Differential water-use efficiency and growth among Eucalyptus grandis hybrids under two different rainfall conditions

Aim of study: To analyze the course of leaf water status, water-use efficiency and growth in Eucalyptus grandis and hybrids throughout seasons with different rainfall. Area of study: The study was conducted in northern Uruguay. Material and methods: A randomized block trial was established containing E. grandis (ABH17), E. grandis × Eucalyptus camaldulensis (GC172), E. grandis × Eucalyptus tereticornis (GT529), and E. grandis × Eucalyptus urophylla (GU08). Predawn leaf water potential (Ψ pd ) and midday leaf water potential (Ψ md ) were measured every six weeks from the age of 16 months, throughout two growing seasons. Stomatal conductance ( g s ), net photosynthetic rate ( A ), and leaf-level transpiration ( E ) were measured once in each growing season, along with leaf carbon isotope discrimination (∆ 13 C ) and tree growth. Stomatal density and distribution were studied. Main results: ABH17 and GU08 had the lowest daily fluctuation of leaf water potential and showed stronger stomatal regulation; they were hypostomatic, and stomata on the adaxial leaf surfaces remained immature. GC172 and GT529 (Red-Gum hybrids) were amphistomatic and transpired more intensively; they were less efficient in instantaneous and intrinsic water use and grew faster under high soil moisture (inferred from rainfall). Under such conditions, GC172 reached the highest gas-exchange rate due to an increase in tree hydraulic conductance. ABH17 and GU08 were hypostomatic and used water more efficiently because of stronger stomatal regulation. Research highlights: Red-Gum hybrids evidenced less water use efficiency due to lower stomatal regulation, different stomatal features, and distinct growth patterns as a function of soil moisture (inferred from rainfall). Abbreviations used: Ψ pd predawn Ψ md midday leaf ΔΨ: daily fluctuation of leaf water potential (ΔΨ = Ψ pd - Ψ md ); A : net photosynthetic rate, E : leaf transpiration rate, g s : stomatal conductance, WUE: instantaneous water-use efficiency; WUEi: integrated water-use efficiency; A/E leaf photosynthesis-to-leaf transpiration ratio; ∆ 13 C leaf carbon isotope discrimination; K : tree hydraulic conductance; E /∆Ψ: ratio between leaf transpiration and daily fluctuation of leaf water potential; δ 13 C natural abundance of 13 C role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results".


