SEARCH

SEARCH BY CITATION

Keywords:

  • venation pattern;
  • bundle sheath extension;
  • transpiration stream;
  • hydraulic compartmentalization

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

We examined the leaf hydraulic design in 10 species based on their rehydration kinetics. In all cases, a biphasic response described the temporal pattern of water uptake, with time constants of ∼30 to 800 s and ∼800 to 8000 s. The time constants of the fast phase were significantly shorter in the six angiosperms (30 to 110 s) compared with the two single-veined conifer species (>400 s) examined, while the two multi-veined gymnosperm species, Gnetum gnemon and Ginkgo biloba, had time constants for the fast phase of ∼150 s. Among angiosperm species, the fast phase constituted 50–90% of the total water absorbed, whereas in gymnosperms 70–90% of the water uptake could be assigned to the slow phase. In the four gymnosperms, the relative water uptake corresponding to the fast phase matched to a good degree the relative volume of the venation and bundle sheath extension; whereas in the angiosperm species, the relatively larger water influx during the fast phase was similar in relative volume to the combined venation, bundle sheath extension, epidermis and (in four species) the spongy mesophyll. This suggests a general trend from a design in which the epidermis is weakly connected to the veins (all four gymnosperms), to a design with good hydraulic connection between epidermis and veins that largely bypasses the mesophyll (four of six angiosperms), to a design in which almost the entire leaf appears to function as a single pool.


INTRODUCTION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

Leaves of terrestrial plants are topologically complex structures charged with the task of creating an internal environment that permits a positive return on the plant's investments in photosynthetic tissues. Optimization of photosynthetic activity requires that leaves be thin, to facilitate light penetration, and porous, to allow CO2 diffusion within the leaf, making leaf structure at odds with the simultaneous demand for water conservation. The inevitability of evaporation during CO2 uptake places tremendous pressure on leaf hydraulic design as seen in the good correlation between leaf hydraulic conductance and assimilation rates (Brodribb et al. 2005). However, in addition to being able to support steady-state levels of transpiration, leaves must also cope with transient changes in water loss rates. The driving force for evaporative loss can change rapidly because of radiative effects on leaf temperature (Singsaas et al. 1999), as well as variation in the thickness of the boundary layer as a result of turbulent mixing of the surrounding air (Schuepp 1993). Although the valve-like action of guard cells provides leaves with the ability to control rates of water loss to the environment, stomatal closure involves the net movement of water out of cells and thus proceeds at a finite rate (Buckley, Mott & Farquhar 2003). Differences in the timescales that characterize short-term variation in water loss rates and the corresponding response in stomatal aperture mean that leaves will frequently experience imbalances in supply and demand. In this paper, we explore how the internal hydraulic design of leaves contributes to their ability to maintain physiological function in the face of rapid fluctuations in transpiration rate.

Transient imbalances between hydraulic supply and demand can negatively impact leaf function in two ways. Loss of xylem conductivity due to cavitation can force stomata to close leading to a decrease in photosynthetic carbon gain, while desiccation can impair the metabolic activities of living cells. Although a direct effect of leaf water content (or leaf water potential) on photosynthetic capacity remains a matter of debate, declines in water content below the turgor loss point have the potential to impair physiological processes (Matthews & Boyer 1984; Berkowitz & Kroll 1988; Tang et al. 2002; Tezara et al. 2002). The challenge for leaf hydraulic design is that ameliorating one risk comes at the price of accentuating the other. For example, the hydraulic capacitance of leaf cells means that the mesophyll can act to buffer the propagation of transient decreases in apoplastic pressure, thus protecting the xylem – but only by exposing these living cells to significant changes in water content. Hydraulically separating the photosynthetic tissues from the transpiration stream will minimize the degree to which short-term variation in transpiration rate affects the mesophyll, but will speed up the rate at which fluctuations in leaf water potential are propagated to the xylem. Despite increasing interest in the movement of water through leaves (Tyree & Cheung 1977; Canny 1993, 1995; Fricke 2000; Sacket al. 2003; Nardini, Gortan & Salleo 2005; Sack, Tyree & Holbrook 2005), our understanding of leaf micro-hydrology remains limited. Here, we use rehydration kinetics to explore the hydraulic design of leaves, focusing on the degree to which leaves function hydraulically as a single pool versus the possibility that they exhibit substantial hydraulic compartmentalization.

