Leafminers help us understand leaf hydraulic design


A. Nardini. Fax: +39 040 568855; e-mail: nardini@units.it


Leaf hydraulics of Aesculus hippocastanum L. were measured over the growing season and during extensive leaf mining by the larvae of an invasive moth (Cameraria ohridella Deschka et Dimic) that specifically destroy the palisade tissue. Leaves showed seasonal changes in hydraulic resistance (Rlamina) which were related to ontogeny. After leaf expansion was complete, the hydraulic resistance of leaves and the partitioning of resistances between vascular and extra-vascular compartments remained unchanged despite extensive disruption of the palisade by leafminers (up to 50%). This finding suggests that water flow from the petiole to the evaporation sites might not directly involve the palisade cells. The analysis of the temperature dependence of Rlamina in terms of Q10 revealed that at least one transmembrane step was involved in water transport outside the leaf vasculature. Anatomical analysis suggested that this symplastic step may be located at the bundle sheath where the apoplast is interrupted by hydrophobic thickening of cell walls. Our findings offer some support to the view of a compartmentalization of leaves into well-organized water pools so that the transpiration stream would involve veins, bundle sheath and spongy parenchyma, while the palisade tissue would be largely by-passed with the possible advantage of protecting cells from short-term fluctuations in water status.


Water flows from roots to leaves through the xylem, which represents over 99% of the overall length of the water pathway in plants (Tyree & Zimmermann 2002). At the two opposite poles of this pathway (i.e. leaves and roots), however, water flows for a short distance through a series of cell layers in an extra-vascular pathway. Water movement through non-vascular tissues in roots is relatively well understood (Steudle 2000), and studies have provided evidence for different pathways including apoplastic, symplastic and transcellular route for water flowing centripetally through the cortex with a site of high hydraulic resistance represented by the endodermis. The very last millimetres of the liquid water pathway before evaporation are located in the leaves which make up a disproportionally high fraction (up to 30%) of the overall plant hydraulic resistance (Yang & Tyree 1994; Sack & Holbrook 2006; Sellin, Õunapuu & Kupper 2008). As a consequence, leaf hydraulic resistance (Rleaf) is one of the major functional constraints imposed on plant gas exchange and photosynthesis (Nardini & Salleo 2003; Brodribb et al. 2005; Brodribb & Jordan 2008). On the basis of the above, it is not surprising that the study of leaf hydraulic architecture has received increasing attention in recent years (Zwieniecki et al. 2002; Salleo et al. 2003b; Cochard, Nardini & Coll 2004; Sack, Streeter & Holbrook 2004; Aasamaa, Niinemets & Sõber 2005; Brodribb, Feild & Jordan 2007).

A detailed description of the partitioning of hydraulic resistances within the leaf is of fundamental importance for understanding leaf hydraulic design and its functional consequences under different environmental conditions. In angiosperms, water enters the leaf lamina through the petiole and the midrib(s) branching into progressively higher-order veins. Water can either exit major veins to enter the surrounding tissues (Salleo et al. 2003b) or it can flow to the minor veins (Canny 1990), which account for the largest fraction of total vein length. Our present knowledge of the actual pathway(s) followed by water outside the vein xylem is still incomplete. There is current agreement that water crosses the bundle sheath – a layer of parenchyma cells wrapping the veins – but the exact route is still unknown. Early studies had suggested that water leaves the xylem through the apoplast of these cells (Boyer 1974), but the demonstrated occurrence of aquaporins at the bundle sheath cells suggests that water movement through cell membranes is plausible as well (Otto & Kaldenhoff 2000; Hachez et al. 2008; Sakurai et al. 2008).

Beyond the bundle sheath, our knowledge of the route followed by water becomes even more dubious. In fact, the leaf is composed of many cell types (epidermal cells, spongy and palisade parenchyma, mechanical tissues), many of which might be involved to different degrees in water transport to the evaporation sites. The hydraulic resistance of the leaf xylem is thought to be approximately of the same order as that of the extra xylem pathways, and species seem to vary in the partitioning of leaf hydraulic resistances with some habitat-specific trend (Nardini, Gortan & Salleo 2005a; Sack, Tyree & Holbrook 2005; Sack & Holbrook 2006).

