Heterogeneity of gas exchange rates over the leaf surface in tobacco: an effect of hydraulic architecture?


A. Nardini. Fax: +39-040-5583875; e-mail: nardini@units.it


Spatial heterogeneity of gas exchange rates in the leaves of Nicotiana tabacum L. (tobacco) was investigated. Leaf conductance to water vapour was higher (by about 18%) at the apical regions of leaves than at the basal ones. Local, small-scale measurements of pressure–volume (PV) parameters and water status (performed with a dewpoint hygrometer) revealed that bulk leaf water potential, osmotic potential, turgor pressure and bulk modulus of elasticity were not significantly different in the leaf apex or base. Hydraulic measurements showed that the apical regions of the leaf blade were about 30% more conductive than the basal regions. Such differences were explained by analogous differences in terms of venation patterns. In fact, vein density turned out to be higher (by about 13%) near the leaf apex with respect to the leaf base. On the contrary, stomatal density was the same both in the apical and basal leaf portions. Our data suggest that spatial stomatal heterogeneity may arise from heterogenous distribution of local hydraulic resistances and would be addressed to maintaining local water potential above critical values, possibly triggering vein cavitation.


Leaves are the site of multiple functions among which light capture and conversion into metabolic energy have most contributed to shape their morphological and anatomical evolution in response to environmental pressure. As a consequence, leaves are characterized by high polymorphism not only among species and habitats, but also within one tree or herb depending on several factors including plant's age, incident light and water balance (Fahn 1990). Such a structural and functional heterogeneity can also be observed within a single leaf (Croxdale et al. 1992; Terashima 1992). As an example, classical and recent studies have repeatedly reported marked heterogeneity of stomatal conductance to water vapour over the leaf surface (e.g. Wong, Cowan & Farquhar 1979; Terashima et al. 1988; Buckley, Farquhar & Mott 1997). This heterogeneity may at least partly depend on variations of stomatal density over the leaf blade. In some cases, stomatal density decreases from the tip to the base of the leaf but opposite patterns have also been described (Pospíšilová & Santrůček 1994). Differences in stomatal density between the edge and the centre of the leaf blade are a common feature in many species (Rowland-Bamford et al. 1990; Weyers & Lawson 1997).

More recently, increasing attention has been devoted to the phenomenon known as ‘stomatal patchiness’, a special case of spatial and temporal stomatal heterogeneity implying stomata appearing mostly closed in small but distinct regions of the leaf blade while stomata in adjacent regions are mostly open (Mott & Buckley 1998, 2000). Although a thorough mechanistic explanation of the stomatal patchiness is still missing, chemical and/or hydraulic signals are among the best candidates for the co-ordination of stomatal behaviour within patches (Terashima et al. 1988; Mott 1995, 2007; Mott, Shope & Buckley 1999; Beyschlag & Eckstein 2001).

Heterogeneity of stomatal behaviour within a leaf implies that rates of water loss are not uniform over the leaf surface (Lawson & Weyers 1999). In fact, recent studies have revealed that the pattern of water supply to different leaf regions is also not uniform across species as well as within species (e.g. Nardini, Gortan & Salleo 2005). Nardini, Tyree & Salleo (2001) have observed that the hydraulic resistance of leaves of Prunus laurocerasus L. (cherry laurel) was highest towards the tip and lowest near the base. Similarly, Zwieniecki et al. (2002) and Cochard, Nardini & Coll (2004) have reported marked heterogeneity in vein hydraulic resistance over the leaf blade of Laurus nobilis L. (laurel) and Juglans regia L. (walnut), with possible species-specific pattern of spatial variation of this parameter. In turn, temporal variation of vein hydraulic resistance has been shown to arise as a consequence of vein embolism and refilling occurring over the short- to midterm (e.g. Salleo et al. 2001; Trifilòet al. 2003; Stiller, Sperry & Lafitte 2005; Salleo, Trifilò & Lo Gullo 2006). A recent study by Marenco et al. (2006) has suggested the possibility that cycles of vein embolism/refilling might be involved in the induction of temporal oscillations in stomatal aperture in Gossypium hirsutum L. (cotton).

