Stem water potential measurements obtained using standard methodologies diverge under extreme drought in tomato and grapevine

Authors: Celia M Rodriguez-Dominguez, Madeline R Carins-Murphy, Jaime Sebastian-Azcona, Timothy J Brodribb

Abstract

Stem water potential (ψstem) is one of the most common metrics used to define plant water status. Accurate measurement of ψstem is therefore essential to quantify critical plant physiological processes in response to water availability. In theory, leaf and stem water potential equilibrate when leaf transpiration is prevented, and leaves and stems are hydraulically connected. Therefore, many studies quantify ψstem by measuring equilibrated leaves with a pressure chamber. However, leaf tissue damage occurring during dehydration due to xylem cavitation events may impact the accuracy of this indirect measurement of ψstem. Here, we present 2 case-studies in which ψstem close to the lethal threshold for leaf mesophyll tissue led to high discrepancies between pressure chamber measurements of equilibrated leaves and direct psychrometric measurements. We dehydrated (i) whole plants of an herbaceous species, tomato, under typical diurnal cycles in a glasshouse, and (ii) branches of a woody species, grapevine, under laboratory conditions. Dehydration beyond the point of xylem cavitation is expected to lead to leaf water potential falling below ψstem, but rather we observed that once ψstem declined to values expected to cause considerable loss of leaf xylem function, indirect ψstem values from leaves (ψleaf-eq) remained higher than direct ψstem measured with psychrometers (ψstem-PSY). A decline in the osmotic potential of the leaf xylem sap (ψπ-sap) was consistent in both species, possibly indicating that the contents of disrupted leaf cells contributed to this decline. These results demonstrate that, at least for tomato and grapevine species, caution should be exercised when using the pressure chamber at water potential levels that may induce leaf tissue damage and support a method for approximating these levels by leaf xylem sap extraction.

Introduction

Stem water potential (ψstem) is one of the most common metrics used to quantify plant water status in studies of plant physiology. It is generally required to contextualize how specific plant traits are linked to different drought responses. This is because water travels from the soil to the leaf, and to the atmosphere through stomata, along a gradient in water potential (Philip 1966) as explained by the cohesion-tension theory (Steudle 2001; Angeles et al. 2004). An accurate knowledge of this gradient is fundamental to improve predictions of plant responses to drought conditions and to connect them to the ecosystem function (Novick et al. 2022). Furthermore, this accurate knowledge permits comparisons to be made among and within species and individuals, and serves as the main indicator to define important traits underlying plant performance, such as the risk of drought-driven hydraulic failure (Choat et al. 2012) that would induce plant tissue mortality (Brodribb et al. 2021).

The precise measurement of ψstem is therefore critical to a wide spectrum of plant sciences studies, e.g. (i) the characterization of the risk of drought-driven xylem cavitation, which induces the formation of gas bubbles (embolisms) within the xylem vessels, or (ii) the identification and determination of plant tissue mortality thresholds. These 2 highly connected processes, i.e. xylem cavitation and tissue mortality (Brodribb et al. 2021; Mantova et al. 2022), are commonly studied by constructing xylem hydraulic vulnerability curves (e.g. Brodribb et al. 2017; Rodriguez-Dominguez et al. 2018; Gauthey et al. 2020) and by conducting detailed measurements of a variety of physiological traits on plants under conditions approaching lethal water stress (Mantova et al. 2023), respectively. In all cases, a precise monitoring of ψstem, to which both cavitation resistance and mortality thresholds will be referred to, is essential for an accurate characterization of plant response to drought as well as for a better comparability among similar studies.

ψ stem can be monitored directly using thermocouple psychrometry (ψstem-PSY), which measures the vapor pressure within a psychrometer chamber in equilibrium with the stem apoplastic water potential. However, in many studies ψstem is monitored indirectly by measuring the water potential of excised nontranspiring leaves with a pressure chamber. When measuring leaf water potential with the pressure chamber, it is assumed that the osmotic potential of the xylem sap is negligible, so the “balancing pressure” of the sample is equal to the tension of the xylem as it was when attached to the plant, neglecting the gravitational and matric components of water potential (Turner 1981; Steudle 2001). This technique has disadvantages, such as its destructive nature, which prevents continuous and automatic measurement. However, due to its easy and extended use, when implemented correctly, it is still the most widely applied technique among plant scientists for measuring leaf water potential (Rodriguez-Dominguez et al. 2022).