Leaf rehydration has long been used to study the movement of water in leaves (Weatherley 1963; Milburn 1966; Boyer 1974, 1977, 1985). Analysis of rehydration kinetics suggests that leaves should not be treated as uniform capacitors because the rehydration of partially dried leaves could not be described using a single exponential (Cruiziat et al. 1980; Tyree et al. 1981). Although the idea that a number of parallel pathways with different resistances and/or spatially distinct water pools must exist to account for the observed kinetics was proposed (Tyree et al. 1981), further discussion of the physiological implications of these patterns has been limited. Leaves are composed of many cell types (e.g. epidermal cells, spongy and palisade mesophyll, vascular parenchyma) which may differ in their individual hydraulic properties as well as the degree to which they are hydraulically connected to neighbouring cells. Thus, it is not surprising that leaves do not behave hydraulically as a single, well-mixed pool. What is unexpected, however, is that the rehydration of these diverse cell types, located within spatially defined regions of the leaf, produces kinetics that can be described by a small number of exponential processes.

The diversity of extant leaf structures and morphologies provides a powerful tool for the comparative analysis of rehydration kinetics. Beginning with the simple designs offered by coniferous leaves characterized by a single vein through to the reticulate, hierarchical venation of most angiosperms, we can use variation in leaf anatomy and vascular supply to help pinpoint structural features associated with the principal characteristics of leaf hydraulic design. We can also examine the role of hydraulic inventions and modifications in leaf anatomy such as the presence of vessels, vein reticulation or bundle sheath extensions. In this paper, we present data on the rehydration of leaves from 10 species (six angiosperms and four gymnosperms), examine the rehydration of leaves dried to various water potentials to test whether the observed patterns reflect a change in the resistance of a single pathway for rehydration versus the existence of distinct pools, compare the spatial patterns of water loss from leaves subject to rapid over-pressurization using a pressure chamber to test the idea that leaves exhibit substantial hydraulic compartmentalization, and use the relative sizes of the two phases in relation to the relative volumes of the various tissue within each leaf to inform our discussion of leaf hydraulic design.

METHODS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

Plant material

A diverse group of 10 gymnosperm and angiosperm species representing a range of venation types was studied (Table 1). In almost all cases, leaves were collected from individuals growing in either the greenhouse or experimental garden on the campus of Harvard University, Cambridge, MA, USA. Leaves of Eucalyptus globulus were collected from trees growing on the California Institute of Technology campus (Pasadena, CA, USA), shipped overnight in double plastic bags filled with moist paper towels, and measured the next day. All plants sampled appeared healthy, without any symptoms of viral or bacterial infection, insect activities or water stress.

Table 1.  List of species examined with the major morphological and anatomical characteristics of their leaves
SpeciesVenation typeXylem typeBundle sheath extensionMesophyllStomata
Gnetum gnemonMulti-veined reticulateVesselsPresentPalisade and spongyHypostomatous
Gingko bilobaMulti-veined paralleledTracheidsPresentUnstructuredHypostomatous
Metasequoia glyptostroboidesSingle veinedTracheidsAbsentUnstructuredHypostomatous
Larix laricinaSingle veinedTracheidsAbsentUnstructuredHypostomatous
Quercus rubraMulti-veined reticulateVesselsPresentPalisade and spongyHypostomatous
Acer rubrumMulti-veined reticulateVesselsPresentPalisade and spongyHypostomatous
Liriodendron tulipiferaMulti-veined reticulateVesselsPresentPalisade and spongyHypostomatous
Populus nigraMulti-veined reticulateVesselsPresentPalisade and spongyHypostomatous
Phyllostachys aureosulcataMulti-veined paralleledVesselsPresentPalisadeAmphistomatous
Eucalyptus globulusMulti-veined reticulateVesselsPresentPalisadeAmphistomatous

Leaf rehydration

Rehydration kinetics was measured using a ‘reverse Polish guillotine’ (Fig. 1) that allows a leaf to be severed underwater and attached to a water-filled tube in a single motion. The name of this device derives from the fact that it is the leaf, rather than the cutting blade that moves, thus avoiding perturbation of the hydraulic connection to the analytic balance used to measure water flow into the excised leaf. The procedure involved cutting a 1- to 2-m-long branch during the morning hours between 0800 and 1000 h (except for E. globulus which was collected at midday). Branches were carried to the lab in a plastic bag with moist towel to limit transpiration. Prior to beginning the rehydration experiment, the water potential of the leaf was estimated by measuring the water potential of two leaves, one located above and one below the selected leaf. The target water potential was between −0.6 and −1.0 MPa so as to avoid major losses in conductivity because of cavitation, but to provide sufficient water deficit so that the patterns of water uptake would be discernable. However, leaves desiccated to a greater range of initial water potentials were measured in two species.

