In vivo dynamic analysis of water refilling in embolized xylem vessels of intact Zea mays leaves

Abstract

Background and Aims The refilling of embolized xylem vessels under tension is a major issue in water transport among vascular plants. However, xylem embolism and refilling remain poorly understood because of technical limitations. Direct observation of embolism repair in intact plants is essential to understand the biophysical aspects of water refilling in embolized xylem vessels. This paper reports on details of the water refilling process in leaves of the intact herbaceous monocot plant Zea mays and its refilling kinetics obtained by a direct visualization technique.

Methods A synchrotron X-ray micro-imaging technique was used to monitor water refilling in embolized xylem vessels of intact maize leaves. Xylem embolism was artificially induced by using a glass capillary; real-time images of water refilling dynamics were consecutively captured at a frame rate of 50 f.p.s.

Key Results Water supply in the radial direction initiates droplet formation on the wall of embolized xylem vessels. Each droplet grows into a water column; this phenomenon shows translation motion or continuous increase in water column volume. In some instances, water columns merge and form one large water column. Water refilling in the radial direction causes rapid recovery from embolism in several minutes. The average water refilling velocity is approx. 1 μm s⁻¹.

Conclusions Non-destructive visualization of embolized xylem vessels demonstrates rapid water refilling and gas bubble removal as key elements of embolism repair in a herbaceous monocot species. The refilling kinetics provides new insights into the dynamic mechanism of water refilling phenomena.

Introduction

Xylem embolism is a critical problem associated with the function and viability of vascular plants. Such events break water chain continuity in xylem vessels and reduce plant hydraulic conductivity. These constraints also threaten the stability of long-distance water transport in plants (Clearwater and Goldstein, 2005), and also limit height during growth (Koch et al., 2004) and cause plant deaths (Tyree and Sperry, 1989).

Xylem embolism is caused by various factors such as drought, freeze–thaw stress or insect-borne damage. To avoid damage from embolism, vascular plants can refill embolized xylem vessels and restore hydraulic conductivity of xylem vessels. Positive root pressures remove embolism in some herbaceous crop species and vines when the soil is saturated and transpiration is low (Tyree et al., 1986; Sperry et al., 1987; Stiller et al., 2003). Embolized xylem vessels can be refilled under tension during periods of active transpiration (Salleo et al., 1996; Canny, 1997; McCully et al., 1998; Zwieniecki and Holbrook, 1998; Hacke and Sperry, 2003; Stiller et al., 2005). Refilling under tension has received particular interest in plant biology and ecology because hydraulic recovery from embolism at midday provides evidence on the survival strategies used by plants against seasonal and environmental changes.

However, it was difficult to study the dynamic process of xylem refilling under tension and understand the mechanism of embolism repair. Most previous studies using destructive experimental techniques were limited to provide unequivocal evidence for water refilling dynamics. Embolism was artificially induced by injecting pressurized air into the xylem vessels of the excised twig segments (Salleo et al., 1996) or by freezing the petiole pieces and the excised roots (Canny, 1997; McCully et al., 1998). Xylem embolism and refilling were mostly measured indirectly from loss of hydraulic conductivity (Zwieniecki and Holbrook, 1998; Hacke and Sperry, 2003) and observed by light microscopy or scanning electron microscopy (McCully et al., 1998; Canny, 2001; Stiller et al., 2005). These measurements on the excised parts of plants after inducing xylem embolism were potentially prone to artefacts in monitoring water refilling dynamics (Sperry, 2013; Wheeler et al., 2013).