According to theory, when a leaf is prevented from transpiring by enclosure in a darkened plastic bag (or similar), an equilibrium between leaf and stem water potential is reached (Waring and Cleary 1967; Shackel et al. 1997). For building hydraulic vulnerability curves, ψstem of a tree branch or an entire sapling or seedling is monitored by either installing a stem psychrometer (ψstem-PSY) or equilibrating leaves and measuring them with the pressure chamber (ψleaf-eq). Validation of both techniques is usually presented in these studies (Skelton et al. 2017; Rodriguez-Dominguez et al. 2018; Gauthey et al. 2020). However, in some cases, like in tomato where leaf cavitation and leaf damage occur within a narrow range in ψstem (see Supplementary Information from Skelton et al. 2017), the use of the pressure chamber for measuring leaves at those levels of ψstem, i.e. with a considerable percentage of leaf xylem embolisms and probably tissue damage, would require users to pressurize damaged leaves, and hence, to possibly extract the cell content through the pressure chamber measurement (Boyer 1967; Turner 1976). Under these circumstances, one of the main assumptions of leaf water potential measurements with the pressure chamber, i.e. that the osmotic potential of the leaf xylem sap (ψπ-sap) is negligible (Turner 1981), may be broken. Moreover, it would also violate the assumption that the apoplastic water volume is constant during measurement. The decrease in ψπ-sap of the collected sap could reflect the contribution of solutes from the damaged cells and/or stem deformation (Boyer 1967). Alternatively, as shown by other studies, this could occur due to the contribution of ions or other solutes that change in the xylem due to other causes, e.g. dynamic changes in xylem sap osmolality in mangrove species or ionic effects on xylem hydraulic conductance (Schill et al. 1996; López-Portillo et al. 2014; Trifilò et al. 2014), although these events are expected to have a smaller influence on ψπ-sap and to occur over long timescales in living plants.

Here, we compared indirect measurements of ψstem made by pressurizing equilibrated whole leaves with the pressure chamber, as a traditional and widespread measure of ψstem (Waring and Cleary 1967; Shackel et al. 1997; Moriana et al. 2012; Corell et al. 2020), with direct measurements of ψstem made with thermocouple psychrometry. This would reveal the range of ψstem that can be accurately measured indirectly with excised, nontranspiring leaves in a simulated drought-induced mortality study of an herbaceous species (tomato), and during the construction of leaf hydraulic vulnerability curves for a woody species (grapevine). These species represent 2 opposite ends of the plant functional spectrum of drought responses. We hypothesize that ψstem should be very similar when measured concurrently using equilibrated leaves with the pressure chamber or directly on stems with thermocouple psychrometers, unless leaves are damaged by xylem cavitation and/or tissue damage. We also hypothesize that higher values from equilibrated leaves measured with the pressure chamber (ψleaf-eq) than ψstem-PSY would likely indicate that sap from locations other than the apoplast, e.g. due to leaf cell membrane disruption and release of symplastic sap, would contribute to the earlier appearance of the endpoint, i.e. ψleaf-eq being less negative than ψstem-PSY. On the contrary, lower values of ψleaf-eq, as suggested in other studies (West and Gaff 1971), would likely reflect the need to overpressurize leaves to refill emptied xylem vessels due to cavitation. Proxy measures of leaf tissue damage at low ψstem, such as the decline of ψπ-sap, would support the extra contribution of the symplastic to the apoplastic sap as a plausible cause for the higher values of ψleaf-eq. We further hypothesize that ψleaf-disc-eq (water potential of leaf discs taken from equilibrated leaves measured using psychrometry) would better approximate ψstem-PSY, since ψleaf-disc-eq avoids the possible erroneous values obtained by pressurizing possibly damaged leaves at low ψstem and corresponds, like ψstem-PSY, to total water potential.

Results

There was not a precisely 1 relationship between paired measurements of stem water potential made indirectly (excised whole leaves, ψleaf-eq, or leaf discs, ψleaf-disc-eq) and directly (stem psychrometers, ψ  stem-PSY) in both tomato and grapevine. The overall linear relationships between ψleaf-eq or ψleaf-disc-eq and ψ  stem-PSY resulted in a slope of 0.41 (r^2^ = 0.54, P < 0.0001) and 0.74 (r^2^ = 0.92, P < 0.0001), respectively, for tomato, and 0.66 (r^2^ = 0.80, P < 0.0001) and 0.76 (r^2^ = 0.54, P = 0.0002), respectively, for grapevine (Fig. 1). These linear relationships indicate that, for both species, the ψleaf-disc-eq − ψstem-PSY relationship was closer to the 1 line than the ψleaf-eq − ψstem-PSY relationship. In grapevine, 2 of the ψleaf-disc-eq data-points were exceptionally lower than ψstem-PSY (Fig. 1B). One of these leaves had a low ψπ-sap (−4.39 MPa), whereas ψπ-sap was impossible to measure in the other leaf due to the difficulty to extract sufficient sap. Thus, these 2 ψleaf-disc-eq points were excluded from the analyses (see Materials and Methods section for details).