Stomatal kinetics are not only responsible for the control of the balance between carbon gain and water loss, but also for maintaining the relative homeostasis of leaf water potential, which is usually kept above the critical threshold triggering vein and/or stem cavitation (Bond & Kavanagh 1999; Salleo et al. 2000; Trifilòet al. 2003). This is thought to be accomplished by tight co-ordination of water transport across the leaf between the liquid and vapour phase, that is, between the vein hydraulic resistance and stomatal conductance to water vapour (Hubbard et al. 2001; Meinzer 2002; Nardini & Salleo 2005; Woodruff et al. 2007). It has to be noted that stomatal responses as described in the literature have shown a poor correlation with bulk leaf water potential (e.g. Sperry & Saliendra 1994; Meinzer et al. 1995; Salleo et al. 2000, 2001; Meinzer 2002). Instead, a feedback has proved to exist between stomatal conductance and the vein (or stem) water potential corresponding to the cavitation threshold (Nardini & Salleo 2003). In other words, turgor changes, turgor-dependent chemical signals and hydraulic variations in the proximity of stomata, for example, changes in the resistance of the minor veins, are thought to be important factors influencing local stomatal behaviour. On this basis, it is tempting to speculate that heterogeneity of stomatal behaviour may arise from heterogeneous distribution or temporal variation of local hydraulic resistances over the leaf blade. In this sense, heterogeneous stomatal behaviour would be addressed at fulfilling the functional need to maintain local water potentials above critical values triggering xylem cavitation. Some experimental evidence supports this view (see Buckley 2005), but up to now, detailed, local and concurrent measurements of stomatal conductance to water vapour, vein hydraulic resistance and leaf water potential components are largely missing.

In the present study, we report measurements of stomatal conductance, hydraulic resistance, bulk water potential and pressure–volume (PV) parameters of discrete leaf regions (leaf discs) from different areas of tobacco leaves. Preliminary attempts have been made in testing the possibility of using a dewpoint hygrometer to measure water potential isotherms (PV curves) of small leaf areas with the purpose of estimating changes in turgor pressure and osmotic potential during dehydration of portions of the leaf blade collected from different leaf zones (see further discussion). Classical studies exist in the literature reporting water potential isotherms measured of leaf portions of angiosperms or of bryophythes using thermocouple psycrometers (e.g. Kyriakopoulos & Richter 1991; Kikuta & Richter 1992; Proctor et al. 1998), but there is only one report on dewpoint hygrometer as a tool to evaluate water relations of drying leaves (Richter 1978), at the best of our knowledge.


Preliminary experiments: water potential isotherms of leaves of Hedera helix L. (ivy) as measured with the dewpoint hygrometer compared to pressure chamber

In order to develop a suitable and reliable technique to measure the PV curves of small leaf areas, preliminary measurements of water potential isotherms were performed using both a pressure chamber (Scholander et al. 1965; Tyree & Hammel 1972) and a dewpoint hygrometer (WP4, Decagon Devices, Inc., Pullman, WA, USA). The WP4 hygrometer is based on the chilled-mirror dewpoint technique (Campbell, Campbell & Barlow 1973). This instrument measures water potentials from 0 to –80 MPa with a precision of ±0.1 MPa. Measurements were made of whole leaves (in the case of pressure chamber) and leaf discs (in the case of the dewpoint hygrometer) of ivy. The leaf discs were 40 mm in diameter, which is approximately the size of the sample holder of the WP4. H. helix was preferred for preliminary experiments because the surface area of the discs (approximately 12.5 cm2) was about 70% of the surface of a typical leaf of H. helix (Fig. 1b). Hence, the leaf discs were considered as representative of the whole leaf lamina of this species, thus, allowing a reliable comparison of the two techniques.

Figure 1.

Typical leaves of Nicotiana tabacum (tobacco, a) and Hedera helix (ivy, b) showing the leaf areas from where leaf discs for physiological measurements were sampled.