Rapid variations in hydraulic behaviour in response to environmental factors represent further complications to the study of leaf hydraulics. For instance, leaf hydraulic resistance is known to be modulated by light (Lo Gullo et al. 2005; Tyree et al. 2005; Scoffoni et al. 2008), temperature (Sack et al. 2004; Sellin & Kupper 2007) and water availability (Nardini & Salleo 2005). Changes in Rleaf have been shown to arise as a consequence of the hydraulic properties of both the vascular and extra-vascular water pathways. Vein hydraulics can change as a function of cavitation-induced embolism (Nardini, Tyree & Salleo 2001; Nardini, Salleo & Raimondo 2003) eventually followed by refilling (Trifilòet al. 2003; Stiller, Sperry & Lafitte 2005; Nardini et al. 2008), whereas water transport properties of the extra-vascular pathway can be modulated via expression and/or regulation of aquaporins (Nardini, Salleo & Andri 2005b; Cochard et al. 2007; Kim & Steudle 2007; Lovisolo et al. 2007).

All the above factors may modify the actual route followed by water in the transpiration stream (and related hydraulic resistances), making the understanding of this important physiological feature quite elusive (Sack & Holbrook 2006). A recent study by Zwieniecki, Brodribb & Holbrook (2007) has provided interesting and stimulating insights into the possible compartmentalization of leaves into well-organized water pools separated by hydraulic resistances of sufficient magnitude to maintain water potential disequilibria between different mesophyll tissues. In four out of the six angiosperms studied by Zwieniecki et al., experimental results suggested a design based on leaf veins being hydraulically connected to the bundle sheath, spongy parenchyma and epidermis. In this scenario, palisade tissue would be relatively isolated and by-passed by the transpiration stream, with the possible advantage of protecting these major photosynthetic cells from short-term changes in their water status.

Zwieniecki et al. (2007) based their conclusions on detailed analysis of rehydration kinetics (Tyree et al. 1981) and anatomical measurements of relative volumes occupied by different mesophyll tissues. Ye, Holbrook & Zwieniecki (2008) have recently provided convincing, but still indirect evidence for a close hydraulic linkage between xylem and epidermal cells. An experimental demonstration of the extent of hydraulic connection between palisade tissue and other leaf components requires Rleaf to be measured after selective removal of palisade cells, a rather challenging task for most researchers, but apparently not for Cameraria ohridella Deschka et Dimic (Lepidoptera, Gracillariidae), an invasive pest of Aesculus hippocastanum L. (horse chestnut) in Europe (Gilbert et al. 2003). The leaf-mining larvae of this moth specifically tunnel through the palisade while leaving the upper and lower epidermis, spongy mesophyll, veins and associated bundle sheath cells largely unaffected (Raimondo et al. 2003). Leaf mining by C. ohridella causes severe damage to plant photosynthetic capacity, growth and reproduction of horse chestnut trees (Raimondo et al. 2003; Salleo et al. 2003a; Thalmann et al. 2003; Nardini et al. 2004; Raimondo et al. 2005). On the other hand, the very selective and peculiar feeding behaviour of larvae of C. ohridella provides a promising possibility to test the hypothesis first advanced by Zwieniecki et al. (2007).

In the study by Raimondo et al. (2003), Rleaf of both healthy and mined leaves of A. hippocastanum was reported to be similar, despite extensive destruction of palisade by larvae. However, the lack of any impact of palisade destruction on Rleaf values could not be explained because information of the partitioning of hydraulic resistances within horse chestnut leaves was missing. In fact, if vascular hydraulic resistance were dominant over extra-vascular one, any perturbation of extra-vascular water pathways would be expected to have negligible effects on Rleaf, provided that the vascular system is not affected as clearly demonstrated by Raimondo et al. (2003) on the basis of vital staining of leaf veins during transpiration.

In the present study, we report measurements of leaf hydraulic resistance and architecture of both control and mined leaves of A. hippocastanum, and provide insights into the possible pathway(s) followed by water within the leaf in this species.