ψ leaf-eq of tomato plants decreased in parallel with ψstem-PSY, i.e. Δψ (leaf) remained close to zero, until the onset of leaf xylem cavitation events (Pe), after which ψleaf-eq remained higher (less negative) than ψstem-PSY, increasing their statistically significant differences with decreasing ψstem-PSY (Fig. 2A). ψleaf-disc-eq also decreased in parallel with ψstem-PSY, but started to be higher than ψstem-PSY as the leaves passed P50 and approached P88, i.e. when a significant amount of leaf xylem embolism should have formed (Fig. 2A). Similarly, in the branches of the woody grapevine species, ψleaf-eq decreased in parallel with ψstem-PSY, but in this case, it started to be higher than ψstem-PSY after leaf P50, and increased their statistically significant differences from then on (Fig. 2B). Due to the higher variability of ψleaf-disc-eq for a given ψstem-PSY in grapevine than in tomato, Δψ (leaf-disc) was not significantly different from zero at any of the leaf xylem cavitation ranges considered. The leaf Pe–P50–P88 values (green, orange, and red solid lines in Fig. 2) corresponded to −1.24, −1.68, and −2.13 MPa, respectively, for tomato (Skelton et al. 2017), and −1.97 ± 0.09, −2.65 ± 0.13, and −3.30 ± 0.35 MPa, respectively, for grapevine (Supplementary Fig. S1). Therefore, binning the data considering these leaf xylem cavitation ranges allowed us to identify the levels of dehydration at which differences between indirect measurements and paired direct measurements of stem water potential (Δψ) became significantly higher than zero (Fig. 2, A and B). Furthermore, in both tomato and grapevine species, ψπ-sap decreased exponentially with ψstem-PSY (Fig. 2, C and D) and more predominantly when approaching and exceeding the leaf Pe–P50–P88 range. Thus, the point of maximum change of slopes occurred at −1.72 MPa in tomato, and at −2.78 MPa in grapevine.

Discussion

Stem water potential (ψstem) is a ubiquitous indicator of plant water status that is used in highly diverse studies of plant physiology. Here, we tested whether an indirect method commonly used to measure ψstem (applying positive pressure to excised equilibrated whole leaves in a pressure chamber) can be used interchangeably with direct measurements of ψstem using stem thermocouple psychrometers over a wide range of sample dehydration in 2 contrasted species (tomato and grapevine). We hypothesized that pressurizing severely water-stressed leaves, likely carrying significant tissue damage caused by dehydration, would result in a divergence from direct measurements of ψstem. In support of this, we found that ψstem measured indirectly using pressurized leaves differed from direct measurements using stem psychrometry after samples had been dehydrated beyond water potentials known to induce significant embolism in the leaves of each species. The fact that additional indirect measurements of ψstem made by measuring the water potential of leaf discs using psychrometry (i.e. not under positive pressure) agreed more closely with direct measurements (slopes closer to 1 than when considering ψleaf-eq) provides further evidence that ψstem measured by pressurizing severely stressed leaves may potentially lead to erroneous values in our study species.

Direct vs. indirect stem water potential measurements

The best agreement among direct and indirect ψstem was found in both species prior to water deficit levels expected to induce xylem embolisms (i.e. down to the leaf Pe–P50–P88 range, from −1.24 to −2.13 MPa for tomato plants, and from −1.97 to −3.30 MPa for grapevine branches). Past these levels of plant water stress, we found that ψleaf-eq was always higher than ψstem-PSY, despite ongoing sample dehydration. Large decreases in osmotic potential of the leaf xylem sap (ψπ-sap) at those levels of water stress suggest that discrepancies likely occurred due to leaf tissue damage and cell membrane disruption. These events are associated with tissue mortality and likely occur within the P50–P88 range of leaf xylem failure (Brodribb et al. 2019). This highlights an important limitation of the pressure chamber to determine very low leaf water potentials, at least for the 2 species tested, an herbaceous and a deciduous woody plant species. This technique appears to be less problematic for evergreen woody species (Rodriguez-Dominguez et al. 2018; Gauthey et al. 2020; Feng et al. 2023). In that sense, it would be plausible to expect that our results may be more likely observed in deciduous, with leaves in general more vulnerable to dehydration and with quicker decomposition rates, than in evergreen woody species, which tend to be thicker, more resistant to dehydration, and with higher longevity (Ye et al. 2022). In our case, when pressurizing highly water stressed leaves with the pressure chamber, higher ψleaf-eq than ψstem-PSY values (i.e. the sap at the petiole cut surface raised earlier than it should) likely indicated that the pressure applied facilitated the extraction of the damaged cell contents or disrupted cells, contributing both to the earlier appearance of the sap at the cut surface and to the increase in solutes of the extracted xylem sap (decrease in ψπ-sap). This interpretation is highly plausible since leaf tissue damage and cell membrane disruption are events linked to water status (Guadagno et al. 2017). The linkage between cavitation, which leads to embolism formation, and tissue damage has been recently demonstrated and represents a current hot topic within the plant science community attempting to reconcile the limits between cavitation, tissue damage, and drought-induced plant mortality (Brodribb et al. 2021; Mantova et al. 2022, 2023). It could be criticized that the first amount of extracted sap should come from the xylem, and hence, its ψπ-sap should be negligible. However, to ensure that enough sap was extracted, we needed to empty the xylem volume, which may have included mesophyll water, through the extraction procedure. Moreover, the damage is likely to occur in all living tissue, so parenchyma tissue in the xylem would also leak immediately into the xylem sap.