Leaves were collected in the evening before the experiments, from a single individual tree growing at the botanical garden of the University of Trieste, and were rehydrated overnight by enclosing them in a plastic film and leaving the petiole in contact with deionized water. The leaves were then inserted into the pressure chamber, and the PV curves were obtained according to the procedure described in detail by Salleo (1983). Measurements were repeated on five different leaves. Water potential isotherms were elaborated (Tyree & Hammel 1972) to obtain the osmotic potential at full turgor (π0), leaf water potential at the turgor loss point (Ψtlp) and the modulus of elasticity at full turgor (εmax).

In the case of measurements with the dewpoint hygrometer, the leaf discs were sampled from rehydrated leaves using a cork borer. The cuticle was gently abraded using a 600-grit sandpaper, and the discs were briefly rinsed by immersing them repeatedly in Petri dishes filled with deionized water. This procedure was aimed at improving the vapour pressure equilibration between the leaf interior and the WP4 chamber during PV analysis. The disc surface was rapidly and carefully dried using filter paper, and the discs were immediately placed in the sample holder of the WP4. Leaf water potential was measured in a continuous mode until the readings became stable, which usually took 10–15 min. Only samples with a starting water potential >–0.2 MPa were further analysed. The leaf discs were first weighed on a digital balance to obtain their initial fresh weight. Then, they were bench-dehydrated in steps of about 2.5 mg. Each dehydration step was followed by measurements of water potential. The procedure was repeated, and the PV curve was generated and used to calculate the π0, Ψtlp and εmax (see previous discussion). The typical time required to generate a complete PV curve using the WP4 was 4–5 h, that is, not much longer than that required when using the pressure chamber.

Growing tobacco plants

Seeds of tobacco were sowed in greenhouse trays. Fourteen days after sowing, seedlings were transplanted (one per pot) into 3 L pots filled with potting mix, and were grown in a controlled environmental chamber where air temperatures were adjusted to vary between 25 and 17 °C (day/night); relative humidity was set at 70%, and light was provided by lamps with a photosynthetically active radiation (PAR) of 400 ± 50 µmol m−2 s−1. The photoperiod was set at 12 h. Plants were irrigated daily with 200 mL tap water. All measurements were performed 6 weeks after transplantation when the plants were fully developed and inflorescence was present but not yet mature.

Measuring leaf water potential isotherms of tobacco leaf discs

In order to detect the eventual spatial variability of water relations of different regions of tobacco leaves, water potential isotherms were measured of leaf discs collected near the apex or the base of leaf blades (Fig. 1a), using the WP4 hygrometer. Basal mature leaves were sampled in the evening before the experiments, and were rehydrated overnight (see previous discussion). The leaf discs were obtained according to the procedure described earlier. In these experiments, the cuticle was not abraded using sandpaper, because the leaves of tobacco are thin and soft, and showed to be damaged by such a treatment. Instead, the leaf discs were pierced with a needle 80–100 times immediately before measurements to reduce cuticular resistances to vapour transfer (Kikuta & Richter 1992). PV curves were generated for three apical and three basal discs (from three different leaves), and π0, Ψtlp and εmax were calculated.

Measuring stomatal density, stomatal conductance and water potential of tobacco leaf discs

In order to quantify the eventual spatial and kinetic variation of stomatal patterning between different regions of tobacco leaves, stomatal densities and conductances (gL) were measured of leaf regions or discs sampled from the tip or the base of leaves. Stomatal conductances were measured prior to the collection of discs that were used for measuring stomatal density and water potential (see further discussion). Measurements of gL were performed at the apical and basal parts of different intact leaves, corresponding to the areas selected for PV analysis (see Fig. 1a). Five leaves from five different plants were used for gL measurements. At the time of measurements, the plants were fully hydrated and had been maintained in the light for at least 5 h. Leaf conductance to water vapour was measured using a steady-state porometer (LI-1600, Li-Cor, Inc., Lincoln, NE, USA) equipped with a 2 cm2 leaf chamber. Immediately after gL measurements, a leaf disc was sampled from the same region of the leaf using a cork borer and was immediately pierced with a needle 80–100 times. The leaf discs were then enclosed in sample holders and were measured for water potential using the WP4 (see earlier discussion).