Introduction
Eucalyptus afforestation occupies 22 million hectares in more than 90 countries, being one of the most popular woody crops in the world (Dharhi et al., 2018). In Uruguay, eucalypt plantations for hardwood timber or pulpwood supply cover 70% of the Uruguayan afforested area (one million hectares) (Martin, 2018), and in the last few years, the increase of climate variability (Linderman et al., 2013) led to the selection of promising clones by crossing Eucalyptus grandis with Eucalyptus urophylla and Red-Gum species (Eucalyptus camaldulensis and Eucalyptus tereticornis). Since there is a positive correlation between carbon fixation and water loss, breeders face the challenge of selecting genotypes for high plant growth and high water-use efficiency (WUE) (Morris et al., 1998;Dye, 2000;Mokotedi, 2013). However, information regarding several commercial genotypes is still lacking.
Generally, eucalypts exhibit high transpiration rates to enhance carbon fixation and growth, even under severe drought conditions (Lewis et al., 2011). Such behavior is commonly known as anisohydric and reflects the occurrence of high transpiration rates all year round (Meinzer et al., 2014). On the other hand, isohydric plants close stomata to reduce gas exchange and maintain high Ψ (Mc Dowell et al., 2008). These patterns represent the two ends of the stomatal regulation sensitivity spectrum (Sade & Moshelion, 2014). Gas exchange can be inferred from leaf water potential fluctuation (∆Ψ), that is, the difference between predawn and midday bulk leaf water potential (Ψ pd and Ψ md , respectively) (Choné et al., 2001;Franks et al., 2007), which varied as a function of soil water content and atmospheric demand (vapor pressure deficit, VPD).
Leaf-level water-use efficiency is frequently estimated by instantaneous and integrated methodologies (WUE and WUE i , respectively). The former is the ratio of leaf photosynthesis to leaf-transpiration rate, while the latter is related to leaf carbon isotope discrimination (∆ 13 C) during gas diffusion and Rubisco-dependent reactions. In C 3 species, such as eucalypts, ∆ 13 C is negatively correlated with the ratio of transpiration efficiency and whole-plant dry-mass accumulation (Farquhar et al., 1989), and a significant correlation was observed between foliar ∆ 13 C and WUE i in E. grandis (Olbrich et al., 1993), E. globulus (Osório andPereira, 1994), E. camaldulensis (Akhter et al., 2005) and E. grandis × E. camaldulensis (Le Roux et al., 1996). However, ∆ 13 C may change when scaling up to stems, shoots, and whole trees (Olbrich et al., 1993), and so may WUE. Moreover, WUE i can be inferred from different plant organs, and it can be estimated as the whole-plant WUE (Seib et al., 2008). Although ∆ 13 C may not be an indicator of WUE, it may be useful as a screening tool (Le Roux et al., 1996). A significant relationship between foliar 13 C signature (δ 13 C) and instantaneous WUE in E. grandis was found, which could be used as a proxy for WUE (Casparus et al., 2018). Water-use efficiency can also be studied from wood rings (from the trunk or branch cores), although in many species ring growth may be caused by carbon synthesized in previous growing seasons (Mc Farlane & Adams, 1998;Skomarkova et al., 2006).
Stem biomass can increase without modifying WUE, as previously found by Battie-Laclau et al. (2016) in E. grandis plantations. Leaf transpiration also depends upon stomatal traits such as density and distribution, both of them varying not only among eucalypt species (David et al., 1997;Morris et al., 1998;Héroult et al., 2013) but also between related hybrids (Eksteen et al., 2013). However, little research has been done in commercial genotypes so far.
The present study was designed to analyze the course of leaf water status and leaf-level WUE in E. grandis, E. grandis × E. camaldulensis, E. grandis × E. tereticornis, and E. grandis × E. urophylla clones, during two consecutive growing seasons with different rainfall. We hypothesized the following: (a) these genotypes differ in stem growth capacity, transpiration activity (ΔΨ), stomatal traits and leaf-level water use efficiency (WUE), as inferred either from their instantaneous WUE or leaf carbon isotope discrimination (∆ 13 C); b) that this behavior is related to different regulation of stomatal opening (g s ) and carbon fixation (A); c) this regulation modifies stem growth and varies between growing seasons as a function of soil water moisture (inferred from effective rainfall).

Plant material and experiment design
Cuttings from three months old sprouts of each clone were collected in April 2012 and rooted in the clonal nursery of Lumin (formerly Weyerhaeuser Uruguay) with controlled humidity (90 to 95%). They were irrigated from May to August with a Biorend® solution (10 cm 3 L -1 ) and then transplanted into 3 L plastic pots containing Carolina Soils® substrate, a mix of sphagnum peat Differential gas-exchange regulation and growth among E. grandis hybrids (58%), vermiculite (40%), and trace minerals (2%). They were fertilized three times a week with 18-18-18 (NPK) for two months.
At the age of four months (September 2012), the plants were planted in a field trial in Tacuarembó (Uruguay; 31º 38`15" S, 55º 54' 17" W), where the regional climate is humid subtropical (Cfa, Köppen classification system) with a mean annual rainfall of 1484 mm and an average temperature of 10.3 ºC in July and 22.3 ºC in January. Soils are sandy to loamy (Typic Hapludult, USDA classification system) with a poorly drained B horizon (0.35 to 0.60 m) and Fe-Mn concretions down to 1.50 m deep.
A randomized complete block design with three blocks was used. Within each of the three blocks, ten trees per clone of E. grandis (ABH17), E. grandis × camaldulensis (GC172), E grandis × tereticornis (GT529), and E. grandis × urophylla (GU08) were planted in rows, at a spacing of 4.75 m × 5 m. Each row was a plot. Two trees per plot were randomly selected to assess ecophysiological variables. From the age of 16 months (December 2013), Ψ was measured every six weeks throughout two consecutive growing seasons (2013-14 and 2014-15). Diameter at breast height and tree height were measured for all trees of the trial (n=72) at the beginning and end of each growing season. Leaf gas-exchange measurements were performed once a season, as well as foliar sampling for carbon isotope analyses.