Plant material

Experiments were performed on leaves sampled from six adult individuals (more than 30 years old) of A. hippocastanum growing in the Botanical Garden of University of Trieste. The study trees have been regularly attacked by C. ohridella since 1997, and since then they have suffered precocious leaf shedding by mid-September (Salleo et al. 2003a). Larvae of C. ohridella feed on the cells of the palisade, firstly widening the mine in circle (dark spot up to 6–8 mm), and then continuing to enlarge it along the second-order veins. The mine can grow up to 30 mm in length and up to 12–15 mm in width, but very rarely individual mines cross a second-order vein. In-between second-order veins, the larvae feed only on palisade, leaving intact the spongy mesophyll, veins and associated bundle sheath cells. In A. hippocastanum, bundle sheath extensions can be clearly observed up to third-order veins. Higher-order veins (fourth and fifth) apparently lack bundle sheath estensions towards the upper epidermis, so that the larvae can easily move between areoles while still leaving the bundle sheath undamaged. Of course, bundle sheath extensions of third-order veins are affected when larvae feed above these veins.

Measurements were performed sequentially during seven different 1 week periods between early April (approximately 15 d after bud sprouting) and the end of August 2008 (when most leaves showed extensive damage caused by the larvae of C. ohridella). During 2008, the first visible sign of attack in our study trees was observed at the end of May because of the very rainy spring which was adverse to the emergence of adult forms of the insect. Additional measurements were performed between April and August 2009 on both attacked trees, as well as on two individuals that had been preliminarly treated with a systemic insecticide that prevented C. ohridella attacks over the growing season (controls). In treated trees, less than 10% of the foliage showed signs of attacks by the leafminer, and by the end of August most leaves were still green and healthy. Seven to ten leaves per study period were measured. Leaves were sampled at about 5–7 m height, and immediately transported to the laboratory for experimental measurements.

Hydraulic measurements

Comparative measurements of leaf hydraulic resistance (Rleaf) were performed using three different techniques [i.e. the vacuum pressure method (VPM), the high pressure method (HPM) and the evaporative flux method (EFM)]. Both healthy leaves from control trees and leaves with 50% surface area covered by mines (see below) were measured.

The VPM was based on the vacuum chamber technique first described by Kolb, Sperry & Lamont (1996) and modified for small samples (Nardini et al. 2001). Leaves were cut off in the field under distilled water, and their petioles were connected while under water to plastic HPLC tubing (PEEK tubing of 0.7 mm i.d. and 1.5 mm o.d., Alltech Associates Inc., Deerfield, IL, USA) with two-way Teflon PTFE fittings (Omnifit Ltd, Cambridge, England). The leaf was placed in a vacuum chamber 8 L in volume, and the tubing was passed through a rubber seal to a beaker containing distilled water (filtered at 0.1 µm), resting on a digital balance (Sartorius AE220, Goettingen, Germany) with an accuracy of ±0.1 mg. A vacuum pump maintained pressures in the chamber at 80, 60, 40 and 20 kPa below atmospheric. At each pressure, a computer recorded the weight of the beaker on the balance at 30 s intervals, thus giving the corresponding flow rates (F). Flow recordings were made at a temperature of 20 ± 1 °C. The average F was computed from at least 6–10 readings after F became stable (i.e. the SD of the mean was <3%). Rleaf was computed as the slope of the vacuum pressure to flow relationship. Leaf lamina was removed, and petiole hydraulic resistance (Rpetiole) was measured following the same procedure. Lamina hydraulic resistance (Rlamina) was calculated as:


At the end of measurements, leaf surface area (Aleaf) was recorded using a leaf area meter (LI-3000A, Li-Cor Inc., Lincoln, NE, USA) and R values were normalized by Aleaf. In the case of attacked leaves, the percentage surface area covered by mines (Amined) was also measured as detailed below.

The EFM adopted by us was similar to that described by Sack et al. (2002). The leaf petiole was removed, leaving only about 1 cm to allow connection to a PEEK tubing through a compression fitting. The tubing was immersed in a beaker containing distilled water resting on a balance (see above). The leaf was maintained in a horizontal position by means of a wood frame. A light source with a photosynthetically active radiation (PAR) of about 800 µmol m−2 s−1 was suspended above a glass container filled with water in order to buffer leaf temperature at 20 ± 2 °C. After the light was turned on, water flow into the leaf typically increased for 20–30 min before stabilizing (coefficient of variation of F < 3% for 3 min). The flow rate was recorded, and within 10 s the leaf was covered with a plastic bag and removed from the tubing. The leaf water potential (Ψleaf) was measured with a pressure chamber (3005 Plant Water Status Console, Soilmoisture Inc., Santa Barbara, CA, USA). Rlamina was calculated as Ψleaf/F and scaled by Aleaf.

