Embolism resistance in petioles and leaflets of palms

Abstract

Background and aims

Hydraulic studies are currently biased towards conifers and dicotyledonous angiosperms; responses of arborescent monocots to increasing temperature and drought remain poorly known. This study aims to assess xylem resistance to drought-induced embolism in palms.

Methods

We quantified embolism resistance via P₅₀ (xylem pressure inducing 50 % embolism or loss of hydraulic conductivity) in petioles and leaflets of six palm species differing in habitat and phylogenetic relatedness using three techniques: in vivo X-ray-based microcomputed tomography, the in situ flow centrifuge technique and the optical vulnerability method.

Key results

Our results show that P₅₀ of petioles varies greatly in the palm family, from −2.2 ± 0.4 MPa in Dypsis baronii to −5.8 ± 0.3 MPa in Rhapis excelsa (mean ± s.e.). No difference or weak differences were found between petioles and leaf blades within species. Surprisingly, where differences occurred, leaflets were less vulnerable to embolism than petioles. Embolism resistance was not correlated with conduit size (r = 0.37, P = 0.11).

Conclusions

This study represents the first estimate of drought-induced xylem embolism in palms across biomes and provides the first step towards understanding hydraulic adaptations in long-lived arborescent monocots. It showed an almost 3-fold range of embolism resistance between palm species, as large as that reported in all angiosperms. We found little evidence for hydraulic segmentation between leaflets and petioles in palms, suggesting that when it happens, hydraulic segregation may lack a clear relationship with organ cost or replaceability.

INTRODUCTION

Global warming is likely to result in a worldwide increase in aridity (Sherwood and Fu, 2014; Huang et al., 2016). Dramatic changes in dry and wet periods are predicted for even the most humid regions (Duffy et al., 2015). Therefore, it is critical to reliably predict the likelihood of a plant succumbing to drought. Interspecific patterns of drought-induced mortality are best predicted on a global scale by xylem resistance to embolism and particularly by the species’ hydraulic safety margin (Anderegg et al., 2016). Although data on xylem vulnerability have increased in recent decades, the available information is strongly biased towards conifers and dicotyledonous angiosperms (Choat et al., 2012). In contrast, the hydraulics of monocots remain poorly understood.

Monocots account for 25 % of angiosperm species, including economically important plant families such as grasses, orchids, lilies and palms (Anderson and Janssen, 2009). Embolism resistance in monocots is known mostly for species in the grass family (e.g. Lens et al., 2016) and, more recently, bromeliads (Males, 2017). Embolism resistance in the palm family has been relatively poorly studied (but see Renninger and Phillips, 2011). Palms are important components of carbon (Fauset et al., 2015) and water cycling (Renninger and Phillips, 2010a) in tropical forests. Yet the role of palms as ecosystem engineers and their potential decline under climate change have been neglected due to a relative lack of understanding of their hydraulic functioning.

The palm family (Arecaceae) includes >2600 species and 181 genera distributed in tropical and subtropical regions worldwide (Baker and Dransfield, 2016). Over 90 % of palm species are restricted to tropical rainforests (Couvreur et al., 2011), though, intriguingly, large arborescent palms can also occur in extremely arid climates such as deserts. Palms differ strongly, both morphologically and functionally, from other species in their ecosystems, including the large woody trees (dicotyledonous and coniferous) that dominate forests worldwide and the small herbaceous grasses that dominate open environments (Tomlinson, 2006). Most of the factors that make palms hydraulically distinct are potentially linked to three key features. First, palms are perennial, long-lived plants lacking dormancy mechanisms to avoid seasonally unfavourable environmental conditions. Second, palms (like all monocots) lack a vascular cambium and both xylem and phloem cells need to remain functional throughout an individual lifespan. Third, the early loss of the primary root that characterizes all monocots (which in dicotyledonous trees develops into a branching tap root) restricts access to deep soil water. These differences – sometimes expressed as constraints (e.g. Alves et al., 2004; Tomlinson, 2006) – are linked to the dominance in monocots of herbaceous life-forms.