image

Figure 1. Schematic and picture of the ‘reverse Polish guillotine’ used to connect a partially dehydrated leaf to a water-filled tube such that the entire period of water uptake by the leaf can be monitored. The ‘reverse’ action of the guillotine derives from the fact that the blade is fixed in the lower position, while the specimen to be cut slides downwards across the blade, thus allowing the water-filled tube (located immediately below the blade) to remain fixed in place. The latter is critical for obtaining accurate measurements of the initial rates of water flow. The seal between the severed petiole and the water filled tube was made using an o-ring. A split block allowed the leaf to be inserted into the guillotine while still attached to the branch. The entire system was submerged under water to avoid air from being drawn into the severed xylem of the petiole, as well as to prevent water loss from the leaf surface. Because the water level on the balance was approximately 5 cm lower than the water level in the basin containing the leaf, any failure of the seal resulted in a backflow of water to the balance. In the few cases in which this occurred (primarily with Phyllostachys aureosulcata which has very thin petioles), the measurements were discarded.

Download figure to PowerPoint

The petiole of the target leaf was wrapped in parafilm and secured in the guillotine holder using Blu-tack (Bostik, Pty. Ltd., Notting Hill, VIC, Australia). A layer of high vacuum silicon grease (Dow Corning, Midland, MI, USA) was added to the petiole to help form a seal with the o-ring. The entire system (target leaf + guillotine) was then submerged in water to prevent evaporation from leaf surfaces during the measurement period. All species were tested for cuticular uptake by submerging dehydrated leaves in water with petiole in the air. In no case was leaf water potential increased by this treatment, precluding the possibility of water uptake via the leaf surface. Any cut-ends of the branch were covered with high vacuum silicon grease to prevent water entry. After the entire setup was prepared, the guillotine was deployed such that the leaf was severed from the branch and simultaneously connected to a tube linked to a water reservoir seated on an analytic balance (Sartorius ±0.01 mg, Goettingen, Germany). In the case of Metasequoia glyptostroboides and Larix laricina, we measured branchlets instead of individual leaves and thus the measurements include the rehydration of a small piece of stem. Water uptake was recorded at 1 s intervals for ∼9000 s, followed by ∼500 s with the leaf (or branchlet) disconnected so as to account for any evaporative losses that might occur from either the tube or the water reservoir. We measured 5 to 10 leaves (or branchlets) for each species. The temperature of the water bath was maintained at 20 °C and leaves were illuminated with 30 µmol m−2 s−1 throughout the measurement period, a fluence rate sufficient to saturate any light-induced changes in leaf hydraulic conductance using this measurement technique (Rockwell et al., unpublished data). After accounting for the evaporation from the measurement system itself, a small linear uptake was still observed in several species. This residual uptake, which we believe was due to capillary infiltration into leaf intercellular spaces, was also accounted for before further analysis.

The weight–time function of water flow from the balance provides a continuous record of leaf rehydration. Because there is a limit for water absorption by the dehydrated leaf (i.e. to zero water potential), one can use an exponential growth to maximum function to describe these data. We used both a single,

  • image

and a double exponential model:

  • image

to describe water uptake by dehydrated leaves. In both models, parameters a and c describe the volume of the respective compartments, and b, d their corresponding time constant.

Anatomy

The relative volumes of different tissue types were determined from analysis of paraffin embedded leaf cross sections. Fresh leaves were vacuumed infiltrated in FAA, dehydrated and embedded in paraffin (Ruzin 1999), sectioned at 10 µm using a microtome, and doubled stained using alcian blue (Sigma-Aldrich, St. Louis, MO, USA) and safranin-O (Sigma-Aldrich). Images from five randomly selected sections from the central lamina of five leaves were made at 100× or 200× magnifications for a total of 25 pictures per species. Image analysis software (ImageJ, a public domain, Java-based image processing program developed at the National Institutes of Health, Bethesda, MD, USA) was used to measure the area within each cross section of the following tissue types: vein (xylem, phloem and parenchyma cells surrounding the vascular bundle), bundle sheath extension (if present, defined as a parenchymatous extension connecting the vein to the epidermis), mesophyll (separating spongy and palisade when morphologically distinct), upper and lower epidermis (measured together), and other types of cells if present. Gnetum gnemon leaves contain large numbers of non-lignified fibres, which were placed into their own category (Tomlinson & Fisher 2005). Using these areas and assuming that the thickness of the sample was uniform in the field of view, we calculated the % contribution of each tissue type to the cellular volume of the leaf.

