New evidence for a role of vessel-associated cells and phloem in the rapid xylem refilling of cavitated stems of Laurus nobilis L.


Sebastiano Salleo. Fax: + 39 040 568855; e-mail:


Xylem recovery from embolism was studied in Laurus nobilis L. stems that were induced to cavitate by combining negative xylem pressure potentials (PX = −1.1 MPa) with positive air pressures (PC) applied using a pressure collar. Xylem refilling was measured by recording the percentage loss of hydraulic conductance (PLC) with respect to the maximum 2 min, 20 min and 15 h after pressure release. Sodium orthovanadate (an inhibitor of many ATP-ases) strongly inhibited xylem refilling while fusicoccin (a stimulator of the plasma membrane H+-ATPase) promoted complete embolism reversal. So, the refilling process was interpreted to result from energy-dependent mechanisms. Stem girdling induced progressively larger inhibition to refilling the nearer to the embolized stem segment phloem was removed. The starch content of wood parenchyma was estimated as percentages of ray and vasicentric cells with high starch content with respect to the total, before and after stem embolism was induced. A closely linear positive relationship was found to exist between recovery from PLC and starch hydrolysis. This, was especially evident in vasicentric cells. A mechanism for xylem refilling based upon starch to sugar conversion and transport into embolized conduits, assisted by phloem pressure-driven radial mass flow is proposed.


Nowadays, it is generally agreed that plant xylem is subject to embolism due to the intrinsic vulnerability to cavitation of xylem conduits (e.g. Milburn 1979; Tyree & Sperry 1989). We also know that xylem cavitation can be triggered by tiny bubbles sucked into functioning conduits from neighbouring air-filled wood compartments (Tyree & Zimmerman 2002) at critical xylem pressures. Xylem embolism is generally regarded as the primary effect of several environmental stresses among which drought (e.g. Salleo & Lo Gullo 1993; Pockman & Sperry 2000; Davis et al. 2002) and freeze (e.g. Just & Sauter 1991; Tyree & Cochard 1996; Nardini et al. 2000). Nonetheless, the threshold pressures triggering cavitation have been reported to be very close to those commonly experienced by plants in the field (Tyree & Sperry 1988, 1989; Nardini & Salleo 2000), so that xylem dysfunction is by no means a rare event in a plant's life. The above consideration strongly suggests that plant species lacking specific adaptations to cavitation avoidance and apparently surviving xylem pressures potentially causing embolization, should have developed some repair mechanism(s) to cope with xylem dysfunction.

Seasonal recovery from winter embolism has been reported to occur in different woody genera such as Vitis (Sperry et al. 1987), and forest trees such as Betula, Acer, Alnus and Juglans (Sperry, Donnelly & Tyree 1988a; Sperry 1993; Hacke & Sauter 1996; Améglio et al. 2002). Other species such as a vine-like bamboo (Cochard, Ewers & Tyree 1994) and rice (Stiller, Lafitte & Sperry 2003) have been shown to recover from xylem embolism on renewed water supply. In all these cases, xylem refilling appeared to be driven by positive (over-atmospheric) root pressures. Root pressure, however, is known not to be a general phenomenon: out of 109 species measured for this variable, in fact, only 61 were shown to develop root pressures (Fisher et al. 1997) and in many cases, pressures were too low to account for xylem refilling (Ewers, Cochard & Tyree 1997). Yang & Tyree (1992) have convincingly computed (and documented) that the xylem pressure (PX) that needs to be exceeded for reversal of embolism to occur (PXR) is given by the Henry's law:

PXR = − 2T/r (1)

where T is the surface tension of water and r is the conduit radius (assuming that the maximum embolus’ radius approximates that of the conduit). Salleo et al. (1996) measured rapid xylem refilling in stems of Laurus nobilis L. (laurel) despite the fact that PX was well below PXR. In fact, laurel twigs when experimentally induced to embolize through air injection into xylem (Salleo et al. 1992), recovered consistently from loss of hydraulic conductivity within 20 min after the injection pressure had been released. The measured reversal of embolism was stimulated by auxin and was independent of root pressure that was not measurable in this species. As laurel twigs when girdled (i.e. with secondary phloem removed) proximally to the embolized stem segment did not recover from embolism, the authors concluded that phloem was involved in xylem refilling possibly by transporting some hormonal signal that might increase phloem loading and induce radial flow of solutes across the rays into the embolized conduits, thus generating the required PXR. Later, Tyree et al. (1999) repeated some experiments on L. nobilis that confirmed rapid xylem refilling occurring in plants at PX = −1.0 MPa (the expected PXR was about −0.05 MPa at 20 °C assuming the maximum radius of laurel conduits to be about 30 µm). These authors also found that: (1) the concentration of inorganic ions such as K+, Cl and Ca2+ increased in xylem sap during refilling and so did sap osmolality; (2) both variables increased under auxin treatment. Although these data were in accordance with the hypothesis above, the authors were aware of three main weaknesses of their ‘osmotic hypothesis’ for refilling, namely (1) the least osmotic potential computed to develop in refilling conduits due to increased concentration of inorganic ions, was about −0.34 MPa; that is the driving force was too little to drive a massive water flow allowing conduits to refill in 20 min. In fact, Vesala et al. (2003) have computed that conduits can refill when at a pressure of −0.4 MPa (see below) but this process would require about 10 h, and Hacke & Sperry (2003) have come to similar conclusions; (2) the reason why water flow was addressed to embolized conduits and not to functioning ones where the pressure gradients were more favourable to inward flows, remained unexplained; (3) changes in sap osmolality were, in fact, measured on solutions flushed from twigs recovering from partial xylem embolism and not on single conduits actively refilling. Summarizing the relatively abundant data on xylem refilling that has appeared in recent years in the literature, three main hypotheses for rapid embolism reversal emerge:

