Leaf physiological and morphological constraints of water-use efficiency in C3 plants

Abstract The increasing evaporative demand due to climate change will significantly affect the balance of carbon assimilation and water losses of plants worldwide. The development of crop varieties with improved water-use efficiency (WUE) will be critical for adapting agricultural strategies under predicted future climates. This review aims to summarize the most important leaf morpho-physiological constraints of WUE in C3 plants and identify gaps in knowledge. From the carbon gain side of the WUE, the discussed parameters are mesophyll conductance, carboxylation efficiency and respiratory losses. The traits and parameters affecting the waterside of WUE balance discussed in this review are stomatal size and density, stomatal control and residual water losses (cuticular and bark conductance), nocturnal conductance and leaf hydraulic conductance. In addition, we discussed the impact of leaf anatomy and crown architecture on both the carbon gain and water loss components of WUE. There are multiple possible targets for future development in understanding sources of WUE variability in plants. We identified residual water losses and respiratory carbon losses as the greatest knowledge gaps of whole-plant WUE assessments. Moreover, the impact of trichomes, leaf hydraulic conductance and canopy structure on plants’ WUE is still not well understood. The development of a multi-trait approach is urgently needed for a better understanding of WUE dynamics and optimization.


Introduction
Water-use efficiency (WUE) reflects a balance between carbon gain and water loss in plants, introduced more than 100 years ago by Briggs and Shantz (1913). Since then, multiple ways and methods to assess WUE at a different level of organization and temporal resolution were developed and conceptualized (Vadez et al. 2014(Vadez et al. , 2023Hatfield and Dold 2019;Brendel 2021). Two WUE parameters reflect a momentary state of leaf carbon and water fluxes: intrinsic water-use efficiency (WUE i ) as a ratio of CO 2 assimilation rate (A n ) to water vapour stomatal conductance (g s ), obtained during gas-exchange measurements at leaf level (Petrik et al. 2022a). Another closely related variant, instantaneous WUE i , is calculated as a ratio of A n and leaf transpiration (Bacon et al. 2004). Other WUE parameters capture the long-term balance between carbon fixation and transpiratory water losses. Biomass-based indices include whole-plant WUE bio as the ratio of biomass accumulation to cumulative transpiration of the plants ( Condon et al. 2004;Brendel 2021). Furthermore, yield WUE is usually calculated as crop yield per hectare divided by total transpiration or evapotranspiration (Hatfield and Dold 2019;Zahoor et al. 2019). The use of growth-based WUE calculated as the ratio of annual basal area increment and cumulative annual transpiration is used in dendrobiology (Szatniewska et al. 2022). Moreover, the carbon isotope ratio (δ 13 C) has been extensively used as a proxy of long-term WUE 13C , because of the preference for the lighter isotope during physical and chemical processes involved in CO 2 uptake and assimilation (Farquhar et al. 1989;Frank et al. 2015;Ma et al. 2023).
Ecosystem-wide WUE derived from eddy-covariance measurements (WUE GPP ) is a ratio between gross primary production (GPP) of the ecosystem and total cumulative transpiration or evapotranspiration (Yi et al. 2019). WUE GPP can be also derived from remote sensing data as the GPP to evapotranspiration ratio (Ahmadi et al. 2019). Overall, the individual-level, long-term (vegetation season) based WUE bio is the most precise assessment of real resource utilization of plants as they capture both assimilatory and respiratory balance with productive and unproductive water losses (Brendel and Epron 2022). WUE bio should thus be more commonly used as the standard WUE estimates in agricultural and plant sciences, instead of the WUE i , which is much easier to measure but represents only one point in time.
The importance of WUE acclimation in plants is due to raising evaporation demands caused by climate change and possible frequent water-deficit stress during seasonal droughts (Ponce-Campos et al. 2013;Schuldt et al. 2020). Plants with higher WUE will have a competitive advantage in natural ecosystems and economic significance for agricultural production. The momentary WUE i of plants can be improved either by lower transpiration losses or higher efficiency of carbon assimilation (Flexas et al. 2016;Hatfield et al. 2019). Understanding of constraining factors of WUE is crucial for crop optimization efforts and the correct assessment of adaptive responses of plant communities (Quan et al. 2020;Kang et al. 2021). WUE variability is affected by multiple morphological and physiological traits (Figure 1). The size and density of stomata affect the maximal stomatal conductance and stomatal responsiveness to environmental changes (Nunes et al. 2022;Pitaloka et al. 2022). As stomatal morphology and anatomy can be altered with biotechnological methods for improved WUE, it is a great target for future research (Caine et al. 2019;Li et al. 2020). The responsiveness of stomata to fluctuating light and drought can also improve long-term WUE bio (Xylogiannis et al. 2020;Zhao et al. 2021a). Several studies have found a negative correlation between WUE estimates and leaf hydraulic conductance (Wedegaertner et al. 2022; Barrera-Ayala et al. 2023;Liu et al. 2023), but these findings are still inconclusive (Corcuera et al. 2012;Sellin et al. 2014;Jin et al. 2016) and we need a better causal explanation of the relationship. Another important constraint of WUE is the mesophyll conductance (g m ) of CO 2 towards Rubisco (Flexas et al. 2016;Zhu et al. 2021). Maximization of the g m /g s ratio was suggested as a possible goal for improving WUE of crops (Flexas et al. 2013a;Fullana-Pericàs et al. 2017). The next step of WUE improvement is an optimization of Rubisco carboxylation efficiency (Flexas et al. 2016). The long-term WUE bio enhancement could be further achieved by the reduction of respiratory losses and residual water losses during night or drought (Escalona et al. 2012;Coupel-Ledru et al. 2016). Finally, leaf anatomy, which influences both mesophyll conductance CO 2 and transpiratory losses, can also alter plant WUE (Bramley et al. 2013;Trueba et al. 2022).
The objective of this review paper was to summarize various morphological and physiological factors, which influence WUE in plants, as a stepping stone for a more holistic approach to the multi-factor assessment of WUE constraints ( Figure 1). We also focused on identifying under-represented physiological and morphological traits in current research, which are needed for understanding WUE optimization in plants. Moreover, this review focuses specifically on WUE i , WUE 13C and WUE bio to provide the most possibly concise overview of this complex topic at a similar spatial scale. It is worth pointing out that environmental factors such as water availability (Amitrano et al. 2019;Zhao et al. 2021b), soil structure (Hatfield et al. 2001;Rabarijaona et al. 2022), air pollution (Hatfield et al. 2001;Rabarijaona et al. 2022) and nutrients (Dijkstra et al. 2016;Gharun et al. 2021;Song et al. 2022) can also have a significant impact on WUE. However, this falls beyond the scope of the study and is therefore not further discussed.