Bulk leaf water potential (Ψ pd , Ψ md )
Leaf water potential (predawn and midday) was measured with a Schölander pressure chamber (Soil Moisture Equipment®, Santa Barbara, CA) on three current-year branchlets of each selected tree. Predawn leaf water potential (MPa) was recorded before sunrise (4:30 to 06:00 a.m.) on the lower branches. Midday leaf water potential (MPa) was measured from 12:00 to 03:00 p.m. (one hour per block) on the sunlit foliage of the upper tree crown. Average clone Ψ was calculated from three measurements of each selected tree, and the daily fluctuation of Ψ was calculated a ∆Ψ = Ψ pd -Ψ md . We checked for block effect when analyzing the data. Atmospheric demand was used as a covariate (Faustino et al., 2013) when analyzing the date effect.

Gas exchange (A, E)
Net photosynthetic rate A (μmol m -2 s -1 ), stomatal conductance g s (mmol m -2 s -1 ), and leaf transpiration rate E (mmol m -2 s -1 ) were measured on three current-year leaves at the top of the crown, between 11:00 a.m. and 03:00 p.m., using a portable gas-exchange device (LiCor 6400, LiCor®, Lincoln, NE) at a photon flux density of 900.78 ± 1.20 μmol m -2 s -1 and an airflow rate of 500.16 ± 0.64 mL min -1 . For E. grandis, this irradiance is saturating (Whitehead and Beadle 2004). Measurements within each block lasted a maximum of one hour and 15 minutes, and we checked for block effect when analyzing the data. Air temperature, CO 2 level, and VPD inside the chamber matched the outside air condition. The purpose was to study gas exchange under water deficits. However, soil moisture (inferred from Ψ pd value) began to decrease in autumn of the second growing season. Measurements were taken on 02/26/2014 (summer) and 05/08/2015 (autumn). The maximum air temperature during measurements was 29.25 ± 1.84 °C in the first growing season and 23.90 ± 0.95 °C in the second. Data were used to calculate instantaneous water-use efficiency (WUE) as the A/E ratio, and K as E/ΔΨ ratio.

C abundance and carbon isotope discrimination
Nine leaves per tree were collected to obtain the 13 C signature (δ 13 C), from the same branchlets selected for gas-exchange measurements, and located on the northern (sunny) side of the trees' upper crown. Total carbon concentration and relative abundance of 12 C and 13 C were measured by mass spectrometry using a Delta Plus® spectrometer (Finnigan MAT, Bremen, Germany). Each sample was analyzed with an elemental Flash EA 112. The standard deviation of total carbon and 13 C were calculated from leucine as a reference. The natural abundance of 13 C (δ 13 C, ‰) was expressed in relation to international standard PDB as [(R sample / R standard ) -1] (Craig, 1957), where R is the 13 C/ 12 C ratio. Data were used to calculate Δ 13 C (‰) as (δ 13 C air -δ 13 C leaf ) / (1 + δ 13 C leaf /1000) , being δ 13 C air the atmospheric δ 13 C abundance and δ 13 C leaf the sample 13 C content (Farquar et al., 1989).