Hypothesis 1: the osmotic hypothesis

In this case, the driving force for refilling would be generated through enriching the thin layer of sap still surrounding emboli (Tyree et al. 1999) with solutes extruded from the parenchyma vasicentric cells. Solutes might consist of inorganic ions (Grace 1993; Tyree et al. 1999) or of sugars resulting from hydrolysis of starch which is ever present in wood parenchyma cells (Canny 1997; Bucci et al. 2003). A different version of the osmotic hypothesis has been recently advanced by Hacke & Sperry (2003), who suggested that the pit membrane itself may act as a semi-permeable barrier against osmotica like oligosaccharides present in the xylem sap that would drive an inward osmotic water flow and promote xylem refilling (the so-called ‘pit membrane osmosis’).

Hypothesis 2: the reverse osmotic hypothesis

This model is based upon cycles of starch-to-sugar hydrolysis occurring in the wood living cells on a daily basis. Osmotically active sugars would consistently decrease the osmotic potential of these cells causing them to become fully turgid and generating a ‘tissue pressure’ that would be contained within stems by external tissues. This positive pressure would cause cells to squeeze water into conduits (Canny 1997; McCully 1999; Facette et al. 2001). According to this model, conduits would refill even in transpiring plants, that is, in plant organs under tension. Tyree et al. (1999) have objected that the turgor-generated tissue pressure would be dissipated in all directions and would not drive a water flow specifically directed to embolized conduits. Furthermore, vasicentric cells would be the site of a contemporary inward (to maintain cell turgor) and outward (to refill conduits) mass flow which is thermodynamically impossible.

Hypothesis 3: the phloem-driven xylem refilling

The involvement of phloem in xylem refilling was first proposed by Salleo et al. (1996). Zwieniecki et al. (2000) have confirmed this hypothesis in that petioles of red maple, tulip tree and fox grape were found to reduce recovery from embolism when girdled. The mechanism was proposed to consist of phloem overloading due to activation of the proton pump in response to an auxin signal possibly transported from the leaves to stems during xylem embolism. In this case, phloem was regarded as the source of positive pressure needed to drive radial water flow that would assist osmotic flow into embolized conduits.

All the three hypotheses above require some conditions to be fulfilled: (1) if a water flow no matter how it can be generated is to be specifically directed to embolized conduits, these should be hydraulically isolated from functioning ones during refilling, otherwise water would flow into the transpiration stream. This condition is required by hypothesis 1 (but not by the ‘pit membrane osmosis’ mechanism, see above) as well as by hypotheses 2 and 3; (2) hypothesis 1 requires osmotic gradients to be large enough to drive massive water flow allowing the observed rapid refilling; (3) hypotheses 2 and 3, alternatively, require that the hydraulic conductance of the interface between an embolized conduit and neighbouring vasicentric cells becomes much higher than that at the interface with functioning conduits so as to facilitate water flow into the formers.

In recent years, some of these questions have received attention and some answers. First, Holbrook & Zwieniecki (1999) and Zwieniecki & Holbrook (2000) have suggested that air pockets may form in the pit chambers of embolized conduits that would isolate these from functioning ones until their refilling becomes complete. Under these conditions, Vesala et al. (2003) have computed that refilling may occur even in conduits with PX < PXR. Secondly, aquaporins have been immuno-localized in wood parenchyma cells of roots (Kirch et al. 2000), petioles and stems (Otto & Kaldenhoff 2000). Asymmetrical distribution of aquaporins at the contact pits between conduits and living cells has also been suggested (Canny 1997) and documented (Siefritz et al. 2001); if these aquaporins would be activated in response to a mechanical or chemical signal generated during conduit embolism, they might greatly increase the hydraulic conductance of vasicentric cell membranes, thus favouring water flow into embolized conduits (Wan, Steudle & Hartung 2004). Finally, plasma membrane proton pumps have been localized in xylem-associated cells (De Boer & Volkov 2003) that would be involved in driving secondary active import/export of sugars between wood parenchyma and xylem conduits (Alves et al. 2004). On the basis of the above, good evidence now exists of different possible steps of the pathway leading to xylem recovery from embolism but our knowledge of the links between them is still very poor. The present study was aimed at investigating: (a) the importance of the energy-dependent component of the embolism reversal in xylem conduits; (b) the eventual contribution of starch-to-sugar conversion in the process; (c) the contribution of the phloem integrity to xylem refilling. In an attempt at elucidating the above points, it was decided to study rapid refilling in the species where this phenomenon has been best documented so far, namely L. nobilis.