On the other hand, leaf discs taken from the same equilibrated leaves and measured with thermocouple psychrometers (Tyree et al. 2003; Kursar et al. 2009) appeared to more closely approximate the ψstem-PSY across the full range of water status measured in both species, supporting that the pressurization of highly stressed leaves was exacerbating the extraction of cell contents from presumably damaged leaf tissue. Still, differences started to appear in tomato at ψstem-PSY lower than leaf P50, whereas we did not find statistically significant differences in grapevine, although a trend of higher Δψ (leaf-disc) was also observed. The explanation for these results lies in the different sampling methodology applied to each species due to their different leaf morphologies. In the case of tomato, leaf discs for ψleaf-disc-eq measurements were taken from a physically separated basal leaflet from the compound leaf, whereas in grapevine, the leaf discs were sampled at the base of the same simple leaf. This sampling methodology, together with the fact that leaf tissue damage occurs heterogeneously in both species (Supplementary Fig. S2), implies a higher variability of ψleaf-disc-eq in grapevine than in tomato, due to the greater difficulty in avoiding sampling damaged leaf tissue. There is evidence that damage within the leaf can occur rapidly and be very heterogeneous in the leaf lamina (Brodribb et al. 2021). This means that some discs may have sampled living and dead tissue in the same leaf, leading to heterogeneous ψleaf-disc-eq. Furthermore, this heterogeneity would also explain that some leaf disc samples resulted in higher ψleaf-disc-eq than ψstem-PSY, although not statistically significant, i.e. during leaf equilibration, water would move from embolized leaf vessels to leaf areas not yet embolized (McElrone et al. 2013) and hold onto this water because of low conductivity to drier regions. Further studies exploring in detail these phenomena will help advance on how this heterogeneity influences water potentials at different regions within the leaf and how this is related to the leaf hydraulic sectoring.

Nevertheless, in previous studies, psychrometric leaf water potentials and pressure chamber leaf water potentials in tomato seemed to correlate better than our results (Barrs et al. 1970; Duniway 1971; Skelton et al. 2017). However, when carefully analyzed, those correlations were presented down to water potential levels where neither leaf xylem cavitation nor tissue damage was expected to be significant. In fact, Skelton et al. (2017) found in the same tomato variety as the one used in our study that leaf damage started to increase from those levels of water potentials (ca. −2 MPa), in agreement with the levels of ψstem-PSY at which ψπ-sap significantly decreased and ψleaf-eq more likely differed from ψstem-PSY.

Some particularities

It could also be suggested that, since ψstem-PSY and ψleaf-disc-eq measure total water potentials, correcting ψleaf-eq values by adding ψπ-sap should give very similar values to ψstem-PSY and ψleaf-disc-eq, being total water potential (ψ) = ψπ + ψp (ψπ, osmotic potential; ψp, pressure potential, or in our case ψleaf-eq). However, there are 2 aspects that explain why this is not reasonable for our results. First, ψstem-PSY is the total water potential measured at the xylem stem of the plant or branch (mainly dead cells), where the apoplast is less influenced by the living cells' content, so the agreement between ψleaf-eq and ψstem-PSY would occur when the osmotic potential of the xylem sap is very close to zero. Note that ψπ-sap refers to the osmotic potential of the leaf, not the stem, xylem sap. And second, ψleaf-disc-eq is the total water potential obtained from a small area of a leaf, whereas ψπ-sap, that we would use to correct ψleaf-eq, is obtained by extracting sap from the entire leaf. For dehydrated leaves in which cell disruption has possibly occurred, this means to extract sap that may contain a larger amount of cell solute contents, and hence, to result in a lower osmotic potential than would be expected for a smaller portion of the leaf (nondamaged leaf disc) and for correcting ψleaf-eq. That is, the values did not add up because cell rupture would add volume and osmolytes to the apoplast in a spatially nonhomogeneous way, violating the assumption of water potential equilibrium between symplast and apoplast within the leaf. It can also be observed that the very low ψπ-sap obtained are hardly common, mainly in tomato species, for which the literature suggests a range from −0.7 to −1.8 MPa for leaf osmotic potential under both well-watered and drought conditions (Garcia et al. 1996; Romero-Aranda et al. 2001; Bloom et al. 2004; Li and Liu 2022; Song et al. 2023). However, our experiments involved subjecting the species to extremely high water stress conditions, which are also uncommon for this species. Cells are shrivelled at zero turgor with low osmotic potentials, while dead cells leak fluid into spaces where it is concentrated by evaporation. Thus, both these events lead to low osmotic potentials in the fluid that is forced into the xylem. Very recently, it has also been reported that very low bud tissue osmotic potentials in pea (Ray et al. 2025). Further studies on these abrupt decreases in ψπ-sap will help to elucidate whether this could be used as an indication of imminent leaf death, similar to the abrupt increase in leaf abscisic acid levels observed across vascular plants on dying leaves (McAdam et al. 2022).