The lack of a vascular cambium in monocots restricts thickening growth and may result in reduced efficiency for secure water transport when achieving greater stature. Many monocots combine a vessel-based root xylem system with an entirely tracheid-based xylem system in the stem and leaf (Carlquist, 2012). The relative lack of vessels in aerial parts makes monocots more efficient in avoiding the spread of air embolism, but also restricts plant size, water flow and transpiration (Carlquist, 2012). However, palms appear to have overcome this limitation by displaying a broad range of size and growth forms, almost equivalent to that of all other tropical plant groups combined (Couvreur et al., 2015). Even though they lack ‘true’ secondary growth by means of a vascular cambium, different palm species can achieve heights as tall as eudicot trees and develop climbing stems that exceed the length of any vascular plant (Tomlinson, 2006). Like dicot trees, arborescent palms taper their vessels from the base to the apex of the plant to overcome hydrodynamic resistance to water flow from root to leaf (Petit et al., 2014). The tallest palm (Ceroxylon quindiuense) can grow up to 61 m in height (Bernal et al., 2018) and there is no evidence for hydraulic limitation in height for large-canopy palms living under non-limiting water conditions (Renninger and Phillips, 2011).

The hydraulic efficiency of the vascular architecture of palms surpasses the theoretical hydraulic conductivity of all tropical and temperate trees (Renninger et al., 2013). This is because palms, like monocots in general, have broader vessels than dicots (Olson et al., 2014). In contrast with other monocots, many palms have broad vessels distributed along all vegetative organs (Carlquist, 2012). Broader vessels make palms efficient in water transport, but this strategy can have its drawbacks. Embolism resistance can decrease with increase in vessel diameter (Hacke et al., 2006), which suggests that palm xylem would potentially be vulnerable to embolism (Carlquist, 2012). Although the trade-off between xylem efficiency and safety is weaker than previously thought, conduit diameter still imposes an upper limit for hydraulic safety (Gleason et al., 2016).

Structurally, xylem safeguarding in palms can be achieved by discontinuity between stem and leaf vessels (Tyree and Zimmermann, 2002). Leaf-trace linkage to the stem xylem occurs entirely via protoxylem elements (Tomlinson, 1990), which in palms are mainly tracheids rather than vessels (Tomlinson et al., 2011). The protoxylem and metaxylem connection occurs deep within the central stem cylinder (Tomlinson, 1990), securing a long distance between transpiring leaves, where embolism is more likely, and the axial xylem system, where embolism must be avoided at all costs. Even within the same vascular bundle, vessels in palms may not share intercellular pathways but can be separated by several layers of parenchymatous cells, potentially representing an adaptation for preventing embolism spread between vessels (Tyree and Zimmerman, 2002). Extreme examples of xylem segmentation between leaf and stem xylem are observed in climbing rattan palms of the genus Calamus, in which there is no connection between protoxylem and metaxylem even within the same bundle (Tomlinson et al., 2001).

It was early hypothesized that palms restrict embolism to their expendable leaves to prevent embolism generating irreversible damage to their long-lived stems and to ensure continued growth of the plant (Zimmermann, 1982; Zimmermann and Sperry, 1983; Sperry, 1985). However, it is unlikely that palm species rely exclusively on hydraulic segmentation for the maintenance of their hydraulic functioning. Palms store large amounts of water in their trunks and nearly 20 % of their daily transpiration can be sustained by this source (Renninger et al., 2009; Renninger and Phillips, 2010b; Sperling et al., 2015). This effect is negligible in some dicotyledonous and coniferous trees with relatively small amounts of parenchyma (Jupa et al., 2016), but it can be significant in plants such as palms, in which parenchyma can occupy >50 % of the cross-sectional area (Renninger and Phillips, 2010b).

Embolism resistance describes the lethal threshold of dehydration in plants beyond which the plant fails to recover (Urli et al., 2013). Thus, embolism resistance sets the upper limit of plant transpiration, with implications for photosynthesis productivity (Brodribb, 2009) and ultimately biome colonization (Choat et al., 2012). Palms have wider vessels, fewer vessel-to-vessel connections and higher proportions of parenchymatous tissue than other plants for which hydraulic data are available. All these features are expected to affect drought resistance, but the extent to which they render palms more or less vulnerable than other plants is unknown.

Palms rely on early-formed xylem vessels for their entire life, which can in some cases exceed 100 years (Tomlinson, 2006). Limited evidence has shown that some palms are resistant to embolism onset (Sperry, 1985; Renninger and Phillips, 2011), but the scarce data on palm hydraulics are insufficient to identify the range of variation in embolism resistance. In this paper, we determine the embolism resistance of six palm species from different biomes and compare them with the broad range of embolism resistance previously reported in angiosperms and conifers. To test the hydraulic segmentation hypothesis, we monitor embolism formation and spread in palm leaves with different amounts of parenchyma and compare P₅₀ values (xylem pressure inducing 50 % embolism or loss of hydraulic conductivity) of leaves and petioles.

