INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Drought is one of the most limiting environmental stresses for plant production (Kramer and Boyer, 1995). Plants respond to drought by closing their stomata, which reduces leaf transpiration and prevents the development of excessive water deficits in their tissues. The drawback of the stomatal closure for plants is that their carbon gain is lowered and their growth is impaired. Therefore, information about plant responses to drought and their underlying mechanisms will have relevance for plant adaptation to new climatic conditions and breeding programs (Schulze et al., 1987).

Many physiological variables are known to be implicated in the stomatal regulation; some of these variables are involved in the mechanics of stomatal movements, others in the signals triggering these movements (Zeiger et al., 1987). A new set of variables has been proposed in the recent literature, all of which are related to the hydraulics of plants. Indeed, it has been suggested that stomata may respond to drought-induced changes in the hydraulic characteristics of the sap pathway, such as whole-plant hydraulic resistance (R_plant; Meinzer and Grantz, 1990; Cochard et al., 1996a; Lu et al., 1996), soil resistance (R_soil), root resistance (R_root; Meinzer and Grantz, 1990; Cochard et al., 2000b), or shoot resistance (R_shoot; Salleo et al., 1992; Sperry et al., 1993). However, during water stress, changes in the above parameters are not independent and, further, correlated to variations in soil water potential (_soil), whose impact on stomatal function has long been recognized (e.g. Hinckley and Bruckerhoff, 1975).

It has also been suggested that stomatal closure during water stress might be associated with the maintenance of xylem integrity (Tyree and Sperry, 1988; Jones and Sutherland, 1991). Sap is transported under negative pressures in plants, and, therefore, is susceptible to cavitation events (Pickard, 1981) that render xylem conduits non-conductive. Cavitation occurs when the negative sap pressure exceeds a threshold value defined by anatomical characteristics (Sperry and Tyree, 1988). Many species have been found to operate very close to the point of embolism. Therefore, stomata control both plant water losses and sap pressure and, thus, may actively control the risk of xylem embolism (Jones and Sutherland, 1991).

The first objective of our study was to experimentally uncouple R_soil, R_root, R_shoot, and _soil to identify the specific hydraulic parameters associated with stomatal regulation during water stress. The experiments were conducted on potted walnut (Juglans regia × nigra) trees in a growth cabinet to minimize the effects of temperature, light intensity, air vapor deficit, and air CO₂ concentration on stomatal behavior. The second objective of our study was to test the hypothesis that stomata control embolism during water stress in walnut.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Vulnerability to Cavitation

Vulnerability curves (VCs) were constructed by plotting the changes in the percentage loss of xylem conductance (PLC) versus xylem pressure. The VCs for the different walnut organs are shown in Figure 1 and the parameters of the logistic fitting (Eq. 4) are given in Table I. Significant differences were found between organs. Leaf rachises were significantly the most vulnerable, roots the least vulnerable, and leaf veins and shoots intermediate. The few measurements obtained on shoot segments were in agreement with our previous study.

View larger version (25K): [in this window] [in a new window]   Figure 1.   VCs for walnut leaf midribs and rachis, and current year shoots and roots. Close symbols refer to measures on air-pressurized segments. White symbols refer to leaf midribs and rachis collected on trees during the different experiments. Error bars are ± SE. Lines are logistic fits through the data. For shoots, the lines are logistic fits through the data published by Tyree et al. (1993) for current year (plain line) and previous year (dashed line) segments.

View this table: [in this window] [in a new window]   Table I.   Xylem vulnerability to embolism Parameters of the logistic function (see equation in the text) fitted to the experimental PLC versus rachis curves. Values are given ± SE. Data having a letter in common are not significantly different at P = 0.01.

Pressure-Volume Analysis

Turgor pressure (P_leaf  0) at full turgor averaged 0.93 ± 0.06 MPa (n = 5, ±SE) and turgor loss point averaged 1.53 ± 0.04 MPa. The relative change in P_leaf (P_leaf/P_leaf  0) with decreasing leaf water potential (_leaf) is shown on Figure 7.