Stomatal density was measured on leaf replicas obtained following the procedure described by Lawson, Weyers & A’Brook (1998). Briefly, Xantopren L Plus silicone impression material and hardener (Beyer Dental, Leverkusen, Germany) were mixed and smeared over the lower surface of the discs sampled from the apical and basal regions of leaves. After hardening, which took less than 5 min, the resulting negative leaf surface replica was peeled off. A 10 × 10 mm square was selected from the negative impression and was placed face down onto a layer of clear nail varnish painted on a microscope slide, which was left overnight to harden. Positive replicas were observed under a light microscope. All the stomata within a 1 × 1 mm square were counted, and stomatal density was calculated as the number of stomata mm−2. Measurements were repeated on 10 apical and 10 basal samples from 10 different leaves.

Measuring vein density and hydraulic resistance of tobacco leaves

Vein density was measured of leaf discs sampled from the apical and basal part of tobacco leaves. Discs were collected as described previously and were cleared using the procedure described by Berlyn & Miksche (1976). Subsamples with a surface of about 60 mm2 were excised from the discs and were observed at a magnification of 32× under a microscope (Leica Wild M420, Leica, Inc., Deerfield, IL, USA) equipped with a digital camera (Leica DC 300, Leica, Inc.). Images were then processed using the software ImageJ (v. 1.32j, National Institutes of Health, Bethesda, MA, USA), and the density of minor veins (of the third to fifth order) was estimated as total vein length per unit sample area.

Measurements of whole leaf hydraulic conductance (Kleaf) were made with a high-pressure flow meter (HPFM, Tyree et al. 1995), which has been widely used to measure leaf hydraulic properties (Gascò, Nardini & Salleo 2004; Nardini et al. 2005) and has been proven to yield correct and consistent values of Kleaf when compared with independent methods (Sack et al. 2002). Leaves were excised and immediately connected to the HPFM via the petiole using compression fittings. Distilled water filtered at 0.1 µm was forced into leaves at a pressure (P) of 0.3 MPa. The corresponding flow (F) was measured, and Kleaf was computed as K = F/P. Hydraulic conductance values were recorded at 16 s intervals until the values became stable (i.e. the coefficient of variation of the last 20 readings was less then 3%). Because Kleaf is known to be modulated by the light conditions (Sack et al. 2002; Tyree et al. 2005; Cochard et al. 2007), during measurements, leaves were illuminated using a fibre optic light source (FL-460 Lighting Unit, Heinz Walz GmbH, Effeltrich, Germany) providing a PAR of 400 µmol m−2 s−1, that is, the same PAR level at which plants were growing. At the end of each experiment, leaf surface area (Aleaf) was measured using a leaf area meter (LI3000A, Li-Cor, Inc.), and Kleaf was scaled by Aleaf.

The relative hydraulic conductance of apical versus basal leaf regions was measured using the leaf discs of 40 mm in diameter. Disc hydraulic conductance (Kdisc) was measured using the water potential relaxation technique first proposed by Brodribb & Holbrook (2003). The capacitance of leaf discs (Cdisc) was preliminarily estimated on the basis of PV analysis performed on the apical and basal regions of leaves (see previous discussion). Fully expanded leaves of tobacco were detached from well-hydrated plants, and were dehydrated on the bench for 20–30 min. Leaves were then enclosed in a plastic film for 30 min to stop transpiration and favour the equilibration of water potential throughout the leaf blade. An apical or basal disc was sampled from one side of the midrib, pierced 80–100 times with a needle and immediately measured using the WP4 hygrometer to obtain the initial value of leaf water potential (Ψ0). During measurements of Ψ0, the leaf (deprived of the excised disc) remained enclosed in the plastic film. A second disc was cut from the opposite side of the midrib while keeping the leaf immersed in distilled water. After cutting, the disc remained immersed in water for 90 s more in order to allow rehydration. The disc was then carefully dried and pierced with a needle as described previously, and the value of leaf water potential following rehydration (Ψf) was measured using the WP4. In all cases, the discs included a second-order vein crossing the disc through its centre. Measurements were performed on both the apical and basal discs sampled from 10 different leaves. Kdisc was calculated from the ratio of the initial to final water potential, and the disc capacitance according to Brodribb & Holbrook (2006):


where Ψ0 = initial water potential (MPa); Ψf = final water potential (MPa); t = duration of rehydration (s); and Cdisc = disc capacitance (kg m−2 MPa−1).