Experiments were conducted on potted 4- to 5-year-old plants of Laurus nobilis L. (laurel) between August 2003 and January 2004 when plants had ceased active growth (most Mediterranean evergreen species restrict their growth period to the spring and early summer). All plants had been obtained by root suckers and were fairly uniform in size. Their height, trunk diameter, length and diameter of twigs of the current year growth as well as the leaf surface area of leaves of the current year stem (one side only) are reported in Table 1. About 50 plants were grown in 3.2 L pots in a room at the Department of Botanical Sciences of Messina (Sicily), at a temperature ranging between 18.0 ± 1.5 and 24.0 ± 1.3 °C, relative humidity of 62.7 ± 4.5% and under artificial lighting provided by lamps at an irradiance of 175 W m−2. The photoperiod was 9 h. Before the experiments, the plants were regularly irrigated to field capacity and leaf water potential (ΨL) as measured on three leaves per plant during the dark period using a pressure chamber (Tyree & Hammel 1972) was higher (less negative) than −0.07 MPa. Plants were divided into groups of five plants each that were serially water stressed (see below) for the experiments. Two leaves per plant were preliminarily measured for water potential isotherms (Tyree & Hammel 1972; Salleo 1983) in order to get the leaf water potential at the turgor loss point (ΨTLP), which was taken as a reference point for applying a known stress level (see below). ΨTLP was found to be −2.3 ± 0.37 MPa.

Table 1.  Plant height (h) and trunk diameter (φT) of Laurus nobilis L. plants used in this study. The twig diameter (φcy), length (Lcy) and total leaf surface area (AL) are also reported. Means are given ± SD
h (m)φT (mm)φcy (mm)AL (cm2)Lcy (cm)
1.29 ± 0.1812.2 ± 4.15.1 ± 0.5513.7 ± 47.457.1 ± 7.6

Previous work (Salleo et al. 1996) had shown that pressure differentials of 1.75 MPa applied to the interconduit pit membranes caused significant xylem embolism in laurel that could be consistently reversed under favourable experimental conditions, within 20 min. Therefore, it was decided to apply a pressure differential of 1.75 MPa obtained by decreasing xylem pressures (PX, Fig. 1) to −1.0 MPa and applying complementary positive air pressures (PC,Fig. 1) of +0.75 MPa using a pressure collar (Salleo et al. 1992, 1996 and see below). In summary, embolism was induced in plants that had been stressed to ΨL ≈ 77% of ΨTLP. Leaf conductance to water vapour (gL) and transpiration rate (E) were measured for five leaves per plant while in the dark using a porometer (LI-1600; LiCor Inc., Lincoln, NE, USA). Under these conditions, gL was found to be 16.9 ± 3.9 mmol m−2 s−1 and E was 0.37 ± 0.12 mmol m−2 s−1, which is near cuticular levels.

Figure 1.

Experimental set-up (not to scale) showing the pressure collar (B) used to induce stem embolism through injection of air from the tank (C). Arrows show the points where stems were girdled (G) in some experiments. GB refers to experiments where plants were girdled at their base. GJ are stems of the current year growth girdled at their junction with older stems. DG are stems girdled at both sides of the pressure collar. ΔP is the pressure differential applied to the interconduit pit membranes as resulting from combining xylem tensions (PX) and air pressures (PC).

Inducing xylem embolism and measuring stem hydraulic conductance

Pre-established xylem pressure potentials (PX) were achieved by depriving plants of irrigation and measuring ΨL every 2 d on three leaves per plant immediately before lights were turned on. The leaves had been enclosed in plastic bags the evening before the experiments so as to allow leaf water potentials to equilibrate to xylem pressure potentials. As leaf water potentials were measured in a ‘pre-dawn’ condition, they can be taken as the maximum ΨL plants could develop under our experimental conditions and did not change during experiments. When ΨL reached a mean value of −1.1 ± 0.22 MPa (7–10 d after suspending irrigation), the stems of plants under study were pressurized by clamping a pressure collar (Salleo et al. 1992) at about two-thirds of the twigs’ length while the plants were kept in the dark. Air pressure was increased at a rate of 70 kPa min−1, maintained at the desired level to get the pre-established pressure differential for 20 min and then decreased at the same rate. The experimental set up and pressures applied are reported in Fig. 1.

Measurements were performed at different times after pressure release, i.e. 2 min, 20 min and 15 h. During this time, plants remained covered with black plastic bags to minimize transpiration and prevent changes in internal pressures during measurements. Previous work, in fact, had shown that at 2 min after pressure release, xylem embolism is still in its ‘initial’ status, namely as induced by pressurization. Twenty minutes after pressure release, on the contrary, was shown to be a sufficient time interval for detecting rapid refilling (Salleo et al. 1996) and 15 h is a time long enough to detect long-term changes in xylem efficiency. It has to be noted that because twigs were cut off at the end of the dark period, when they were tested for K, 15 h after pressure release, they had in fact remained in the dark for about 30 h (15 h in the dark before experiments plus 15 more hours after pressure release). At the end of the pre-established time intervals for recovery, the twigs were cut off at their junction plane, under distilled water filtered to 0.1 µm to prevent conduit clogging with spurious emboli and debris. The twigs were re-cut under filtered water using new razor blades and connected to the hydraulic apparatus first described by Sperry, Donnelly & Tyree (1988b) and slightly modified by us (Lo Gullo & Salleo 1991). About 10 mm of stem were removed from the distal end, so that samples used for hydraulic measurements were about 0.5 m long. Preliminary K measurements (one twig per plant, n = 5) were performed so as to get the ‘native’ PLC that turned out to be about 6.0 ± 2.0%. The PLC reported in the present work are all net PLC.