Soil Dehydration

When plants were exposed to different light levels while soil watering was stopped (procedure 1), only a moderate water stress was induced. Figure 2 shows the concurrent changes in E_plant and P_rachis for one tree. The other trees behaved the same. After 4 d, P_rachis measured on the plants placed in the dark (predawn P_rachis) was reduced only by 0.1 MPa (see y intercepts on Fig. 2). However, E_plant, stomatal conductance (g_s), and the P_rachis at maximum light intensity were considerably reduced at the end of the treatment. This was consistent with a sharp increase in R_plant. This increase could not be attributed to xylem cavitation because the xylem pressure never drops below the threshold for cavitation for all organs measured. The change in R_plant was more probably due to increase in R_soil and/or R_root.

View larger version (24K): [in this window] [in a new window]   Figure 2.   Typical changes of plant transpiration and xylem water pressure during the first stage of a soil water stress for one of the studied tree. The four different symbols correspond to 4 consecutive d. Four different light intensities were used on each day to vary Eplant. Lines are linear regressions through the data for each day. The slope of the lines represents the Rplant.

When plants were continuously exposed to a constant and high light intensity for 1 week (procedure 2), higher levels of water stress were obtained. Figure 3 shows the results for one tree, the other trees behaving the same. E_plant and g_s dropped close to zero whereas P_rachis leveled at approximately 1.4Mpa. At this point, the degree of xylem embolism in the leaf rachis was still less than 10 PLC (not shown).

View larger version (25K): [in this window] [in a new window]   Figure 3.   Typical time course of plant transpiration and xylem water pressure for a non irrigated tree exposed to a constant light intensity.

Root Chilling

Decreasing soil temperature from ambient to near zero temperature provoked a significant reduction in g_s, E_plant, and P_rachis (Fig. 4). These changes were reversed when soil temperature was returned to ambient.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Typical time course of xylem water pressure and soil temperature (top panel), and plant transpiration and gs (bottom panel) during a soil chilling experiment. Error bars are ± SE (n = 5).

Stem Pressurization

Application of a positive pneumatic pressure around the trunk of a plant induced a sharp decrease of E_plant, g_s, and P_rachis only for pressures greater than 2.5 MPa (Fig. 5). At this pressure, about 70 PLC was induced in the stem (see the VC for stems plotted on top of Fig. 5). For air pressure applications higher than 3 MPa, more than 95 PLC was induced in the stem, leaf wilting was noticed, and nearly 100 PLC was measured in the leaf rachis.

View larger version (20K): [in this window] [in a new window]   Figure 5.   Changes in gs (top panel) and xylem water pressure (bottom panel) of trees exposed to increasing pneumatic pressures around their trunk (x axis). The curve represents the change in embolism in the trunk versus the applied pressure.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

The first objective of our study was to combine different experimental approaches to unravel the effect of plant hydraulics on stomatal closure. The second objective was to analyze the effect of stomatal closure on plant hydraulics and to test the hypothesis of a stomatal control of embolism.

Unraveling the Effects of _soil, R_soil, R_root, and R_shoot in the Response of g_s to Water Stress

In this study, we focused our analysis on some of the endogenous physiological parameters likely to be altered during a water stress and that have been reported in the literature to be associated with stomatal responses. These parameters are the _soil, the R_soil, the R_root, and the R_shoot; all of these parameters are strongly correlated under natural drought conditions. The experiments were designed to alter R_plant in very different ways, which probably had a primary influence on different parts of the pathway.

