, vessels may end at the stem–petiole transition, or vessels may be particularly short and thin-walled in this region, increasing the number of vessel ends to be crossed for a given distance. Such places of segmentation or compartmentalization may slow down gas diffusion and embolism spreading (Salleo et al., 1984; André et al., 1999; Guan et al., 2021a).

Conclusions

This study provides evidence that interconduit pit membranes may play a role in embolism resistance and vulnerability segmentation of leaf xylem in tropical rainforest species. The observation of narrow pore constrictions ˂50 nm, and the occurrence of polar lipids in conduits of leaf xylem are in line with earlier work on stem xylem, and raise questions about the mechanisms behind embolism propagation from an embolized to a functional conduit. More research is needed to investigate the multiphase interactions between microfibrils of pit membranes, gas, xylem sap, and polar lipids as surfactants. The functional importance of the T_pm,max/D_pm ratio also requires a better understanding of the mechanical properties of pit membranes.

Materials and methods

Study site, species, and sampling procedure

The experiment was conducted in French Guiana at the Paracou experimental station (https://paracou.cirad.fr/website; 5°16′26″N, 52°55′26″W), which represents a lowland tropical rainforest (Gourlet-Fleury et al., 2004). The warm and wet tropical climate of French Guiana is highly seasonal due to the North–South movement of the Inter-Tropical Convergence Zone. Mean (±se) annual air temperature is 25.7°C ± 0.1°C, and the mean annual precipitation is 3,102 mm ± 70 mm (data between 2004 and 2014; Aguilos et al., 2019). There is a dry season lasting from mid-August to mid-November, during which rainfall is <100 mm month⁻¹.

Overall, a first sampling was held between January and July 2017 to measure stem embolism resistance and pit membrane traits in stem xylem (Ziegler et al. 2019; Levionnois et al. 2021a, 2021b). A second sampling was held between November 2018 and March 2019 to measure leaf embolism resistance as published in Levionnois et al. (2020). During this second sampling, (1) measurements of pit membrane traits, (2) a gold perfusion experiment to determine pit membrane pore sizes, and (3) a dye experiment for polar lipids were conducted. All trees measured during the second sampling were part of the first sampling.

We sampled only dominant canopy trees from terra firme. A total of 18 tree species and 50 trees were sampled by professional tree climbers in the canopy. For P₅₀ measurements, we measured three trees per species for 14 species, and two trees per species for four species (Table 2). The success rate in obtaining P₅₀ values of stems was low for some species, explaining why there are species with only two individual replicates per species (Ziegler et al., 2019; Levionnois et al., 2020). The studied species covered a broad phylogenetic diversity such that the main clades of the flowering plants were represented, that is, magnoliids, rosids, and asterids (Table 2).

Leaf xylem measurements

Leaf embolism resistance was measured as the leaf water potential inducing 12%, 50%, and 88% loss of conductivity, respectively (P_12,leaf, P_50,leaf, and P_88,leaf, respectively; MPa; Table 1; Pammenter and Van der Willigen, 1998; Domec and Gartner, 2001). We generally sampled three trees per sampling day, during the morning and before solar midday, in order to avoid too negative leaf water potentials and possibly native leaf embolism. Only sun-exposed branches were sampled as much as possible. A single 1-meter-long canopy branch was sampled per tree, with approximately 50 leaves or leaflets to monitor water potentials as described below. To measure leaf xylem embolism resistance, we relied on the optical light transmission method (Brodribb et al., 2016b; Brodribb et al., 2016a). For details about the method and data for each species, we refer the reader to Levionnois et al. (2020).

One leaf midrib sample per tree and per species was conserved in water, and placed in a fridge at 5°C to keep the midrib fresh and to avoid pit membrane shrinkage by chemical treatment or dehydration (Kaack et al., 2019; Kotowska et al., 2020). Midrib samples were then shipped to Ulm University (Ulm, Germany). These samples were used to measure pit membrane thickness and diameter based on TEM. For pit membrane traits, we measured one tree per species and one leaf per tree for a total of 18 species. C.schomburgkianus, G. glabra, and G. hexapetala were sampled more intensely (∼35-cm long branches with 6–10 leaves) for gold perfusion experiments to estimate pore constriction diameters in pit membranes of leaf xylem (Supplemental Method S1). Perfusion of an FM1–43 dye solution was also applied to observe polar lipids in leaf xylem (Supplemental Method S1). These three species were selected to be representative of species with low, middle, and high embolism resistance of leaf xylem. For the gold perfusion experiment and the dye experiment, we also measured one tree per species.