Temporal and spatial patterns of water extraction from over-pressurized leaves

Leaves of Populus nigra were cut off under water and allowed to rehydrate for several hours while supplied with a 0.01% aqueous solution of sulforhodamine (Sigma). Leaves were placed in a pressure chamber and pressurized to 2.0 MPa. Pressurization took place over 5–10 s to avoid excessive (>35 °C) increases in leaf temperature. Exudation from the petiole was collected into pre-weighed Eppendorf tubes (0.5 mL) at intervals of 30 s during the initial phase, and subsequently at 60 s intervals. Each Eppendorf tube was then weighed and the expressed mass per unit time interval calculated.

Leaves that were frozen in liquid nitrogen (LN) were prepared in the same way as described above, before they were placed in the pressure chamber containing a 0.25 L paper cup filled with LN at the bottom of the chamber (Fig. 2). Two thin wires attached to the rim of the cup protruded through the rubber fitting at the top of the chamber. The petiole was enclosed in a plastic tube to avoid crushing before being sealed in the chamber lid. A thermocouple placed on the leaf was used to avoid either overheating or overcooling the leaf, as well as to later confirm that the leaf was successfully submerged in LN. Once the leaf and all the wires were fixed in the chamber top, the chamber was sealed and a pressure of 2.0 MPa applied for 30 s, after which the LN-filled cup was pulled up until the thermocouple registered that the leaf had frozen (Fig. 2b). The cup containing the frozen leaf was then carefully removed from the pressure chamber and the leaf transferred to a LN storage dewer. During the pressurization, special attention was needed because pressurization of the chamber causes the temperature to rise, thus increasing evaporation of LN with consequent effects on chamber pressure (as well as temperature). With practice, we were able to maintain leaf temperature within the range of 17–28 °C, while still achieving pressurization of 0.1 to 2.0 MPa within ∼10 s. Fully hydrated leaves that were not pressurized were frozen as a control.

image

Figure 2. Diagram of apparatus used to freeze-pressurize leaves. After pressurization for the specified time interval, a plastic, non-compressible cup filled with liquid nitrogen (LN) initially placed at the chamber bottom (a) is pulled up to the top of the chamber using fine wires inserted through the pressure chamber gasket surrounding the leaf petiole (b). A thermocouple attached to the leaf allowed the pressurization rate to be regulated so as to avoid either overheating or cooling, as well as to confirm that the leaf had successfully been immersed in LN before the pressure was released.

Download figure to PowerPoint

The frozen leaves were then prepared for observation in a fluorescent microscope using previously described techniques (Cochard et al. 2004; Brodribb & Holbrook 2005). Pictures of leaf cross sections were taken using a long-distance (7 mm) objective (magnification 50×; Zeiss). The presence of sulforhodamine in the bundle sheath allowed us to analyse the shapes of vein parenchyma cells, while chlorophyll autofluorescence from photosynthetic mesophyll cells was sufficient to allow analysis of their shape. Both the cross-sectional surface area of cells and their perimeter were measured using Image J software. A unitless circularity index, [(cell circumference)2/(cell cross-sectional area)]/4π, was used to quantify deformation. A total of 42 bundle sheath cells and 28 palisade mesophyll cells were measured from three pressurized and three control leaves.

RESULTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

Rehydration kinetics

For the 10 species examined, a double exponential function better fit the measured rehydration kinetics than did the single exponential function (Fig. 3), although the discrepancy between the two models varied between species. The adjusted R2 of the single exponential model ranged from 0.75 to 0.96, while for the double exponential model R2 values ranged from 0.92 to 0.99. The only species for which a significant improvement of fit between the single and double exponential model was not observed was E. globulus, with the R2 for single and double exponential models being respectively 0.96 and 0.99.

image

Figure 3. Representative rehydration kinetic curve for an Acer rubrum leaf dehydrated to ∼ −1 MPa. The experimental data (solid dashed line being the raw data from the balance) have been fit with both a single and double exponential rise to maximum. Inset shows water uptake partitioned between the fast and slow phase, calculated from the double exponential model.

Download figure to PowerPoint

The double exponential model describes leaf rehydration as consisting of two phases: the fast phase has time constants ranging from 30 to 800 s, while the slow phase has time constants ranging from 800 to 8000 s. Time constants for the fast phase were shorter (i.e. rehydration occurred more quickly) in all of the angiosperm species examined compared to all of the gymnosperms (Fig. 4). The time constants for this phase were significantly longer in the two conifers, than in all other species. The two multi-veined gymnosperms (Gnetom gnemon and Ginkgo biloba) had time constants of the fast phase that were longer than in all of the angiosperms (>150 versus <110 s), but the difference was not significant. In contrast, there was no obvious difference in the time constants of the slow phase between gymnosperms and angiosperms. Most species had time constants between 3000 to 5500 s. Eucalyptus globulus had a significantly shorter slow phase time constant, consistent with the fact that this species was equally well fit by a single exponential model. The slow phase time constant of Larix laricina was significantly longer than for all other species, although it included rehydration/resistance of small portion of the stem because we could not measure single leaves.