MATERIALS AND METHODS

Plant material

We selected 7 palm species for the experiments (Supplementary Data Table S1), mostly belonging to the two most widely cultivated palm subfamilies, Arecoideae and Coryphoideae. We targeted species that show a wide range of natural habitats (Supplementary Data Fig. S2). Plants of Chamaedorea radicalis, Howea forsteriana, Rhapis excelsa and Trachycarpus fortunei were used for direct micro-CT visualization of embolism formation. Plants 0.6–1.2 m in height, depending on species, were sourced from a commercial nursery (À L’Ombre des Figuiers, Pluguffan, France). A second set of plants was used for experiments at the UMR-BIOGECO laboratory (University of Bordeaux, France). Additionally, for the experiments at UMR-BIOGECO, excised leaf petioles from adult plants of Calamus longipinna, Chamaedorea seifrizii, Dypsis baronii, Dypsis decaryi, Dypsis lutescens, Phoenix canariensis, Rhapis subtilis and Salacca ramosiana were sourced from the living collections of the Royal Botanic Gardens, Kew (Richmond, UK).

Embolism visualization in intact plants (micro-CT)

A first set of 12 plants was shipped to the SOLEIL synchrotron (Paris, France) in April 2017 for direct micro-CT visualization of embolism formation 2 d before the dehydration experiments started. The first scan was performed on petioles of well-watered plants while they were still potted to quantify the presence of any native embolism events. After the first scan, plants were removed from the pots and the soil was carefully removed from the roots. Uprooted plants were allowed to dry on the bench for the duration of the experiments (3–6 d). For each scan, the intact plant was placed in the rotating stage and their petioles were fixed to minimize the risk of movement during tomography.

Observations were conducted at the SOLEIL micro-CT PSICHE beam line (King et al., 2016). The petioles of three 1.5-m tall intact palms were scanned using a high-flux (3.1011 photons mm⁻²) 25 keV monochromatic X-ray beam, while being rotated from 0° to 180° using a continuous rotation mode. Thanks to the high load capacity and free central aperture of the rotation stage, imaging cross-sections were also selected near the middle of the petiole. Dehydration was progressively induced in the three bare-rooted plants per species by stopping irrigation and exposing them to a fan to accelerate the process. X-ray projections were collected with a 50-ms exposure time during rotation and recorded with an Orca-Flash sCMOS camera (Hamamatsu Photonics, Naka-ku, Japan) equipped with a 250-μm thick LuAG scintillator. The scan time was 120 s for each sample and yielded a stack of 2048 TIFF image slices. Each petiole was scanned once throughout the dehydration of each individual and up to 11 petioles were scanned per species. Tomographic reconstructions were conducted using the Paganin method (Paganin et al., 2002) in PyHST2 software (Mirone et al., 2014) and resulted in two-bit volumic images with 3.02-μm voxel resolution.

The water potential of the plant was measured from a detached leaf with a Scholander-type pressure chamber (DG Meca, Gradignan, France) immediately after each scan. The plant was then allowed to dry until the next scan. The scans were spaced 6–24 h apart, depending on the species and dehydration stage. A maximum of four scans were taken per plant and we did not observe any evidence of damage during the experiments. The experiment ended when the plant embolism percentage was near 100. Subsequently, an additional scan was performed after removing the segment immediately above the scanned area (~1 mm) to embolize all the vessels artefactually. This final cut scan was then used as the reference scan to estimate the maximum area of embolized vessels.

The areas of embolized and functional vessels were determined in each scan image using Image J software (NIH). Conduit diameter (D) for each vessel was calculated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle D=2\times \surd (A/\pi )}\end{array}}$$ (1)

where A is the area of the vessel. The sum of conduit diameter from the reference scans was used to calculate the maximum theoretical conductivity and from the scans taken during the bench-dry experiment used to calculate the theoretical conductivity at each water potential (K_th) according to the Hagen–Poiseuille equation:

$${\displaystyle \begin{array}{l}{\displaystyle K_{\mathrm{t}\mathrm{h}}=\mathrm{\Sigma }\pi D^4/128\eta }\end{array}}$$ (2)

where D is the vessel diameter and η is the viscosity of water (1.002 MPa s⁻¹ at 20 °C). The percentage loss of xylem conductance (PLC) for each scan was then calculated from the conductivity of the functional vessels observed in each scan as a function of the maximum theoretical conductivity:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{P}\mathrm{L}\mathrm{C}=100\times (1-K_{\mathrm{\Psi }}/K_{\mathrm{m}\mathrm{a}\mathrm{x}})}\end{array}}$$ (3)

where K_Ψ is the theoretical conductivity of the functional vessels and K_max is the maximum theoretical conductivity.