We also observed that ψπ-sap values before the P50–P88 range were not exactly zero, and some variability appeared for both species. This was likely due to the difficulty of accurately separating only the xylem from other living tissues. The leaf petioles of both species tested have large portions of parenchyma cells, and the xylem vessels are organized in bundles, which prevents the extraction of sap only from the xylem when the leaves are pressurized (Hochberg et al. 2016; Fathy et al. 2021; Cohen et al. 2022). Nevertheless, this limitation does not affect the reliability of our results since, once the leaf tissue reaches the P50–P88 range, leaf cell shrinkage and turgor loss point (TLP) have already occurred, so the contribution of petiole parenchyma cells to ψπ-sap in comparison to the extracted xylem sap would be negligible (Scoffoni et al. 2014).

Implications

At moderate levels of ψstem, both direct and indirect measures, as those used in the present study, would be acceptable to accurately approximate ψstem. However, once leaves begin to develop xylem embolisms and possibly tissue damage, the values of ψleaf-eq and, to a lesser extent, ψleaf-disc-eq become more inconsistent compared to ψstem-PSY, which may have important implications for the determination of xylem resistance metrics that are expected to occur at low ψstem, such as P88. We are certainly not attempting to discredit any of these methodologies for measuring ψstem, and even more when it does not occur in other species (see below). However, our data clearly illustrate the limitations that must be considered when using (i) leaf disc psychrometric measurements in plant species where leaf tissue damage during dehydration is heterogeneous, and (ii) the pressure chamber close to the leaf damage point. In this latter context, the appearance of large quantities of osmolytes in the xylem sap as cells presumably become damaged due to the high levels in xylem embolisms, clearly violates the assumptions of the balance pressure measurement. This is a critically important consideration for the growing number of plant scientists working with highly stressed plants, and it is not only relevant for indirect ψstem but also for direct measures of leaf water potential in herbaceous and deciduous woody species, which are highly likely to be affected.

Nevertheless, although the discrepancies between ψstem-PSY and ψleaf-eq occurred in grapevine, a woody deciduous species, they do not seem to occur in other woody species (Rodriguez-Dominguez et al. 2018; Gauthey et al. 2020; Feng et al. 2023), and thus, our results do not imply that all previous indirect ψstem measurements (equilibrated leaves measured with the pressure chamber) are wrong in the dry range for all species. In the case of studies performed in the past in other herbaceous species, such as rice, sunflower, or even tomato, linear and significant relationships between pressure chamber and psychrometric water potential measurements have been presented (Boyer 1967; Barrs et al. 1970; Duniway 1971; O’Toole and Moya 1981). However, comparisons were made using leaves (pressure chamber) vs. leaf discs (thermocouple psychrometer) with the objective of calibrating the former one, and, in the case of tomato or sunflower, measurements ranged from levels of water potentials in which leaf tissue damage was not expected. On the other hand, other studies have shown, in apple species, the opposite of what we observed in tomato and grapevine, i.e. pressure chamber leaf water potentials appeared to be lower than psychrometric leaf water potentials at a certain level of water stress (West and Gaff 1971). These authors suggested, as we also hypothesized but did not observe in our results, that the need to overpressurize highly stressed leaves came from the necessity to refill emptied xylem vessels due to cavitation, and leaf tissue disruption would presumably have occurred at lower water potentials. Similarly, as has been reported in grapevine, leaf petiole xylem cavitation occurs at higher water potentials than stem xylem cavitation (Hochberg et al. 2016), so it could be argued that petiole xylem cavitation could disconnect leaf and stem xylems, preventing equilibrium. However, in our case, grapevine ψleaf-eq was higher, not lower as hypothesized before, than ψstem-PSY, so the extra contribution of cell content would still be the most plausible explanation, even though the leaf disconnection could accelerate leaf dehydration. Therefore, we recommend being cautious when measuring highly dehydrated leaves (very low water potentials) with the pressure chamber in species in which leaf cavitation and tissue damage may occur within narrow ranges of water potentials, and to avoid measuring leaves with damaged tissue.