In the case of HPM, Rleaf was measured using a high-pressure flow meter (HPFM, Tyree et al. 1995). Leaves were connected to the instrument via the petiole using compression fittings. Degassed water filtered at 0.1 µm was forced into the petiole at a pressure of 0.25 MPa, and R was measured as the pressure-to-flow ratio at 8 s intervals until values became stable (i.e. the coefficient of variation of the last 20 readings was less than 3%), which usually took 10–20 min. Measurements were made on leaves while immersed in a water bath where the temperature was maintained at 20 °C.

During preliminary HPM measurements, leaves were first maintained under usual laboratory irradiance (PAR < 10 µmol m−2 s−1), and then illuminated at PAR = 1200 µmol m−2 s−1 using a fibre-optic light source (FL-460 Lighting Unit, Heinz Walz GmbH, Effeltrich, Germany). This procedure was aimed at checking eventual decrease of Rleaf upon illumination as observed in several species (Sack et al. 2002; Tyree et al. 2005). Because Rleaf was found to be insensitive to light conditions during experiments, all subsequent measurements were performed under laboratory irradiance.

Because the three techniques tested gave similar results (see below), we decided to use HPM for all subsequent measurements, because this technique has been proven to be suitable for determination of the partitioning of hydraulic resistances within the leaf (Sack et al. 2004). After Rleaf determination, the hydraulic resistance of the leaf vasculature (Rvascular) was estimated using the technique first proposed by Sack et al. (2004) consisting of serial measurements of Rleaf after increasing numbers of minor veins of the fourth order or higher had been cut open in order to by-pass the extra-vascular leaf compartment. Minor veins were cut at random locations throughout the lamina by making 1.5–2 mm incisions with a scalpel. In the case of horse chestnut leaves, up to 500 cuts per leaf were necessary to achieve stable and relatively invariant R values. At the end of experiments, the leaf lamina was cut close to the insertion of petiole, and petiole hydraulic resistance (Rpetiole) was measured. The hydraulic resistance of the leaf lamina (Rlamina) was calculated as described above (see Eqn 1).

The hydraulic resistance of leaf venation (Rvascular) was calculated as:


where Rleaf after cut is the hydraulic resistance of the leaf after vein cuttings (see above).

The hydraulic resistance of the extra-vascular water pathway was calculated as:


All R values were scaled by Aleaf. The extension of mines in the attacked leaves (Amines) was estimated using paper replicas, and the extent of damage suffered by leaves was expressed in percentage of total leaf surface area [Amined = (Amines/Aleaf) × 100]. R values measured on mined leaves were normalized by total Aleaf (i.e. by the surface area of the whole leaf lamina including the mined areas).

Temperature dependence of leaf hydraulic properties

The effect of temperature on leaf hydraulic conductance (Klamina) of control and mined leaves (Amined = 50%) and on vascular hydraulic conductance (Kvascular) was assessed with the HPM by varying the temperature of the water bath while maintaining constant the delivery pressure, and recording changes in flow rates (Sack et al. 2004). The aim of this procedure was to discriminate apoplastic water flow which can be expected to be mainly influenced by temperature because of changes in water viscosity (Nobel 1999) from that across the symplast which is considerably increased at higher temperature. To test this, the Q10 was measured as change in hydraulic conductance for control and mined leaves (see below). During measurements, only the leaf lamina (not the petiole) was immersed in the water bath. Leaves either intact or pre-treated with 500 cuts at the minor vein level (see above) were first measured for K at a temperature of 15 °C. Once flow into the leaf had stabilized, the temperature of the water bath was increased to 25 °C at a rate of 0.5 °C min−1. As expected, the hydraulic conductance increased during warming, and soon stabilized as the final temperature was reached. At the end of measurements, Rpetiole was recorded as described above, and Klamina and Kvascular at both temperatures were calculated according to Eqns 1 and 2. The values of Q10 for hydraulic conductance of five to seven leaves per treatment were calculated as the flow rate recorded at 25 °C divided by that recorded at 15 °C.