The effects of the different experiments conducted on the same plant material can be analyzed according to their main impact on _soil, R_soil, R_root, and R_shoot. Soil dehydration provoked mainly a drop in _soil and an increase of R_soil. The resistance of the interface between the soil and the root probably also increased (Nobel and Cui, 1992). R_root was modified to the extent that the radial flow into the root xylem was altered. R_shoot was probably not altered, because the level of xylem embolism remained low during these experiments. However, if the ionic composition of the sap changed dramatically as a result of the drought, then R_shoot may have varied (Zwieniecki et al., 2001). Root chilling altered mainly R_root and R_soil. The soil being watered to maximum capacity in this experiment, R_soil was very low compared with R_root, and its variations were not quantitatively significant. _soil remained constant because the temperature dependence of _soil is very low (Muromtsev, 1981). Stem pressurization provoked only an increase in R_shoot when the pressure exceeded the point of embolism induction (about 2.0 MPa; Tyree et al., 1993; Fig. 5). If the air was propagated along the xylem flow path significantly beyond the injection point, R_root and R_leaf may also have being altered. Therefore, combining the results of all these experiments, it is possible to determine whether g_s and E_plant were specifically responding to main changes in _soil, R_soil, R_root, and/or R_shoot or not. Because air humidity, air temperature, and light intensity were maintained constant in our experiments, leaf to air vapor deficits and leaf boundary layer conductances were also constant. Therefore, the g_s and E_plant patterns corresponded in our study. The relationship between g_s and hydraulic parameters are likely to depend on these environmental conditions, contrary to the relationships with E_plant (Monteith, 1995).

Our results showed that the different experiments all significantly reduced E_plant and g_s. Therefore, the response of g_s to _soil, R_root, R_soil, and R_shoot was neither specific nor exclusive. An alternative analysis of the problem is not to consider _soil, R_root, R_soil, and R_shoot individually but rather to examine their combined effect on P_rachis or _leaf. The relationship between P_rachis, _soil, R_root, R_soil, E_plant, and g_s under steady-state conditions is well described by the Ohm's law analogy (Tyree and Ewers, 1991):

(1)

where SF_plant is the plant leaf area and D the air vapor pressure deficit, two parameters that remained constant during our experiments. The gravity term and the xylem sap osmotic potential are assumed negligible in equation 1. A similar relationship is obtained with _leaf if we further include the leaf blade hydraulic resistance. The dependency of g_s or E_plant on P_rachis and _leaf is shown on Figure 6 for all of the experiments. The responses were similar whatever the experiments. This would suggest that in these experiments, g_s was not correlated to changes in _soil, R_soil, R_root, or R_shoot per se but rather to P_rachis and/or _leaf. Another illustration on the combining effects of R_soil R_root, and R_shoot on E_plant is shown on Figure 8. An identical relationship was obtained between E_plant and C_plant (defined as [R_soil + R_root + R_shoot]¹). Data for plants exposed to a high soil water depletion (procedure 2) are not included on Figure 8. These results are in agreement with the finding of Saliendra et al. (1995), Sperry (2000), and Hubbard et al. (2001). Our data do not enable us to test whether g_s was more specifically responding to changes in _leaf or P_rachis because both values were highly correlated and differed only by a few 0.1 MPa. Experiments where P_rachis and _leaf are uncoupled are awaited.

View larger version (37K): [in this window] [in a new window]   Figure 6.   Dependence of gs (top panels) and transpiration (Eplant) on xylem water pressure (left panels) and leaf. The different symbols represent the different experiments conducted in this study. Transpiration was normalized by the transpiration of each tree before treatment (Eplant/Emax).

Evidence for a Stomatal Control of Xylem Embolism in Walnut

From the above discussion, it could be concluded that the effect of the stomatal closure in our study was to maintain P_rachis above a threshold value around 1.4 MPa and _leaf above approximately 1.6 MPa. To further understand this behavior, we have to identify a major physiological trait that would threaten plant integrity at lower P_rachis and/or _leaf values. The answer to this question is obviously very complex, because many traits are probably involved and correlations between them probably exist. We will argue that, in our study, xylem cavitation was a trait correlated with the stomatal closure.