image

Figure 4. Calculated time constants from two exponential rise to maximum model of rehydration kinetics for (a) fast phase and (b) slow phase. The grey area denotes gymnosperm species. Error bars denote SE of the mean. Small letters above the bars denote significance differences (P < 0.05) resulting from one-way analysis of variance. G. gnemon, Gnetum gnemon; G. biloba, Gingko biloba; M. glyptostroboides, Metasequoia glyptostroboides; L. laricina, Larix laricina; Q. rubra, Quercus rubra; A. rubrum, Acer rubrum; L. tulipifera, Liriodendron tulipifera; P. nigra, Populus nigra; P. aureosulcata, Phyllostachys aureosulcata; E. globulus, Eucalyptus globulus.

Download figure to PowerPoint

The relative volume of water absorbed during the fast and slow phases showed consistent differences between angiosperms and gymnosperms. The dominant pool of water in the gymnosperms was associated with the slow phase, with 70 to 90% of the total water flowing into the leaf being attributed to this slower pool. In contrast, in the angiosperm species the relative volume associated with the slow phase was much smaller, ranging from approximately 10 to 50% of the total amount of water absorbed during rehydration (Fig. 5).

image

Figure 5. Relative water uptake calculated from the double exponential rise to maximum model of rehydration kinetics for fast phase and slow phase. Error bars denote SE of the mean.

Download figure to PowerPoint

Underlying basis for biphasic kinetics

Having established that biphasic rehydration kinetics occurs broadly across vascular plants, an important question is to interpret these patterns in terms of the underlying physical and biological phenomena. One potential explanation is that the biphasic response reflects an increase in the resistance of a single pathway for rehydration during the period of water uptake because of generation of intercellular pressure differentials (Fricke 2000). If this is true, then the relative size of the fast and slow phases should depend upon the water deficit. To test this idea, we measured leaf rehydration for two species (Acer rubrum and Liriodendron tulipifera) across a broad range of initial leaf water potentials (0.2 to 1.4 MPa). Both species exhibited biphasic rehydration kinetics across the full range of initial leaf water potential (see also Tyree et al. 1981), and the relative sizes of the two phases were independent of the degree of desiccation experienced by the leaf prior to rehydration (Fig. 6).

image

Figure 6. Relative volume of the fast phase (rehydrating in less than 100 s) for two species (Acer rubrum and Liriodendron tulipifera) dehydrated to different initial water potentials.

Download figure to PowerPoint

This led us to explore an alternative scenario in which the biphasic kinetics reflects the rehydration of two spatially distinct regions within the leaf, by examining the spatial distribution of water loss from within leaves of Populus nigra. The analysis started with the measurement of water efflux from hydrated leaves pressurized to 2.0 MPa using a pressure chamber. There was significant efflux over the first 40 s (accounting for ∼40% of water forced from the leaf), followed by slower efflux extending for approximately 10 min (Fig. 7a). The temporal dynamics of water efflux was consistent with prior observation of rehydration kinetics of poplar leaves (Fig. 4). The time constant of the fast compartment was calculated to be ∼30 s, a value slightly lower then calculated from the rehydration curves. The slow phase time constant, however, was in the range of ∼300 s – much lower than in the case of the rehydration kinetics. This difference may result from the fact that over-pressurization forces very dry air through the petiole, accelerating evaporation from the collection vials. Underestimation of water exudation due to evaporation will have a disproportionate effect on the calculation of the time constant of the slow phase because of the increasingly smaller exudation volumes as the leaf approaches equilibrium with the applied pressure.

image

Figure 7. (a) Cumulative water lost from leaves pressurized to 2.0 MPa. Values are scaled relative to the total leaf water content to take into account variation in leaf size. (b) View of frozen vein cross section from pressurized leaf (arrows indicate typical cell deformation; image is 150 µm on a side). (c) View of vein cross section of fully hydrated leaf. (d,e,f) Shape and calculated circularity index for palisade mesophyll cells, while (g,h,j) show the same vein parenchyma cells. *** indicates a significant difference (P < 0.001) between hydrated (control) and pressurized leaves.