Microscopic observations

Leaf samples were fixed with a solution containing 50% distilled water and 50% formalin:ethanol:acetic acid (1:1:1, v/v). Samples were then dehydrated in an ethanol series, and embedded in paraffin. From five leaves, cross-sections 10 µm thick were made using a microtome (Slee Mainz, Mainz, Germany) and stained with the lipophilic dye Sudan III to localize eventual suberin in the walls of bundle sheath cells. Microscopic observations were made using a Leitz (Laborlux D, Leitz GmbH, Stuttgart, Germany) optical microscope (at total magnification of 1000×, Leica Camera AG, Solms, Germany) equipped with a digital camera (Leica DC-300F) connected to a PC. A fluorescence microscopic system (Aristoplan; Leitz) equipped with a UV filter (340–380 nm) and a digital camera (Optronics DEI-470, with cooling head, Optronics, Goleta, CA, USA) was also used for suberized cell wall fluorescent imaging both in fresh leaves or in leaves pre-stained with Sudan III. Digital images were acquired with commercial photographic analysis software.


During the early days of April 2008 (15 d after bud sprouting), leaf surface area was about 300 cm2 (i.e. approximately one-third of the final leaf size (Fig. 1). Leaf expansion was complete between late April and early May. By the end of May, the first small mines were observed on leaves in the lower part of the crown. By the end of June, about 10% of leaf surface area showed mines, and this percentage increased to about 30 and 50% by the end of July and August, respectively (Fig. 1).

Figure 1.

Seasonal changes in the surface area of Aesculus hippocastanum L. leaves. The areas of both whole leaf (black) and mined portions (grey) of leaves are reported. Mean values are reported ±SD (n = 7–10).

The three methods tested for measuring Rlamina produced similar results. In fact, Rlamina values as measured in August 2009 with VPM, HPM or EFM were not significantly different from one another, both in control leaves sampled from treated trees and in leaves with 50% surface area covered by mines (Fig. 2). In all cases, the Rlamina of fully mature and expanded leaves turned out to range between 20 and 25 e3 MPa s m2 kg−1.

Figure 2.

Leaf lamina hydraulic resistance as measured on Aesculus hippocastanum L. leaves with three different methods [i.e. vacuum pressure method (VPM), high pressure method (HPM) and evaporative flux method (EFM)]. Measurements were performed in July–August 2009 on control leaves sampled from trees treated with a systemic insecticide (see text for details) and on leaves with 50% of the surface area attacked by a leafminer. Means are reported ±SD (n = 5–7).

The hydraulic resistance of the leaf lamina showed an apparent seasonal trend (Fig. 3), with values steadily increasing from early April (7 e3 MPa s m2 kg−1) to early June (20 e3 MPa s m2 kg−1). Thereafter, it remained constant at about 20–25 e3 MPa s m2 kg−1. Interestingly, Rlamina did not change even after larvae of C. ohridella started to destroy the palisade tissue. In fact, Rlamina was not statistically different in June, July and August, despite the fact that leaves appeared to be mined by 10, 30 and 50%, respectively. The apparent lack of impact of leaf mining on Rlamina was confirmed by comparison of control with attacked foliage using the three tested methods (see above).

Figure 3.

Seasonal changes in hydraulic resistances of leaf lamina (Rlamina, black columns), leaf vasculature (Rvascular, light grey columns) and extra-vascular compartment (Rextra-vascular, dark grey columns) as measured in Aesculus hippocastanum L. Means are reported ±SD (n = 7–10). The percentage of mined leaf surface area is also reported, where applicable.

Not only the overall leaf hydraulic resistance changed during leaf ontogeny, but also the partitioning of Rlamina into vascular and extra-vascular components showed marked seasonal changes. Both Rvascular and Rextra-vascular increased from early April until the end of May (Fig. 3). In particular, Rvascular was about 4 e3 MPa s m2 kg−1 when Aleaf was only 30% of its final value (early April, Fig. 3), and reached values of about 10 e3 MPa s m2 kg−1 by May which remained unchanged until the end of August. In turn, Rextra-vascular increased from about 3 e3 MPa s m2 kg−1 as recorded in April to about 12 e3 MPa s m2 kg−1 as recorded by the end of May. It is worth noting that Rextra-vascular remained constant until the end of August despite severe damage to the palisade tissue caused by the parasite. In August, control leaves showed the same values of hydraulic resistance of leaf lamina, vasculature and extra-vascular pathway as those recorded in foliage sampled from attacked trees (Fig. 3).