Due to the lack of in vivo observations of embolism repair, many of the hypotheses regarding water refilling of embolized vessels under tension remain unsubstantiated (Zwieniecki and Holbrook, 2009). Embolism repair requires positive pressure to dissolve gas bubbles in sap water, expel gas from embolized vessels or pull water toward air-filled xylem vessels. Most scenarios involve osmotic refilling, which can be generated from solute secretion by surrounding living cells or phloem, and water inflow from adjacent water-filled xylem vessels or surrounding living cells (Holbrook and Zwieniecki, 1999; Tyree et al., 1999; Zwieniecki and Holbrook, 2000; Hacke and Sperry, 2003; Konrad and Roth-Nebelsick, 2005; Salleo et al., 2006). The putative role of tissue pressure has been also proposed in embolism repair (Canny, 2001; Bucci et al., 2003; Clearwater and Goldstein, 2005). Several major concerns need to be investigated to elucidate the mechanism of embolism repair (Clearwater and Goldstein, 2005; Wheeler and Holbrook, 2007; Nardini et al., 2011; Stroock et al., 2014): the driving force that removes gas phase and draws water into embolized vessels, the pressure required for refilling, the overall speed of the refilling process, and the source and pathway of water for refilling.

To resolve these questions, researchers should obtain reliable experimental evidence from intact plants, using non-destructive measurements with high temporal resolution. In the last decade, magnetic resonance (MR) and X-ray imaging techniques as non-destructive alternatives have been applied to studies monitoring xylem refilling. However, the spatial and temporal resolution of MR images was previously insufficient to study the dynamic refilling process in individual small vessels of intact plants. MR imaging presented in vivo direct measurements of xylem refilling in intact plants having xylem vessels larger than 50 μm in diameter with low temporal resolution (Holbrook et al., 2001; Clearwater and Clark, 2003; Scheenen et al., 2007; Kaufmann et al., 2009). X-ray computed tomography has also been used to demonstrate the spatial patterns of embolism repair on the order of hours, and to study refilling in woody plants (Brodersen et al., 2010; Suuronen et al., 2013; Choat et al., 2015). X-ray radiography has allowed the direct visualization of xylem sap flow and refilling phenomena with higher temporal resolution but has still been applied only to excised parts of plants (Lee and Kim, 2008; Kim and Lee, 2010; Lee et al., 2013; Yanling et al., 2013; Hwang et al., 2014; Ryu et al., 2014).

This study aimed to reveal real-time, dynamic phenomenon of water refilling in embolized xylem vessels of intact maize leaves via a non-destructive and non-invasive synchrotron X-ray micro-imaging technique. We artificially induced embolism in xylem vessels by stabbing the xylem vessel wall with a sharp glass capillary tube; we then monitored water influx into embolized vessels along the radial direction and quantitatively analysed water-refilling kinetics with high spatial and temporal resolution. We also investigated the dynamic behaviour of gas-bubble removal during refilling.

Materials and methods

Plant sample

Wild-type maize (Zea mays L.) was used in all the experiments. Plant samples ranging from 1·0 to 1·5 m in height were hydroponically grown for 4–8 weeks in a cultivation room at constant room temperature of 25 °C, relative humidity of 70 % and photosynthetically active radiation (PAR) of approx. 300 μmol m⁻² s⁻¹ (10 h daily). PAR was monitored using a PAR sensor (E90/Jauntering International Corp., Taipei, Taiwan).

Synchrotron X-ray imaging

Synchrotron X-ray imaging was used to obtain phase-contrast images of water refilling in xylem vessels of maize leaves. Experiments were performed at the 6C Biomedical Imaging Beamline (PLS-II) of Pohang Accelerator Laboratory (PAL). Stable environmental conditions in the experimental hutch of PAL were maintained during X-ray imaging. The temperature in the hutch was held constant at 25 °C at a relative humidity of 18–43 % based on the desired experimental conditions. PAR intensity was maintained at 1000 μmol m⁻² s⁻¹ except in the cases illustrated in Figs 3 and 4 where it was maintained at 250 μmol m⁻² s⁻¹ for active transpiration of plants. The X-ray beam through a test sample was converted to visible light by passing through a scintillator crystal. An sCMOS camera (Andor Zyla 5.5) with a 10× objective lens was used to capture the X-ray images with a field of view (FOV) of 1700 ×1400 μm at a frame rate of 50 f.p.s. and exposure time of 20 ms. The corresponding spatial resolution was 1·3 μm per pixel. Preliminary tests using an X-ray beam of 11–30 keV were performed to minimize X-ray exposure and its damage in acquiring X-ray images of plant leaves, and the photon energy of the X-ray beam was then fixed at 14 keV with an X-ray photon flux density of approx. 10¹¹ photons s⁻¹ mm⁻² at 300 mA per run. The beam flux density at Beamline 6C of PAL is the same as other fluxes used in X-ray studies on live plants (Brodersen et al., 2010, 2011, 2013a, b; Choat et al., 2015). A mechanical shutter and attenuating plates were also utilized to minimize physiological damage caused by exposure of the test sample to the X-ray beam. The total exposure time to continuous X-ray radiation was less than 5 min.

