Vulnerability to cavitation of leaf minor veins and stems of Laurus nobilis L. was quantified together with that of leaflets, rachides and stems of Ceratonia siliqua L. during air-dehydration of 3-year-old branches. Embolism was estimated by counting ultrasound acoustic emissions (UAE) and relating them to leaf water potential (ΨL). The threshold ΨL for cavitation was less negative in L. nobilis than in C. siliqua according to the known higher drought resistance of the latter species. Leaf minor vein cavitation was also quantified by infiltrating leaves with fluorescein at different dehydration levels and observing them under microscope. Distinct decreases in the functional integrity of minor veins were observed during leaf dehydration, with high correlation between the two variables. The relationship between leaf conductance to water vapour (gL) and ΨL showed that stomata of L. nobilis closed in response to stem and not to leaf cavitation. However, in C. siliqua, gL decreased in coincidence to the leaf cavitation threshold, which was, nevertheless, very close to that of the stem. The hypothesis that stem cavitation acts as a signal for stomatal closure was confirmed, while the same role for leaf cavitation remains an open problem.
Cavitation in plant organs has been recorded using different techniques, ranging from measuring changes in hydraulic conductance to direct observations of cavitated samples under light and electron microscopy (Lo Gullo & Salleo 1991; Lo Gullo et al. 1995; Utsumi et al. 1998; Pate & Canny 1999) and, more recently, nuclear magnetic resonance imaging (NMR; Rokitta et al. 1999; Scheenen et al. 2000). A recent study (Kikuta et al. 1997) had attempted to detect cavitation in leaves of nine woody species using the acoustic method (Tyree & Dixon 1983), consisting of counting ultrasound acoustic emissions (UAE) from the leaf blade during air dehydration of leafy shoots. While this study confirmed the relevance of UAE to leaf dehydration, two problems remained open: (i) which leaf tissue had produced the majority of the UAE counted (conduits, fibres or even living cells), and (ii) if the large numbers of the UAE counted (e.g. over 2000 UAE recorded within 360 min from leaves of Ilex aquifolium) had any impact on leaf hydraulics and stomatal response.
Detecting cavitation in the leaf veins implies several methodological problems. In fact, the hydraulic methods commonly used to estimate xylem embolism in more-or-less cylindrical organs such as stems, roots and petioles are not suitable for leaf blades. As an example, the high pressure flow meter (HPFM), developed recently by Tyree et al. (1995) for measuring the hydraulic conductance of plant organs, operates at pressures of over 0·3 MPa (above the atmospheric pressure), which are expected to dissolve all the emboli eventually present in the xylem conduits. On the other hand, the vacuum technique (Kolb et al. 1996), operating at pressures up to + 0·02 MPa above vacuum (when used for single detached leaves of different Mediterranean species) proved to drive flows too small to be measured with the required accuracy, even if using balances with 0·01 mg resolution (Nardini, Tyree & Salleo 2001). Cryo-scanning electron microscopy, used recently for detecting cavitation in petioles of sunflower (Canny 1997), has been questioned as overestimating xylem embolism (Cochard et al. 2000). In comparison with the above methods, the acoustic method has the main advantage of being non-destructive, so that continuous UAE countings can be performed for several hours. Furthermore, a linear relationship between the cumulative number of UAE (cUAE) and the loss of hydraulic conductance has been reported to exist in young stems of Laurus nobilis L. (Salleo et al. 2000). Unfortunately, the acoustic method does not allow the discrimination of signals produced by cavitating conduits from those eventually emitted by other tissues (Sperry & Tyree 1988), so that the reported linearity between cUAE and the loss of hydraulic conductance of stems may not be verified in the leaves and cannot therefore be generalized. A further possibility for detecting cavitation in the leaf veins that was explored in the present study (see below) is infiltrating leaves with fluorescent dyes, allowing the entire functional vein network in a leaf and eventual changes in its distribution as caused by cavitation to be monitored.