Time-dependent, ontogenetic changes of the hydraulic efficiency of both vascular and extra-vascular water pathways resulted in permanent changes in the partitioning of hydraulic resistances within the leaf blade. In April, during the phase of leaf expansion, the largest share of hydraulic resistance was located in the vascular compartment which represented over 70% of Rlamina. By early May, the contribution of Rvascular to Rlamina decreased to about 60%, and by the end of the same month, Rvascular was a minor fraction of Rlamina. At this time, Rextra-vascular became dominant, representing about 60% of the overall leaf hydraulic resistance. From this time onwards, the partitioning of hydraulic resistances within the leaf remained unchanged until the end of August, when measurements ended.

The Q10 values for the hydraulic conductance of leaf venation (Fig. 4) were not statistically different from 1.3 (the expected value for temperature-dependent changes in water viscosity), suggesting that the cutting procedure effectively by-passed the extra-vascular compartment allowing accurate measurements of vascular hydraulic properties accurately. By contrast, the Q10 values for Klamina of both intact and healthy leaves (controls) turned out to be 1.64 ± 0.15, which was significantly larger than the value expected from effects of temperature on the viscosity of water. When the temperature dependence of Klamina was measured of leaves mined by 50%, the Q10 value turned out to be not statistically different from that recorded for healthy leaves.

Figure 4.

Temperature dependence (Q10) of hydraulic conductance of Aesculus hippocastanum L. leaves as measured on: (a) leaf vasculature (i.e. leaves measured after cutting minor veins, see text for details); (b) intact healthy leaves (control); (c) leaves with mines over 50% of surface area. Mean values are reported ±SD (n = 5–7). Different letters indicate significant differences (P < 0.05). The horizontal dotted line represents the Q10 value expected for changes in water viscosity.

Microscopic observations revealed that larvae of C. ohridella specifically destroyed the palisade tissue (Fig. 5a,b). In particular, Fig. 5b shows a Sudan III-stained leaf with the margin of the excavated tunnel through the palisade and the yellow-stained bundle sheath, suggesting that the walls of these cells were at least partially suberized. It can also be noted that the spongy parenchyma appeared to be intact and showed intact chloroplasts. Figure 5c,d shows healthy leaves observed either in autofluorescence (Fig. 5c) or in fluorescence after staining with Sudan III. The veins with associated bundle sheath were apparently intact with the bundle sheath cells fluoresceing in both cases. This suggests that the source of the fluorescence was the native lipophilic walls of the bundle sheath cells, not Sudan III.

Figure 5.

(a) Cross-section of a leaf of Aesculus hippocastanum L. showing a mined portion where palisade had been destroyed by larvae of Cameraria ohridella. Note that veins, bundle sheath and spongy parenchyma were unaffected. (b) A detail of a cross-section of a leaf stained with Sudan III. Note the yellow-stained walls of the bundle sheath cells. The upper margin of the section is the border of a mine. (c) Cross-section of a leaf observed in autofluorescence. Note the two fluoresceing bundle sheaths. (d) Cross-section of a leaf first stained with Sudan III and then observed under fluorescence microscopy. Note the fluoresceing walls of bundle sheath cells.


Over the study period, leaves of A. hippocastanum showed marked seasonal changes of both Rlamina and partitioning of leaf hydraulic resistances, which were apparently related to leaf ontogeny. After leaf expansion was complete, the hydraulic design of leaves remained unchanged throughout the growing season. Extensive disruption of palisade tissue by larvae of C. ohridella did not change Rlamina or Rvascular and Rextra-vascular. This finding suggests that water flow from the petiole to the evaporation sites might not directly involve the palisade tissue. The analysis of temperature dependence of Rlamina revealed that a transmembrane pathway was involved in water transport outside the leaf vasculature. Anatomical analysis suggests that this symplastic step may be located at the bundle sheath level where the apoplast is apparently interrupted by hydrophobic thickening of cell walls surrounding the veins.

