Early onset of stomatal closure confounds current interpretations and applications of iso‐/anisohydry theory

We critically assessed the iso-/anisohydry concept, which builds on the assumption of tight stomatal regulation of water potential during tree dehydration. We found a consistent pattern of early stomatal closure in mature trees of eight tested species that precedes the decline of water potential in advanced stages of tree dehydration. This calls for a revision of the iso-/anisohydry concept in which early stomatal closure prevents trees from entering an advanced stage of dehydration. Global observations of drought-induced tree mortality have encouraged intense research on plant hydraulic traits and derived metrics that can be used to quantify a tree's drought response strategy and define functional stress limits (Hartmann et al., 2018; Walthert et al., 2021; McDowell et al., 2022). Besides the discovery of various thresholds of plant hydraulic integrity (e.g. percentage loss of hydraulic conductance), the more than 80-year-old iso-/anisohydry theory (Walter, 1931) has received new attention. Essentially, it builds on a plant's dilemma of absorbing CO2 through opening its stomata, thereby running the risk of excessive water loss through transpiration. Based on this, plants have been divided into iso- or anisohydric drought response strategy types (e.g. Leuschner et al., 2021 and references herein). Broadly, isohydric trees adopt a conservative water use strategy at the expense of carbon gain due to strict stomatal regulation during hydraulic stress. Anisohydric trees, on the other hand, favour CO2-uptake with the risk of excessive water loss and hydraulic failure. As a result, isohydric trees target less negative water potentials (Ψ) than anisohydric trees and are therefore considered to be more resistant to drought (McDowell et al., 2008). In recent years, however, divergent definitions and terminology of isohydrocity have emerged (Martinez-Vilalta & Garcia-Forner, 2017; Ratzmann et al., 2019), leading to a variety of coexisting interpretations and thus complicating practical applications in comparative tree ecology (Hochberg et al., 2018; Leuschner et al., 2021). The concept of iso-/anisohydry is based on the popular assumption that gradual stomatal adjustments mediate a species-specific trade-off between transpirational water loss and CO2-uptake on a tree's path to dehydration (Tardieu & Simonneau, 1998; McDowell et al., 2008). This process is orchestrated by a multitude of chemical and hydraulic signals, acting on different levels of plant organization and environmental interaction on stomatal aperture and thus leaf transpiration and CO2-uptake (Comstock, 2002). It is further assumed that tight coordination of stomatal conductance (gs) and Ψ is the critical component in this regulatory network (Joshi et al., 2022), although the stringency of this interaction was recently questioned (Martinez-Vilalta & Garcia-Forner, 2017). Based on the role of stomata in regulating transpiration, species-specific gs vs Ψ relationships were constructed, which eventually led to the current theory of a continuum rather than a dichotomy of iso-/anisohydric behaviour across species (Klein, 2014). This work was an attempt to quantitatively describe the iso-/anisohydric behaviour of trees and develop physiologically based criteria for assessing their drought response strategy (Peters et al., 2023). However, to make the iso-/anisohydric concept operational, fully quantitative metrics for iso-/anisohydric behaviour are needed. Over the last decade, progress has been made in developing a conceptual framework that proposes new quantitative metrics of isohydrocity and thus drought resistance (Martinez-Vilalta et al., 2014; Meinzer et al., 2016; Hartmann et al., 2021). It assumes that the extent to which minimum water potentials at midday (Ψmd) deviate from preceding predawn water potentials (Ψpd) reflects stomatal regulation, with large deviations indicating low stomatal control. This raised the idea that the slope of the Ψpd vs Ψmd regression is a useful metric to describe the operational range for stomatal regulation under progressing drought (Fig. 1). Furthermore, this concept assumes stomatal closure at the point where Ψmd equals Ψpd. Based on these assumptions, a ‘hydroscape area’ (HA) is defined to quantify the stringency of stomatal control over Ψmd. The advancement of the HA metric lies in the fact that it integrates all processes controlling Ψmd over a large range of tree dehydration. Global analysis of Ψpd vs Ψmd relationships has shown that HAs can separate species across a broad range of functional groups and biomes (Fu & Meinzer, 2019; Salvi et al., 2022). Despite these promising results, there are reasonable concerns about the applicability of the isohydrocity concept for assessing a tree's drought response strategy, as the fundamental assumptions of tight stomatal regulation of Ψmd are not fully tested and studies on mature trees under natural droughts are largely lacking. This could explain the discrepancies in studies using