Embolism visualization in detached leaves (optical vulnerability technique)

The onset and accumulation of embolism during palm leaf desiccation were monitored on five to six individuals per species using the optical vulnerability (OV) method (Brodribb et al., 2016; full details are available at http://www.opensourceov.org). Briefly, all individuals were kept well hydrated prior to measurements. During the experimental set-up, where palms had to be uprooted, both roots and non-scanned leaves were kept bagged to avoid water loss by transpiration. Only after complete installation of the scanner and psychrometer was the targeted leaf excised from the plant, at least 30 cm from the scan location. The first embolism event was observed at least 12 h after excision, confirming that no detectable embolism was caused by leaf excision.

For each individual, two leaflets were placed on a scanner (Perfection V800 Photo, Epson, Suwa, Japan) and fixed with a transparent glass and adhesive tape to minimize leaf movement during dehydration. Scan magnification, brightness and contrast and leaf area were set to optimize visualization of embolisms in both main veins and cross-veins. Each leaflet was automatically scanned every 5 min using computer automation software (AutoIt 3) until mesophyll cells turned brown, indicating cell death.

Water potentials were recorded every 30 min using a stem psychrometer (ICT International, Armidale, NSW, Australia) that was attached to the petiole a few centimetres below the scanned leaflets. Accuracy of the readings was double-checked by periodically measuring xylem water potentials on neighbouring leaflets using a Scholander pressure chamber (DG Meca, Gradignan, France). Once completed, stacks of images were analysed using ImageJ (National Institutes of Health, New York, NY, USA) to highlight embolism events that occurred throughout dehydration (for the detailed procedure see Brodribb et al., 2016 and the website https://github.com/OpenSourceOV/image-processing-instructions). The cumulative percentage embolized area dataset obtained was then coupled with the corresponding water potential sequences. Vulnerability curves were constructed as a percentage embolism function of water potential.

Hydraulic measurements of embolism resistance (Cavitron)

Resistance to drought-induced embolism in the petioles was determined using an in situ flow centrifuge technique (Cavitron, Cochard et al., 2005) at the GENOBOIS hydraulic platform.

As techniques that build vulnerability curves from plant sections can suffer from open-vessel artefact (Cochard et al., 2013; Torres-Ruiz, 2017), three samples per species were used to test for open-vessel artefact (Torres-Ruiz et al., 2017) by injecting air at 2 bar at one end. Eight species failed the open-vessel artefact test, produced unsatisfactory vulnerability curves and were removed from our analysis. Unfortunately, we were unable to obtain Cavitron data for any of the species for which we collected data using micro-CT and OV techniques, limiting cross-validation among all techniques.

No open-vessel artefact was detected for four palm species: D. baronii, D. decaryi, D. lutescens, T. fortunei. For these species, petioles were recut under water with a razor blade to a standard length of 0.27 m and infiltrated with a reference ionic solution of 10 mm 25 KCl and 1 mm CaCl₂ in deionized ultrapure water. Centrifugal force was used to generate negative pressure into the xylem and induce embolism with a rotor of 0.27 m diameter. This method allows measurement of xylem conductance under negative pressure using the custom software Cavisoft 4.0 (University of Bordeaux, Pessac, France). Initially, the maximum conductance of petioles (in m² MPa⁻¹ s⁻¹) was calculated under low xylem pressures. The percentage loss of xylem conductance of the petioles was calculated at different xylem pressures from −0.8 to −9 MPa using eqn (3), except for T. fortunei, where only a negligible loss of conductance was measured even in the most negative water potentials.

Vulnerability curves

Embolism resistance in plants is commonly expressed by the xylem pressure inducing 50 % loss of conductivity or its equivalent in embolized area (P₅₀; Choat et al., 2012). This embolism resistance parameter can be calculated from the xylem vulnerability curve, a 2-D graph relating percentage of xylem conductance loss and xylem pressure (Cochard et al., 2013). Vulnerability curves were obtained with the three different techniques (micro-CT, OV and Cavitron). For each sample, the relationship between PLC/percentage of embolized pixels and xylem water pressure was fitted with the following sigmoidal equation (Pammenter and Vander Willigen, 1998):

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{P}\mathrm{L}\mathrm{C}=100/(1+\mathrm{e}\mathrm{x}\mathrm{p}(S/25\times (P_{\mathrm{i}}-P_{50})))}\end{array}}$$ (4)

where P_i is the xylem pressure at each point, P₅₀ is in MPa and S (% MPa^–1) is the slope of the vulnerability curve at the inflection point.