Fig. 3.

Merging of individual water columns. (A) Refilling water columns (W₁ and W₂) are simultaneously formed at multiple points in the FOV. A gas column entrapped between these columns shrinks into a small spherical bubble and finally disappears, whereas the two surrounding water columns merge. (B) Temporal variations in the heights of the water columns W₁ and W₂ and the merged column (W₁ + W₂). W₁ (black squares) and W₂ (red circles) represent the upper and lower ends of the two separate water columns.

Fig. 4.

Shrinkage of gas bubbles during water refilling. (A) Two gas columns B₁ (marked with a red arrow) and B₂ are entrapped between emerging adjacent water columns. The perforation plate is marked with an arrowhead, P-P. (B) Temporal volume variation in B₁. The gas column B₁ is transformed into a spherical gas bubble, as indicated before and after the dotted line.

Embolism induction

A whole maize plant grown in a pot containing a hydroponic solution was moved into the experimental hutch of PAL. Without excision of leaves, one leaf of the intact maize plant was affixed at the sample holder on a motorized 3D translation stage (Supplementary Data Fig. S1). The adaxial side located 15 cm from the leaf apex was positioned in the pathway of the X-ray beam for experimental manipulation and image recording.

A glass capillary tube was installed on a micro linear stage (Newport M-ILS100CC) and moved toward the adaxial leaf surface to stab the wall of a targeted xylem vessel. The glass capillary tube was thermally pulled at the tip opening with a diameter of approx. 6–10 μm (Supplementary Data Fig. S2). The open-end capillary tube was then pierced to a specific point of a metaxylem vessel to induce artificial embolism without tearing the leaf. The approaching speed of the capillary tube was carefully adjusted to 10 μm s⁻¹. The stabbed xylem vessel was likely to be exposed to atmospheric pressure. Similar xylem pressure would be induced in intact plants when insects insert their stylets into the xylem vessel for sap feeding or when herbivores bite the leaves (Kim, 2013). Since the tip size of the capillary is much smaller than the dimension of xylem vessels, this pricking is a minimally invasive method. We continuously monitored individual xylem vessels around the target vessel and positional variation of the capillary using X-ray micro-imaging, simultaneously controlling a 3D motorized stage attached to the capillary. If the neighbouring vessels were stabbed together, air would be also inserted inside the vessels. However, the neighbouring xylem vessels adjacent to the stabbed vessel were undamaged. As a consequence, no change of water content was observed inside the neighbouring xylem vessels. Therefore, the undamaged neighbouring vessels were speculated to maintain their water under tension, while being exposed to light irradiation. Wei et al. (1999) directly measured xylem pressure in leaves of intact maize plants, and was approx. −0·35 to − 0·6 MPa, depending on light irradiation (150–260 μmol m⁻² s⁻¹). Although the xylem pressure was not exactly the same as in the present study, the xylem vessels adjacent to the punctuated vessel were assumed to be under tension. In addition, the effect of positive root pressure on water refilling can be neglected, because there was no guttation in leaves during the present experiments. Water refilling in embolized xylem vessels was recorded consecutively.