In the present study, values of Rlamina as recorded using three different methods (VPM, HPM and EFM) were not significantly different from one another (Fig. 2), and were comparable to values measured by Raimondo et al. (2003) for the same species using the VPM. This finding is in accordance with previous and more extensive comparisons of different techniques for measuring Rlamina as reported by Sack et al. (2002). The repeatedly reported accordance between EFM and HPM is surprising, because HPM is expected to flood mesophyll air spaces, thus probably reducing the extra-vascular hydraulic resistances and possibly underestimating Rlamina. Because this was not the case, we can only conclude, in accordance with Sack et al. (2002), that the mesophyll pathways that are by-passed in leaves during HPM measurements contribute to a relatively negligible extent to the overall Rlamina. In other words, the substantial agreement among the three different techniques provides strong, even if circumstantial, evidence for identifying the main hydraulic bottleneck in the leaf lamina as located up-stream the mesophyll, either in the vascular system or at the bundle sheath level.

Our data show a clear seasonal trend from low to high Rlamina values as observed from early April until the end of May (Fig. 3). Changes were concurrent with the time-course of leaf expansion and maturation (Fig. 1). During early phases of leaf expansion, Rextra-vascular was quite low compared to values recorded after leaf maturation was complete. On the contrary, Rvascular underwent much smaller increase from April to June, so that changes in the partitioning of hydraulic resistances within the leaf blade were largely dictated by modifications of Rextra-vascular. Aasamaa et al. (2005) have reported ontogenetic changes of Rleaf in Populus tremula L., and suggested that changes in Rlamina were related to changes in membrane-associated traits like the number of cells per unit volume, which influences the frequency of plasmalemmas and/or tonoplasts to be crossed by water. In a recent study, Hachez et al. (2008) have reported that aquaporins were more abundant in the elongation zone of maize leaves, suggesting a role for these membrane proteins in facilitating cell expansion (Heinen, Ye & Chaumont 2009). On the basis of the above, we suggest that low Rextra-vascular during leaf expansion in A. hippocastanum might be caused by high water permeability of membranes of growing cells, possibly because of high expression of aquaporins during leaf ontogeny.

The hydraulic resistance of leaves mined by up to 50% was not different from that measured in intact leaves (Figs 2 & 3), despite the fact that a large fraction of the palisade tissue had been completely destroyed by the larvae of the parasite (Fig. 1). Similar data have been reported by Raimondo et al. (2003). In accordance with the relative invariance of Rlamina, Raimondo et al. had observed that gas exchange rates in the still healthy portions of mined leaves were similar to those recorded in control leaves, suggesting that the evaporation sites continued to be supplied with water even in severely mined leaves.

Our data confirm the above findings and further show that even the partitioning of Rlamina into vascular and extra-vascular components was not affected by substantial disruption of the palisade. The lack of impact of C. ohridella mines on Rvascular is not surprising, in accordance to the fact that leaf veins were not affected by feeding larvae (Fig. 5c,d; see also Raimondo et al. 2003). On the contrary, the lack of impact of palisade disruption on Rnon-vascular is surprising, because treatments inducing non-selective killing of mesophyll cells, like freezing or boiling, have been reported to induce large decrease in leaf hydraulic resistance (Tyree, Nardini & Salleo 2001; Cochard et al. 2004; Gascò, Nardini & Salleo 2004). Because a substantial fraction of Rlamina is located outside the xylem, the selective removal of palisade tissue would be expected to induce a drop in Rlamina, unless the palisade is largely by-passed by water as suggested by Zwieniecki et al. (2007) for intact leaves. This figure is not easy to visualize in the consideration that a large fraction of the mesophyll volume is occupied by the palisade. On the other hand, classical studies have shown that spongy mesophyll cells are in a much closer hydraulic contact to one another than palisade cells are (Wylie 1943, 1946), and recently Ye et al. (2008) have shown that significant hydraulic connection exists between leaf xylem and epidermis through a cell-to-cell pathway.