Download figure to PowerPoint

To determine the source of water of the fast phase, we froze leaves in LN after 30 s of exposure to 2.0 MPa; as a control we froze fully hydrated leaves. All leaves had been previously infiltrated with sulforhodamine to allow us to quantify changes in cell shape in sections of frozen tissues using fluorescence microscopy (Fig. 7b,c). We used a simple ‘circularity index’ to describe changes in cell shape in terms of their deviation from a circular cross section: [(cell circumference)2/(cell cross-sectional area)]/4π. A smaller value of this parameter indicates greater deviation from a circular shape. There was no difference in the shape of palisade mesophyll cells measured from control and pressure-dehydrated leaves (Fig. 7d–f). However, there was a significant decrease in the circularity index of parenchyma cells surrounding minor leaf veins (Fig. 7g–j), suggesting preferential loss of water from these cells. Cross-sectional area of the bundle sheath cells was smaller, but due to variation in cell size, the difference was not significant.

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

Interpretation of biphasic kinetics

Nine of the 10 species studied exhibited biphasic leaf uptake kinetics, suggesting that this is a widespread phenomenon among vascular plants (Fig. 4). More detailed studies on a subset of these species (Figs 6 & 7) support the hypothesis that this reflects underlying hydraulic compartmentalization, consistent with earlier studies using stable isotopes demonstrating that leaves are not well-mixed pools (Yakir, DeNiro & Rundel 1989; Yakir, DeNiro & Gat 1990; Wang & Yakir 1995). Our findings agree with pioneering studies by Tyree and colleagues in terms of the existence of complex (bi- or multiphasic) rehydration kinetics (Cruiziat et al. 1980; Tyree et al. 1981), but differ in the values of the reported time constants. Comparisons between our data and the earlier works are complicated by the fact that in these earlier studies, leaves were measured in a pressure bomb prior to their uptake rates being recorded. Pressurization results in marked decreases in the subsequent ability of leaves to absorb water (Brodribb & Holbrook 2003) and likely contributed to the differences in uptake kinetics reported in earlier studies and those presented here.

The uptake of water by partially dehydrated leaves occurs along a pathway that includes both the vasculature of the petiole and leaf veins, as well as the living tissues separating the xylem from each rehydrating cell. Because the actual amount of water uptake attributed to the fast phase is too large to be accounted for by refilling of embolized conduits, differences in the uptake kinetics between gymnosperms and angiosperms (Fig. 4) are likely to reflect structural aspects of the xylem that influence its hydraulic conductivity (Pittermann et al. 2005; Hacke et al. 2006). Careful comparative work demonstrates that estimates of leaf hydraulic conductance (Kleaf, mmol m−2 s−1 MPa−1) calculated from the fast uptake phase agree well with estimates of Kleaf determined using other approaches (Rockwell, unpublished data). This suggests that the fast phase corresponds with the pathway used by the transpiration stream. Estimates of Kleaf calculated from the data in this paper range from 0.17 to 10.6 [mmol(H2O) MPa−1 m−2 s−1], in good agreement with values reported for temperate angiosperm and gymnosperm leaves (Brodribb et al. 2005; Sack & Holbrook 2006).

The tentative association of the fast phase or compartment with the pathway involved in the transpiration stream highlights the question of the identity and underlying mechanisms responsible for the existence of the slow rehydration phase. The fact that a substantial component of the leaf hydraulic system is significantly decoupled from a faster phase suggests that an entirely new way of thinking about the internal hydraulic design of leaves is needed (Boyer 1985). In particular, the marked difference in time constants between the two phases raises the possibility that leaves may function internally as low pass filters, effectively buffering (protecting) critical tissues from transient changes imposed by variation in water loss rates.

Mechanisms of hydraulic compartmentalization

The idea that cells in close proximity might rehydrate at substantially different rates raises the question of ‘what aspects of leaf structure and/or physiology might allow such compartmentalization?’ The extent to which cells are hydraulically coupled will depend in part upon their physical contact. Thus, some degree of hydraulic compartmentalization could be achieved solely on the basis of the internal architecture of leaves. The fact that Larix laricina had the longest time constant for the slow compartment may, in part, reflect the relatively small degree of physical contact between the stele and the mesophyll (Fig. 8). It is unlikely, however, that geometry can account for the full range of hydraulic patterns exhibited here. Thus, differences in the hydraulic conductance governing the movement of water between cells and different tissue types are also likely to play an important role.

image

Figure 8. Typical cross section of each analysed species used to calculate relative distribution of volumes between different leaf tissue types. Sections are 10 µm thick and stained with alcian blue and safranin-O. Scale bars are 200 µm.