Data were analysed with the SigmaStat 2.0 (SPSS, Chicago, IL, USA) statistics package. Student's t-test was used to test differences between experimental groups.


Preliminary experiments on H. helix

Whole leaves and leaf discs of H. helix, when measured for water potential isotherms using the pressure chamber or the dewpoint hygrometer, respectively, showed considerably similar changes of leaf water potential (Ψleaf) during controlled, increasing dehydration of leaf samples (Fig. 2). In particular, changes of leaf osmotic potential between the two reference conditions of full turgor and zero turgor (Fig. 3) were remarkably similar independently on the method used. When checking changes in the Ψleaf components, however, some differences appeared to exist between the two techniques for turgor pressure (PT) decrease in response to leaf dehydration. The pressure chamber measured a slightly sharper drop of PT compared to the dewpoint hygrometer at equal percentage sample dehydration. As an example, for a 10% water loss, the pressure chamber measured a drop in PT from 1.7 to 0.9 MPa (–53%) versus a PT drop to 1.05 MPa (–38%) as calculated on the basis of measurements performed with the dewpoint hygrometer. Accordingly, a water loss of about 27% was recorded at the turgor loss point in the case of pressure chamber versus about 33% in the case of measurements obtained with the Decagon WP4. Leaf water potential at the turgor loss point and osmotic potential at full turgor, however (Ψtlp and π0, respectively), showed no statistically significant differences between the two techniques (Table 1). Discrepancies in the PT drop between the two techniques produced analogous differences in the bulk modulus of elasticity (εmax) that turned out to be higher when calculated on the basis of pressure chamber compared to the dewpoint hygrometer (8.9 versus 6.2 MPa, respectively). Such differences, however, were not statistically significant.

Figure 2.

Relationship between the inverse of leaf water potential (Ψleaf) and percentage water loss as measured during pressure–volume (PV) analysis of leaf discs or whole leaves of Hedera helix (ivy), using the WP4 dewpoint hygrometer (solid circles) or the pressure chamber (open circles), respectively. Means are reported ± SD (n = 5).

Figure 3.

Höfler diagrams reporting changes of leaf water potential (–Ψleaf, circles), osmotic potential (–π0, triangles) and turgor pressure (PT, squares) as a function of percentage water loss as measured in leaves of Hedera helix (ivy), using the WP4 dewpoint hygrometer (solid symbols) or the pressure chamber (open symbols), respectively. Mean values are reported (n = 5).

Table 1.  Leaf water potential at the turgor loss point (Ψtlp), osmotic potential at full turgor (π0) and bulk modulus of elasticity (εmax) as calculated on the basis of pressure–volume (PV) analysis performed on whole leaves or leaf discs of Hedera helix (ivy) using the pressure chamber or the WP4 dewpoint hygrometer, respectively
 Ψtlp, MPaπ0, MPaεmax, MPa
  1. Means are reported ± SD (n = 5). Differences between means are not statistically significant.

Pressure chamber−2.29 ± 0.15−1.69 ± 0.158.91 ± 1.73
Dewpoint hygrometer−2.46 ± 0.40−1.66 ± 0.316.22 ± 1.83

Measurements on tobacco leaves

When water potential isotherms of the leaf discs sampled from the apical or basal portion of tobacco leaves were elaborated and compared, no statistically significant difference appeared to exist between the two leaf regions in terms of Ψtlp, π0 or εmax (Table 2) because all the three variables resulted to be remarkably similar over the leaf blade independently on the leaf region measured.