Stomatal density and distribution
The observation of these traits was not part of the original work and was considered only for the second growing season. Three one-year-old leaves per tree of similar leaf age were collected to study stomatal density (No. mm -2 ) and distribution. Three epidermal impressions of adaxial and abaxial leaf surfaces were prepared with nail varnish; dried layers were peeled off with tweezers (modified from D'Ambrogio, 1986). Stomatal density was calculated from images taken with a Dino Eye 2.0 digital camera added to a Nikon E100 light microscope (10X). Five fields (0.15 mm 2 ) on each leaf surface were analyzed using DinoCapture® software.

Tree growth
Tree height (Ht, m) and diameter at breast height (Dbh, m) of all the trees in a plot were measured at the beginning of each growing season. Data were used to calculate diameter and height relative growth as RG = (G n -G n-1 )/ Gn -1 , being G the tree Dbh or Ht, and standing tree volume (V, m 3 ) as V = Dbh 2 × π ⁄4 × Ht × FF , being FF the form factor An average form factor of 0.5 was used (Da Silva et al., 1999).

Statistical analysis
The effect of the differences and interactions between clone and date on the variances was tested using a two-way analysis of variance (ANOVA). Data were analyzed for homoscedasticity and normality (Shapiro Wilks, p>0.05), and when these conditions were met, the ANOVA was performed using InfoStat® software (UNC, Córdoba, Argentina). Mean comparison was performed by Tukey's post hoc test (p<0.05). Clone-date interaction was studied for Ψ within each growing season using VPD as a covariate. The proposed model was the following: yijk = µ + αi + βj + τk + (ατ)ik + γl +εijkl where y ijk is the response variable, µ is the overall mean, α i is the effect of the i th clone, β j is the effect of the j th block, τ k is the effect of the k th date, (ατ) ik is the effect of the i th clone and k th date, γ l is the effect of the covariate, and ε ijkl is the residual error with εij ~ Ν (0, σ2ε).

Bulk leaf water potential and rainfall
Effective rainfall was higher during the second growing season (2014-15) ( Fig. 1), and Ψ pd reflected such variation (Tables 1 and 2). In that season, Ψ pd attained the highest average (-0.21± 0.01 MPa) and no clonal effect was observed. However, differences between clones appeared in the first growing season (the driest one), as E. grandis (ABH17) averaged the lowest value (Ψ pd = -0.66 ± 0.03 MPa) (p<0.0149).
Unlike Ψ pd , Ψ md did show the clone effect (p<0.0001) in both seasons, and Red-Gum hybrids had the lowest average. GT529 showed this behavior throughout both growing seasons (Ψ md = -1.96 ± 0.08 in the first season, and -2.45 ± 0.05 MPa in the second), whereas GC172 showed it during the wettest one (Ψ md = -2.34 ± 0.05 MPa). ABH17 and GU08 behaved similarly, except at the beginning of the wettest season (2013-14), when GU08 averaged the minimum value.
Clone-date interaction was significant for Ψ md in both growing seasons (p<0.0001), although for Ψ pd , this interaction appeared during the driest season (p<0.0001) (Tables 1 and 2). The daily fluctuation of leaf water potential varied among genotypes (p<0.0001) and was lowest in the Red-Gum hybrids (Fig. 2). During summertime, it was 30 to 50% higher than that of the other clones.

Gas exchange and instantaneous water-use efficiency
Net photosynthetic rate (A) differed among hybrids in both growing seasons (p<0.0068 and p<0.0001), with the GU08 clone showing the lowest A in both study periods (12.80 and 11.91µmol m -2 s -1 , first and second season respectively). Despite being variable, carbon fixation was similar among the other genotypes (Table 3). Stomatal conductance (g s ) also varied among clones across both growing seasons (p<0.0001), when Red-Gum hybrids displayed the highest average and, therefore transpired more intensively (higher E) (Table 3). These results revealed a stronger relationship between A and g s for the ABH17 clone during both growing seasons (r 2 = 0.76 in the first growing season and 0.8 in the second), and for GU08 during the first growing season (r 2 = 0.83) (Fig. 3). Unlike Red-Gum hybrids, these genotypes attained higher carbon fixation under low stomatal conductance (Fig. 3), therefore showing higher instantaneous WUE (A/E ratio). In addition, there was no relationship between overall instantaneous WUE and Ψ md , so WUE remained relatively constant for a wide range of Ψ md .