Stems were all perfused with a 50-mm KCl solution and the axial flow (F) through the stems was measured under a pressure of 10 kPa until the flow was stable. The initial hydraulic conductance (Ki) was calculated as Ki = FP where ΔP is the pressure difference across the stem and was related to the maximum conductance (Kmax) obtained by flushing the stems at a pressure of 175 kPa to remove emboli. This procedure was repeated until Kmax became stable. The percentage loss of conductance (PLC) was then computed as:

PLC = (1 − Ki/Kmax) × 100 (2)

In summary, twigs were cut off from plants enclosed in black bags and remained in the dark during pressurization as well as during the times established for recovery and during the subsequent K measurements. The entire technique for measuring PLC has been previously described in detail by some of us (Lo Gullo & Salleo 1991; Salleo et al. 1996).

Preliminary hydraulic measurements were undertaken in leafy twigs with intact epidermis and in twigs in which three symmetrical ‘windows’ were opened in the epidermis 6 mm2 each, prepared immediately before the twigs were pressurized. Pieces of filter paper were wetted with about 6 ×  10−5 L of different solutions (see below) and applied to the exposed cortex. Plastic sheets were then fixed to these areas to maintain the wetted paper in situ. This procedure had previously shown (Salleo et al. 1996; Tyree et al. 1999) that radial penetration of external solutions into stems was possible during pressurization without altering Ki or PLC changes. Additional experiments were undertaken on twigs with all the leaves excised 30 min before pressurization (leafless twigs). In most experiments, a 50-mm KCl solution was supplied to stems during pressurization. In other experiments addressed at checking the energy-dependence of the refilling process, two other solutions were radially supplied, namely sodium orthovanadate (Na3VO4) and fusicoccin (FC). Vanadate has been classically used to inhibit the plasma membrane proton pump (e.g. Marrè & Ballarin-Denti 1985). Fusicoccin, on the contrary, is a specific stimulator of the plasma membrane H+-ATPase (e.g. Sze, Li & Palmgren 1999; Zingarelli et al. 1999). In particular, Na3VO4 and FC were used at concentrations of 1.0 mm and 20 µm, respectively. The pH of all the solutions supplied as previously measured using a pH-meter (HI 8417; Hanna Instruments, Norfolk, VA, USA), were found to be 6.6 ± 0.7 for KCl solutions, 6.4 ± 0.15 for Na3VO4 (as corrected using 1 m MES buffer) and 6.7 ± 0.05 for FC. Measurements of PLC in stems treated with different solutions were repeated in at least four stems from different plants at the three time intervals after pressure release (2 min, 20 min and 15 h).

Experiments on girdled stems

In order to investigate the possible contribution of phloem to xylem refilling, experiments were designed with stems girdled at different distances from the pressure collar used to cause xylem embolization. In particular, the secondary cortex of stems (containing epidermis, cortical parenchyma and secondary phloem) was removed in a ring 5 mm long, 30 min before stem pressurization. The exposed wood was immediately covered with a thin layer of silicone grease to prevent desiccation. Stems were girdled either at the plant base (10 mm from the ground, GB) or at the junction between stems of the current year growth and older ones (GJ). In other experiments, stems were girdled at both sides of the pressure collar (20 mm apart, double girdling, DG). These procedures had the aim of preventing phloem transport of eventual signals from roots to stems (GB), dissipate phloem pressure (GJ and DG) and check whether the process of refilling was generated in situ (DG). All girdled stems were radially supplied with a 50-mm KCl solution during pressurization. At least four girdled stems from different plants were measured for K at each of the three time intervals after pressure release.

Checking dynamic changes in wood cell starch content

Changes in starch content of the wood parenchyma cells were estimated on cross-sections of intact unpressurized stems (‘native’ starch content) and on stems subjected to various treatments (see above), 2 min, 20 min and 15 h after pressure release. Stems were cross-sectioned at the middle plane of the segment enclosed in the pressure collar, using a microtome (Technik GmBH, Münich, Germany). Sections, 30 µm thick were first briefly rinsed with distilled water to remove debris and then immersed in a Lugol solution (iodine-potassium iodide) that stains starch in dark blue. After immersion for 1 min, sections were rinsed with distilled water to remove excess staining, mounted on a slide and observed under a microscope (Laborlux S-Leitz Esselte; Leitz GmbH, Stuttgart, Germany) equipped with a digital camera (Leica Camera AG, Solm, Germany). Digital images were acquired through connection of the camera to a computer. Because cell starch content can hardly be quantified visually, cells were taken as ‘with high starch content, HSC’ if starch granules filled more than 50% of the cell lumen as appearing in cross-section (a typical HSC cell can be seen in Fig. 5e). The number of these cells was counted using an image analysis software (Image Tool for Windows version 3.00) and related to the total number of wood parenchyma cells counted per microscopic field observed. Eight to 15 different microscopic fields per section were observed for the number of HSC cells and at least four stems per treatment and per time interval after pressure release were studied. To get more detailed information of starch hydrolysis within wood parenchyma, cells were counted for rays and vasicentric parenchyma, separately.

Figure 5.