Evaluation of water refilling

The temporal variation of water columns in the embolized portion of xylem vessels was analysed quantitatively during refilling. The geometry of water columns captured in 2D X-ray projection images was analysed with the digital image processing software ImageJ (Schneider et al., 2012). The inner diameter of each xylem vessel was determined by averaging five measured diameters along the section of interest. Distinctive structures in xylem vessels were designated as fixed reference points in relative coordinates to evaluate the relative height or displacement of water columns. The volume of each water column was then evaluated by reconstructing a simple 3D structure based on the projected 2D image. Supplementary Data Fig. S3 shows how 2D projection images were converted into a 3D structure. During image reconstruction, xylem vessels were assumed to be cylindrical, and the meniscus between air and water was regarded as a part of a sphere. The temporal variation of each refilling water column was evaluated from X-ray images, which were consecutively recorded at 50 f.p.s.

Results

Real-time analysis of the dynamic process of water refilling in intact maize leaves

We observed the dynamics of water refilling in xylem vessels of intact maize leaves where embolism was artificially induced. Water refilling is a continuous process within a short time interval, but we classified the water-refilling process into three distinct sequential stages according to water shape. First, water droplets are formed in contact with the wall surface of the embolized xylem vessel and then radial water influx leads to continuous growth of the droplets (stage I). The water droplets develop into discontinuous water columns, and the volume of each water column gradually increases (stage II). Afterwards, the elongated water column moves along the xylem vessel without noticeable volume change (stage III). Several water columns are sometimes merged together, and the embolized xylem vessel is refilled with water.

The water-refilling process starts with the formation of droplets on the left side wall of the embolized metaxylem vessel near the perforation plate, followed by growth of droplets (Figs 1 and 2, Supplementary Data Videos S1 and S2). The contact angle of water droplets formed on the wall surface is variable (15–50°). The volume of water droplets increased because of continued water supply and bulged toward the other side wall of the vessel at stage I (Fig. 1A: –119 s from the start of water refilling). Two droplets were formed on the antipodal side walls of the embolized metaxylem vessel at the position of the perforation plate within a shorter time (Fig. 2A: –8·8 s, stage I). The water droplets coalesced in a water bridge or water column with a contact angle in the range 30–90° during the transition from stage I to stage II. The average refilling flow rate increased abruptly during the transition between stages I and II (Fig. 1B). A short water column was formed at stage II, and the short, irregular and asymmetrical water column became long and cylindrical as its volume increased (Fig. 1A: –144 s; Fig. 2A: –16·6 s).

Fig. 1.

Rapid radial influx and growth of an individual water column in water refilling. (A) Sequential kinetics of water refilling in the embolized xylem vessel. The perforation plate is marked with an arrowhead, P-P. (B) Temporal variations of refilling water volume by radial influx from the start of water refilling (Time* = 0). Time* is the non-dimensional time normalized by the time interval from the start of water refilling to the transition in stage II, i.e. the duration of stage I.

Fig. 2.

Growth and translational movement of refilling water columns. (A) Antipodal refilling was observed in the form of water droplets on the opposite sides of the embolized xylem vessel near the perforation plate (marked with an arrowhead, P-P) at the initial period of water refilling (–8·8 s, stage I). These two droplets coalesce into a short water column (–16·6 s, stage II). Instantaneously, the water column stops increasing and moves downward along the vessel (16·62 s, stage III). (B) Temporal variations in height (cross-hatched bars) and volume (squares) of the refilling water column. Crosses indicate the lower locations of water influx at the interior xylem vessel walls, and hatched bars indicate the heights of separate water columns.

We estimated the increase in volume of water droplets and separated water columns during the water-refilling process, and obtained the radial influx rate of water in the embolized vessels for three cases, representing three plants (Table 1). The appearance of water droplets and separated water columns could be attributed to the continued supply of refilling water in the radial direction. Radial water inflow was continuously observed in the embolized xylem vessel on the order of minutes. The average refilling flow rates during stages I and II were estimated to be 9·6 × 10² and 1·39 × 10³ μm³ s⁻¹, respectively. At the onset of water refilling in stage II, the volume of sap water increased abruptly with a high instantaneous refilling velocity (up to 8·8 × 10² μm s⁻¹). In general, refilling flow rates during stage II were faster than those during stage I.

Table 1.