Carbon isotope discrimination and water-use efficiency
Red-Gum hybrids showed higher ∆ 13 C (p<0.0001), suggesting less efficiency in leaf-level intrinsic water use (A/E integrated over time). In contrast, ABH17 and Means within columns followed by different letters are significantly different (Tukey, p < 0.05), p VPD and r VPD for VPD covariance also shown GU08 showed the opposite behavior, as inferred either from their higher instantaneous WUE or their lower foliar ∆ 13 C (Table 3).

Stomatal density and distribution
Stomatal density showed no clonal difference and ranged between 428 ± 82 and 340 ± 56 stomata mm -2 (Fig. 4). The stomatal distribution pattern did vary among clones, since Red-Gum hybrids were amphistomatous, whereas ABH17 and GU08 were hypostomatic and stomata on their upper epidermis remained entirely immature (Fig. 5).

Tree growth
Clonal difference (p<0.0001) in diametric relative growth was observed in the second year of the study (i.e., during the wettest growing season). By then, the E. grandis × camaldulensis (GC172) clone had the highest diametric relative increment (0.23 ± 0.01), whereas E. grandis × urophylla (GU08) showed the lowest one (0.07 ± 0.01). On the other hand, differences in height growth appeared during the driest year of the study (the first growing season), but they did not affect volumetric tree growth (all clones averaged 0.065 ± 5.8 ×10 -3 m 3 ; data not shown). However, differences in stem volume were observed at the end of the study, when GU08 averaged the lowest value (Tables 5 and 6).