A physiological trait associated with a stomatal closure during water stress should meet at least the following three main conditions. First, its impairment should represent a serious threat to plant functioning. This results from the consideration that the cost associated with stomatal closure is high (reduced carbon gain, reduced growth, reduced reproductive success, etc.) so the gain associated with the regulation should overcome the loss. Cavitation is a serious threat for plants because it impairs the xylem conductive capacity and may eventually lead to leaf desiccation and branch mortality (Rood et al., 2000). Indeed, leaf desiccation was not observed in our study as long as the xylem integrity was maintained. Leaf desiccation was noticed only when high levels of embolism were measured in the leaf petioles. The gain associated with stomatal closure was thus the maintenance of leaf vitality, which largely overcomes the drawbacks cited above.

The second condition is that the impairment of the trait should be water deficit dependent because the effect of stomatal closure is precisely to prevent excessive leaf dehydration. The mechanism of water stress-induced cavitation has been well documented (Sperry and Tyree, 1988; Cochard et al., 1992; Jarbeau et al., 1995). Air is sucked into the xylem lumens through pores in the pit wall when pressures in the sap exceed the maximum capillary pressures that can sustain the pores. Therefore, the likelihood of cavitation occurrence is directly determined by the degree of water deficit in the xylem, more precisely by P_rachis. The maintenance of leaf turgor above cell plasmolysis is another physiological trait that might also satisfy these first two conditions.

The third condition is that the impairment of the trait should have the same water deficit dependence as stomata. Stomata were completely closed in walnut trees when P_rachis reached about approximately 1.4 MPa and _leaf about approximately 1.6 MPa. The impairment of the trait associated with stomatal closure should therefore occur at comparable P_rachis or _leaf values. On Figure 7 (top panel) is represented the dependencies of E_plant and rachis embolism on P_rachis. The leaf rachis was the most vulnerable organ along the sap pathway in the xylem and was also exposed to the lowest xylem pressure values. Leaf rachis is therefore the Achilles' heel of the walnut tree sap pathway. Segmentation in xylem vulnerability to cavitation has been demonstrated for several other species (Cochard, 1992; Sperry and Saliendra, 1994). A lot of variation exists between species, and occasionally the roots appear to be the most cavitation sensitive organs in the plant (Sperry and Ikeda, 1997). From Figure 7, it is clear that the dependencies of leaf rachis xylem embolism and transpiration on water deficit were very similar. Stomata were completely closed at the incipience of xylem embolism in the leaf rachis. Figure 7 (bottom panel) shows the concurrent variations of E_plant and leaf turgor pressure (P_leaf) with bulk _leaf. It is also clear from this graph that stomata were completely closed at the incipience of leaf cell plasmolysis (turgor loss point). The maintenance of xylem integrity and leaf turgor were closely associated with stomatal closure during water stress in walnut.

View larger version (24K): [in this window] [in a new window]   Figure 7.   Dependence of plant transpiration and xylem rachis embolism on rachis xylem pressure (top panel) and dependence of plant transpiration and leaf turgor pressure on leaf (bottom panel). Error bars are ± SE. The VC was replotted from Figure 1, and data for transpiration rate were averaged from Figure 7.

The data from Figure 7 suggest that stomatal closure was rather preemptive in avoiding cavitation. This behavior might be explained by the potential for "catastrophic xylem failure" (Tyree and Sperry, 1988). There is a feedback between xylem conductance and xylem pressure during cavitation. Cavitation decreases xylem conductance, which in turn decreases xylem pressure and thus provokes more cavitation. Tyree and Sperry (1988) and Jones and Sutherland (1991) have computed that catastrophic xylem failure occurs at the expense of some xylem conductance and at a critical transpiration rate (E_crit) only slightly greater than the actual maximum E. The hypothesis of a stomatal control of catastrophic xylem failure was evaluated with an hydraulic model of a walnut tree explicitly taking into account the feedback between xylem pressure and xylem conductance. Our simulations confirmed the results of Sperry et al. (1998) and Comstock and Sperry (2000). Transpiration was maximized (E_crit) at the expense of all conductance in the distal leaf rachis segment. E_crit was therefore much higher than the actual E_plant (Fig. 8, plain line). Using the same model, we have computed E_plant provoking 1% (E_1PLC) and 10% (E_10PLC) loss of rachis conductance (Fig. 8, dotted and dashed lines, respectively). The onset of tree water loss regulation occurred when E_plant reached E_1PLC and E_plant tracked E_10PLC when plant conductance was further reduced. This model suggests that the risk of catastrophic xylem failure was not associated with the stomatal regulation in walnut. g_s was not maximized at the expense of all xylem conductance. Rather, xylem conductance was maximized at the expense of all g_s. To experimentally validate these computations, we have tried, without success, to feed stressed plants with fusiccocine, a drug supposed to promote stomatal opening. The use of mutants lacking efficient stomatal regulation is probably a better way to test such hypotheses (Cochard et al., 1996b).