Kinetics of water refilling in embolized xylem vessels

Plant sample		Vessel diameter (μm)	Method	Refilling stage	Average refilling flow rate (μm3 s−1)	Average refilling velocity (μm s−1)	Max. refilling velocity (flow onset in stage II) (μm s−1)
Zea mays leaf	Case 1	32·9	X-ray imaging	Stage I	3·2 × 102	3·8 × 10−1	4·0 × 10−1
				Stage II	3·3 × 102	3·9 × 10−1
	Case 2	36·4		Stage I	2·4 × 103	2·3 × 100	1·2 × 101
				Stage II	3·6 × 103	3·4 × 100
	Case 3	29·9		Stage I	1·50 × 102	2·2 × 10−1	8·8 × 102
				Stage II	2·4 × 102	3·4 × 10−1

Refilling velocity is a normalized value obtained by dividing the refilling flow rate by the cross-section area of the xylem vessel.

The volume of water column increased without translational movement during stages I and II, whereas the refilling column exhibited translational movement with minimal volume change at stage III (Fig. 2 and Video S2). Only the separated water column sequentially moved downward along the xylem vessel. Translational movement of the water column suggests that substantial pressure differences occurred in the regions above and below it. No additional change of refilling water volume at stage III also indicates that water could not be continually supplied to the embolized vessel along the radial direction.

Removal of gas phase during the water-refilling process

Xylem refilling involves the coalescence of individual water columns forming inside the embolized xylem vessels. During refilling, water droplets appeared at two sites on the vessel wall; these grew into individual water columns, W₁ and W₂ (Fig. 3 and Supplementary Data Video S3). In a short period of time (<1 min), separate gas columns were formed in the FOV of the embolized vessel. While these two water columns expanded and merged into each other (W₁ + W₂), the gas column entrapped between them was compressed into a small gas bubble in contact with the vessel wall and subsequently removed. These processes occurred for approx. 420 s.

The water-refilling process also revealed shrinkage of a gas bubble surrounded by water (Fig. 4). The volume of gas column B₁, which was formed between other water columns in the same vessel illustrated in Fig. 3, decreased progressively; the gas column eventually shrank into a spherical-shaped bubble isolated from the walls on either side of the xylem vessel for less than 700 s. The size of the bubble also decreased gradually, and the bubble finally disappeared after 900 s from the start of the water-refilling process (Fig. 4 and Supplementary Data Video S4).

Disappearing water columns near the perforation plate

Individual water columns were moved along the embolized vessel and drained at the position around the perforation plate (Fig. 5 and Supplementary Data Video S5). Water columns sequentially ascended toward perforation plates. In this case, radial influx of sap water into the embolized metaxylem vessel started exceptionally near the piercing point, not near the perforation plate. Each individual water column (W₁ and W₂) appeared at the height of the lower white arrow in Fig. 5; the water column started to move in an ascending direction during stage III after its height reached approx. 100 μm. The translational movement of the water column stopped at the position near the perforation plate until the volume of the water column decreased to 0, indicating that sap water was completely drained. The kinetics of water column growth, translation and disappearance was subsequently repeated (not shown in the figures).

Fig. 5.

Migration of water columns around the perforation plate. (A) Water columns (W₁ and W₂) form at the height of the lower white arrowhead, ascend upward along the xylem vessel, and drain at the height of the upper black arrowhead, where a perforation plate (P-P) exists. (B) Temporal variation in the height of water columns W₁ and W₂. Here, W₁ (black squares) and W₂ (red circles) represent the upper and lower ends of the water columns. Dotted lines represent the positions of the perforation plate and the piercing point.

Discussion

In this study, water refilling in embolized xylem vessels was directly visualized and analysed via a synchrotron X-ray imaging technique. The formation and growth of water droplets from the initial radial influx to the embolized vessel and coalescence of water droplets led to embolism repair (Fig. 6). In vivo dynamic analysis of the water-refilling process demonstrates the spatial and temporal characteristics of water refilling and their implications, suggesting further consideration of the mechanism for embolism repair.

Fig. 6.