TEM measurements were conducted based on well-established and previously described protocols (Kotowska et al., 2020; Zhang et al., 2020b; Kaack et al., 2021; Levionnois et al., 2021a). The detailed protocol is described in the Supporting Information (Supplemental Method S1). As storage in water might cause a slight pit membrane shrinkage (Schuldt et al., 2016; Levionnois et al., 2021a), we paid attention to remove shrunken, aspired, deformed, or ruptured pit membranes. Shrunken pit membranes become thin, and are characterized by a dark, homogeneous color. Nonshrunken pit membranes, however, typically show a granular appearance after treatment with OsO₄, and are much more transparent, from almost invisible to grey (Li et al., 2016; Zhang et al., 2020b; Levionnois et al., 2021a). Nevertheless, we assumed that the thickest pit membranes, with granular and transparent appearance, might reflect the hydrated pit membrane thickness in the field, as previously tested and discussed by Levionnois et al. (2021a) for stems. We measured for each wood sample the mean pit membrane thickness (T_pm, nm) and the maximum pit membrane thickness (T_pm,max, nm; Figure 1, A and B). About 30 bordered pits were measured per leaf midrib sample (with one leaf midrib sample per tree and one tree per species), using one to three sections per sample. While most pits were intervessel pits, it is likely that some vessel-tracheid pits were included, as it was sometimes impossible to distinguish tracheids from narrow, fibriform vessels under TEM. From this population of bordered pits measured for T_pm, we estimated the pit membrane diameter (D_pm, µm, Figure 1D) by retrieving pit membrane observations with apertures, as this criterion proves that sections were made through the center of the pit border where the diameter is largest. Since TEM images were based on transverse sections, D_pm represented the horizontal diameter of an intervessel pit border. We used T_pm,max and D_pm to compute the pit membrane thickness-to-diameter ratio. All image analyses and measurements were conducted with ImageJ software version 1.43u.

In the gold perfusion experiment, the penetration of colloidal gold particles of different sizes (5, 10, 20, and 50 nm) in pit membranes was qualitatively analyzed. We, therefore, selected pit membranes that showed gold particles of all sizes on one side of the pit membrane. Penetration was defined as gold particles that entered 30 nm or deeper into a pit membrane (Supplemental Methods S1). Ten pit membranes in one or two transverse sections of each species were analyzed. The presence of polar lipids within vessels was also qualitatively assessed by the presence of FM1-43 signals following Schenk et al. (2018; Supplemental Methods S1). At least three sections of each stem of a species were analyzed. To evaluate results obtained from the gold perfusion experiment (estimation of pore constriction sizes) and the dye experiment (observation of polar lipids), we calculated a theoretical air-seeding pressure based on a modified Young–Laplace equation (Kaack et al., 2019):

$$\Delta \mathrm{P}=\varkappa 2 \gamma \hspace{0.17em}\text{cos}\hspace{0.17em}(\alpha )/\mathrm{R}$$

where ΔP is the pressure required to induce air-seeding (the pressure difference between the gas phase, including water vapor, and the xylem sap pressure), ϰ is a dimensionless shape correction factor which was assumed to be equal to 0.5 (Schenk et al., 2015; Kaack et al., 2021), γ is the surface tension of xylem sap (mN m⁻¹), α is the contact angle of the gas–xylem sap interface with the solid cellulose microfibril and was assumed to be equal to 0 (Kaack et al., 2021), and R is the pore constriction radius (nm). We assumed that γ was either 72 or 25 mN m⁻¹, which corresponded to the surface tension of pure water and was close to the bulk surface tension of xylem sap (Christensen-Dalsgaard et al., 2011), and the equilibrium surface tension of xylem sap based on dynamic surface tension (Schenk et al., 2017; Yang et al., 2020).

A complete list of traits with their abbreviations is presented in Table 1.

Stem xylem measurements from published data for the assessment of vulnerability segmentation

To investigate the linkage between leaf–stem variation in xylem anatomy and vulnerability segmentation, data on stem embolism resistance (measured as the stem water potential inducing 12%, 50%, and 88% loss of conductivity) and stem anatomical traits were retrieved from previously published studies (Ziegler et al., 2019; Levionnois et al., 2021a), where detailed descriptions of the method and the data for each species were provided. Vulnerability curves were measured with the flow-centrifugation technique using a Cavi1000 system (DGMeca, BIOGECO lab, Gradignan, France).