The present study was aimed at: (i) measuring the extent to which cavitation-induced embolism occurs in the minor veins of leaves during dehydration, and (ii) discriminating the contribution of stem cavitation from that of leaf to stomatal response. For this purpose, two Mediterranean sclerophylls, L. nobilis (Laurel) and Ceratonia siliqua L. (Carob tree), were studied. The former species typically grows in relatively humid areas (Pignatti 1982) and has been reported to be rather vulnerable to cavitation (Salleo & Lo Gullo 1993; Kikuta et al. 1997; Salleo et al. 2000). Carob tree, meanwhile, grows in arid areas of all the Mediterranean Basin region (Pignatti 1982) and is a drought-avoiding water spender (sensuLevitt 1980), typically resistant to drought-induced cavitation (Salleo & Lo Gullo 1993).
Materials and methods
All of the experiments were conducted between June and September 2000 on 30 well-irrigated plants of L. nobilis and C. siliqua (15 plants per species). Plants were 5–7 years old and had been grown in the field near the Botanical Institute of Messina (Sicily, Italy) at sea level. Samples induced to cavitate consisted of 3-year-old branches of the two species, about 1 m long, 90 mm basal diameter and 0·8 m2 total leaf surface area. Branches were preferred to young twigs because they had proved to be large enough to ensure that internal water reservoirs were sufficient for sustaining transpiration during experiments (Salleo et al. 2000), thus buffering the ΨL drop, which would otherwise have been too rapid and difficult to stop at the desired values. On the evening before the experiments, plants were enclosed in black plastic bags to minimize transpiration and favour full turgor. In the early morning, branches were cut off (while still covered) under distilled water filtered to 0·2 µm and recut in the laboratory, where they remained in the dark and in contact with water until they were ready for measurements (see below). Under these conditions, base values of gL and ΨL were measured for at least five leaves per species using a steady-state porometer (LiCor 1600; Li-Cor Inc., Lincoln, NE, USA) and a pressure chamber (PMS 600; PMS Instrument Company, Corvallis, OR, USA), respectively. Plants were near full turgor (i.e. ΨL was ≥ −0·02 MPa). While branches were still in the dark and in contact with water, gL andΨL were measured every 10 min for about 1 h to check reference values of the two variables in well-hydrated branches. Plastic bags and water were then removed and branches were exposed to light, provided by mercury halide and high-pressure sodium lamps with an irradiance of 175 W m−2. All the experiments were conducted at a temperature of 27 ± 1 °C and relative humidity of 60 ± 5%, corresponding to common diurnal climatic conditions occurring during Sicilian summers (Salleo & Lo Gullo 1993). Air dehydration of branches continued until ΨL reached about −2·5 MPa for L. nobilis and about −3·0 MPa for C. siliqua, corresponding to minimum diurnal levels recorded in the field in Sicily during the summer (Lo Gullo & Salleo 1988; Salleo & Lo Gullo 1993). The time course of air dehydration of branches is reported in Fig. 1. The pre-established ΨL values were reached within 110 and 130 min for C. siliqua and L. nobilis branches, respectively. During branch dehydration, gL andΨL were measured on one leaf every 2·5 min to record rapid changes in the two variables (Salleo et al. 2000). More precisely, gL was measured first, then leaves were cut off and measured immediately for ΨL while in plastic bags to minimize water loss.
Estimating leaf and stem cavitation and embolism
Xylem cavitation was estimated using two different methods: the acoustic method (Tyree & Dixon 1983; Salleo & Lo Gullo 1989; Kikuta et al. 1997) and the ‘visual’ method, consisting of infiltrating leaves with fluorescein and observing them under light microscope before and after air dehydration (see below).
The acoustic method
The acoustic method consisted of counting UAEs (about 300 kHz in frequency) using two UAE transducers (RI 151; Physical Acoustics Corporation, Princeton, NJ, USA) connected to two different UAE counters (4615 Drought Stress Monitor, Physical Acoustics Corporation). One UAE transducer was positioned on the proximal third of the leaf blade (adaxial side) and the second was clamped on a proximal internode of twigs of the current year’s growth so that simultaneous UAE recordings were obtained from leaves and twigs (Salleo et al. 2000). In the case of C. siliqua, having long pinnate-compound leaves, simultaneous UAE recordings were obtained either from an apical leaflet and a twig or from the proximal third of the leaf rachis and a twig. The rachis was about 150 mm long, showed a secondary growth and might undergo significant cavitation, thus influencing water supply to leaflets. A thin layer of silicon grease was interposed between the transducer and the sample surface or, in the case of twigs, the exposed wood (a ‘window’ of about 16 mm2 was opened in the cortex) to secure better contact and prevent wood dessiccation. The UAE transducers were positioned while detached branches were still in the dark and the absence of UAE production was checked for at least 30 min. UAEs were amplified by 72 dB (52 dB by the main amplifier and 20 dB by the built-in transducer amplifier) and recorded every 60 s using a stopwatch (accuracy ± 1 s). To estimate ΨCAV, the water potential of one leaf (L. nobilis) or leaflet (C. siliqua) was measured every 2·5 min. In every case, experiments lasted about 90 min.