Table 2.  Leaf water potential at the turgor loss point (Ψtlp), osmotic potential at full turgor (π0) and bulk modulus of elasticity (εmax) as calculated on the basis of pressure–volume (PV) analysis performed on leaf discs sampled from the apex or base of Nicotiana tabacum (tobacco) leaves
 Ψtlp, MPaπ0, MPaεmax, MPa
  1. Measurements were performed using the WP4 dewpoint hygrometer. Means are reported ± SD (n = 3). Differences between means are not statistically significant.

Apex−0.97 ± 0.21−0.70 ± 0.154.65 ± 0.42
Base−0.96 ± 0.06−0.72 ± 0.044.40 ± 0.18

Leaf conductance to water vapour as measured under steady-state conditions was significantly higher in the apical region of the leaf with respect to the basal one (by about 18%, Fig. 4a). In fact, gL near the leaf apex was about 200 mmol m−2 s−1 versus about 170 mmol m−2 s−1 as recorded at the base of the same leaf. Such a difference in gas exchange rates was not paralleled by analogous differences in terms of stomatal density between the two leaf regions because the number of stomata per unit leaf surface area was 67 ± 15 and 71 ± 21 stomata mm−2 for the leaf apex and base, respectively (the difference recorded was not statistically significant). Spatial variability of gL across the leaf blade did not translate into different water potential values between the corresponding leaf regions tested. In fact, the water potential of transpiring leaves was about –0.45 MPa when measured on the leaf discs collected both from the leaf apex and base (Fig. 4b).

Figure 4.

Leaf conductance (a) to water vapour and water potential (b) as measured of the apical or basal regions of transpiring tobacco leaves. Means are reported ± SD (n = 5). The P value is also reported; n.s., not significant.

Whole leaf hydraulic conductance (Kleaf) as measured with the HPFM turned out to be 2.20 ± 0.74 e-4 kg s−1 m−2 MPa−1. When Kdisc was measured using the water potential relaxation technique, the initial values of water potential were –1.43 ± 0.21 MPa and –1.48 ± 0.30 MPa for the apical and basal leaf discs, respectively. The final values of water potential (after 90 s rehydration) were –0.67 ± 0.22 MPa and –0.91 ± 0.24 MPa for the apical and basal leaf discs, respectively. The capacitance of the leaf discs as calculated on the basis of PV analysis in the corresponding range of water potential values was 4.02 ± 0.79 e-4 kg m−2 MPa−1, with no statistically significant difference between the apical and basal leaf regions in this regard. Kdisc estimated on the basis of Eqn 1 was 3.37 ± 1.33 e-4 kg s−1 m−2 MPa−1 for the apical region of the leaf versus 2.29 ± 0.83 e-4 kg s−1 m−2 MPa−1 for the basal one (Fig. 5a), that is, the discs sampled from the leaf apex were about 30% more conductive than the discs sampled from the leaf base. Total vein length per unit leaf surface area (vein density, Fig. 5b) was about 1.45 mm mm−2 for the apical leaf discs, and about 1.26 mm mm−2 for the basal ones with a difference between the two leaf regions of about 13%.

Figure 5.

Hydraulic conductance (a) and vein density (b) as measured of leaf discs sampled from the apical or basal regions of tobacco leaves. Means are reported ± SD (n = 10). The P value is also reported.


To the best of our knowledge, the present study represents the first attempt at dissecting the leaf blade into different zones and checking the variability of hydraulic properties, water status and leaf vasculature as related to stomatal conductance. Gas exchange rates were found to be not uniform over the surface of tobacco leaves, in that gL values recorded at the apical regions of the leaf were about 18% higher than at the basal ones (Fig. 4). In spite of likely analogous differences in the water loss rates, leaf net water potential (Ψleaf), Ψleaf at the turgor loss point (Ψtlp), osmotic potential at full turgor (πo) and bulk modulus of elasticity (εmax) were remarkably constant across the leaf (Table 2). This suggests that the apical and basal leaf regions had similar mean water status and cell wall mechanical properties. By contrast, measurements of total length of the leaf vein network per unit surface area and hydraulic conductance revealed that leaf areas near the apex where gas exchange rates were higher were better supplied with water than the basal areas because the former had a higher vein density and higher hydraulic conductance than the latter. This finding provides evidence that heterogeneity of stomatal aperture may be mechanistically linked to the efficiency of water supply changing along the leaf blade.