In our study, we have not only identified the ψstem levels at which caution should be exercised when measuring them indirectly with the pressure chamber (i.e. leaf cavitation, and possibly leaf tissue damage, thresholds), but also a method for identifying these thresholds by measuring ψπ-sap, which can be useful for a relatively quick species assessment. Importantly, these effects also have profound implications for the performance of pressure–volume (P–V) curves and on the metrics obtained from them. P–V curves are commonly performed by dehydrating leaves and measuring them repeatedly with the pressure chamber, together with their loss of water (Koide et al. 1989). Important metrics are derived from their analyses beyond the TLP, such as osmotic potential, apoplastic fraction, or turgor pressure (Bartlett et al. 2012). Since the probability of leaf cell damage increases after TLP (Mantova et al. 2023), accurate measures of leaf water potentials at those levels are fundamental. Alternatively, other plant sensors, such as the use of thermocouple psychrometers for measuring leaf discs at low leaf water potentials, the use of mechanical or optical dendrometry (Bourbia et al. 2021) that appears to be a very appropriate and recommended solution for monitoring precise leaf water potential dynamics (Bourbia and Brodribb 2023), or highly sophisticated techniques that use hydrogel nanoreporters (Jain et al. 2021), are potentially suitable and will help to deal with the limitations raised from other methodologies for measuring water potentials.

Conclusions

In tomato and grapevine, direct stem water potentials (ψstem-PSY) agreed with indirect water potentials of equilibrated leaves measured with the pressure chamber (ψleaf-eq) and leaf discs measured with thermocouple psychrometers (ψleaf-disc-eq) at moderate ψstem. However, when water potential fell below leaf xylem cavitation thresholds, which increase the probability of leaf tissue damage events, stem water potential could not be accurately estimated from leaves measured with the pressure chamber. Less error was evident when using leaf disc psychrometric measurements, but direct ψstem-PSY is still recommended. In our study, we have identified (i) levels of ψstem at which caution should be exercised when measuring leaves with the pressure chamber (i.e. levels of significant leaf xylem cavitation, and possibly leaf tissue damage), and, more practically, (ii) a method for approximating these levels by measuring the osmotic potential of the leaf xylem sap (ψπ-sap). As mentioned, since ψleaf-disc-eq measurements correlated better with ψstem-PSY, they appeared to be a suitable alternative for measuring low leaf water potentials, or indirect ψstem (only when the leaf remains hydraulically connected), in these species, considering, however, the occurrence of heterogeneous leaf tissue damage when interpreting the results. This study highlights the importance of understanding the complex mechanisms that occur during leaf dehydration when selecting the most appropriate method for measuring stem water potential at any water status level.

Materials and methods

Plant species and experimental design

The experiments were performed in an herbaceous species, Solanum lycopersicum L. (var. Rheinlands Ruhm), and a woody species, Vitis vinifera L. (cv. Cabernet Sauvignon). S. lycopersicum (tomato) measurements were performed at the glasshouse and laboratory facilities of the University of Tasmania (Australia) from May to September 2019, while V. vinifera (grapevine) measurements were performed at La Hampa experimental orchard and laboratory facilities of IRNAS-CSIC in Seville (Spain), from June to August 2022. Entire tomato plants were removed from their pots and dehydrated under glasshouse conditions simulating a drought-induced mortality experiment, and grapevine branches were bench dried in the laboratory. To make paired indirect and direct measurements of stem water potential, target compound leaves of tomato or simple grapevine leaves were maintained under nontranspiring conditions for at least 1.5 h while still attached to the plant or branch to allow leaf and stem water potential to equilibrate. This equilibration procedure was carefully applied by covering the leaves with an opaque plastic bag with a small piece of wet paper towel inside, avoiding contact with the leaf (Rodriguez-Dominguez et al. 2022). In the literature, it is commonly reported that the samples are only covered with aluminium foil. However, this may not fully stop transpiration, particularly on the tomato leaves under glasshouse conditions, where the leaf-to-air vapour pressure difference surrounding them may keep stomata open even in the dark (Caird et al. 2007). Equilibration times were selected as the most commonly reported in the literature to ensure equilibration between leaf and stem water potentials (Hochberg 2020; Knipfer et al. 2020; Li et al. 2021; Rodriguez-Dominguez et al. 2022). After this equilibration time, water potential comparisons between psychrometric stem water potential (ψstem-PSY), psychrometric leaf disc water potentials from equilibrated leaves (ψleaf-disc-eq), and pressure chamber water potentials of equilibrated leaves (ψleaf-eq) were performed (Table 1). In addition, the osmotic potential of the extracted xylem sap of the equilibrated leaf (ψπ-sap) was also measured as a proxy for leaf tissue damage.

Tomato drought-induced mortality experiment

Four tomato plants were grown under glasshouse conditions in pots with a potting mix medium comprising a 4 mix of composted fine pine bark and coarse washed river sand with added Scott's Osmocote Classic 14-14-14 fertilizer (Scotts-Sierra, Marysville, OH, USA). Glasshouse conditions were 15 °C (day:night temperatures), a photoperiod of 14 h with sodium vapor lamps continuously illuminating the plants from 00 to 00 h local time, and relative humidity matched the ambient (∼40%). Pots were irrigated daily until the beginning of the experiment. One plant was monitored at a time. The night before the monitoring started, the plant was fully watered. On the day of the experiment, the plant was removed from its pot to accelerate the dehydration process, keeping the roots intact with some soil attached.