First growing season (2013-14)
Second growing season (  Differential gas-exchange regulation and growth among E. grandis hybrids rainfall). During the wettest growing season, Ψ pd showed a three-fold increase in comparison to the first seasons, and all genotypes averaged a similar value. In addition, ∆Ψ was 1.88 times larger, which suggests a higher leaf-level water loss. Red-Gum hybrids (GC172 and GT529) transpired more intensively in both growing seasons, with an increase of 67% in their ∆Ψ under well-watered conditions (2014-15). Stomatal opening can be quite sensitive to atmospheric demand in some eucalypt species (White et al., 2000). However, soil moisture modifies the relationship between g s and VPD (Leuning et al., 1991). Under well-watered conditions, as in the second growing season, the magnitude of the increase in water loss (E) depends on the sensitivity of decreasing g s with increased VPD. We observed an increase of E in Red-Gum hybrids during the summer (high ΔΨ), and higher g s when measuring instantaneous gas exchange (measured once a season). Therefore, midday leaf water potential decreased more steeply because of transpiration (Tables 1 and 2), which reveals a lower stomatal sensitivity in these clones. E. camaldulensis and E. tereticornis seem to display higher Figure 3. Relationship between net photosynthetic rate (A) and stomatal conductance (g s ) of (a) E. grandis (ABH17), (b) E. grandis × E. camaldulensis (GC172), (c) E. grandis × E. tetereticornis (GT529) and (d) E. grandis × E. urophylla (GU08) clones in the first and second growing season (circles and squares, respectively); regression equations for both growing seasons also shown (y 1 and y 2 , first and second respectively) g s at saturating irradiance (Whitehead & Beadle, 2004), similarly to GC172 and GT529 in the present work.
Although Red-Gum hybrids had the highest E (and ΔΨ) during the summer, GC172 was more sensitive to soil water content and displayed stronger regulation of stomata opening (Table 3). In addition, gas exchange results for this clone revealed a 40% reduction in g s during that period. Stomatal conductance in E. camaldulensis decreases steeply as a function of soil water status (White, 2000), the same behaviour recorded for the GC hybrid. However, transpiration also depends on K (Tyree & Ewers, 1991;Bond & Kavanagh, 1999). Our ABH17 0.07 ± 3.1* a 0.13 ± 4.2* a 0.16 ± 4.7* b 0.48 ± 0.01 a 0.17 ± 0.02 b 0.13 ± 0.01 a GC172 0.07 ± 2.0* a 0.14 ± 2.6* a 0.17 ± 3.0* a 0.49 ± 0.01 a 0.23 ± 0.01 a 0.14 ± 0.01 a GT529 0.06 ± 2.0* b 0.12 ± 3.5* b 0.14 ± 4.2* b 0.50 ± 0.01 a 0.20 ± 0.01 ab 0.11 ± 0.01 ab GU08 0.07 ± 3.1* a 0.14 ± 3.1* a 0.14 ± 4.  (Table 3), mainly due to an increase in K (Table 4) to deal with rising xylem tension (lower Ψ md ). Under such conditions, this species is known to produce extensive adventitious roots, to display stem hypertrophy (increased stem diameter), and to increase root porosity to sustain higher g s and K, which modify tree growth (Argus et al., 2015). Some of these features could explain the behavior of the GC clone in the wettest season, mainly if stem growth is considered (Tables 5 and 6).
In recent work, we studied the clonal variation of hydraulic traits and wood anatomy of one-and-a-half-year specimens of GC172, GT529, and ABH17, grown without water limitations under greenhouse conditions. We found that GC172 had the highest leaf-specific hydraulic conductivity without embolism (k L max = 7.53 ± 0.53 kg m -1 MPa -1 s -1 ) and wider xylem vessels (0.04 ± 0,001 mm). This clone showed the highest ΔΨ and was also the most vulnerable to cavitation (higher percentage loss of hydraulic conductivity; unpublished data). These results are consistent with the high transpiration of Red-Gum hybrids in the present study, and the higher hydraulic conductance of GC172 under well-watered conditions.

Water-use efficiency and growth
The analysis of instantaneous WUE (A/E) and ∆ 13 C led to similar conclusions. Red-Gum hybrids were the least efficient in intrinsic leaf-level water use, most likely due to lower stomatal regulation over time (long term A/E ratio) ( Table 3, Fig. 3). This outcome confirms the information provided by leaf water potential and suggests that Red-Gum hybrids adjust ∆Ψ to sustain carbon gain (Le Roux et al., 1996). Conversely, ABH17 and GU08 seemed to be more VPD sensitive and would therefore be more suitable for withstanding drought conditions by reducing leaf gas exchange. These clones were highly efficient in carbon fixation (high A), although GU08 was more water-use efficient in during the driest period (as inferred by ∆ 13 C data) ( Table  3). This behavior could have led to higher WUE by decoupling stomatal conductance and carbon gain (A), as previously reported for humid-zone eucalypt species (Héroult et al., 2013).
To our knowledge, hybrids with E. urophylla appear to be highly sensitive to VPD, which implies a reduction of carbon gain under high VPD. However, WUE is not a constant characteristic of a given genotype, and it varies according to a combination of site conditions, weather and tree age. The frequency and duration of soil water deficits are crucial in determining WUE, which is sensitive to annual variations in rainfall amount as well as to rainfall distribution throughout the year (Dye, 2000). Under soil water deficits, E. urophylla is known to show higher WUE (lower ∆ 13 C) due to stable stomatal