different metrics of isohydrocity in different experimental setups to compare species-specific drought response strategies (Martinez-Vilalta & Garcia-Forner, 2017; Fu & Meinzer, 2019; Li et al., 2019). Here, we use extensive observations of Ψpd and Ψmd and midday measurements of gs over 3 yr in five angiosperm and three coniferous species in a mature temperate forest at the Swiss-Canopy-Crane-II site (Supporting Information Table S1; Peters et al., 2025a). The observation period covered the full spectrum of moisture conditions from wet to exceptionally dry (Table S2) in which trees came close to their previously reported thresholds of hydraulic integrity (Kahmen et al., 2022). Using these data, we constructed species-specific relationships of Ψpd vs Ψmd over a wide range of tree hydration (Fig. 2). In line with current theory (Fig. 1), the observed relationships reflect three phases of dehydration (Meinzer et al., 2016). Increasing deviations of Ψmd from preceding Ψpd, linearly disconnecting from the 1 : 1 line in the initial phase (I), declining deviations, linearly approaching the 1 : 1 line in the second phase (II) and small deviations following the 1 : 1 line in the third phase (III) (cf. Carpinus betulus and Fagus sylvatica in Fig. 2). Data from the second phase and data linearly extending into the first phase were selected to delineate HAs, assuming the slope of declining deviations of Ψmd from Ψpd defines the operational range of stomatal control over Ψmd under progressing drought (Table S3; the Materials and Methods section; Meinzer et al., 2016). In all species, the second phase started at a Ψpd of c. −1.0 MPa, while the endpoints, where Ψmd becomes nearly equal to Ψpd, varied. In the ring-porous angiosperms (Fraxinus excelsior, Quercus spec.), we were confronted with some uncertainty in defining the second phase of dehydration and delineating HAs, as the endpoints were not reached during severe drought, and data showed strong scatter or no linearity. Random subsampling analyses confirmed this uncertainty but showed little effect on relative species comparisons of HAs (Notes S1; Fig. S1; Tables S3, S4). Despite the uncertainty described above, we nevertheless calculated HAs to quantify drought response strategies for the investigated species (Fig. 2; Table S3). We were particularly interested in whether HAs were associated with wood functional classification or previously assessed ecological characteristics (Niinemets & Valladares, 2006). We obtained the largest HAs for the ring-porous angiosperms Quercus spec. and Fraxinus excelsior. Conversely, the diffuse-porous angiosperm Carpinus betulus and the conifer Pinus sylvestris showed the lowest HAs. Regarding ecological classification, the calculated HAs barely reflected the reported characteristic of the studied species. For example, the two species classified as the most drought-tolerant species showed the highest (Quercus spec.) and lowest (Pinus sylvestris) HA of all species (Table S3); and Fagus sylvatica and Picea abies, which severely suffered and died during recent drought spells (Schuldt et al., 2020; Arend et al., 2021, 2022), both displayed HAs in an intermediate range. This inconsistent species ranking shows, in agreement with previous concerns (Li et al., 2019), that the drought tolerance classification of trees goes far beyond a description of their iso-/anisohydric behaviour. To gain deeper insights into the mechanisms that shape the obtained HAs and better explain the inconsistent species ranking, we tested in each species how the Ψpd vs Ψmd trajectories related to stomatal regulation (Fig. 2). Similar to observations of Ψmd, we found large variations of gs in the initial phase of tree dehydration, which likely reflect stomatal sensitivity to daily fluctuations of incident light and vapour pressure. During the transition to the second phase of tree dehydration, gs dropped rapidly to low values and then remained at a level close to the instrumental detection limit. All eight species showed this early steep decline of gs, a mechanism that we have recently linked with growth regulation at the onset of drought (Peters et al., 2025a). To quantitatively describe these exponential relationships for each species separately, we employed asymptotic regressions and derived species-specific metrics that define a critical point at which stomata begin to close in response to declining Ψpd (Table S3). This point of Ψpd, after which gs declined rapidly, ranged between −0.4 and −0.8 MPa across the studied species. Below a Ψpd of c. −1 MPa, gs came close to the instrumental detection limit (which is within the range of previously detected stomatal closure; Peters et al., 2025a). Notably, the decline of gs was very steep in some species (Carpinus betulus, Abies alba), while it was more gradual in others (Acer pseudoplatanus; Fig. 2). Regardless of these differences in stomatal sensitivity (quantified as ρ; Table S3), we show that stomata closed much earlier than the equalization of Ψpd and Ψmd suggested by current iso-/anisohydry theory. With the finding described above, we