Schematic of water refilling in an embolized xylem vessel. In the initial stage of water refilling (stage I), radial influx forms droplets on one side wall or both side walls of an embolized xylem vessel near the perforation plates. With continuous water supply in the radial direction, droplets merge into water columns (stage II). The volume of these individual water columns increases continuously; individual water columns also combine, forming a larger water column. In this process, the gas phase between the two water columns is dissolved in water. Water columns show vertical translational movement in xylem vessels (stage III).

Water refilling phenomena observed in this study occurred within several minutes. The overall refilling flow rate was approx. 1·2 × 10³ μm³ s⁻¹, and the average water refilling velocity was in the order of 1 μm s⁻¹. To the best of our knowledge, the water refilling kinetics obtained in this study was faster than previously published mathematical predictions (Vesala et al., 2003). Previous studies (Scheenen et al., 2007; Brodersen et al., 2010) examined the kinetics of water refilling in intact plants. Scheenen et al. (2007) reported that the water refilling in vessels was presumably caused by root pressure. Using MR imaging, they directly estimated the water refilling flux in the roots of intact Cucuis sativus plants to be 1·2 × 10⁻¹ μm³ s⁻¹. Brodersen et al. (2010) also investigated water refilling in the stem of Vitis vinifera under tension, depending on water status, using synchrotron X-ray computed tomography. The water influx rate from surrounding living tissues was evaluated to be 6 × 10⁻⁴ μm³ s⁻¹. It is difficult to directly compare the present results with these previous studies because the experimental set-up, embolism inducing stress and species-specific anatomical structures are clearly different. Nonetheless, it is noticeable that the embolized vessels of Zea mays were refilled within a few seconds or minutes, while Vitis vinifera took several hours or days to refill them.

The distinct rapid water refilling may implicitly provide a clue to develop a new mechanism for embolism repair. First, we consider if the osmotic refilling from surrounding parenchyma cells (McCully et al., 1998; McCully, 1999; Brodersen et al., 2010) can explain the rapid water refilling phenomena observed in this study as a dominant mechanism for embolism repair under tension. The flow rate of water refilling in a radial direction per unit area, J, is expressed as follows:

$$J=L_{\text{p}}(\mathbf{\Delta }P–\Delta \Pi ),$$

where L_p is hydraulic permeability, ΔP the difference of hydrostatic pressure (turgor pressure), and ΔΠ the osmotic pressure difference between the embolized xylem vessel and the surrounding living cells or water-filled vessels. If the water-refilling process starts at a cell bordering on the xylem vessel, we can assume ΔP as − 0·03 to 0·4 MPa, given the appreciable turgor pressure in parenchyma cells in the midrib of intact Zea mays leaves under transpiration (Kim and Steudle, 2007). The hydraulic permeability (L_p) in leaf parenchyma cells of maize leaves ranged from 0·3 × 10⁻⁶ to 2·5 × 10⁻⁶ m MPa⁻¹ s⁻¹ (Westgate and Steudle, 1985). It is difficult to estimate the membrane surface involved in water release, but the average water refilling velocity in the FOV of xylem vessels can be evaluated based on the calculated refilling flow rate and the cross-section area of the xylem vessel. The osmotic pressure should reach approx. 50 MPa to explain the rapid water-refilling kinetics observed in this study. However, the osmotic pressure of sap pressed out of mature leaves of most vascular plants is known to be 0·6–3·0 MPa (Nobel, 1999). The osmotic pressure in vascular plants is insufficient to drive the radial influx of sap water. Thus, we conjecture that the osmotic mechanism cannot explain xylem refilling adequately. Recently, Secchi and Zwieniecki (2012) reported that sugars and ions accumulated in embolized xylem vessels can explain the driving force for the water-refilling process. In addition, the co-transport of water and solutes across the cell membrane may assist the water refilling of embolized xylem vessels (Wegner, 2014). These alternative mechanisms need not exclude the present suggested mechanisms to explain xylem refilling, but a new mechanism is required to explain the rapid embolism repair in vascular plants.