Forest Systems
August 2020 • Volume 29 • Issue 2 • e006 conductance as leaf water potential decreases (Zhang et al., 2016). In our work, this mechanism was displayed by GU08 throughout the less rainy season (2013-14), as inferred from WUE i ( Table 3). All genotypes had similar diametric relative growth and volumetric stem growth during the first year of the study (the less rainy one). During the wettest year, the GC clone attained the highest diametric relative growth (0.23 ± 0.01), whereas GU averaged the lowest (0.07 ± 0.01). The latter also showed the lowest stem volume at the end of the study (Table 6) and was the most efficient in water use (lower ∆ 13 C) during the driest growing season. This behavior is consistent with the so-called "hare strategy," which means that growth increases rapidly with soil water availability, as observed for diameter growth in 2014-15 (Table 6). Drew et al. (2009) reported this pattern for some GU clones in South Africa. These authors found E. grandis × camaldulensis hybrids to keep growing even under severe drought conditions, referred to as the "tortoise" growing pattern. In our study, this hybrid grew faster than other clones in the driest season (Table 6), mainly due to a higher gas exchange (as inferred by ∆Ψ, WUE, and WUE i ), which promoted carbon fixation.

Stomatal traits and water status
Transpiration was influenced by stomatal density and distribution. ABH17 and GU08 had hipostomatous leaves and showed stronger stomatal control that reduced gas exchange. In GU clones, Eksteen et al. (2013) reported that stomatal sensitivity to drought could be a consequence of both stomatal closure and the absence of stomata on adaxial leaf surfaces. These traits are highly variable among closely related clones, and we did not study stomata development or their functionality. Leaves were sampled at the end of the second growing season on the same branchlets used for gas-exchange measurements, therefore reflecting the effect of environmental conditions on stomatal anatomy during this time (the preceding days or even weeks).
On the other hand, Red-Gum hybrids transpired more intensively and were amphistomatic. E. grandis × camaldulensis hybrid seems to have higher K under well-watered conditions to sustain carbon gain, which is consistent with gas-exchange results. Conversely, ABH17 and GU08 enhanced their net carbon fixation. Eksteen et al. (2013) observed that GC hybrids had higher growth than GU clones after 12 months of chronic water stress. This behavior was related to a higher gas exchange and different stomata density and distribution.
A noteworthy observation was that stomata on adaxial surfaces of ABH17 and GU08 remained entirely immature, similarly to some tropical herbs (Kagan & Sachs, 1991). England & Attiwill (2011) studied stomata development in Eucalyptus regnans and reported that they evolved progressively from margin to midrib in juvenile leaves. However, in the current study, stomata remained immature on the whole adaxial surface of adult leaves. This pattern could have reduced leaf gas exchange and it probably prevented leaf water potential from falling (i.e., from being more negative). Hydraulic isolation in amphistomatic leaves can induce surface-specific stomatal closure in E. globulus (Richardson et al., 2017), and this could explain why Red-Gum hybrids attained the highest gas exchange in our work, even in the driest growing season. Nevertheless, further research is needed to study leaf hydraulics and the functionality of such immature stomata.

Conclusions
We found differences in gas-exchange regulation, water status, and growth among genotypes as a function of soil water moisture (inferred from rainfall) in two growing seasons. Red-Gum hybrids (E. grandis × E. camaldulensis, GC172, and E. grandis × E. tereticornis, GT529) transpired more intensively (higher daily ∆Ψ), and were therefore less water-use efficient as confirmed either by instantaneous (A/E) or intrinsic (∆ 13 C) WUE. These clones were amphistomatic and displayed lower stomatal control over time (higher ∆ 13 C). The E. grandis × E. camaldulensis hybrid showed the highest gas exchange under well-watered conditions (wettest growing season), mainly due to an increase in tree hydraulic conductance (K). This behavior granted this hybrid constant g s , therefore promoting carbon gain and stem growth. Conversely, E. grandis (ABH17) and E. grandis × E. urophylla (GU08) genotypes were more water-use efficient over time because of stronger stomatal regulation that increased the A/E ratio. Therefore, they were more VPD sensitive across both growing seasons (and therefore in different soil water moisture conditions). These clones were hypostomatic, and stomata on their adaxial leaf epidermis remained entirely immature.