Cross-sections of 1-year-old twigs showing wood ray and vasicentric cells containing starch granules stained in dark blue with Lugol solution. (a) non-pressurized twigs maintained in the dark for 30 h. (b) The same as (a) but at higher magnification. (c) Detail of wood parenchyma cells from twigs radially supplied with KCl during air pressurization to induce xylem embolism and cut off 20 min after pressure release. Note that some ray and vasicentric cells have lower starch content than those in (b). (d) Ray and vasicentric cells from twigs radially supplied with fusicoccin (FC). Note that wood cells are almost devoid of starch. (e) Typical wood cells with high starch content (note that starch filled more than 50% of the cellular lumen). (f) Typical cells with low or no starch content.


In Fig. 2, we report percentage loss of conductance (PLC) measured in intact leafy stems (i.e. in twigs pressurized to induce embolism but with intact epidermis, black columns) as well as in leafy stems (grey columns) and in stems deprived of leaves prior to pressurization (white columns) both radially supplied with KCl during pressurization. Two minutes after pressure release, there was no statistically significant difference in terms of PLC between intact twigs (black columns) and twigs with ‘windows’ opened in the epidermis (grey and white columns), so that we concluded that the partial removal of the epidermis and radial supply with solutions did not cause per se any change in xylem embolism as induced by air pressurization. Furthermore, the excision of leaves had no effect on ‘initial’ PLC (2 min after pressure release) in that PLC of leafless twigs was equal to that of leafy ones. Twenty minutes after pressure release, the initial PLC that was about 40% was found to decrease to 25–29% with somewhat higher recovery from loss of conductance measured in KCl-supplied twigs with respect to intact ones. Again, leafless and leafy twigs showed similar PLC values at this time. PLC further decreased to about 15% over a period of 15 h in the dark for all the twigs studied.

Figure 2.

Percentage loss of hydraulic conductance ± SD (n = 4) as measured in pressurized twigs cut off 2 min, 20 min and 15 h after pressure release. Black columns refer to intact leafy twigs. Leafy twigs (grey columns) and leafless twigs (white columns) were radially supplied with KCl during pressurization through three areas where the epidermis was removed. Different letters indicate significant differences (P < 0.05) for Tukey pairwise comparisons.

Short-term (20 min) recovery from PLC was inhibited completely in twigs treated with vanadate (PLC was still about 40%, Fig. 3) and only 15 h after pressure release was PLC partially recovered (PLC decreased to 25%, similarly to that measured in KCl-treated stems, 20 min after inducing embolization). FC induced an impressively rapid and massive refilling of conduits and, in fact, only 20 min after pressure release PLC was as low as about 13% and further decreased to only about 3%, 15 h later. In other words, xylem recovery from embolism was practically complete at this time which did not happen in stems treated with KCl or in intact ones (Fig. 2).

Figure 3.

Percentage loss of hydraulic conductance ± SD (n = 4) as measured in pressurized twigs cut off 2 min, 20 min and 15 h after release. Twigs were radially supplied with 50 mm KCl or with 1 mm Na3VO4 (sodium orthovanadate) or with 20 µm fusicoccin (FC) during pressurization. Different letters indicate significant differences (P < 0.05) for Tukey pairwise comparisons.

Stems girdled at different distances from the pressure collar (Fig. 4), showed various responses in terms of changes in PLC as a function of the time intervals after pressurization. It can be noted that all girdling treatments gave similar PLC values 2 min after pressure release but 20 min later twigs from plants girdled at their base (GB) recovered similarly to twigs from not-girdled plants (PLC decreased from about 40% to about 22%) whereas twigs girdled at their junction plane (GJ) and twigs with double girdling at both sides of the pressure collar (DG) did not recover at all. Although GJ twigs showed small but significant recovery from PLC, 15 h after pressure release (PLC was still about 30%), DG stems maintained the initial PLC and even when treated with FC they were shown to be totally unresponsive in this regard.

Figure 4.

Percentage loss of hydraulic conductance ± SD (n = 4) as measured in previously girdled (G) pressurized twigs, radially supplied with KCl during pressurization and cut off 2 min, 20 min and 15 h after pressure release. GB, twigs from plants girdled at their base; GJ, twigs girdled at their junction with older twigs; DG, twigs girdled at both sides of the pressure collar used for inducing xylem embolism. DG + FC, twigs as above radially supplied with fusicoccin (FC). Different letters indicate significant differences (P < 0.05) for Tukey pairwise comparisons.

Dynamic changes in starch content of wood parenchyma cells

Wood parenchyma cells appeared to be full of starch granules in the stems in their ‘physiological’ status, i.e. non-pressurized stems (NP, Figs 5a, b & 6). In fact, over 90% of these cells were found to be filled with dark blue-stained starch, even after plants had remained in the dark for 30 h (for simplicity sake we have omitted to report similar data for stems observed 15 and 48 h after plants were put in the dark). In other words, plants remaining in the dark for up to 2 d showed no visible changes in the cell starch content and therefore no or very little starch hydrolysis could be assumed to occur in them. All air-injected stems regardless of which solution had been radially supplied during pressurization, showed the same percentages of HSC cells, 2 min after pressure release. Therefore, they were grouped together and labelled as P (= pressurized), in Fig. 6. Twenty minutes after pressure release, KCl-supplied stems showed consistently lower starch content than before (Fig. 5c) as indicated by the fact that only about 30% of parenchyma cells appeared to be filled with starch and similar percentages were counted 15 h later. The response of stems to FC in terms of wood cell starch content was impressive because this toxin caused less than 10% of wood parenchyma cells to appear filled with starch and most of them were not stained at all by the Lugol solution (Figs 5d and 6) indicating that nearly all starch was hydrolysed to sugars. Fifteen hours later, however, about 35% of wood living cells appeared to have regained their starch content.