failed to confirm the fundamental assumption of current iso-/anisohydry theory that the Ψpd vs Ψmd trajectory reflects the path to stomatal closure, which also contradicts the interpretation of HAs as an integrative measure of stomatal regulation. It is particularly remarkable that dynamic stomatal regulation of Ψmd was limited in all species to the beginning of tree dehydration when trees are not yet experiencing severe hydraulic stress and risk of hydraulic failure. This seems to be counter-intuitive, considering the function of stomata to regulate leaf transpiration and tree water status. However, similar observations have already been made in other stress-physiological studies (Arend et al., 2013; Bréda et al., 1993; Li et al., 2019; Arend et al., 2021; Aranda et al., 2024), and most recently, we found that early stomatal closure coincides with drought-induced growth cessation (Peters et al., 2025a). These observations of early stomatal closure during tree dehydration are scattered throughout the extensive plant-hydraulic and stress-physiological literature and thus received little attention in the current iso-/anisohydry theory. This may explain why our results conflict with the definition of the HA metric as a measure of stomatal function under progressing drought. If no further stomatal regulation is possible during tree dehydration, what other mechanisms shape the Ψpd vs Ψmd trajectories after stomatal closure? And why does Ψmd still deviate from Ψpd? Here, nonstomatal controls of Ψ may become relevant, for example cuticular transpiration (Duursma et al., 2018; Wang et al., 2024) or capacitive water release from shrinking stems (Salomón et al., 2017; Peters et al., 2023) or embolized sap wood (Salleo et al., 2000; Hölttä et al., 2009). Although not actively regulated, their effect on Ψ is closely linked to daily fluctuations of atmospheric water demand, which makes it plausible that Ψmd still deviates from Ψpd after stomatal closure. Given the large amount of water that trees save in their voluminous stems, capacitive water release from sapwood could play a particular role in buffering Ψ against rapid uncontrolled decline (Scholz et al., 2011). In this context, it is noteworthy that we obtained the largest HAs, and thus least stringent control of Ψmd, for the ring-porous angiosperms (Fraxinus excelsior, Quercus spec.), which have lower sapwood proportions than the other studied species. This may underline the importance of capacitive water release to stabilise Ψ in advanced stages of tree dehydration. Thus, the variability of HA in different species may depend not only on their specific stomatal regulation but also on their ability to buffer water loss by nonstomatal controls. Quantifying these nonstomatal controls would be a next step to truly disentangle the processes that shape Ψ during tree dehydration. Continuous stem diameter readings, from which tree water deficits and estimates of Ψ can be extracted, could facilitate this research (Ehrenberger et al., 2012; Peters et al., 2025b). Taken together, our observations provide physiological evidence for early onset of stomatal closure during tree dehydration. We found that stomata start to close rapidly when Ψpd falls below a value of −0.4 to −0.8 MPa. We therefore conclude that stomatal regulation serves to prevent the trees from entering an advanced stage of dehydration rather than to shape the further decline of Ψ afterwards. This contrasts sharply with current interpretations of iso-/anisohydry that assume dynamic stomatal regulation across the full range of tree dehydration but supports the recent finding of low correlation between gs and isohydrocity definitions (Martinez-Vilalta & Garcia-Forner, 2017). The limited range for dynamic stomatal regulation of Ψ calls for rethinking the physiological foundation of iso-/anisohydry theory. Therefore, the HA metric derived thereof should be interpreted with care and not used as a sole index of stomatal control over Ψ during tree dehydration, at least when comparing temperate species. This may finally explain the still existing discrepancies in attempts – including our own – to utilize quantitative metrics of iso-/anisohydry as measures of stomatal function and tree ecological classification. However, further studies on trees in other forest biomes and climate regions are required to derive a generally valid readjustment of iso-/anisohydry theory. All measurements were carried out at the Swiss-Canopy-Crane II (SCC II) research site in Hölstein/BL, Switzerland (47°26′17″N, 7°46′37″E; 550 m asl), which is situated in a mixed temperate forest of the Swiss Jura mountains. The site conditions are characterized by a calcareous loamy soil and a climate with 9.0°C annual temperature and 1009 mm annual precipitation. The technical infrastructure includes a canopy crane with a height of 45 m and a jib of 50 m. The upper tree layer of the site is dominated by adult individuals of European beech (Fagus sylvatica L.) and Norway spruce (Picea abies L.) with co-occurring oak (Quercus petraea Liebl. x robur L.), hornbeam (Carpinus betulus