Interestingly, we observed that the rapid water refilling in the embolized vessel frequently began with the formation of droplets on the wall surface near perforation plates (Figs 1 and 2). Based on our direct observations, we assumed that the water refilling might be closely related to porous structures of the vessel end, such as perforation plates. Individual water columns disappearing around perforation plates (Fig. 5) also support the presence of a radial pathway near perforation plates to supply water in embolized xylem vessels or to drain water from these vessels. Given radial hydraulic permeability of 2·26 × 10⁻⁷ m MPa⁻¹ s⁻¹ in xylem vessels of Fraxinus americana (Zwieniecki et al., 2001), water refilling velocity is estimated to be approx. 0·1 μm s⁻¹. This estimated velocity is comparable with our data. Based on our direct observations, we suggest the existence of any direct pathway near perforation plates through vessel walls connecting to the adjacent water-filled vessel to explain the rapid water refilling. It is plausible that the radial permeability of xylem vessels contributes to rapid influx of the xylem sap into the embolized vessel. The presence of this water transport pathway may explain the rapid water refilling in embolized vessels.

The fate of gas phase in the water-refilling process remains an enigma. When the embolized xylem vessel is filled with water by its radial influx from the intact neighbouring vessels under tension, the gas phase inside the embolized vessel dissolves by surrounding water or may be expelled from the embolized vessel through other radial pathways. At the end of gas bubble removal (Fig. 4A; t = 810–900 s), we observed that gas bubble B₁ is detached from either of the side walls of the xylem vessel. This implies that the chance of a pressurized gas bubble escaping through radial pathways is low in this specific case, and the gas bubble is directly dissolved in an embolized vessel. The detachment of a gas bubble in water refilling may be attributed to the surface morphology and wettability of the inner conduit wall. High wettability is known to promote the water-refilling process (Kohonen, 2006; Kohonen and Helland, 2009). The gas phase can escape through gas-filled hydrophobic routes such as the borders of neighbouring pits (Zwieniecki and Holbrook, 2009; McCully et al., 2014). Nevertheless, the exact pathway through which gas escapes during the water-refilling process remains unknown even with the use of advanced flow visualization techniques.

From artificially induced embolism in the targeted xylem vessel, we observed the kinetics of the gas phase in a water refilling cycle and estimated local overpressure. A gas bubble entrapped between two columns shrinks and eventually disappears when two individual water columns that grow in a xylem vessel in stage II merge into a larger column (Fig. 3A). The dynamic behaviour of gas bubble removal (Fig. 4) suggests that overpressure might be applied to the gas phase. Continuous water supply and radial water influx may compress the gas bubble. Cell-wall debris resulting from pricking might plug the damaged site and facilitate positive pressure on the gas phase. The kinetics of gas removal (Fig. 4) shows the speed of embolism recovery and the amount of pressure required to dissolve gas bubbles in sap water. The gas phase is often removed in a relatively short period of time. Small gas plugs and bubbles can also be removed in a few minutes; by contrast, water refilling in large stems lasts several hours (Yang and Tyree, 1992). On the basis of these relatively different time scales, we estimated the hydrostatic pressure exerted in the refilling water column. We assumed that the gas bubble detached from the xylem wall exhibits a spherical symmetry and the properties of sap are the same as those of pure water. We then applied the following Rayleigh–Plesset equation (Plesset and Prosperetti, 1977):

$\frac{P_B(t)-P_{\infty }(t)}{{\rho }_L}=R\ddot{R}+\frac{3}{2}{\left(\dot{R}\right)}^2+\frac{4{\mu }_L}{{\rho }_L}\frac{\dot{R}}{R}+\frac{2{\sigma }_L}{{\rho }_{L}R}$