The ‘visual’ method
At least five 3-year-old branches of each species were collected and dehydrated in the light as described earlier. Starting from near full turgor (ΨL ≥ −0·02 MPa), different levels of dehydration were tested: leaves were collected at ΨL = −0·5, −1·0, −1·5, −2·0 and −2·5 MPa (for L. nobilis) and at ΨL = −1·0, −1·5, −2·0, −2·5 and −3·0 MPa (for C. siliqua), corresponding to commonly recorded levels in the field (Salleo & Lo Gullo 1993). At these pre-established ΨLs, leaves were cut off under distilled water, filtered to 0·1 µm to prevent vein clogging with spurious emboli or debris, and connected immediately to the apparatus described by Kolb et al. (1996). This consisted of a vacuum flask (2 L Pyrex) connected to a vacuum pump. The leaves were inserted into the flask with their petiole fitted tightly to rigid peek tubing, passing through the rubber seal of the vacuum flask. The opposite end of the tubing was immersed in a fluorescein solution (0·7 mm fluorescein, sodium salt, dissolved in 100 mm KCl). Leaf infiltration with fluorescein was performed under a subatmospheric pressure of + 0·02 MPa (above vacuum), which was maintained constant for 5 h. Preliminary experiments had shown that such a procedure infiltrated well-hydrated leaves of the two species without any visible embolism in the leaf veins, as revealed by leaf areas with non-fluorescing veins. After infiltration, leaves were removed from the apparatus and leaf (or leaflet) blades were put under a binocular (Wild M8; Leica Camera AG, Solms, Germany) equipped with a mercury vapour lamp with a UV excitation filter (360/40 nm). The microscope was then connected via a telecamera (Leica DC 100) to a computer with dedicated software for image analysis. Images were processed and assembled using Adobe Photoshop software (Adobe Systems Inc., San José, CA, USA)
To estimate the extent to which leaf veins became non-functional due to cavitation and consequent embolism, leaf images were modified in grayscale using Paint Shop Pro software (Jasc Software Inc., Eden Prairie, MN, USA) and imported in a software designed for image processing (Erdas Imagine; Erdas Inc., Atlanta, GA, USA). An unsupervised classification was performed on each leaf image, using the isodata algorithm. The isodata clustering method uses the minimum spectral distance formula to form clusters. It begins either with arbitrary cluster means or with means of an existing signature set; each time the clustering repeats, the means of these clusters are shifted. The new cluster means are used for the next iteration and the calculations are repeated until a maximum percentage of unchanged pixels has been reached between two iterations. This classification produced 10 classes in each leaf image: classes one to five were non-fluorescent areas of the leaf, while classes six to 10 were fluorescent ones. Erdas image was then used to calculate the whole-leaf surface area as well as the percentage of fluorescent area with respect to total leaf area. Five to seven leaves per species and per water potential level were observed.