Stomatal density and trichomes
Plants can influence their transpiratory losses and therefore potentially their WUE via stomatal regulation (Hetherington and Woodward 2003;Bertolino et al. 2019). The stomatal adjustment could include changes in stomatal density (SD), stomatal anatomy (size, shape) and stomatal control mechanisms (Sack and Buckley 2016;Petrik et al. 2022b). Multiple recent studies, which used genetic manipulation methods to alter SD, have reported improved WUE i connected to the reduction of SD. A genetic manipulation (EPF2OE) approach in a study by Franks et al. (2015) led to Arabidopsis mutants with lower SD that showed higher WUE i and long-term WUE 13C due to lower stomatal conductance of water vapour (g s ) but unchanged photosynthetic capacity. Similarly, a combination of high-yield rice cultivars with overexpressed OsEPF1 epidermal patterning factor (EPF) led to a reduction of SD, lower g s , improved WUE i and overall drought tolerance (Caine et al. 2019). The EPF overexpression in bread wheat has led to similar results of reduced SD and improved WUE i , without yield losses (Dunn et al. 2019). Guo et al. (2019a) have reported the genetic pathway of EDT1/ HDG11, ERECTA, and E2Fa loci, which regulates WUE i of Arabidopsis via modulation of SD. Overexpression of SlTLFP8 (Tubby-like F-box protein 8) reduced SD by 10-20 % in tomatoes and was connected to enhanced WUE i . Similarly, repression of PuGTL1 via Pu-miR172d overexpression led to a reduction of SD and higher WUE i in Populus ussuriensis (Liu et al. 2021). On the other hand, overexpression of STOMAGEN led to higher SD, greater photosynthetic activity (+30 %), but also greater transpiration (+100 %), which resulted in reduced WUE i (Tanaka et al. 2013). Contrary, the study by Bhaskara et al. (2022) also reported a positive relationship between SD and WUE bio derived from natural variation in Arabidopsis accessions. Moreover, other leaf structures such as trichomes (trichomes/SD ratio) can play a significant role in WUE i and WUE bio enhancement via lower transpiratory losses due to leaf-air boundary layer resistance (Mo et al. 2016;Galdon-Armero et al. 2018). For example, Chen et al. (2022) observed a doubling in trichome density and a decline in g s by 85 % between droughted and well-watered Shepherdia × utahensis plants. Single gene manipulation efforts, such as EPF2OE, could have negative pleiotropic effects on other metabolic processes and should be further explored to avoid these negative side effects (Flexas et al. 2016;Husaini et al. 2022). It seems that the reduction of SD for improving WUE i and WUE 13C/bio could be a viable option for plant breeding initiatives. Additionally, the incorporation of further leaf structures, such as trichomes, in combination with SD can improve our understanding of WUE constraints.