Statistical analysis

Individual vulnerability curves were obtained for each sample from OV and Cavitron data. To calculate the value of embolism resistance for each species, the values of P₅₀ of all the samples were averaged. A single vulnerability curve was obtained based on the micro-CT method. In this case, the value of embolism resistance for each species is given by the P₅₀ of the single curve adjusted from multiple individual data points. Embolism resistance differences between petiole and leaflets was tested using the one-sample t-test, since only a single vulnerability curve was obtained for petioles in most cases. We compared embolism resistance in petioles by analysis of variance followed by Tukey’s multiple comparisons with 95 % family-wise confidence level. Pearson’s linear correlations were calculated to test for correlations between embolism resistance and conduit size.

Vulnerability curves were fitted using the NLIN procedure in SAS 9.4 (SAS Institute, Cary, NC, USA). Visualizations of 3-D reconstructions of plant petioles were generated with the Drishti volume exploration and presentation tool (Limaye, 2012). All remaining data manipulation, analysis and plots were performed in R v.3.5.1 (R Core Development Team, 2018) unless otherwise stated.

RESULTS

Embolism resistance in palm leaves and petioles

Embolism resistance was characterized in the petioles of six palm species that differed in habitat and phylogenetic relatedness (Supplementary Data Table S1) using three different techniques: in vivo X-ray-based micro-CT, the in situ flow centrifuge technique and the OV method. We found a large and significant variation in P₅₀ between species (F = 71.3, P < 0.01, Table 1), ranging from −2.24 ± 0.4 MPa in D. baronii to −5.82 ± 0.3 MPa in R. excelsa (mean ± SE). Embolism resistance in R. excelsa differed significantly from that in all other species (Table 1). Moreover, palm petioles showed an embolism resistance range equivalent to that observed for angiosperms across all the biomes in which palms grow naturally (Fig. 1).

Table 1.

Comparison of vulnerability curves parameters between species and leaf section

Species	Leaf section	Method	P 12	P 50	P 88	Slope
Chamaedorea radicalis	Blade	Optical vulnerability	−3.12 ± 0.15A	−3.3 ± 0.16A	−3.47 ± 0.18 A	320.05A
Howea forsteriana			−3.75 ± 0.29A	−4.08 ± 0.38A	−4.41 ± 0.49A	213.32 A
Rhapis excelsa			−5.3 ± 0.18B	−5.85 ± 0.13B	−6.4 ± 0.25B	151.01A
Dypsis baronii	Petiole	Cavitron	−1.17 ± 0.27 a	−2.24 ± 0.16a	−3.31 ± 0.12a	50.08a
Dypsis decaryi			−1.51 ± 0.23a	−2.79 ± 0.12b	−4.06 ± 0.09b	42.09a
Dypsis lutescens			−1.58 ± 0.21a	−2.59 ± 0.09ab	−3.6 ± 0.09ab	56.29a
Chamaedorea radicalis	Petiole	Micro-CT	−2.23 ± 0.29ab	−2.63 ± 0.1ab	−3.03 ± 0.1a	325.78a
Howea forsteriana			−2.9 ± 0.18b	−3.14 ± 0.04b	−3.38 ± 0.12a	497.66a
Rhapis excelsa			−5.56 ± 0.18c	−5.82 ± 0.2c	−6.09 ± 0.33c	616.51a

P ₁₂, P₅₀ and P₈₈ are the xylem potentials associated with 12, 50 and 88 % loss of hydraulic conductance, respectively.

Values are mean ± s.e.

Superscript letters (capitals for leaf blade and lower-case for petiole comparisons) for P₅₀ represent significant differences between species (Tukey’s multiple comparisons with 95 % family-wise confidence level; Supplementary Data Table S4). Note that comparisons were made only within the same leaf section. For results of comparisons between leaf sections, see Fig. 4.

Fig. 1.

Embolism resistance in palm species occurring in different biomes in comparison with other plants. (A) Embolism resistance in palm species. (B) Embolism resistance of angiosperms and gymnosperm species for each biome (Choat et al., 2012). Note that only the six (out of 14) terrestrial biomes where palms can occur are represented here with same colors used in (A) and (B) to identify biomes. Biome classification follows Olson et al. (2001). The length of the bars (except for Trachycarpus fortunei) represents species P₅₀ with their standard error as error bars. Data on Iriartea deltoidea are from Renninger and Phillips (2011). The asterisk (*) denotes that we could not obtain P₅₀ values for this species as little embolism was observed at −8 MPa (see Results section for details).