where P_B(t) is the internal pressure inside the gas bubble and is assumed to be equal to atmospheric pressure, P_∞(t) is the external pressure of the bubble, ρ and µ_L are the density and dynamic viscosity of the surrounding liquid, respectively, and R(t) and σ_L are the radius and surface tension of the gas bubble, respectively. The temporal variation of the radius of the gas bubble can be expressed as $R(t)=-{10}^{-6}{t}^3-2\times {10}^{-4}{t}^2+5.1\times {10}^{-3}t+15.981$ (R²=0·9918; Supplementary Data Fig. S4). Based on the change in radius of a gas bubble, the external pressure applied on the surface of the shrinking spherical gas bubble was estimated to be approx. 70–92 kPa based on the Rayleigh–Plesset equation. The estimated pressure for direct dissolution of a gas bubble is comparable with atmospheric pressure. Interestingly, although the adjacent xylem vessels were speculated to be under tension, the disappearance of gas bubbles, accompanied by radial influx of water into the embolized vessel, in the embolized vessel implies local overpressure in the adjacent intact vessel. Measuring the temporal variation of xylem pressures in the embolized xylem vessel and its neighbouring vessels would allow us to understand the water-refilling phenomena under tension.

The effect of heat addition caused by X-ray irradiation has been considered as a possible factor that affects the dynamic behaviour of gas bubbles. However, no damage was retained on the tested maize leaves after X-ray imaging experiments were completed. Leaves were partly burned after exposure to the X-ray beam for several hours. To avoid any damage to the sample plants, the total irradiation time of the X-ray beam was limited to be less than 5 min during the entire experiment for each sample plant, and there was no discernible damage to the sample plants by eye. Also, there was no phase change including gas bubble generation in the FOV of the water-filled xylem vessels adjacent to the embolized vessel, indicating that X-ray beam irradiation in this study did not generate any significant heating effect. If heating caused by X-ray irradiation had distorted the gas removal process during natural refilling, the gas bubble would have expanded, which we did not observe. Although the decrease in air solubility might considerably affect the water refilling kinetics, a positive pressure much greater than that estimated in Fig. 4B would be required to remove the gas phase inside the embolized vessel. In addition, under similar experimental conditions using the same X-ray facility, a very small water temperature increase of < 1·0 °C after X-ray irradiation was reported in capillary-tube tests (Weon et al., 2008). The beam flux density employed in this study is also similar to those used in previous X-ray imaging experiments on embolism in plant samples (Brodersen et al., 2010, 2011, 2013a, b; Choat et al., 2015). The experimental conditions used in previous studies support the suggestion that the X-ray irradiation used in this study did not to disturb the refilling phenomena by heating.

In summary, a 2D synchrotron X-ray imaging technique with high temporal resolution was used to investigate water refilling dynamics in intact maize leaves. We qualitatively analysed the step-by-step refilling process (Fig. 6) and estimated the kinetics of embolism recovery. The embolized xylem vessel of intact monocotyledon crop plant samples was refilled rapidly with water in the radial direction. Radial influx of water near perforation plates implies the existence of a hydraulic connection of xylem vessels to radial pathways located near perforation plates. Due to this hydraulic connection, discontinuous water columns appear in the embolized vessel and embolism repair occurs. Detailed anatomical studies of the structural features of porous xylem vessels and interconnected xylem vessel networks are required as the future research to understand the source of water for refilling. The present direct observations of water refilling in intact plants provide a stepping stone for further understanding hydrodynamic phenomena in xylem vessel networks and disclosing the dynamic mechanisms for embolism repair.

Supplementary Data

Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Figure S1: schematic diagram of the experimental set-up for X-ray micro-imaging experiments. Figure S2: artificial embolism is induced in xylem vessels of intact maize leaves with a sharp glass capillary tube. Figure S3: X-ray images of water droplets and water column formed in a xylem vessel during water refilling are numerically reconstructed into 3D structures to estimate the corresponding volumes. Figure S4: a gas bubble, B₁, in the water column (Fig. 4) shrinks and eventually disappears as the radius of the gas bubble decreases. The external pressure around the gas bubble is estimated during gas removal. Video S1: water refilling process shown in Fig. 1 (total t = 94 s; ×6·7). Video S2: water refilling process shown in Fig. 2 (total t = 20·8 s; ×1·0). Video S3: merging of multiple water columns shown in Fig. 3 (total t = 900 s; ×53). Video S4: removal of gas bubble during refilling shown in Fig. 4 (total t = 960 s; ×50·6). Video S5: movements of water columns near perforation plates shown in Fig. 5 (total t = 86·4 s; ×2·0)