4
AoB PLANTS, 2023, Vol. 15, No. 4 Stomatal size and responsiveness Stomatal control mechanisms include reaction to atmospheric vapour pressure deficit (Grossiord et al. 2020), plant water potential (Buckley 2005(Buckley , 2019Dayer et al. 2020), light conditions (Lawson et al. 2010;McAusland et al. 2013) and CO 2 concentration (Franks and Beerling 2009). Photosynthetic activity of C 3 plants can adjust in seconds to changes in irradiance, but the lag in stomatal responses limits the CO 2 uptake and therefore constrains photosynthesis and limits WUE (Lawson et al. 2012). Several studies have reported that smaller stomata respond faster than larger stomata to changes in environmental conditions (Drake et al. 2013;Lawson et al. 2014;Kardiman and Raebild 2018;Durand et al. 2019). Faster stomatal response in the study by Lawson et al. (2014) has been linked to higher WUE i values under naturally changing irradiance levels. Theoretical maximal stomatal conductance (g max ) showed a negative correlation with stomatal size, but smaller stomata showed faster response time to variable irradiance in five Banksia species (Drake et al. 2013). A study by Lei et al. (2023) also found that larger stomata of domesticated rice showed slower response time to fluctuating light and overall lower WUE 13C . The genetic manipulation study in rice has found that mutants with small stomatal size showed higher WUE i , in comparison to mutants with greater stomatal size (Pitaloka et al. 2022). Des Marais et al. (2014) found that Arabidopsis genotypes with larger stomata due to AtMPK12 substitution showed lower WUE i compared with the common allele. The improved WUE i of wheat cultivars under waterdeficit stress was linked to smaller stomatal size, lower SD and reduced transpiration rates (Li et al. 2017). A study by Amitrano et al. (2021) showed that a 49 % increase in WUE i and WUE bio of lettuce has been associated with a reduction of stomatal size under different vapour pressure deficit (VPD) treatments. Moreover, drought stress exposure inhibited stomatal development (smaller stomata) and increased the WUE i in cotton (Dubey et al. 2023). On the other hand, a study by Xiong and Flexas (2020) on ferns, gymnosperms and angiosperms found a negative correlation between stomatal size and g m , therefore possibly limiting WUE. A comparison of Quercus robur genotypes has found a positive correlation between guard cell length and WUE 13C , contradicting the majority of results suggesting that smaller stomata promote higher WUE 13C (Roussel et al. 2009). Liu et al. (2018a) have found a quadratic relationship between stomatal size and WUEi at the community level, across forest ecosystems along the latitudinal transect, with an optimal stomatal size of approximately 400 μm 2 . Smaller stomatal size could be connected to higher WUE in plants, presumably due to faster response to environmental conditions. Nevertheless, there is probably an optimal stomatal size and further reduction can be detrimental due to CO 2 limitations of photosynthesis.