No difference or weak differences in P₅₀ were observed between petioles and leaflets across the three studied species (Fig. 2). The difference was significant only in C. radicalis (t = −3.463, P = 0.013), in which leaflets were more resistant than petioles, but not in R. excelsa (t = −0.121, P = 0.909) or H. forsteriana (t = −2.456, P = 0.068). Embolism resistance in petioles was not correlated with conduit size across species (r = 0.37, P = 011; Supplementary Data Fig. S3). Larger vessels embolized first in R. excelsa, but did not embolize first at in C. radicalis and H. forsteriana (Fig. 3, Supplementary Data Table S4).

Fig. 2.

Drought-induced xylem embolism resistance for three palm species: (A) Chamaedorea radicalis, (B) Howea forsteriana and (C) Rhapis excelsa. Vulnerability curves for each species were obtained by direct observation of petioles of intact leaves (X-ray micro-CT, dashed lines) and leaflets of detached leaves (OV method, continuous lines). Black circles correspond to individual micro-CT sample values; grey shading represents the 95 % CI.

Fig. 3.

Frequency distribution of vessel area for three species and each embolism stage.

Micro-CT observation of embolism formation in palm petioles

High-resolution micro-CT images allowed visualization of vessel embolism in petioles of intact plants (Fig. 4). We observed the first embolism events within 1–3 d after the start of the bench-dry experiments. After onset of embolism, the embolism spreads towards most vessels following small changes in water potential. The first embolism events were observed below −2 MPa in C. radicalis, −2.2 MPa in H. forsteriana, −5.3 MPa in R. excelsa and −5.8 MPa in T. fortunei. Within 6–24 h after the first events, most vessels were embolized in all species except T. fortunei. The largest amount of embolism observed for T. fortunei with micro-CT was ~4 % at −8 MPa (Supplementary Data Fig. S5) and no vulnerability curve was obtained for this highly resistant species.

Fig. 4.

2-D digital cross-section X-ray micro-CT images of petioles in desiccating intact plants. Species [Chamaedorea radicalis (A–C), Howea forsteriana (D–F), Rhapis excelsa (G–I) and Trachycarpus fortunei (J–L)] are oriented in rows and desiccation stages (all vessels full of sap, onset of embolism and advanced stages of embolism) in columns. Embolized vessels and air spaces are shown in black. Water potential of the plant at the moment when the image was obtained is shown in the top-right corner of each image. Insets show a single vascular bundle isolated from the full image (not to scale). Image sequences show contrasting amounts of petiole shrinkage and parenchyma damage among species. Scale bars = 500 µm.

In well-hydrated plants, water is stored not only in parenchyma cells but also in intercellular spaces. We observed high amounts of water in intercellular spaces in C. radicalis and H. forsteriana, and smaller amounts in R. excelsa and T. fortunei (Fig. 5). A high proportion of hydrated petiole tissue consists of parenchyma (85 % in C. radicalis, 75 % in H. forsteriana and 74 % in R. excelsa for the samples shown in Fig. 4). When leaves dry out, parenchyma tissue contracts and petioles shrink. Petiole shrinkage displayed marked differences between species (Fig. 4). We observed a reduction in petiole cross-sectional area of 14 % in C. radicalis, 40 % in H. forsteriana and 60 % in R. excelsa at advanced stages of embolism when most of the vessels were embolized. Differences in xylem embolism resistance between species coincided with trends of shrinkage rather than tissue proportion. Expansion of air space within the C. radicalis petiole reduced area shrinkage for this species.

Fig. 5.

2-D digital longitudinal section X-ray micro-CT images of petioles in desiccating intact plants of (A) Chamaedorea radicalis, (B) Howea forsteriana, (C) Rhapis excelsa and (D) Trachycarpus fortunei. Arrows point to parenchyma cells (PA), vascular bundles (VB), intercellular air spaces (black) and water-filled vessels (dark grey shading). Scale bars = 100 µm.

Optical visualization of embolism spread in palm leaflets

Embolism spread in leaflets was visualized using the OV technique (Fig. 6). We observed the first embolism within 12 h (−4.0 MPa in R. excelsa) to 30 d (−3.41 MPa in C. radicalis) after the start of the experiments. Within 1 (−10.11 MPa in R. excelsa) to 18 d (−3.66 MPa in C. radicalis) after the first events, all vessels were embolized. The time to embolism spread can be up to 10-fold in palms (Supplementary Data Fig. S6, left). In our experimental design, C. radicalis took from 20 to 40 d to dehydrate, compared with <4 d in R. excelsa. Large parallel veins were the first to be embolized (Fig. 4A, D, G). Subsequently, embolism spread to the secondary minor parallel and cross-veins.

Fig. 6.