Leaves and stems of the two species under study emitted several acoustic signals during dehydration (Table 1). As expected, the cumulative number of UAE (cUAE) counted at the reference values of about −2·5 MPa for L. nobilis and −3·0 MPa for C. siliqua were about 80% in the leaves of both species with respect to stems. This was likely to depend on a smaller number of conduits in the former with respect to that in the latter. In Figs 3 and 4, UAE are reported as percentages of the maximum (Table 1) and plotted versus corresponding leaf water potentials. The cavitation threshold (ΨCAV) for stems and leaves was set arbitrarily at 10% UAE with respect to the maximum of each species (Salleo et al. 2000). In every case, such a UAE percentage corresponded to the flex point of the asynthotic curve, which has been reported typically for the relationship of percentage UAE to ΨL in several woody species (e.g. Kikuta et al. 1997; Nardini & Salleo 2000). In L. nobilis(Fig. 2), ΨCAV of leaves turned out to be significantly less negative than that of stems (−0·59 versus −1·23 MPa). In other words, stems began to produce UAEs when ΨL reached −1·23 MPa, while at ΨL = −0·59 MPa, leaves of Laurel emitted UAEs but stems did not. Leaflets and rachides of C. siliqua(Fig. 3) had equal ΨCAV levels (about −1·75 and −1·78 MPa, respectively). Stems, on the other hand, began to emit acoustic signals at ΨL levels of about −2·25 MPa. Although stem ΨCAVs in this species were only slightly more negative than those for leaflets or rachides, the difference between the two organs, in this respect, was statistically significant (P < 0·01, Student’s t-test).
Table 1. Cumulative ultrasound acoustic emissons counted on twigs, rachides and leaves or leaflets of L. nobilis and C. siliqua (means ± SD) (n = 10)
694 ± 117
1085 ± 260
299 ± 137
140 ± 35
152 ± 23
Fully hydrated leaves of L. nobilis infiltrated with fluorescein (Fig. 4) fluoresced in the finely reticulate network of minor veins in detail. Here, veins of the first order (i.e. branching from the midrib) were quite evident, together with those of increasing orders up to the very minor veins. At first sight, leaves dehydrated to ΨL = −0·5 MPa showed a reduction in the functional fluorescent veins in that most of the very minor veins were no longer visible. In leaves at ΨL levels of −1·0 and −1·5 MPa, only veins of the first, second and third order were visible, with some marginal connections between them still fluorescing. The distribution of the vein network was less and less visible in further dehydrated leaves (ΨL = −2·0 and −2·5 MPa).
Analogous progressive decrease in the fluorescing vein network was visible in progressively dehydrated leaflets of C. siliqua(Fig. 5). However, leaflets dehydrated up to ΨL = −1·0 MPa appeared to be the same as those at full turgor. In more severely dehydrated leaves (to ΨL levels of −1·5, −2·0 and −2·5 MPa), minor veins were less and less visible, thus indicating that they were no longer conducting. The marginal connections between the veins of the first order, however, were still visible even in leaves dehydrated up to ΨL = −2·5 MPa. The most severely dehydrated leaves (ΨL = −3·0 MPa) showed no visible veins of orders higher than the first one.
An estimate of the loss of the conducting system in the leaf blades due to dehydration-induced embolism is reported in Fig. 6 and expressed in terms of the measured decrease in the percentage of fluorescent area with respect to the total leaf surface area. In both species, a positive correlation was found to exist between the decrease in the fluorescent area in the leaf blades and the decrease in leaf water potential. In other words, leaf dehydration was accompanied by a progressive decrease in the amount of the functional xylem conduits. The correlation existing between the two variables was fairly good and statistically significant in both species (r2 = 0·94 and 0·93 for L. nobilis and C. siliqua, respectively).
The relationship between gL and ΨL (both measured every 2·5 min) of dehydrating leaves is reported in Fig. 7. In both species, such a relationship was expressed by a Gaussian with an r2 value of 0·77 and 0·51 for L. nobilis and C. siliqua, respectively. In the latter case, the larger scattering of data intrinsic to gL measurements was likely to be responsible for the lower correlation calculated for the relationship of gL to ΨL. After branches were exposed to light, gL began to increase up to peaks of 125–150 mmol m−2 s−1 in L. nobilis and 250–350 mmol m−2 s−1 in C. siliqua. When the water potential of Laurel leaves reached the threshold for leaf cavitation (ΨCAV leaf, Fig. 7), gL continued to increase and gL decreased distinctly, reaching base cuticular levels at ΨLs of about −2·5 MPa. Although the ΨCAV levels for leaves and stems were significantly different from one another in the case of C. siliqua (Fig. 7), they were sufficiently close to one another to make it difficult to discriminate clearly between the contribution of leaf cavitation to stomatal response and that of stem. Even in this species, however, leaves exposed to light and deprived of external water supply showed increasing gL levels up to a peak that coincided with the ΨL range, triggering cavitation in the leaflets and rachides.