Stomatal control and light sensitivity
Excessive water loss under an impaired state of photosynthetic apparatus (drought, salinity stress) can negatively affect the WUE of plants. Timely stomatal closure is then another major component of WUE optimization of plants under water-deficit stress Hartmann et al. 2021). A study by Yi et al. (2019) showed that WUE 13C of isohydric species was generally more sensitive to environmental change due to their conservative water potential regulation strategy than WUE 13C of the anisohydric species and increased significantly with rising VPD during periods of water stress. The accumulation of abscisic acid (ABA), which drives the stomatal closure of plants under water deficit, can be considered a key factor for both WUE i and WUE 13C/ bio improvement in plants (Negin and Moshelion 2016;Guo et al. 2019b;Mukarram et al. 2021). Plants capable of finetuning their stomatal control with ABA can possess an enhanced WUE i with sustained biomass or yield gains (Yoo et al. 2009;Yao et al. 2021). Improved WUE i in the presence of elevated ABA levels has been demonstrated in transgenic Arabidopsis (Zhang et al. 2008) and tomato (Thompson et al. 2007;Lamarque et al. 2020). Exogenous application of ABA showed enhanced WUE i and WUE 13C in Populus davidiana (Li et al. 2004) and Marsilea crenata fern (Tai-Chung et al. 2020). French bean and sugar beet plants pretreated with ABA also showed improved WUE i under water-deficit stress (Pospíšilová and Baťková 2004). Enhanced stimulation of ABA signalling of Arabidopsis via distinct ABA receptors can result in constitutively high WUE i . WUE 13C of Arabidopsis and wheat was also enhanced by modulating ABA responses either by using overexpression of specific ABA receptors or deficiency of ABA coreceptors (Yang et al. 2019). ABA receptors from Populus canescens were stably introduced into Arabidopsis in a study by Papacek et al. (2019), which led to enhanced WUE i . Moreover, overexpression of PeJAZ2 increased WUE i of poplar under drought stress by regulating ABA signalling rather than ABA synthesis (Rao et al. 2023). Partial root-zone drying can generate a root-toshoot pressure signal from the dry part of the root zone that also promotes stomatal closure via a drop in cell turgor and enhances WUE i via ABA utilization (Davies et al. 2002;Pérez-Pérez et al. 2012;McAdam and Brodribb 2016;Zhang et al. 2018;Xylogiannis et al. 2020). These results, therefore, suggest great opportunities for WUE optimization in crops with the use of transgenic methods, breeding efforts and biotechnological tools for ABA utilization.
Stomatal sensitivity to light could be another important determinant of plant WUE i by adjusting the magnitudes of change in g s as a function of the environment . Part of the stomatal response involves the balance between photosynthetic electron transport and carbon reduction either in guard cells, chloroplasts, or in the mesophyll (Messinger et al. 2006). Overexpression of Photosystem II Subunit S in tobacco led to lower stomatal opening in response to light, which resulted in a 25 % reduction of water loss and improved WUE i (Glowacka et al. 2018). The desynchronization of A n and g s can lead to a surplus in transpiration when A n is low but g s is high (e.g. transition from high to low light), hence reducing WUE i (McAusland et al. 2016;Coupel-Ledru 2021). The introduction of a blue light-activated K + ion channel, named BLINK1, to Arabidopsis, led to a faster reaction of stomatal aperture under both increasing and decreasing irradiance, which ultimately enhanced the plants' biomass accumulation and WUE bio (Papanatsiou et al. 2019). Dynamic plant response to VPD and light fluctuations under natural conditions were suggested to increase plants WUE bio (Gosa et al. 2019). Lower stomatal openness and lower g s under short-term light transitions led to higher WUE i in chilli pepper treated with "smart glass" compared to the control group (Zhao et al. 2021a). A study by  found that overexpression of OE-PtrVCS2 in Populus trichocarpa led to smaller stomatal aperture under drought stress and overall higher WUE i than in the wild type. Greater WUE i of isohydric Pine species has been also linked to lower stomatal openness under increasing light, while anisohydric Oak species behaved more opportunistically with lower WUE i (Renninger et al. 2015). Reduction of stomatal openness as a reaction to light changes can probably improve the WUE of plants but can lead to a reduction of the total growth and yield of crops. Nevertheless, improving stomatal response time to changing irradiance levels can improve the plants' WUE without a negative impact on assimilation and growth.

Residual and nocturnal conductance
When the stomata are closed (night, drought), plants are still losing water via their cuticle, bark or incompletely closed stomata (Duursma et al. 2019;Lintunen et al. 2021). Cuticular transpiration has been recognized as a significant factor affecting drought survival rates (Duursma et al. 2019) and might affect WUE 13C/bio due to residual transpiration (Ni et al. 2012;Ávila-Lovera et al. 2019). Minimum leaf conductance (g min ) incorporates water loss across the leaf cuticle, bark and through the incompletely closed stomata (Schuster et al. 2017;Blackman et al. 2019;Duursma et al. 2019;Lintunen et al. 2021). Minimization of these residual losses during periods of reduced assimilation rate due to stomatal limitations can therefore lead to improved long-term WUE 13C/bio (Sevanto 2020). The water loss from leaves of plants under drought is dominated by g min after stomatal closure. This has been related to the thickness of the cuticular wax layer, which increases in response to water deficit (Jeffree 2006;Shepherd and Wynne Griffiths 2006;Bueno et al. 2020). However, a relationship between the thickness of the cuticular wax layer and g min can be insignificant, both within (Anfodillo et al. 2002;Bueno et al. 2020) and across species (Riederer and Schreiber 2001). The variability of g min can be also driven by stomatal morphology (leaky stomata) or chemical composition of cuticle (Duursma et al. 2019;Machado et al. 2021). In a recent study across 23 genotypes of wheat, cuticular transpiration showed a strong positive correlation with water loss per dry mass unit, which the authors considered as a proxy for WUE bio (Gašparovič et al. 2021). A modelling simulation approach by Duursma et al. (2019) revealed a theoretical reduction of WUE i under increasing g min of plants using the general Ball-Berry model of stomatal conductance. Moreover, hydroponically grown Festuca arundinacea exposed to salinity treatment showed enhanced WUE i and lower g min compared to the control group (Vandegeer et al. 2021). On the other hand, eucalyptus clones under water-deficit treatment showed significant intra-specific differences in cuticular conductance but not in WUE i (Carignato et al. 2019). A study by Clarke et al. (1991) also found no significant correlation between minimal conductance and long-term WUE bio in wheat under drought stress. The impact of cuticular conductance or g min on WUE has not been yet properly quantified and is therefore a great target for future research.
The analogical parameter, nocturnal conductance, is also critical for optimization of long-term WUE 13C/bio (Coupel-Ledru et al. 2016;Even et al. 2019). Excessive water losses during the night (Dawson et al. 2007;Forster 2014) decrease long-term WUE as there is no photosynthetic gain during the night. It has been suggested that the low nocturnal conductance of shade-tolerant plant species is consistent with their conservative water-use strategy (Resco de Dios et al.