Embolism spread in leaves imaged using a transmission light scanner. Species (Chamaedorea radicalis [A-C], Howea forsteriana [D-F] and Rhapis excelsa [G-I]) are oriented in rows and embolism spread stages (onset of embolism [A,D,G], propagation to minor veins [B,E,H] and advanced stages of embolism [C,F,I]) in columns. Embolism events are coloured in relation to the time of occurrence after the first embolism event, with cooler colours (dark blue/purple) representing earlier events than warmer colours (red/orange/yellow). The first embolism event normally happens in a large parallel vein and then extends to minor parallel and cross-veins. A time comparison of embolism propagation can be observed in Supplementary Data Movie S12.

We observed leaflet shrinkage during the bench-dry experiments using the OV method for all samples. Most of the shrinkage in palm leaflets occurred longitudinally as a result of the parallel disposition of the major veins. Within species, longitudinal shrinkage varied from 3–8 % to almost 40 % (Supplementary Data Table S7). Interestingly, most leaflet shrinkage occurred after embolism onset in all three palm species (Supplementary Data Movies S8, S9 and S10). This shrinkage was most evident in C. radicalis (Supplementary Data Movie S8), where it was possible to observe embolism spread (as change in reflectance inside the main vessels), tracking the advance of the dehydration front from the left (tip) to the right (base) of the leaflet.

DISCUSSION

Broad range of embolism resistance in palms

This study provides the first estimate of drought-induced xylem embolism resistance in palms originating from a range of different biomes. Our results add novel information to previous studies demonstrating that palms encompass a relatively broad range of embolism resistance, from low in D. baronii to high in R. excelsa. To our knowledge, embolism resistance has previously been investigated in only four palm species (Sperry, 1985; Cobb, 2006; Renninger and Phillips, 2011). Sperry (1985) found that in R. excelsa embolism induction commences when xylem pressure falls below ~−3 MPa. Similarly, our study of R. excelsa found no embolism at relatively high water potentials in both leaves and petioles. At the opposite end of the spectrum, Renninger and Phillips (2011) found low resistance to embolism in the Amazonian species Iriartea deltoidea (P₅₀ = −1.4 MPa). All the palms in our study were more resistant to embolism than I. deltoidea, demonstrating that palms can be as resistant as other angiosperm species, including both herbs and trees.

Embolism resistance and monocot hydraulic architecture

The mechanisms by which palms achieve such a wide range of embolism resistance remain unclear and merit further investigation. The hydraulic segmentation hypothesis predicts that vulnerability to embolism will be relatively high in plant parts that are more expendable (Tyree and Zimmerman, 2002). However, we found little evidence for segmentation between leaflet and petioles in palms. This result is consistent with the fact that palms do not drop leaflets under drought stress. Instead, leaflets tend to dry out from the tip to the base (Fig. 6A–C). These observations are in agreement with the results of previous studies that show little difference in embolism resistance between organs of a herbaceous plant (Skelton et al. 2017), though in woody angiosperms hydraulic segmentation seems to be species-specific (Dominguez-Rodriguez et al., 2018; Klepsch et al., 2018; Losso et al., 2018). Surprisingly, we observed leaflets being less vulnerable to embolism than petioles in the sole species where we observed significant differences between proximal and distal parts of the leaf (C. radicalis). This result offers novel evidence showing that hydraulic segregation can occur without a clear relationship with organ cost or replaceability.

Pit membrane thickness, an important predictor of embolism resistance in woody plants (Lens et al., 2013; Li et al., 2016), is also unlikely to play a major role in palm embolism resistance. Most palms possess only one to few metaxylem vessels per vascular bundle in transverse section (Thomas and Boura, 2015). Therefore, connections between metaxylem vessels (and therefore embolism spread) occur via transverse commissures (Tomlinson, 1990). Those transverse commissures are smaller in diameter than metaxylem vessels and may contribute to embolism resistance. Our micro-CT observations on the most resistant species, T. fortunei, support this hypothesis. For this species, we observed that two vessels can be connected through one of these commissures without spread of embolism from the embolized to the functional vessel (Supplementary Data Fig. S11).

The relatively unusual co-occurrence of tracheids and vessels in palms has also been suggested as a strategy to cope with drought, as found in another monocot, the Australian resurrection plant Borya, in which vessel elements can offer a rapid supply of water in the rainy season, while thick-walled tracheids maintain the water column under higher tension during the dry season (Carlquist, 2012). However, our data do not offer support for this hypothesis, as we did not observe functional tracheids in the presence of embolized metaxylem vessels.