Simultaneous recordings of ultrasound acoustic signals emitted by leaves and stems of L. nobilis and by leaflets, rachides and stems of C. siliqua revealed that L. nobilis was more sensitive to cavitation than C. siliqua. In fact, ΨCAV levels estimated for the former species were, on average, significantly less negative than those for the latter. In particular, the estimated ΨCAV for leaves of L. nobilis was about three times less negative than that estimated for leaflets and rachides of C. siliqua (−0·59 versus −1·75 MPa). Stems of Laurel also showed ΨCAV significantly less negative than that estimated for Carob tree. This is in agreement with the typical habitats where the two species grow (the habitat of L. nobilis is relatively humid, that of C. siliqua is typically arid). In both species, leaves showed lower threshold xylem pressures for cavitation (higher ΨCAV) than stems. Previous studies of the water relations of Laurel and Carob tree plants (Lo Gullo & Salleo 1988; Salleo & Lo Gullo 1993) had shown that ΨL varied between about −0·3 and −2·3 MPa in Laurel plants and between about − 0·4 and − 2·8 MPa in Carob trees. Therefore, our data – while confirming analogous data reported previously for Laurel (Salleo et al. 2000) – suggest that even in Carob tree, leaves may cavitate earlier in the morning than stems.
Kikuta et al. (1997) have monitored leaf cavitation of several woody species using the acoustic method. In this study, cUAE counted on leaves near the turgor loss point ranged from about 200 UAE in Phillyrea angustifolia L. up to over 2000 UAE in I. aquifolium L., with reasonably higher cUAE recorded in the most drought-sensitive species (e.g. Acer campestre L. and I. aquifolium) compared with the most resistant ones (P. angustifolia). While the authors were aware of the difficulty of discriminating UAEs emitted by the midrib (on which the UAE transducer was positioned) from those emitted by the conduits of the minor veins – and even from those emitted by other leaf tissues – they computed the ratio of the cUAE to the number of conduits on the basis of the total number of conduits of the midrib alone. Our data show that the leaf minor veins are very sensitive to cavitation-induced dysfunction (Figs 5 and 6) and that, therefore, an unknown fraction of the cUAE (counted by Kikuta et al. 1997) could have been emitted by these. Some intervenous areas were likely to be no longer supplied with water in leaves at ΨL levels of −0·5 MPa in L. nobilis and of −1·5 MPa in C. siliqua, which corresponded more or less to ΨCAVs of the two species as estimated in terms of 10% UAE with respect to the maximum. Our data apparently contradict the results of Canny (2001), who found no significant embolism in the small veins of the maize leaf lamina while larger tracheary elements were gas-filled between 1000 and 1800 h. This unexpected pattern was explained by the author in terms of higher rate of refilling of narrow conduits with respect to larger ones. While we agree that some refilling is likely to occur in the leaf veins immediately after cavitation (see below), we feel that Canny’s data might be influenced by spurious emboli induced by the freezing procedure, causing gases to escape xylem sap and filling conduits (Cochard et al. 2000). The well-known positive relationship existing between conduit diameter and vulnerability to freeze-induced embolism (Tyree & Sperry 1989; Sperry & Sullivan 1992; Lo Gullo & Salleo 1993) might be a more parsimonious explanation of the pattern of leaf vein embolism reported by Canny (2001). We are aware that leaf infiltration with fluorescein has an intrinsic limit, i.e. that the major veins containing several conduits appear to fluoresce even if some conduits are lost due to embolism. This is especially true for the midrib that has been reported to contain about 400 conduits in L. nobilis and about 250 conduits in C. siliqua (Kikuta et al. 1997). In other words, the ‘visual’ method suffers from some underestimation of the impact of cavitation on leaf xylem. The major scope of our work was, however, to monitor the vulnerability to cavitation of the minor veins that are responsible for the very fine water supply in a leaf. Leaf minor veins are known to contain only a few or even single tracheids (Fahn 1990) supplying water to no more than 10–15 living cells, so that their functional integrity could be detected easily under a microscope on the basis of their fluorescence. Leaf infiltration with fluorescein revealed that the midrib and veins of the first and second order remained