Stem wood samples for anatomical measurements were taken at the most distal part of the branches that were also used to measure embolism resistance. These wood samples were conserved in water and stored in a fridge at 5°C to avoid wood decay and pit membrane shrinkage (Kaack et al., 2019; Kotowska et al., 2020). TEM measurements of pit membrane thickness (nm) were conducted the same way as described for midribs above, with the detailed protocol provided in the Supporting Information (Supplemental Method S1). Measurements of the pit membrane diameter (µm) were conducted with a laser microscope (VK-9710K Color 3D Laser Microscope), with a 100× CFPlan Nikon objective to measure mean intervessel pit membrane diameter. Detailed descriptions of the method and the data are provided by Levionnois et al. (2021a, 2021b).

Statistical analyses

All statistical analyses were performed with R software (R Core Team, 2018). Data were tested for normality (Shapiro–Wilk test; α = 0.05). For correlations between anatomical traits and embolism resistance, we used Pearson or Spearman correlation analyses. Comparison tests were conducted with Student’s, Welch’s, or Mann–Whitney–Wilcoxon’s test, depending on the parameters of the samples (effectives, normality of distribution, and variance).

We observed vestured pits in five species (Bocoaprouacensis, Eperuafalcata, E. grandiflora, Qualearosea, and Tachigalimelinonii; Figure 1C). Pit morphology (vestured versus nonvestured) may affect the relationship between pit membrane traits and embolism resistance (Levionnois et al., 2021a, 2021b). Accordingly, we applied multiple linear models for any effect of pit morphology, for the relationships between P_50,leaf and T_pm,max, P_50,leaf and D_pm, and P_50,leaf and the T_pm,max-to-D_pm ratio. Log-transformation was applied to meet assumptions of normality when necessary.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Table S1. Multiple models to test if the relationship between pit membrane traits and leaf embolism resistance (P_50,leaf) is affected by the presence or absence of vestured pits, based on trait values at the individual level.

Supplemental Table S2. Pearson and Spearman correlations between leaf embolism resistance and anatomical traits based on trait values at the individual level.

Supplemental Method S1. Details for anatomy.

P.H., S.J., and S.L. designed the study. S.L, C.Z., S.C., C.S., and P.H. collected field samples. S.L. and P.H. managed anatomical cross-sections of leaves with the support of the SILVATECH platform, INRAE. L.K. N.A., S.L., and S.J. performed TEM, gold perfusion, and lipids surfactant observations. S.L., L.K., N.A., P.H., and S.J. performed image analysis. S.L. performed data analysis. S.L. wrote the manuscript with contributions from S.J., P.H., and L.K. All authors discussed the results and contributed to the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions.) is Sébastien Levionnois ([email protected]).

Acknowledgments

We thank the climbing team for canopy sampling: Jocelyn Cazal, Valentine Alt, Samuel Counil, Anthony Percevaux, and Elodie Courtois. We thank the following colleagues for technical assistance during field work: Oscar Affholder, Louise Authier, Anne Baranger, Maxime Bellifa, Benoit Burban, Maëlle Cario, Bruno Clair, Maxime Corbin, Elia Dardevet, Alexandre De Haldat Du Lys, Emilien Fort, Frederic Fung Fong You, Eva Gril, Thomas Saint-Germain, and Ruth Tchana Wandji.

We thank Géraldine Derroire and Aurélie Dourdain for permission to use the facilities at Paracou. Paracou Forest Research Station in French Guiana is managed and supported by CIRAD, UMR EcoFoG (https://paracou.cirad.fr/), and benefit from financial support by a French Investissement d’Avenir program (Labex CEBA ANR-10-LABX-25-01).

We would like to thank SILVATECH (Silvatech, INRAE, 2018. Structural and functional analysis of tree and wood Facility, doi: 10.15454/1.5572400113627854E12) from UMR 1434 SILVA, 1136 IAM, 1138 BEF, and 4370 EA LERMAB EEF research center INRAE Nancy-Lorraine for support with sectioning wood. We also would like to thank the PHENOBOIS platform for embolism resistance measurements (PHENOBOIS, Bordeaux France).

Funding

This study was funded by the GFclim project (FEDER 20142020, Project GY0006894). This work has benefited from an “Investissement d’Avenir” grant managed by Agence Nationale de la Recherche (CEBA, ref. ANR-10-LABX-25-01). S.L. was supported by a doctoral fellowship from CEBA. SJ and LK acknowledge financial support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation, project number 383393940).

Conflict of interest statement. There is no conflict of interest.

Author notes

†

Joint senior authors