View larger version (37K): [in this window] [in a new window]   Figure 8.   Plant transpiration versus whole plant hydraulic conductance. Different symbols represent different experiments. The plain, dashed, and dotted lines represent the critical transpiration rate provoking 100%, 10%, and 1% loss conductance in the leaf rachis, respectively.

Mechanisms Triggering Stomatal Closure

The mechanisms by which stomata could sense changes in P_rachis to adjust g_s and avoid cavitation remain hypothetical. It seems improbable that P_rachis itself was the triggering parameter, so we must identify a parameter correlated to P_rachis during water stress. Our stem pressurization experiments dismiss the implication of parameters associated with soil or root water stress (such as _soil or abscisic acid production by roots [Tardieu et al., 1992]). Because the onset of cavitation events in leaf blades was correlated with the onset of stomatal closure in laurel, it has been suggested that stomata were responding to hydraulic signals generated by cavitation (Salleo et al., 2000; Nardini et al., 2001). However, in our study, stomatal closure occurred before the onset of cavitation in leaf blades and midribs (Fig. 6) and only after 70% loss of conductance in the trunk (Fig. 5). The possibility that stomata were responding to _leaf cannot be rejected because of the high correlation between P_rachis and _leaf in our study. _leaf could have a direct impact on guard cells turgor (Meidner and Edwards, 1996) or an indirect impact on their sensitivity to abscisic acid (Tardieu and Davis, 1993). A stomatal response to changes in leaf water status has long been suggested (Cowan, 1977) and has recently been supported by root pressure chamber experiments on woody plants (Fuchs and Livingston, 1996; Comstock and Mencuccini, 1998). These experiments demonstrate that stomatal closure caused by soil drought or decreased air humidity can be partially or wholly reversed by root pressurization.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

In conclusion, combining different experimental procedures, we have established that stomata were not responding to changes in _soil, R_soil, R_root, or R_shoot per se but rather to their impact on P_rachis or _leaf. Cavitation avoidance was a likely physiological function associated with stomatal regulation during water stress in these experiments. However, the mechanism triggering the stomatal closure probably resides in the leaf mesophyll itself as stomatal closure was also correlated to bulk leaf turgor. The significance of cavitation avoidance seems crucial to understand plant water relations during water stress, probably more than the occurrence of cavitation itself. The possibility still exists that the striking relations that were observed between cavitation and stomatal function were only correlations and that the main physiological trait involved in the regulation was elsewhere. However, it seems to us that cavitation avoidance fits the three conditions we have identified for such a trait better that any other trait so far.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION CONCLUSIONS MATERIALS AND METHODS LITERATURE CITED

Plant Material and Growth Conditions

Experiments were conducted on 3-year-old walnut (Juglans nigra × regia cv NG38 grafted on Juglans regia rootstocks) trees grown in 35-L pots containing one-third peat and two-thirds natural clay soil from the Limagne region (France). The plants, grown in a nursery, were transferred to growth cabinets where air temperature (25°C) and air humidity (60% relative humidity) were constant. The light intensity at foliage level was adjusted between 0 and 480 µmol m² s¹. A total of 17 trees were used in the different experiments. Because the experiments were time consuming, we chose to combine multiple experiments with a rather limited number of repetitions.