Download figure to PowerPoint

A number of mechanisms that might contribute to hydraulic compartmentalization in leaves are known. For example, variation in the number and/or activity of plasmodesmata (Sowinski, Rudzinska-Langwald & Kobus 2003) or plasma membrane aquaporins (Chrispeels & Maurel 1994; Chrispeels et al. 2001), may underlie differences in the hydraulic linkages between tissues. The fact that the symplasm plays an important role in the post-xylary movement of water through leaves (Canny 1993, 1995; Sack, Streeter & Holbrook 2004) is consistent with the potential contribution of membrane level processes to leaf hydraulic design. Further studies characterizing the hydraulic linkages within leaves, as well as the contribution of both aquaporins and plasmodesmata, are needed to advance our understanding of the internal hydraulic properties of leaves.

Relation between uptake kinetics and leaf anatomy

There were large differences between species in the relative contributions of each tissue type to overall leaf volume (Figs 8 & 9). We explored the idea that leaves consist of spatially distinct hydraulic pools by comparing the relative volumes of water absorbed during the two phases with the relative volumes of different tissue types. This analysis assumes that the relative water deficit is the same for all tissues, such that the amount of water absorbed during rehydration scales linearly with their relative volumes. In the case of the gymnosperm species examined, the relative size of the fast phase compares well with the total volume of the veins and bundle sheath extension. In all but two of the angiosperm species examined, the relative size of the fast phase best matched the relative volume of veins, bundle sheath extensions, both lower and upper epidermides, and spongy mesophyll. For the two angiosperm species that lack a spongy mesophyll layer (Phyllostachys aureosulcata and E. globulus), the relative volume of the fast phase was so large that it had to be assigned to nearly the entire leaf. In contrast, the relative size of the slow phase corresponds to the relative size of the epidermis and mesophyll tissues in gymnosperms, and with only the palisade mesophyll in four of the six angiosperm species (i.e. excluding Phyllostachys and Eucalyptus).

image

Figure 9. Relative contribution of each tissue type for each species. Thick lines with vertical error bars indicate the per cent water corresponding to the fast phase (below the line) and the per cent water corresponding to the slow phase (above the line). Error bars denotes SE of the mean.

Download figure to PowerPoint

Significance of hydraulic compartmentalization

Leaf hydraulic design is likely to play a significant role in leaf photosynthetic performance, in part because species with high photosynthetic rates typically operate with small hydraulic margins. For example, the volume of water passing through a transpiring Quercus rubra leaf in 30 min can exceed the total amount of water in the leaf when it is fully hydrated. Stated another way, a rapidly transpiring Q. rubra leaf will use up its entire water content above the turgor loss point in only 75 s (Zwieniecki, Boyce & Holbrook 2004). The extent to which this substantial and potentially time-varying flux of water through the leaf might affect the performance of the photosynthetic cells depends both on the efficiency with which the xylem is hydraulically coupled to the epidermis, as well as on the degree to which the mesophyll is isolated from the transpiration stream. Although further measurements are needed to quantify the degree of hydraulic isolation between specific tissues, we explore here the potential consequences of such compartmentalization.

We focus on two aspects of leaf hydraulic design: the hydraulic linkage between the xylem and epidermis, and the degree to which the photosynthetic tissues are hydraulically uncoupled from the transpiration stream. The former has implications for stomatal control of xylem tensions, while the latter bears on the effects of transient imbalances in supply and demand on the water status of the mesophyll. Three scenarios for the connection between xylem, epidermal layers and photosynthetic cells are presented in Fig. 10. In the first scenario, the vein is hydraulically separated from the rest of the leaf; the second presents a mixed design in which the epidermis is hydraulically linked to the veins, but the mesophyll remains separated; while the third scenario describes leaves in which all tissues are equally well coupled. Each of the hydraulic designs proposed here has distinct physiological implications and leads to predictions for leaf behaviour in response to water management (Pickard 1982; Canny 1993, 1995; Zwieniecki et al. 2002; Sack et al. 2003).

image

Figure 10. Schematic of three scenarios for leaf hydraulic design describing the hydraulic linkages between different tissues. Solid lines depict water flow, dashed lines describe diffusion of water vapour, and Ø denotes high resistance between tissue types.

Download figure to PowerPoint

The 10 species examined here support the existence of all three designs. Based on the relative volumes of the two phases, we propose that the four gymnosperm species examined have a relatively weak hydraulic connection between the vein and the rest of the leaf, corresponding to the first scenario. In these species, we suggest that the fast phase is limited to the vein (including the transfusion tissue) and bundle sheath extension (if present) and that the slow phase includes the mesophyll and epidermis. This hypothesis is supported by the lack of specialized non-photosynthetic tissue connecting the vein to the epidermis in Larix laricina and M. glyptostroboides. In the other two gymnosperms, Ginkgo biloba and Gnetum gnemon, anatomical separation of the vein from the mesophyll and epidermis is not obvious, although in both cases there is an endodermal-like cell layer surrounding the vein that might provide the hydraulic separation of vein from the rest of the leaf (Esau 1977; Fahn 1990).