Figure 6.

Percentage of wood parenchyma cells with high starch content with respect to the total ± SD (n = 4) as counted on cross-sections of intact non-pressurized twigs (NP) after plants remained 30 h in the dark. Pressurized twigs (P) cut off 2 min after pressure release were grouped together regardless the various treatments, i.e. they were either with intact cortex (KCl) or girdled at plant base (GB) or girdled at the junction with older twigs (GJ) or girdled on both sides of the pressure collar (DG). All twigs were radially supplied with KCl during pressurization. Counting of cells with high starch content were made 20 min and 15 h after pressure release. Different letters indicate significant differences (P < 0.01) for Tukey pairwise comparisons.

Stems from plants girdled at their base (GB, Fig. 6) did not differ from not-girdled ones in terms of changes in the percentage of HSC cells, 20 min or 15 h after pressure release (HSC cells were 30–38% of the total). Stems girdled at their junction plane (GJ) and those girdled at both sides of the pressure collar (DG) appeared not to change their wood starch content, 20 min after pressure release (HSC cells were in both cases 82–90% of the total). GJ stems, however, showed significant decrease of percentage of cells filled with starch, 15 h after pressure release (HCS cells decreased from about 90 to about 55%). In DG stems, on the contrary, the starch content appeared not to change even over the long term.

When the percentage of HSC cells was plotted versus the percentage loss of hydraulic conductance (combining all experimental data obtained 20 min and 15 h after pressure release), a positive linear correlation appeared to exist between the two variables (Fig. 7) with a correlation coefficient r2 of 0.801 and high statistical significance (P < 0.01). In particular, when HSC cells of embolized stems remained between 80 and 95% of the total, PLC was over 40% whereas when the percentage of HSC cells decreased by a half or more, PLC was found to range between 4 and 20%.

Figure 7.

Percentage loss of hydraulic conductance (PLC) measured in all twigs tested, cut off 20 min and 15 h after releasing the air pressure causing xylem embolism plotted versus percentage of wood cells with high starch content with respect to the total. The solid line represents the linear regression with a correlation coefficient r2 = 0.801; the two dashed lines are the 99% confidence intervals. The P-value is also reported (Pearson product moment correlation).

Table 2 reports percentages of wood parenchyma cells with high starch content counted for ray cells and vasicentric cells, separately in stems subjected to the various treatments tested. It can be noted that in every case when a significant decrease in the percentage of HSC cells with respect to the initial one was recorded, vasicentric cells appeared to undergo a larger decrease in their starch content than cells in the rays. As an example, KCl-supplied stems showed 37.51 ± 2.98% of ray cells with high starch content versus only 16.26 ± 3.77% of vasicentric cells, 20 min after pressure release, with a difference between the two cell types of about 50%. Similar differences were recorded for FC stems (both 20 min and 15 h after pressure release) and for GB stems (20 min) but not for GJ stems (20 min) and DG ones (both times tested).

Table 2.  Percentages of ray and vasicentric cells with high starch content with respect to the total ± SD
 Ray-parenchyma cells
with high starch content
Vasicentric-parenchyma cells
with high starch content
  1. Counts were made on cross-sections of intact non-pressurized twigs after remaining 30 h in the dark or in twigs air-injected (P = pressurized) to induce xylem embolism and cut off 2 min after pressure release. Measurements were also made in twigs radially supplied with KCl or with fusicoccin (FC) during pressurization as well as in twigs girdled (G) prior to pressurization. Here, twigs were girdled either at the plant base (GB) or at the junction plane between 1- and 2-year-old twigs (GJ) at both sides of the pressure collar used to inject air into stems (DG). Percentages of wood cells with high starch content were counted in twigs cut off 20 min and 15 h after pressure release. Different letters indicate significant differences (P < 0.05) for Tukey pairwaise comparisons.

Dark 30 h (n = 4)87.7 ± 8.5 a91.0 ± 3.3 a
P 2 min (n = 20)90.1 ± 3.5 a82.4 ± 9.8 a
KCl 20 min (n = 4)37.5 ± 3.0 b16.3 ± 3.8 c
KCl 15 h (n = 4)39.5 ± 4.3 b27.2 ± 4.6 ce
FC 20 min (n = 4)13.9 ± 6.3 c 3.2 ± 2.8 f
FC 15 h (n = 4)46.4 ± 5.9 bd25.1 ± 2.0 ce
GB 20 min (n = 4)36.4 ± 3.3 b17.2 ± 2.6 ce
GB 15 h (n = 4)39.5 ± 6.1 b29.4 ± 3.5 e
GJ 20 min (n = 4)85.3 ± 4.6 a83.7 ± 7.3 a
GJ 15 h (n = 4)61.1 ± 6.1 d48.9 ± 4.2 ba
DG 20 min (n = 4)89.7 ± 3.6 a83.2 ± 4.1 a
DG 15 h (n = 4)91.2 ± 4.8 a84.2 ± 3.6 a