_leaf and Plant Transpiration

_leafs were measured with a Scholander-type pressure chamber on leaflets randomly sampled on the trees. The xylem water pressure at the base of the leaflet blades (P_rachis) was measured on leaflets enclosed for at least 2 h in an air proof aluminum bag (Turner and Long, 1980). To minimize the reduction in tree leaf area due to sampling, only one replicate of P_rachis and _leaf were performed each time.

Plant water loss (E_plant) was estimated with two independent techniques. Sap flow through the trunk was monitored by the heat balance technique of Valancogne and Nasr (1989). E_plant was also measured gravimetrically with an electronic balance (model ID1, resolution 1g, Mettler, Viroflay, Switzerland). In addition, leaf transpiration and leaf g_s were measured with a steady-state porometer (model 1600, LI-COR, Linclon, NE). The measurements were performed on five leaflets randomly chosen in each trees.

Xylem Embolism

The degree of xylem embolism due to the presence of air in the vessels was assessed following the procedure introduced by Sperry et al. (1988) and recently substantiated (Cochard et al., 2000a). We used the new XYL'EM apparatus (Xylem Embolism Meter, Instrutec, Montigny les Cormeilles, France), based on a high-resolution liquid mass flowmeter, to determine the PLC in the leaf rachis. For each PLC determination, one leaf was cut from the trees under tap water. Six segments 15 to 20 mm long were excised, still under water, from the leaf rachis, and four segments were detached from the midribs of four leaflets. The segments were attached to the tubing of the XYL'EM apparatus and their initial hydraulic conductance (K_init) determined with a hydrostatic pressure gradient of approximately 3 kPa. The samples were then flushed with water pressurized to 0.1 MPa and the maximum conductance (K_max) determined as above. The PLC was calculated as follows:

(2)

R_plant

The R_plant was computed as the ratio between the water potential drop along the sap pathway and the whole-plant transpiration:

(3)

It should be noted that, as we have defined it, R_plant does not include the leaf blade hydraulic resistance. _soil was estimated by the P_rachis value measured on the plants placed in the dark (Améglio et al., 1999).

VCs

The vulnerability of the xylem conduits to air embolism was determined following the pressure chamber dehydration technique of Cochard et al. (1992). The technique consisted in inserting a plant segment in a pressure chamber with only the cut end protruding. The pressure in the chamber was then adjusted to a target value and maintained constant until sap exudation ceased (usually after 15-30 min). The pressure was then released and xylem segments were excised under water and their PLC value determined as above. The plot of the PLC value versus the applied air pressure represents the VCs of the plant segment. VCs were first obtained for leaflet midribs, leaf rachises, and fine roots. A few measurements were also made on current year stem segments to confirm our previous results (Tyree et al., 1993).

For midrib and rachis VCs, entire leaves were pressurized. Four to six rachis segments were measured on each leaf. The basal 20 cm of the leaf rachis was discarded to avoid artifacts caused by air entry into cut vessels. On the same leaves, midribs were excised from the four apical leaflets. Root VCs were obtained on lateral roots excised from entire root systems gently washed with tap water. A total of 13 leaves and 13 roots from three different trees were used to construct the VCs. The following logistic function was fitted to the different curves (Cochard et al., 1999).

(4)

where P_rachis50 is the water pressure inducing 50 PLC and s is a slope parameter.

Pressure-Volume Analysis

Pressure-volume curves were obtained for five individual leaves using the bench top dehydration method (Ritchie and Hinckley, 1975). Leaves were harvested, enclosed in a plastic bag, and let to rehydrate overnight with their petiole immerged in distilled water. Bulk tissue estimates of turgor pressure (P_leaf) and turgor loss point were made using the method of Schulte and Hinckley (1985).