2019
). Nocturnal conductance is usually dominated by cuticular transpiration, but incomplete stomatal closure during the night has been observed in C 3 plants (Caird et al. 2007;Escalona et al. 2012). Reduction of night transpiration can theoretically improve the WUE bio of crops without growth penalties (Tardieu et al. 2022). A study by Dayer et al. (2021) has shown that night transpiration was linked more to the specific circadian rhythm of the wine cultivars rather than environmental conditions, suggesting strong genetic control. Night transpiration also had a significant impact on total transpiration and WUE bio in a study by Medrano et al. (2017) and was recognized as one of the under-explored factors affecting whole-plant WUE. Nocturnal conductance also showed a significant negative correlation with WUE bio among black poplar genotypes exposed to drought stress (Bogeat-Triboulot et al. 2019). Differences in the night transpiration between Pinus contorta thinning treatments corresponded to differences in WUE under water-deficit stress . Further quantification of the night transpiration effect on the long-term WUE of plants is needed for a proper understanding of the phenomenon. Selection for plants with low cuticular conductance and conservative stomatal control (avoiding leaky stomata) can greatly improve their WUE and drought resistance.

Leaf hydraulic conductance
Leaf hydraulic conductance (K leaf ) can be coordinated with higher WUE i , as observed in several studies (Fichot et al. 2009;Andrade et al. 2022;Wedegaertner et al. 2022). Nevertheless, it is still unknown if the plants with higher WUE develop smaller xylem vessels causing lower K leaf (but greater xylem embolism resistance, cf. Isasa et al. 2023) as they have lower hydraulic requirements to maintain leaf gas exchange, or the lower K leaf leads to greater WUE by constraining water supply in leaves. K leaf is tied to leaf assimilation and stomatal conductance rate in a positive linear fashion (Santiago et al. 2004;Sellin et al. 2014). Reduction of leaf hydraulic conductance via gene manipulation can lead to lower water losses but is also tied with a proportional reduction of assimilation rates and therefore non-significant changes in WUE bio (Zsögön et al. 2015). The environmental response of K leaf and its impact on WUE has received more attention in recent studies and has been identified as a major trait that could constrain WUE under changing VPD (Flexas et al. 2013a;Xiong et al. 2018). However, no consensus has been reached to date regarding the direction of the relationship between K leaf and WUE. On one hand, Yao et al. (2021) reported that raising WUE i of Caragana sp. with decreasing water potential was coordinated with decreasing K leaf but also rapid biosynthesis of ABA. The Solanum species with significantly lower K leaf showed also significantly higher WUE 13C under well-watered conditions (Barrera-Ayala et al. 2023), while WUE i of Ginkgo biloba was also negatively correlated with K leaf . Warming treatment in four subtropical tree species led to higher K leaf but lower WUE 13C ). On the other hand, Jin et al. (2016) reported a positive relationship between K leaf and WUE i among 10 temperate tree species. Similarly, a positive correlation between WUE 13C and K leaf was reported for Pinus pinaster populations exposed to drought stress (Corcuera et al. 2012). Moreover, Sellin et al. (2013Sellin et al. ( , 2014 found no significant correlation between WUE i and K leaf in birch and aspen trees. In conclusion, the direction of the K leaf -WUE relationship is unclear, and further work must be conducted to assess whether breeding for lower K leaf to reduce water losses possibly leads to improved WUE without a significant reduction of growth. Future experiments with gene manipulation techniques that will not affect other physio-morphological traits are needed to understand the causal link of these correlations.