In accordance with previous studies by some of the present authors (Salleo et al. 1996; Tyree et al. 1999), our present data further confirm that xylem refilling can occur in L. nobilis plants under low transpiration and at Px < −2T/r, within a few minutes after xylem conduit embolism has been induced. Hacke & Sperry (2003) have raised doubt about the possibility that positive air pressures applied to stems might favour what they called ‘novel refilling’; that is, rapid refilling occurring in laurel stems under tension. They based their hypothesis upon the fact that: (1) potted plants of the same species subjected to nearly the same water stress level as in the present work (Salleo & Lo Gullo 1993) were reported to recover only partially from loss of hydraulic conductivity after about 24 h and one irrigation; and (2) in their experiments, Hacke & Sperry (2003) noted that laurel plants stressed to PX < −2.0 MPa showed no evidence of refilling until PX was a few tenths of MPa (negative P’s). Furthermore, naturally droughted Laurel plants studied by Hacke & Sperry (2003) showed much slower (24 h) and variable refilling with respect to the air-injected plants studied by Salleo et al. (1996). This finding was in agreement with Cochard (2002) who reported no evidence of refilling in laurel stems previously centrifuged to PX < −2.0 MPa. We oppose their objections for three reasons: (1) laurel plants studied by Salleo & Lo Gullo (1993) were water stressed, tested for PLC and then re-tested for PLC 24 h later, while transpiring freely (at 1400 h). This protocol did not allow us to check whether laurel stems had recovered overnight and were cavitating again the next day. In the present case, plants were studied for short-term xylem refilling while at low transpiration. We recognize that water stressed plants are a more ‘realistic’ system than single stems induced to embolize and that plants under natural conditions might recover from embolism more slowly than under our experimental conditions. However, in the present study we were most interested in learning more about the mechanism of refilling of embolized stems under tension and assumed that this aspect could be studied more accurately in a better controlled system than in whole plants; (2) if stem pressurization per se induced rapid refilling, then we may expect to record significant (or at least some) reduction in PLC in all the stems measured for K, regardless how they had been treated. On the contrary, stems radially supplied with Na3VO4 (Fig. 3), those girdled at their junction plane (GJ, Fig. 4) or at both sides of the pressure collar (DG, Fig. 4) did not recover from PLC at all, 20 min after pressure release. DG stems even if treated with FC (DG + FC, Fig. 4) did not recover from embolism, 15 h after pressure release. This strongly suggests that stem pressurization per se did not affect PLC or its changes; (3) laurel stems studied by Hacke & Sperry (2003) as well as those measured by Cochard (2002) had been stressed to PX < −2.0 MPa. Salleo et al. (1996) have reported several pressure differentials applied to the interconduit pit membranes as resulting by combining different PX and air-injection pressures. Whenever PX was less than −1.75 MPa, rapid refilling was inhibited over the short-term and at PX of −2.0 MPa, also refilling over the long-term was prevented. Therefore, data by Hacke & Sperry (2003) and by Cochard (2002) are not in contrast with our previous or present findings. The excision of part of the twig epidermis (the ‘windows’ opened to supply twigs with different solutions) as well as the radial supply of solutions had no effect on changes in PLC (Fig. 2). Therefore, we conclude that our protocol did not cause artefacts, in this regard.

The excision of leaves prior to pressurization did not reduce xylem refilling (Fig. 2) because leafless twigs recovered from embolism in a similar manner to leafy ones. Our previous hypothesis was that leaves might export some hormonal signal (auxin) to stems through phloem inducing overloading which would drive radial flow to embolized xylem (Salleo et al. 1996). This was not the case in the present study, because the presence or the absence of leaves did not affect changes in PLC. Therefore, we are inclined to rule out the involvement of leaves in the refilling process.

The opposite effects on recovery from embolism revealed by twigs supplied with Na3VO4 and with FC (Fig. 3) clearly indicate that xylem refilling is the result of energy-dependent processes specifically requiring ATP hydrolysis and protein phosphorylation. In fact, in addition to inhibiting the plasma membrane proton pump (Marrè & Ballarin-Denti 1985), Na3VO4 is also known to be an unspecific inhibitor of many cellular ATP-ases and phosphatases (e.g. Schwartz, Illan & Assmann 1991). FC, in turn, is a very specific stimulator of the plasma membrane H+-ATPase (e.g. Sze et al. 1999). This consideration is of physiological importance, in our opinion. In fact, it implies that the proposed diurnal cycles of xylem embolism/refilling (e.g. Zwieniecki et al. 2000) would not be cost-free in terms of metabolic energy expended. It has been suggested that xylem cavitation may represent a pre-signal for stomatal closure, thus representing an important step of the stomatal control of transpiration (Salleo et al. 2000; Sperry 2000). The proposed pathway would consist of some cavitation of xylem causing stomatal closure that, in turn, would buffer xylem pressure potentials above the cavitation threshold for other conduits. The nocturnal refilling of cavitated conduits would be required to re-generate the signal. On the basis of our present data, we now think that this mechanism is only possible while PX remains above about −1.50 MPa, in laurel plants. Over this threshold, conduits can hardly be recovered and dynamic stomatal control of transpiration would change into more permanent stomatal closure impairing leaf gas exchange. Only different wood anatomies based upon smaller pit membrane pores and/or narrower conduits (Tyree & Sperry 1989; Milburn 1993; Lo Gullo et al. 1995) can, in this case, prevent cavitation and the permanent loss of wood efficiency.