In the case of horse chestnut leaves, our data suggest that water flow through palisade tissue might not be a determinant of Rextra-vascular with the consequent question of the possible location of the extra-vascular resistance which represents up to 60% of Rlamina in this species (Fig. 3). In principle, water may move outside the leaf xylem through three potential pathways: an apoplastic (i.e. in the cell walls), transcellular (i.e. through cell membranes) or symplastic route (i.e. via plasmodesmata). Our measurements of temperature dependence of Klamina rule out the possibility that water moves outside the xylem following a purely apoplastic pathway, because the Q10 values for intact leaves were greater than expected from changes in water viscosity (Fig. 4). Values of Q10 of about 1.7 as reported in the present study are consistent with a water pathway including one or more transcellular components, as already suggested by Sack et al. (2004) on the basis of similar analysis in leaves of two angiosperms. Anatomical observations revealed the presence of lipophylic components in cell walls at the bundle sheath. Similar apoplastic barriers in bundle sheath cells have been reported previously in different plant species (Evert, Russin & Bosabalidis 1996; Lersten 1997; Dalla Vecchia et al. 1999; Hachez et al. 2008). On the basis of similar considerations, Fricke (2002) has suggested that the main hydraulic constriction in the leaf lies at the bundle sheath-to-mesophyll layer, a conclusion that would be in accordance with considerations previously reported on the basis of the agreement of Rlamina values as recorded with different techniques (see above). Such a function of the bundle sheath is supported by studies showing high abundance of aquaporins in these cells (Frangne et al. 2001; Martre et al. 2002; Hachez et al. 2008).

Our data are consistent with the above findings and further reinforce the view that the main hydraulic resistance involved in water movement outside the leaf xylem is located at the bundle sheath where water is forced to cross cell membranes (possibly through aquaporins) because of the presence of an apoplastic barrier similar in function to the Casparian strips of the root endodermis. Our data further suggest that beyond the bundle sheath, water may not flow to a significant extent through the palisade tissue. In fact, similar Q10 values for both healthy and mined leaves (Fig. 4) suggest that the number and/or type of membranes crossed by water outside the xylem was the same both in presence and in absence of the palisade. It is also conceivable that water may be transported directly to the epidermis through bundle sheath extensions or, alternatively, through spongy mesophylls, eventually via plasmodesmata. The exact contributions of different possible pathways to the transpiration stream can be expected to depend on the degree of hydraulic compartmentalization within the leaf, but our data do not allow a resolution of this problem.

Overall, our data provide some support to the hypothesis originally advanced by Tyree et al. (1981)) and recently confirmed by Zwieniecki et al. (2007) that extra-vascular tissues of leaves are composed of spatially and hydraulically distinct pools. In particular, horse chestnut leaves apparently include at least two main compartments: (1) bundle sheath plus epidermis and spongy mesophyll; and (2) palisade tissue. According to this hydraulic design, the palisade cells (i.e. the cells mostly involved in photosynthesis) would be by-passed by the transpiration stream. At first sight, uncoupling photosynthetic cells from the transpiration stream might not seem to represent a functional advantage. However, partial hydraulic isolation of such cells from rapidly changing water flows would ‘protect’ them from large, short-term changes in their water status. Because photosynthetic activity is influenced by cell water status, buffering changes in cell water content might facilitate plants to maintain steady photosynthetic rates despite short-term changes in water flow through the leaf.

The feeding behaviour of larvae of C. ohridella has offered an unexpected tool to obtain an experimental demonstration that removing the palisade tissue does not change leaf hydraulic resistance. In the past, the use of parasites feeding on specific tissues has greatly expanded our understanding of the physiology of plants. As an example, aphids feeding on sap from single sieve elements have allowed us to obtain information about phloem sap composition (e.g. Weatherley, Peel & Hill 1959; Will & van Bel 2006). Nymphs of xylem-feeding spittlebugs have offered the possibility to measure changes in xylem sap chemistry (Andersen, Broadbeck & Mizell 1992; Watson, Pritchard & Malone 2001). More recently, insects cutting major and/or minor veins have been used to investigate the impact of vein injury on leaf gas exchange and photosynthesis (Aldea et al. 2005; Delaney & Higley 2006). Insects feeding on mesophyll living cells might help resolve the hydraulic design of leaves, provided that their feeding behaviour is selective enough to impact specific tissue types. We feel that future studies aimed at elucidating the water pathway(s) inside the leaf will greatly benefit from such an experimental approach.


We are grateful to Dr Giselher Grabenweger (Institute of Plant Protection, University of Agricultural Sciences, Vienna, Austria) for helpful information on leafminers.