A stem psychrometer (PSY1; ICT International, Armidale, NSW, Australia) was installed on the main stem of the plant to monitor ψstem-PSY. To do this, a segment of cuticle and bark, sufficient for fitting the psychrometer, was carefully removed by scraping the stem surface with a sharp razor blade. The psychrometer was then clamped to the stem, sealed airtight with high-vacuum silicone grease (Dow Corning Corp, Midland, MI, USA), and insulated with foam-rubber and aluminum foil to minimize the effect of temperature fluctuations on the psychrometric reading. Even so, as recommended by the manufacturer, differences in temperature between the sample and the chamber thermocouple (ΔT) lower than −1 and higher than 1 μV were not considered. The Peltier cooling time was adjusted to 5 s for the entire process, since the minimum ψstem-PSY measured was −3.25 MPa, ensuring a sufficient volume of water to be condensed onto the thermocouple, and then evaporated to produce a stable reading of the wet-bulb depression temperature.

During the dehydration process of the plants, leaves were covered every morning as described above, and measurements were performed from 00 to 00 (local time). One or more replicate measurements were performed per day on the same plant, depending on the speed of plant drying. The equilibrated leaf was collected by excising at the petiole, inserting it in the opaque bag that covered the leaf, and placing it inside another zip-lock plastic bag with an abundant wet paper towel to prevent any desiccation during storage (10 to 15 min). At the laboratory, a leaf disc was sampled with a cork borer (0.28 cm^2^) from one of the basal leaflets (Supplementary Fig. S2). The leaf disc was used for measuring ψleaf-disc-eq and the entire same leaf from which it was taken was used for measuring ψleaf-eq. We first took the leaf disc and placed it immediately into a second psychrometer chamber other than the one installed on the stem of the plant, sealed it with high-vacuum silicone grease to prevent moisture loss, and put the chamber within a polystyrene box to avoid temperature fluctuations. Equilibration times of 2 to 12 h were needed for stabilizing the readings, depending on whether the ψleaf-disc-eq values were lower or higher, respectively. We determined these times by measuring continuously every 30 min until the readings stabilized (Supplementary Fig. S3). While ψleaf-disc-eq stabilized and before measuring the leaf with a pressure chamber, we used a dissecting microscope to carefully and rapidly remove the bark (and the phloem) 0.5 to 1 cm back from the petiole cut using a razor blade to expose the xylem. This procedure was needed to prevent collecting sap from tissues other than the xylem. The cut end was then gently cleaned with distilled water and wiped dry to remove spilled cell contents and avoid contamination of solutes. This procedure took less than 30 s, and the compound leaf lamina remained covered within a plastic bag with a wet paper towel to prevent any extra desiccation. ψleaf-eq was then measured with a pressure chamber (Model 1515D; PMS Instrument Company, OR, USA), following the recommendations of Rodriguez-Dominguez et al. (2022). Once the endpoint (i.e. xylem sap appearing at the cut surface) corresponding to ψleaf-eq was reached, the leaf was slowly over-pressurized to extract the xylem sap. The over-pressurization value was on average 0.7 MPa higher than the endpoint value, with no apparent need to over-pressurize more, as ψstem-PSY and ψleaf-eq declined (Supplementary Fig. S4). Leaf xylem sap was collected by saturating a filter paper disc in contact with the xylem, ensuring the paper disc was totally soaked, which needed higher over-pressurizations in some cases. If this procedure (soaking the paper disc) was not possible, this measurement was discarded. The saturated paper disc was immediately placed into a third PSY1 psychrometer chamber, sealed with high-vacuum silicone grease to prevent moisture loss, and its osmotic potential measured (i.e. ψπ-sap). The Peltier cooling time was adjusted to 5 s (when ψπ-sap was higher than ca. −4 MPa) or to 30 s (when ψπ-sap was lower than ca. −4 MPa). This cooling time was determined after the first reading. However, in 4 out of 27 samples it was impossible to measure, probably due to their very low values (these ψπ-sap samples corresponded to ψstem-PSY between −2 and −3 MPa). Equilibration times of 20 to 60 min appeared to be enough for stabilizing the readings. To avoid fluctuations in temperatures during these equilibration times, the psychrometer chamber was placed within a polystyrene box.

Grapevine bench-drying experiment

We worked with 5-yr-old grapevines under field conditions at La Hampa experimental orchard in Seville (Spain). These grapevines were fully irrigated to cover their water needs during the entire 2022 season according to the crop coefficient approach (Allen 1998).

A similar procedure to that undertaken for the tomato experiment, with some modifications, was used for these vines. In this case, one ca. 1.5- to 2-m-long branch per vine was monitored at a time. The day of the experiment, a grapevine branch was collected early in the morning (from 00 to 00 local time), stored in a dark plastic bag with abundant wet paper towel, and transferred to the laboratory. A total of 18 grapevine branches were measured.