Embolism resistance describes the lethal threshold of dehydration beyond which the plant fails to recover (Urli et al., 2013). Many other hydraulic traits (e.g. rooting depth, transpiration rate and tissue capacitance) will dictate how fast a plant approaches this threshold (Brodribb, 2017). For the first time in intact palms, we imaged leaf petiole parenchyma shrinkage and tissue collapse while the leaf water potential decreases (Fig. 1). Changes in size and shape of palm petioles are noticeably greater than those observed in other species studied previously using the same methods (e.g. Cochard et al., 2015; Nolf et al., 2017). Significant amounts of petiole shrinkage occur even before the plant reaches a water potential that induces lethal embolism. The proportion of parenchyma that we observed in palm petioles compares closely with that observed in succulent plants (Morris et al., 2016). Together, these observations indicate that water storage in palm petioles (as well as in stems) could play an important role in preventing embolism while maintaining a high transpiration rate.

Petiole capacitance in palms extends the threshold of drought tolerance conferred solely by xylem structural features, rendering palm xylem more resistant than was previously predicted based on their large vessels. This mechanism was already described for tropical tree species with high capacitance that can avoid embolism by releasing stored water from parenchyma to the xylem (Meinzer et al., 2008). The role of water storage in drought tolerance and mortality may be especially important when the drought results from increasing seasonality rather than merely decreasing rainfall. The ability to store water that can sustain transpiration until water again becomes available could ultimately be as important in changing climates as embolism resistance.

Challenges and perspectives in estimating embolism resistance in palms

One of the palm species in this study, T. fortunei, appeared to be so resistant to embolism that even achieving the water potential where the plant has lost 50 % of function was challenging. Even in less extreme cases, estimating embolism resistance in palms is difficult. Cobb (2006) measured embolism resistance in the climbing palm Calamus caryotoides using the single-vessel air-seeding technique, but concluded from the low mean air-seeding pressure and wide variation that this technique is unsatisfactory for species with monocot anatomy where vessels occur in isolation. Renninger and Phillips (2011) used both an air-pressure method (Washingtonia robusta) and a bench-dry method (I. deltoidea) to hydraulically measure changes in the xylem conductivity of palms. However, the exponential shape of their vulnerability curves for W. robusta should be interpreted with caution, as it has recently been shown that this shape is most likely to result from an open-vessel artefact (Torres-Ruiz et al., 2014).

The weak differences in embolism resistance between leaflets and petioles found in our study indicate that in palm species P₅₀ can be characterized from either leaf petioles or leaflets. In this sense, the novel non-invasive optical method (Brodribb et al., 2016) appears particularly promising for phenotyping drought-induced embolism resistance in palms.

Conclusions

Knowledge of drought resistance strategies in a plant group that is as ecologically and economically important as palms represents a significant step towards understanding how plants respond to drought in both natural and agricultural systems. This study provides the first estimate of drought-induced xylem embolism in palms across biomes. Although the mechanisms by which palms achieve embolism resistance in their wide vessels are not entirely clear, it is likely that they protect their large xylem vessels from embolism by mobilizing water from parenchyma tissues to the vessels, thus preventing achievement of embolism-inducing tensions. This investigation provides the first step towards understanding hydraulic adaptations in palms by showing how high embolism resistance can be sustained in long-lived arborescent monocots. Further investigations of mechanisms of embolism resistance across palm lineages should improve our understanding not only of species distribution but also susceptibility to drought, a key factor that underpins the role of palms in tropical ecosystems.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: species studied and their distribution and habitat. Figure S2: palm species occurrence optimum along the gradients of water-table depth and aridity. Figure S3: relationship between xylem pressure and mean area of embolized conduits. Table S4: observed and estimated vessel mean for each micro-CT scan, species and xylem water potential. Figure S5: 2-D digital cross-section X-ray micro-CT of Trachycarpus fortunei. Figure S6: accumulation of embolisms over time for three palm species (Chamaedorea radicalis, Howea forsteriana and Rhapis excelsa). Table S7: leaflet shrinkage percentages for each sample. Movies S8–S10: examples of leaf dehydration and shrinkage. Figure S11: transverse section of Trachycarpus fortunei. Movie S12. Time comparison of embolism propagation between species.

FUNDING

This work was supported by European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement number 706011 (PalmHydraulics); SOLEIL synchrotron facility, PSICHE beamline (grant number 20160990); the programme ‘Investments for the Future’ (grant number ANR-10-EQPX-16, XYLOFOREST) from the French National Agency for Research, IdEx Bordeaux International Post-doctoral Program (grant number UB101 CR1024-R s/CR1024-6M to L.J.L.); and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) –Finance Code 001.