Preliminary experiments testing the WP4 dewpoint hygrometer as a tool for measuring water potential isotherms of leaves of H. helix proved to yield data in accordance with the same variables obtained with the pressure chamber (Table 1, Figs 2 and 3). Moreover, the time required to measure a PV curve of leaf discs using the dewpoint hygrometer was similar to the typical time required for measurements using the pressure chamber. We suggest that the dewpoint hygrometer is a suitable instrument to measure water potential isotherms of small portions of plant organs or of leaves that cannot be easily inserted into a standard pressure chamber as in the case of tobacco, which has wide, soft leaves. Local measurements of Ψleaf and its components have been performed with thermocouple psychrometers to generate water potential isotherms of leaves, thallus of bryophytes and lichens (Beckett 1997; Proctor et al. 1998). In most cases, however, the equilibration time required for each Ψ reading has been reported to be quite long (up to 4 h), making the measurement of a complete PV curve a rather time-consuming procedure.

The use of the WP4 allowed us to generate the PV curves for discrete portions of leaf blade. Our aim was to detect the eventual spatial variability of tissue water relations across the leaf blade. In the case of tobacco, no difference was recorded between the leaf apex and base in terms of Ψtlp, π0 and εmax. This indicates that relevant features of cell water relations like rigidity of living cell walls and osmotic potential were not directly related to stomatal behaviour. Of course, PV curves give the mean values of water relations parameters so that local changes of these variables at the stomatal level can be masked.

The higher gL recorded at the leaf apex compared to the base might be, in principle, a consequence of heterogeneous stomatal density or kinetics or of both. Our data show that stomatal density was similar at the two leaf regions studied. In our opinion, the recorded differences in gL were likely to be due to different levels of stomatal aperture. On the basis of the hydraulic analog of the Ohm's law, water loss, water potential and hydraulic resistance in a leaf are linked to one another by the


where E is transpiration rate, ΔΨ is the water potential drop from petiole to mesophyll, and R is the hydraulic resistance of the water pathway from petiole to leaf cells. As a consequence, different water loss rates among distant leaf regions would imply different steady-state water potentials, but only if the efficiency of water supply to the different leaf regions is the same, which was not the case in tobacco (Fig. 5). Previous studies have also advanced the hypothesis that heterogeneity of stomatal kinetics is caused by heterogeneous water status in different parts of the leaf (Beyschlag & Eckstein 2001). In our case, water potential was basically constant over the leaf with no statistically significant difference between the leaf apex and base in this regard. Because the basal and apical regions of the leaf did not differ in terms of water potential isotherms (Table 2), we conclude that also turgor pressure of mesophyll cells was approximately the same over the mesophyll despite the larger gL recorded at the leaf apex. In other words, local changes in the hydraulics of the leaf blade as due to differences in the vein hydraulic architecture can be interpreted to be a determinant of the observed heterogeneity of the stomatal behaviour in tobacco.

The hydraulic conductance of the leaf discs sampled from the apical or basal portion of the leaf was estimated on the basis of the water potential relaxation technique. To the best of our knowledge, this is the first attempt to apply this technique to discrete portions of the leaf blade. Values of Kdisc obtained on the basis of Eqn 1 ranged between 2.3 and 3.4 e-4 kg s−1 m−2 MPa−1 for the apical and basal regions of the leaf, respectively, which were very close to the K values obtained for the whole leaf (Kleaf) using the HPFM (2.2 e-4 kg s−1 m−2 MPa−1). This suggests that the water potential relaxation technique provides reliable values of leaf hydraulic conductance even when applied to isolated leaf portions. Moreover, Kdisc measured using this technique provided interesting insights into the possible role of leaf hydraulic architecture in determining different water loss rates across the leaf blade.