Laurel stems have been reported (Salleo et al. 1996) to undergo inhibition to refilling when girdled about 19 min before pressurization but they recovered from embolism if girdled after pressurization. This suggested that the functional integrity of the phloem was required at the time of stem embolism and not thereafter. Our present data confirm that phloem has to be efficient for xylem recovery from embolism to occur at least in laurel, in accordance with Zwieniecki et al. (2000). In fact, the nearer to the embolized twig segments stems were girdled, the larger was the inhibition to refilling (GB stems recovered in a similar manner to those with intact phloem, GJ stems showed some refilling only after 15 h and DG stems did not recover at all, Fig. 4). We interpret these effects of girdling as caused by the progressively larger dissipation of the typical positive pressures in the phloem, the nearer to the embolized xylem the phloem was wounded. So, a possible conclusion might be that phloem pressures have to be maintained for reversal of embolism to occur.

Several reports exist in the literature suggesting that some starch-to-sugar conversion occurs in wood parenchyma cells (e.g. Fahn 1990) especially before bud-break in early spring (Alves et al. 2001, 2004) and in some cases (Canny 1997; Bucci et al. 2003) starch-to-sugar conversion has been related to xylem refilling through a reverse osmosis mechanism (see above). Bucci et al. (2003) were the first to present evidence for a role of starch–sugar conversion in embolism reversal. To the best of our knowledge, however, our data are the first attempt at quantifying this variable at the single wood cell level as affected by xylem embolism and refilling. Stem pressurization per se did not induce starch hydrolysis. In fact, cells of GJ and DG stems did not show any decrease in starch content in terms of percentage of HSC cells, 20 min after pressure release and the same was observed for DG stems, even 15 h later (Fig. 4). A large decrease in the percentage of HSC cells (from over 90 to about 30% in most cases and to 13% in the case of FC-treated stems) was recorded in stems that were also shown to refill (Figs 6 & 4, respectively) while no change in the number of HSC cells was observed in stems where refilling was inhibited. As a consequence, a close positive correlation was found to exist between PLC and percentage of cells with high starch content (Fig. 6). This strongly suggests that the two variables may be linked to each other with a causal relationship. The signal for starch-to-sugar conversion was likely to be generated during conduit embolism as suggested by the larger decrease of starch content in vasicentric cells (in close contact with conduits) than in ray cells (Fig. 5c and Table 2). In fact, the percentage of HSC cells was about 50% less in vasicentric cells than that in ray cells, in coincidence of xylem refilling. We did not measure the intraphloem pressures during xylem embolism and refilling (and this may reveal to be a hard task) but we note that whenever phloem was wounded (and, hence, inactivated) near the embolized twig segment, both refilling and starch hydrolysis were inhibited (Figs 4 & 6). Therefore, we can reasonably assume that changes in xylem conductance, starch-to-sugar hydrolysis and phloem pressure are parts of the refilling mechanism.

We know that the abrupt expansion of cavitation bubbles, develops large pressure changes in conduits that are likely to be transmitted to neighbouring cell membranes. To the best of our knowledge, there is no literature dealing with the effect of pressure changes on the activity of amylolytic enzymes such as α-amylase (which brings about the initial attack on starch). The plasma membrane H+-ATPase, by contrast, has been reported to be activated by mechanical signals like pressure (e.g. consequent to hyperosmotic stress, Zingarelli et al. 1999) and we also know that the proton pump energizes solute transporters among which are sugar antiporters. If this were the case, vasicentric cells might export sugars into the cavitated conduits through an active secondary transport (Alves et al. 2004), thus decreasing the conduit osmotic potential and contemporarily activating starch hydrolysis in living vasicentric cells through a mechanism based on increased sugar demand, in turn activating a sugar-sensing mechanism (Rolland, Moore & Sheen 2002). If sugars exported into cavitated conduits exceeded the amount of sugars produced by starch hydrolysis, starch hydrolysis would be further sustained and the signal would be propagated inducing similar processes in progressively more peripheral ray cells. This possible scenario would imply a centripetal flow of sugars towards the vasicentric cells where the proton pump is mostly activated, and from them into the embolized conduits. Enhanced sugar consumption, for example, by increased respiration (Alves et al. 2004) and exportation out of wood parenchyma cells would make them act as sinks to phloem, whose strength would be transiently higher than that of other common, physiological sinks (e.g. roots). As a consequence, a phloem-driven mass flow would be directed to xylem as sustained by positive phloem pressures. Whenever these were dissipated, in fact (DG stems, Fig. 4), no xylem refilling was recorded even 15 h after pressure release. In GJ stems, on the contrary, where phloem pressure dissipation was likely to be less brutal than in DG stems due to longer distance of girdling from the experimentally induced xylem embolism, the hydraulic conductance of xylem was recovered by about 25% in 15 h.

We are fully aware that our proposed mechanism for xylem refilling is based on speculative links of our experimental data to data of the literature. Nonetheless, we feel that a more than circumstantial evidence now exists that xylem refilling is the result of several actively maintained processes. Our knowledge of the extent and the kinetics of the metabolic control over the function modes of the entire xylem system clearly is at its dawning and will require more refined work.


The authors are grateful to Professor M.I. De Michelis, Department of Biology, University of Milano, Italy, for helpful discussion on the role of vanadate in plant cells.