functional enough to appear clearly fluorescent, i.e. sufficiently conductive in leaves dehydrated to ΨL levels commonly recorded in the field (Salleo & Lo Gullo 1993). Moreover, it is worth noting that the marginal veins, i.e. those ‘running’ along the leaf margins and connecting veins of the first order, were still functional in leaves of Laurel at ΨL = −2·0 MPa and in leaves of Carob tree up to ΨL = −2·5 MPa. This might be the reason why the overall leaf hydraulic conductance is only slightly affected, even when the midrib is clogged (Nardini et al. 2001). In both species studied, well-hydrated leaves infiltrated with fluorescein showed a total fluorescent area that corresponded to that of the vein network and that was of the order of 35–45% of the total leaf surface area (Fig. 7). Leaf dehydration, as estimated in terms of decreasing water potential, induced significant decreases in the percentage fluorescent area, which dropped to 17% (i.e. by over twice). Even if this method underestimates the actual loss of leaf conduits, it gives a clear idea of the extent of the potential damage to the low-resistance pathway in a leaf subject to cavitation.
The highest hydraulic resistance in a leaf has been reported to reside in the non-vascular water pathway – i.e. in the living tissues (Tyree & Cheung 1977). As a consequence, the diffuse blockage of the venous system, which is a low-resistance pathway, can be expected to not have a great effect on the overall leaf hydraulics (Nardini et al. 2001). However, this is only true over the short term. If the leaf veins are not recovered promptly from embolism, the water supply to living cells will be severely impaired. Because our data show extensive damage to the leaf minor veins at ΨL levels commonly recorded in the field and in the natural habitats of the two species under study, cavitated leaf veins can be expected to be refilled efficiently and rapidly. The mechanism and the kinetics of refilling of cavitated conduits are still largely unknown (Tyree et al. 1999; Holbrook & Zwieniecki 1999). The method of leaf infiltration with fluorescent dyes for monitoring vein cavitation may be informative in this respect.
In previous studies, some of us had hypothesized that leaf vein cavitation might represent the signal for gL regulation (Salleo et al. 2000; Nardini & Salleo 2000). We have no clear-cut evidence that this is the case. Laurus nobilis leaves did not react to leaf cavitation with stomatal closure (Fig. 7) because gL dropped only when stems began to cavitate. In the case of C. siliqua, meanwhile, the gL peak coincided with ΨL levels that triggered cavitation in the leaflet. However, the large SDs of the mean ΨCAVs for leaves did not allow us to discriminate clearly between the contribution of cavitation in the leaf and that in the stem to gL decrease (stomatal closure), and we doubt that more replications would give better results because of the largely differing gL levels typically recorded in this species. While we do not rule out the possibility that diffuse cavitation in the leaflet and rachis of Carob tree can induce stomatal closure, we feel that the hydraulic signal generated in the cavitated stem is much more likely to be responsible for gL regulation. In fact, the loss of some wide conduits in the stem is expected to lead to a noticeable drop in water supply to leaves, thus triggering the cascade of integrated chemical and physical events that is known to lead to stomatal closure (Tardieu & Davies 1993). It has to be noted, however, that branch air dehydration caused ΨL to reach the cavitation threshold for leaves (ΨCAV, Fig. 2) about 20 min after branches were exposed to light and deprived of water supply (Fig. 1). This might cause the overlapping of two opposite signals, i.e. the light signal for stomatal opening and the hydraulic signal for stomatal closure. If this were the case, the cavitation-induced hydraulic signal might be masked by the opposite light signal.
In conclusion, more research, extended to other species with different stomatal responses to dehydration, is needed to achieve a better insight into leaf cavitation as a possible signal for stomatal control of transpiration.
Financial support was provided by the Italian Ministry for University and for Scientific and Technological Research (National Projects). We are grateful to Dr Paola Giacomich (Department of Biology, University of Trieste) for help in the processing of leaf images.
Received 30 January 2001;received inrevised form 4 May 2001;accepted for publication 4 May 2001