The hydraulic conductance measured of the leaf apical region, that is, of the area with higher gL, was about 30% higher than that of the basal leaf portion (where gL was 18% lower). Because the water potential was the same in the two leaf regions, we conclude that stomatal aperture was likely to be related to the efficiency of water supply and was regulated to maintain a target water potential relatively invariant over a short temporal scale. In this sense, our data are in accordance to analogous conclusions by Bond & Kavanagh (1999) and Salleo et al. (2000). Previous studies have suggested that the target water potential in a leaf would correspond to the critical threshold beyond which xylem cavitation is triggered (Trifilòet al. 2003), but the exact mechanism leading to the tight control of stomatal aperture and water potential remained (and still is) elusive. It could be argued that upon stomatal aperture, water loss rate increases and water potential decreases until some cavitation is triggered in peripheral veins leading to a sudden drop in local water supply to neighbouring cells and to a corresponding rapid drop of turgor pressure of guard cells (Zwieniecki, Brodribb & Holbrook 2007). In turn, this would trigger partial stomatal closure, thus buffering vein water potential within the critical threshold. In the case of Laurus nobilis L. (laurel) and Ceratonia siliqua L. (carob), the vulnerability to cavitation of leaf veins has been reported not to vary across the leaf blade (Salleo et al. 2001). In this case, we should expect gL to be lower in areas with lower hydraulic conductance, which was the case for tobacco leaves. Because refilling of cavitated veins is known to occur rapidly (Trifilòet al. 2003) and even under substantial tension (Stiller et al. 2005), cycles of vein embolism and refilling might lead to a continuous regulation of stomatal aperture around a steady-state value (Marenco et al. 2006).

Within a leaf, water travels along the vein system in the so-called ‘vascular pathway’. At the minor vein level, water leaves the xylem and moves to the ‘extravascular’ pathway, which consists of an apoplastic and a symplastic (cell-to-cell) route. The relative contribution of the vascular and extravascular pathway to Rleaf is variable according to species-specific patterns (Nardini et al. 2005; Sack, Tyree & Holbrook 2005) and to eventual responses to environmental factors (Tyree et al. 2005). In particular, the hydraulic resistance of the vascular pathway is mainly influenced by structural features like vein density and diameter of xylem conduits (Nardini & Salleo 2005) as well as by possible cycles of vein embolism/refilling, while the extravascular water pathway can be modulated via expression and/or activation of aquaporins (Cochard et al. 2007). Hence, the measured differences in Kdisc between the leaf apex and base in tobacco might be a consequence of either structural differences in the vein architecture or different membrane water permeabilities due to different levels of aquaporin expression. The tobacco aquaporin NtAQP1 has been well characterized (Siefritz et al. 2001), and its role in enhancing root water transport has been demonstrated (Siefritz et al. 2002). The analysis of tissue-specific expression of NtAQP1, however, has revealed a minor presence of this aquaporin in leaves (Otto & Kaldenhoff 2000), so that it is unlikely that differential expression of NtAQP1 between the apex and base of tobacco leaves was responsible for the observed differences in hydraulic resistance.

The apical regions of tobacco leaves appeared to have a higher vein density (by about 13%) than the basal regions. Leaf hydraulic conductance has been reported to be positively correlated to vein density across species (Sack & Frole 2006). In one species, similar relationships have been observed between leaf hydraulic conductance and the density of functional veins, that is, of the veins not impaired by embolism (Nardini, Salleo & Raimondo 2003; Trifilòet al. 2003) or pathological conditions like fungal infection (Raimondo et al. 2007). On this basis, it is plausible that the recorded difference in Kdisc between the leaf apex and base was to be attributed to the different vein densities recorded in the two leaf regions.

In conclusion, our data provide evidence for a role of leaf hydraulic architecture in determining the spatial variation of gas exchage rates in tobacco leaves. The possibility that even temporal variation of stomatal aperture is under hydraulic control as a consequence of cycles of vein embolism/refilling deserves further studies.