A stem psychrometer (PSY1; ICT International) was installed in the middle of the branch to monitor ψstem-PSY by the same procedure used for the tomato plants. The branch was then left to dehydrate under laboratory conditions. Since the minimum ψstem-PSY measured was higher than −4 MPa, the Peltier cooling time was adjusted to 5 s for the entire process. At different dehydration levels, grapevine water potential comparisons (ψstem-PSY, ψleaf-disc-eq, and ψleaf-eq) were performed following the same procedure as that described above for the tomato experiment. Since grapevine leaves are simple, leaf discs for ψleaf-disc-eq measurements were sampled from the base of the equilibrated leaf laminas (Supplementary Fig. S2). In this case, 2 ψleaf-disc-eq data-points were found to be exceptionally lower than ψstem-PSY (see Results section). According to methods for measuring osmotic potential of leaf discs (Kikuta and Richter 1992; Callister et al. 2006), these 2 samples, in which damage of leaf cells occurred naturally during dehydration, would more likely reflect leaf osmotic potentials, and hence, they were excluded from the analyses. For ψπ-sap measurements, the extracted leaf xylem sap was obtained by over-pressurizing the leaves 0.3 MPa on average. In this case, the Peltier cooling time was adjusted to 30 s for values below −4 MPa. Like in the tomato plants, no apparent dependence of this over-pressurization on the level of ψstem-PSY and ψleaf-eq was observed (Supplementary Fig. S4).

Leaf optical vulnerability curves

Leaf xylem vulnerability curves (i.e. the relationship between cumulative xylem embolism and water potential) for the same tomato cultivar used in this study have been previously published (Skelton et al. 2017). Since we did not have this information for the grapevine species, we constructed leaf xylem vulnerability curves in the present study using the optical method. During the dehydration of 4 grapevine branches, optical vulnerability (OV) curves (Brodribb et al. 2016) were also performed to derive leaf Pe, P50, and P88 values (Supplementary Fig. S1) from one leaf per branch, i.e. ψstem-PSY at the air-entry point (i.e. the point at which air begins to enter water-filled xylem vessels or the first cavitation events occur, leading to embolism formation), and ψstem-PSY causing 50% and 88% cumulative embolisms, respectively. Briefly, transmitted light images of intact grapevine leaves were captured using a custom-built OpenSourceOV (http://www.opensourceov.org) clamp. Leaves were securely clamped in the 3D-printed enclosure, and images were captured over time (every 5 min) through a 20× hand lens magnifier using a small 8-megapixel Raspberry Pi camera illuminated by 6 bright light-emitting diodes. The capture sequence was orchestrated using a PYTHON script (Python Software Foundation, Python Language Reference, version 2.7, Wilmington, DE, USA) running on a connected Raspberry Pi microcomputer (Raspberry Pi Foundation, http://www.raspberrypi.org). Image sequences were then analyzed using ImageJ according to (Rodriguez-Dominguez et al. 2018). Briefly, linear regressions were fitted to the psychrometer data (ψstem-PSY vs. time) to derive ψstem-PSY at the time of each image capture. ψstem-PSY was then plotted against cumulative embolisms (% of total) to construct the OV curves. Vulnerability curve data of each leaf were fitted to a sigmoidal regression as follows (Pammenter and Vander Willigen 1998): Cumulative embolisms = 100/(1 + exp(a (ψstem-PSY − P50))) (a, fitted parameter related to the slope of the curve). P88 was calculated by solving the equation for P88. Pe was calculated as the x-intercept of the tangent line drawn through the midpoint of the sigmoid function (Domec and Gartner 2001).

Data analyses

To relate the discrepancies between ψstem-PSY and either ψleaf-eq or ψleaf-disc-eq during dehydration with the decrease in ψπ-sap, we first calculated the difference (Δψ) between ψstem-PSY and ψleaf-eq, Δψ (leaf), and between ψstem-PSY and ψleaf-disc-eq, Δψ (leaf-disc). Then, we binned together the Δψ data within hydraulically and functionally relevant ranges of ψstem-PSY for each species, from no cavitation to Pe; from Pe to P50; from P50 to P88; and from P88 to the lowest ψstem-PSY achieved. We analyzed these binned Δψ data by performing a linear model with Δψ as the dependent variable, the bins as the independent variable, and no intercept, so the estimates of the bin effects had individual standard errors and P-values. This methodology allowed us to identify when Δψ was significantly different from zero.

Finally, we fitted exponential curves to the relationships between ψπ-sap and ψstem-PSY, and the ψstem-PSY at the maximum change of slopes of the ψπ-sap − ψstem-PSY exponential relationships were calculated using the lm.segmented function in the package Segmented (Muggeo 2008; RStudio v.3.4.0).

Supplementary Material