Mesophyll conductance
Improving CO 2 diffusion to the sites of carboxylation without increasing stomatal conductance can enhance WUE i . This requires improving mesophyll conductance to CO 2 (g m ) and it has been proposed that the ratio g m /g s is a relevant breeding trait for improving WUE (Galmés et al. 2011;Flexas et al. 2013b;Tomás et al. 2014a;Flexas 2016). The g m has been recognized as one of the main limiting factors of WUE in both crops (Leakey et al. 2019) and tree species (Zhu et al. 2021), potentially due to the close coupling of g m and K leaf as both share the same pathways of water movement in leaves (Flexas et al. 2013b;Xiong et al. 2017). A close positive relationship has also been observed between g m and g s although the reason for this remains speculative (Guiliani et al. 2013;Barbour and Kaiser 2016). However, a study by Fullana-Pericas et al.
(2017) showed a strong positive correlation between g m /g s and WUE i in Mediterranean tomato landraces. Similarly, WUEi showed a strong positive correlation with g m /g s in tobacco under chloride nutrient treatments (Franco-Navarro et al. 2019). The variability of g m has been linked to leaf anatomy, where cell wall thickness, membrane permeabilities, cytosol and stromal conductance were constraining factors of g m (Terashima et al. 2011;Tomás et al. 2013;Ouyang et al. 2017). The cell wall conductance to CO 2 can be influenced by cell wall thickness, porosity and tortuosity Ellsworth et al. 2018). A study by Roig-Oliver et al. (2020) found a strong negative correlation between cellulose and g m in grapevine. The hemicellulose to pectin ratio of the cell wall correlated positively with the g m of tobacco exposed to drought and salinity stress (Clemente-Moreno 2019). Tholen et al. (2008) manipulated the chloroplast arrangement in Arabidopsis and thus modified g m through changes in the surface of chloroplasts exposed to the intercellular air spaces (Sc/S). The positive impact of Sc/S on g m and A n has been observed also for Mediterranean oak species (Peguero-Pina et al. 2017), rice (Xiong et al. 2017) and tobacco (Clarke et al. 2021). A recent study by Baillie and Fleming (2020) has found that coordination of stomatal and mesophyll development is crucial for the optimization of g m and therefore WUE. Findings to date suggest that certain stomatal development signalling components, such as TMM, ER and STOMAGEN, may be required for interlayer coordination, and that gas exchange may also regulate mesophyll structure (Dow et al. 2017). Acclimation of g m to changing environmental conditions has been linked to aquaporins and carbonic anhydrase (Flexas et al. 2006;Warren 2007). The g m can be affected by specific genes (e.g. aquaporin NtAQP1, HvPIP2, AtBBX21) and thus targeted by genetic manipulation of crops ). Overexpression of aquaporin genes led to increased g m (Hanba et al. 2004) and inhibition of lower g m in various crops (Flexas et al. 2006). Tobacco aquaporin NtAQP1 aids the trans-membrane transport of CO 2 in plants and thus contributes to the CO 2 permeability of the plasma membrane of the mesophyll cells (Uehlein et al. 2003). Carbonic anhydrase activity has been positively correlated to g m (Price et al. 1994;Momayyezi and Guy 2017) and chloroplast fraction of g m (Gillon et al. 2000). Carbon anhydrase accelerates the interconversion of the dissolved inorganic carbon species, CO 2 and HCO 3 -, which helps optimize the initial stages of photosynthesis. A recent study by Gómez-Ocampo et al. (2021) found that overexpression of AtBBX21 led to enhanced g m and J max , coupled with higher WUE in potato plants under drought. Moreover, manipulation of heterotrimeric G protein signalling can improve plants' WUE i and productivity due to higher g m rates under drought conditions (Zait et al. 2021). More specifically, the canonical Gα (RGA1) subunit gene of G protein regulated gm in rice, which was reflected in improved photosynthetic capacity and overall WUE (Wang and Botella 2022). The optimization of g m and therefore WUE is multifaceted and incorporates multiple organizational levels from cell biochemistry to whole leaf anatomy. There is also great intra-specific variability of g m across crops (Tomás et al. 2014a;Chen et al. 2021) and trees (Momayyezi and Guy 2017;Peguero-Pina et al. 2017) and therefore, it is a reasonable target for breeding efforts which aim at maximizing WUE. Nevertheless, the practical performance of the population/individual's selection could be hindered by the low reliability of current g m measurements (Pons et al. 2009;Lundgren and Fleming 2020). The development of more precise g m measurement techniques (Márquez et al. 2023) could greatly improve the understanding of WUE constraint by g m . Furthermore, the strong coupling of g m with K leaf (Flexas et al. 2013;Xiong et al. 2017) and g s (Guiliani et al. 2013;Barbour and Kaiser 2016) might impede efforts to improve WUE i through modification of g m . As shown by Pathare et al. (2023) using rice cell wall mutants, modifying g m indeed increases photosynthetic capacity but at the cost of simultaneously increasing g s , resulting in no overall change in WUE i .