What are the physiological consequences of design 1? The most obvious prediction is that in transpiring leaves, the water potential of the epidermis, and thus the stomata, will be significantly lower than that of the xylem. One consequence of this is that stomata could be forced to shut before the xylem experiences a significant drop in pressure. Stomatal closure that significantly preceded the threshold for cavitation has been reported in ferns (Brodribb & Holbrook 2004). This pattern of early stomatal closure has been described as providing leaves with a ‘safety margin’, but might simply reflect the presence of a significant hydraulic discontinuity between xylem and epidermis. Thus, leaves in which the epidermis is hydraulically disconnected from the xylem are not efficient in utilizing their hydraulic system to maximize photosynthetic activity. In addition, under fluctuating atmospheric conditions, their stomata may be forced to limit gas exchange despite abundant soil water availability.

In design 2, we propose that the epidermis and bundle sheath extensions form part of the fast phase, while the palisade and possibly some of the spongy mesophyll belong to the slow phase. Based on the relative volumes of the fast and slow phases, we suggest that this design corresponds to four of the six angiosperm species examined (i.e. excluding Eucalyptus and Phyllostachys). In this design, the easiest route for the water is to flow out of the vein, through the well-connected cells of the bundle sheath extension, to the epidermis where water evaporates from the internal surfaces of stomata or nearby cells. This results in the mesophyll being largely bypassed by the transpiration stream. Whether such hydraulic separation of the mesophyll could result solely from leaf geometry, or requires some degree of restricted water movement between specific cell types (e.g. between the adaxial epidermis and palisade parenchyma cells) is not known.

The physiological consequence of design 2 derives from the epidermis and stomata being hydraulically well linked to the xylem. This allows the water potential of the epidermis to closely follow that of the xylem and thus the threshold for stomatal closure can be much closer to the cavitation threshold than in design 1. Consistent with this, a small ‘safety margin’ between xylem cavitation and stomatal closure has been observed in a number of angiosperm species (Brodribb & Holbrook 2004). Hydraulic separation of the mesophyll from the transpiration stream allows these cells to be buffered from rapid changes in the transpiration rate. One could argue that disconnecting mesophyll cells from the transpiration stream reduces the capacity of the leaf to deal with sudden increases in water loss rates given that the mesophyll cells themselves could act as water capacitors. However, the amount of water in these cells could support transpiration for no more than a few tens of the seconds before their turgor loss point is reached. Moreover, minimizing short-term changes in the water status of mesophyll cells might be advantageous given their photosynthetic role.

In design 3, all major tissue types belong to the fast phase. Based on results from the rehydration kinetics, we believe that such a design is highly probable for E. globulus and Phyllostachys aureosulcata where only a small per cent of the total leaf volume corresponds to the slow phase. This suggests that both the mesophyll and epidermis in these two species are well linked hydraulically to the vasculature. This interpretation is consistent with the very high density (packing) of the mesophyll, the presence of stomata on both surfaces, and the lack of morphological separation between spongy and palisade mesophyll. From a physiological perspective, design 3 has similar properties as design 2, in that leaves would be expected to operate near the cavitation threshold. However, the mesophyll would not be protected from sudden changes in transpiration and thus the photosynthetic cells could be exposed to higher temporal variation in water potential.

The three hydraulic design scenarios discussed here result from a consideration of rehydration kinetics in conjunction with anatomical studies. Although assignment of the two phases to spatially distinct tissues within the leaf remains a hypothesis, our goal is to stimulate discussion concerning the hydraulic designs of leaves and to motivate research that will provide a better understanding of both the biological and physical principles governing water flow in leaves. We have shown that rehydration kinetics provides a powerful tool for describing the movement of water within leaves at negative water potentials. The data presented here represent a significant step in our understanding of leaf hydraulic design. In essence, they motivate a conceptual shift from the idea that leaves function hydraulically as a single pool of water that simply evaporates from the internal surfaces of mesophyll and/or epidermis, to one in which leaves are understood to consist of well-organized water pools separated by hydraulic resistance of sufficient magnitude to maintain water potential disequilibria for sufficient time to have a significant impact on leaf performance.

ACKNOWLEDGMENTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES

We would like to thank Andrea Leigh and Anna Zwieniecka for providing crucial assistance in collecting data from the reverse Polish guillotine, and Anna Zwieniecka for all aspects of microscope slide preparation and image analysis. This work was supported by NSF competitive grant number IOB-0517071 and the Andrew W. Mellon Foundation.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGMENTS
  8. REFERENCES