L.), Sycamore maple (Acer pseudoplatanus L.), European ash (Fraxinus excelsior L.), Scots pine (Pinus sylvestris L.) and Silver fir (Abies alba Mill.). All measurements were carried out in the summer months (June to August) 2020 to 2022. This period included strongly contrasting weather conditions as the exceptionally warm and dry summer months in 2020 and 2022, as well as the rather cool and wet summer of 2021 (Table S2). Our study thus covered the full spectrum of moisture conditions and tree hydration. Weather data for this period and the corresponding long-term trend (2003–2022) for comparison were collected from a nearby climate station operated by the Swiss meteorological service (station Rüneberg; 47°26′04″N, 7°52′45″E; 611 m asl). Water potentials were measured with a Scholander's type pressure chamber (PMS Instrument Company, Albany, OR, USA) on two excised twigs per tree, collected predawn (4:00 h to 6:00 h CET) and midday (12:00 h to 14:00 h CET) in the upper, sun-exposed canopy of the study trees. Hydroscape areas as metrics of isohydrocity were determined according to previously published work (Meinzer et al., 2016; Fu & Meinzer, 2019). In brief, Ψpd vs Ψmd relationships were inspected visually to preselect a range of the second phase of the dehydration trajectory in which the deviation of Ψmd from Ψpd dynamically declines with progressing drought. As statistical approaches were not applicable to the scattered and partly nonlinear data, we followed the following rules to define the final range for delineating the HA: the dry border of the second phase of the dehydration trajectory was set by the point where Ψpd became closest to or equal to Ψmd under severe drought. It was taken as the starting point for running a set of linear regressions with an increasing number of data pairs towards the wet border of the Ψpd vs Ψmd relationship where Ψmd shows large deviations from Ψpd. The regression yielding the highest R2 was used to describe a linear relationship between Ψpd and Ψmd. This best-fit approach allowed us to find a regression line for delineating the HAs that covers the wet to dry range of the dehydration trajectory (first and second phase of the dehydration trajectory) but excludes data in the wet range that was likely influenced by fluctuations of incident light, temperature or atmospheric vapour pressure deficit (vertically oriented trend at the beginning of the dehydration trajectory). The area of the triangle bound by the regression line, 1 : 1 line and the y-axis of the Ψpd vs Ψmd plot was expressed as HA (MPa2). It was dependent on the slope of the obtained regression line, which was used as an additional test metric of isohydrocity (Table S3; Martinez-Vilalta et al., 2014). In the two ring-porous species Quercus spec. and Fraxinus excelsior, the delineation of HAs by linear regression was subject to some uncertainty as the dry endpoint of the second phase of the Ψpd vs Ψmd relationship was not reached or showed strong scatter. In Quercus spec., nonlinearity of the Ψpd vs Ψmd relationship additionally compromised the employment of linear regression. We therefore conducted additional random subsampling analyses with the full data (except the few data points in Phase III of the Ψpd vs Ψmd relationship; Notes S1; Fig. S1; Table S4) to test the uncertainty of the HAs and the resulting species ranking. We found low-to-moderate uncertainties in conifers and diffuse-porous angiosperms, respectively, but high uncertainty in the two ring-porous angiosperms Fraxinus excelsior and Quercus spec. However, the ranking of the HAs across the species was robust (Fig. S1; Tables S3, S4). Instant leaf gas exchange was measured using a portable photosynthesis system (LI-COR 6800, LI-COR Inc., Lincoln, NE, USA) equipped with an illuminated leaf or needle chamber. All measurements were taken in the upper, sun-exposed canopy at the same time when Ψmd was measured (12:00 h to 14:00 h CET). The conditions inside the chamber were kept at 1000 μmol photon flux density and 500 μmol s−1 flow rate with ambient CO2, temperature and humidity. Variable needle mass inside the chamber was corrected by measurements of needle area using a bench-top leaf area metre (LI 3100C–LI-COR Inc., Lincoln, NE, USA). We thank Andre Kühne for operating the canopy crane and several internship students for helping us with the fieldwork. This work was supported by funding from the Swiss Federal Office for the Environment. None declared. MA, GH and AK designed the MA, and collected the data in the MA, and the the with from all All data are included in the and Information Table Fig. Random subsampling analysis of Notes Random subsampling analysis of Table tree water potential and stomatal Table Weather conditions in the observation Table and ecological classification of the studied tree species to metrics of isohydrocity and stomatal regulation. Table from the subsampling analysis across all subsampling is not for the or of Information by the than should be to the The is not for the or of by the than should be to the corresponding for the The with to in and in