Carboxylation rate
Another target to achieve improved photosynthesis is to improve the biochemical capacity for CO 2 assimilation, that is, improving the carboxylation efficiency of Rubisco for C 3 species Flexas et al. 2016). Optimizing the efficiency of RuBP carboxylation by Rubisco has the potential of improving WUE by decreasing the concentration of CO 2 required to achieve high photosynthetic rates (Carmo-Silva et al. 2015). The maximum carboxylase activity of Rubisco (V cmax ) and the capacity for photosynthetic electron transport (J max ) can constrain the WUE from the carbon assimilation side. Maintenance of functional electron transport under drought stress led to higher WUE i in Magnolia grandiflora (Vastag et al. 2020). Reduction of V cmax under ozone treatment caused decoupling of photosynthesis and stomatal conductance, which led to lowered WUE i in rice (Masutomi et al. 2019) and poplar clones . V cmax and therefore photosynthetic capacity increases with leaf maturation, thus young spring foliage can experience reduced WUE 13C , which can be critical, especially during spring droughts (Cernusak 2020). Enhanced WUE i of common bean genotypes under heat stress was linked to higher V cmax (Suárez et al. 2021). Additionally, V cmax /g s ratio has been suggested as a useful trait to characterize WUE i variability (positive correlation) across multiple plant species . Acclimation of WUE i and WUE 13C was coupled to V cmax and J max across Arabidopsis genotypes in a study by Easlon et al. (2014). Moreover, the improvement of WUE i in Brassica juncea was linked to higher carboxylation efficiency (A/C i ) under biochar treatment (Silva Gonzaga et al. 2019). Photosynthesis and therefore WUE i can be limited by Rubisco and RuBP regeneration, especially under high irradiance conditions . Plants with simultaneous stimulation of RuBP regeneration and electron transport can improve their WUE i due to better photosynthetic capacity (López-Calcagno et al. 2020). Other alternatives to improve the WUE 13C/bio would be decreasing photorespiration by means of higher Rubisco efficiency for CO 2 (Whitney et al. 2011;Parry et al. 2013) or altering the photorespiratory CO 2 release by adjusting metabolic pathways in leaves (Peterhansel and Maurino 2011). Total leaf N content shows a significant positive impact on the carboxylation capacity of plants (Wright et al. 2003;Paillassa et al. 2020). The identification of specific amino acids affecting Rubisco kinetics (Orr et al. 2016) may provide suitable targets for improving CO 2 assimilation and consequently WUE i (Nadal and Flexas 2019). Further exploration of optimization of Rubisco activity can positively influence the WUE of plants without any direct trade-off with growth capacity and yield of crops.

Respiration
Carbon loss through respiration is another process that decreases WUE bio (Seibt et al. 2008;Gago et al. 2014;Tortosa et al. 2016). Plants with lower maintenance respiration rates can maintain higher WUE bio . Moreover, respiration could be considered the main factor behind the gap between WUE i and whole-plant WUE bio (Medrano et al. 2017). High respiratory losses were linked to lower WUE bio of C 4 Miscanthus x gigantus located in USA drylands (Maleski et al. 2019). Greater night-time respiration (i.e. high nocturnal transpiration) has been also recognized as one of the major factors behind the reduction of WUE bio under magnesium deficiency of barley (Tränkner et al. 2016). High VPD fluxes led to larger reductions in photosynthesis in comparison to respiration, which decreased the overall productivity and WUE bio of plants from a semi-arid ecosystem (Roby et al. 2020). The higher stability of mitochondria and susceptibility of chloroplasts, especially PSII, to abiotic stress can negatively influence the balance between carbon assimilation and respiration towards lower WUE i (Dahal and Vanlerberghe 2017). Root respiration explained around 40 % of WUE bio reduction in both wellirrigated and non-irrigated treatments of grapevine (Tomás et al. 2014b). Root respiration might be a major component of total plant respiration and thus an important target for further exploration for WUE bio optimization (Escalona et al. 2012). Leaf development (maturation) connected with greater respiratory losses could be seen as an additional constraint to long-term WUE 13C (Zufferey 2016;Hernández-Montes et al. 2019). There is a still lack of precise quantification of day respiration or night-time respiration effect on whole-plant WUE bio and further research is needed. Nevertheless, respiration is connected with plant growth and fruit ripening. Therefore, plant breeding or genetic manipulation efforts that would aim at reducing respiration rates would probably lead to a significant reduction of growth and/or yield. Higher respiratory losses could be also linked to the upregulation of antioxidant systems and artificial reduction of respiration could be therefore defective. The inclusion of respiration for WUE calculation creates a more robust estimate, which improves the correlation with whole-plant WUE bio (Cernusak et al. 2007;Zhang et al. 2019). For example, Senbayram et al. (2015) have shown that the 9.8-48.6 % beneficiary effect of nitrogen fertilization on daytime WUE i was lost when nocturnal stomatal conductance and night-time respiration were taken into consideration. Therefore, the respiratory aspect of carbon balance should not be neglected for correct total plant WUE bio estimates.