Soil Dehydration

The stomatal response to a soil water deficit was studied on six different trees following two complementary procedures. The water deficit was induced by withholding irrigation. For the first procedure, the light intensity in the chamber was increased by steps during the day. Four light intensities (0, 170, 250, and 450 µmol m² s¹) were applied for 2 h each, to obtain different transpiration rates so the slope of E versus P_rachis could be determined for getting R_plant. After 2 h, plant transpiration was stable, enabling determination of _leaf, P_rachis, and g_s. During the night, the light level was maintained at 170 µmol m² s¹ to progressively lower the soil water content. The procedure was repeated for 4 consecutive d for each of the three trees used in this experiment. In a second procedure, plants were continuously exposed to 450 µmol m² s¹ during 7 consecutive d. _leaf, P_rachis, and g_s were determined twice a day. Leaves were periodically sampled for embolism assessment. The second procedure was repeated on three different trees.

Stem Pressurization

The aim of this experiment was to analyze the stomatal response to an increase in trunk hydraulic resistance. We used the air injection technique (Cochard et al., 1992; Salleo et al., 1992) to induce air embolism in the trunk and thus increase the trunk resistance. A 20-cm-long pressure chamber was fastened to the trunk of three different trees exposed to a constant light intensity (450 µmol m² s¹). The pressure chamber was similar to the chamber described by Salleo et al. (1992), except that it was divided in two parts and could be clamped on intact trees (Améglio et al., 1994). To facilitate air entry into the xylem conduits, six 1.3-mm-diameter holes were drilled radially through the trunk inside the chamber. The holes were aligned vertically to minimize the damage to the xylem conduits. In a preliminary experiment, we verified that fastening the pressure chamber to the tree had no effect on tree transpiration and _leaf. The trunk segment in the chamber was exposed to increasingly higher air pressures (up to 3.5 MPa). Each target pressure was maintained constant for 15 min. Two hours after each pressurization, two leaflets were sampled, one to measured P_rachis, and one to measure _leaf with a pressure chamber, and g_s was measured on five leaflets. Leaves were also periodically sampled for embolism assessment.

Root Chilling

This experiment aimed at altering root and soil hydraulic resistance by changing soil temperature (Brodribb and Hill, 2000; Cochard et al., 2000b). A well-watered tree was placed in the growth chamber with 450 µmol m² s¹ light intensity. _leaf, P_rachis, and g_s were measured when E_plant became stable. Solutions with different temperature were obtained by mixing tap water with ice. The solutions were successively poured on the soil to lower its temperature (T_soil) in a step-wise manner from ambient to approximately 0.5°C. T_soil was measured with a thermocouple inserted near the center of the pot. After approximately 75 min, _leaf, P_rachis, and g_s were measured, and a cooler solution was poured on the soil. T_soil was then returned to 20°C by pouring water at ambient temperature on the soil. The whole procedure was repeated three times on two different trees.

Modeling

The relationship between plant hydraulic parameters and water loss regulation was analyzed with a model similar to the ones developed by Tyree and Sperry (1988) and Sperry et al. (1998). The soil-plant continuum is represented by hydraulic resistances in series: R_soil, R_root, R_stem, R_trunk, R_branch, R_rachis, and R_leaf. The partitioning of resistances within a tree was measured with an high pressure flowmeter (H. Cochard, unpublished data; Tyree et al., 1994). The relationships between the different xylem resistances and the xylem pressure were derived from the logistic fits of their VCs. R_soil was supposed negligible and not _soil dependent. Therefore, our model pertains only for the understanding of stomatal behavior during our root chilling and stem pressurization experiments. Sperry et al. (1998) and Comstock and Sperry (2000) have pointed out that the accuracy of such models depends on how finely the continuum is discretized. In our model, each xylem resistance was divided into 10 tiers. A test with 1,000 tiers yielded only 1.7% higher E_crit values. E_crit was determined to the nearest 0.001 mmol s¹ by progressively increasing E_plant until 100 PLC was obtained in one tier. The model was run with increasing R_root and R_trunk values to simulate the experimental observations. The model, called "RER," was developed on an Excel (Microsoft, Redmond, WA) spreadsheet.