Leaf Anatomy and Plant Crown Architecture
Leaf anatomy can affect the mesophyll diffusion conductance to CO 2 , carboxylation capacity and ultimately WUE in plants (Tomás et al. 2013;Carriquí et al. 2015). Increasing internal air volume might have positive effects on WUE i (Mediavilla et al. 2001), probably due to enhanced internal CO 2 conductance to the site of carboxylation. Similarly, Guerfel et al. (2009) reported more efficient water use associated with thicker palisade parenchyma in olive trees. The leaves' architecture can influence the WUE i due to variable mesophyll porosity and SD to intercellular airspace volume ratio in coniferous tree species (Trueba et al. 2022), and cell wall properties such as cell wall thickness (T cw ) might influence g m and thus WUE i (Flexas et al. 2021;Pathare et al. 2023). Mutant rice populations with higher leaf mass per area (LMA) showed improved whole-plant WUE bio under both control and water-limited conditions (Reddy et al. 2020a). In the study by Horike et al. (2021), WUE i of five shrub species covaried with LMA under drought stress. LMA differences explained WUE 13C variance across rice mutants through its influence on carbon gain (Reddy et al. 2020b). A study by Medrano et al. (2009) also reported a positive correlation between WUE i and LMA in Mediterranean herbs and shrubs. Similarly, LMA was positively correlated with WUE 13C among trees (Betula, Larix, Pinus) in the boreal forest (Ge et al. 2022). A thicker leaf can be associated with a thicker boundary layer, which lowers transpiratory losses and ultimately improves WUE bio (Bramley et al. 2013). The manipulation of leaf anatomy has been proposed as a potential theoretical target for improving photosynthetic capacity and WUE in plants ( Tholen et al. 2012). The development of plants with thicker leaves and high internal air volume can theoretically improve their WUE.
Further macro-morphological constraint, which affects the whole-plant WUE bio , is plant crown architecture (Christina et al. 2016;Medrano et al. 2017;McNeil et al. 2023). A more complex crown architecture creates shade for inner leaves, which can reduce evaporative demand and therefore improve WUE balance. A positive effect of shading treatment on leaf-level WUE i has been observed for Actinidia chinensis (Chartzoulakis et al. 1993), A. deliciosa (Montanaro et al. 2009), Citrus aurantium (García-Sánchez et al. 2015, C. sinensis (Jifon et al. 2003;, Coffea arabica (Liu et al. 2018b) and Fragaria ×ananassa (Cordoba-Novoa et al. 2022). It is notable to say that shade leaves are optimized for low irradiance and if exposed to direct sunlight (crown damage) they can show decreased WUE i (Dai et al. 2009). Moreover, the leaves of Pinus taeda in the lower parts of the crown showed significantly higher WUE i in comparison to the upper part during the peak of the vegetation season (Blazier et al. 2004). The WUE 13C derived from wood in Fagus crenata and Quercus crispula showed a positive correlation with tree height, crown depth and crown width (Osada et al. 2004). Furthermore, Glenn et al. (2015) showed that the less complex pillar form of Prunus persica had lower WUE i due to higher canopy transpiration in comparison to the common crown form. Leaf area index (LAI) as an indicator of crown density also shows a positive impact on WUE bio across various terrestrial ecosystem types Luo et al. 2022). The raising WUE bio of Alpine grasslands has been also linked to increasing LAI (Ma and Zhang 2022). Nevertheless, higher LAI and therefore greater total transpiration can be detrimental for arid regions where it can have a negative impact on WUE bio (Malone et al. 2016). More complex crown architecture and higher LAI can enable plants to optimize and improve their whole-plant WUE due to the shading effect and probably also due to better microclimatic conditions within the crown.

Conclusion and Future Prospects
The WUE balance of plants is multifaceted and affected at multiple levels of organization from molecular to whole-plant level. The main constraining factors identified in this review were stomatal morphology and control, minimal and nocturnal conductance, mesophyll conductance, carboxylation efficiency, respiration rates, leaf anatomy and crown architecture. The traits are usually analysed in research papers separately or in specific combinations (e.g. stomatal morphology and gas exchange). We suggest that future research should include multi-trait analyses with the aim of WUE optimization, thereby deepening our understanding of the coupling and decoupling of carbon uptake and water-use traits. The technological progress of phenotyping platforms can lead to more robust experimental designs that could handle multitrait analysis. The night-time transpiration and respiration seem to be under-developed major aspects of long-term WUE optimization, which could be further investigated. The effect of leaf hydraulic conductance and canopy structure on WUE is also not very well understood and can be improved. A better understanding of morpho-physiological constraints of WUE can help us to effectively develop more droughtresilient crop and tree species.

Sources of Funding
PP was supported by the Federal Ministry of Education and Research, BMBF project BioWaWi grant number 16LW0093.

Contributions by the Authors
PP conceived the paper idea. PP, APP and MM wrote the first draft. BS and LJL supervised the process and helped with the editing of the manuscript.

Conflict of Interest Statement
None declared.

Supporting Information
The following additional information is available in the online version of this article -

Data Availability
No original data was used in this commentary. The discussion and synthesis are based on already published studies.