Triticum aestivum L., cultivar ‘Tasman’, was used to address the role of leaf age on leaf ultrastructure and plant physiological response to water stress and water stress recovery. The cultivar ‘Tasman’ is one of the parents in a quantitative trait loci (QTL) mapping population studied in depth for Δ¹³C of leaf tissue by Rebetzke et al. (2009); the cultivar has low stomatal conductance to CO₂ (g _sc) and high leaf-intrinsic water-use efficiency (A/g _sw) (Jahan et al., 2014).

Two or three of the youngest fully expanded leaves per plant were selected for gas exchange measurements (weeks 5 to 9 after plant emergence). Leaves were labelled on the week of their full expansion (leaf number 5 on week 5, up to leaf 9 on week 9), and not by their position or tiller number. Under the experimental growth conditions used, leaves took between 5 and 8 days to expand, so that the ‘fifth leaf’ was 5-8 days older than the ‘sixth leaf’. Same-aged leaves from the same plant were placed side-by-side in a 6 cm² (2×3) leaf chamber of a LI-6400XT portable photosynthesis system fitted with a red-blue light (LI-COR, Lincoln, NE, USA). During the time of measurement, all the plants were at the tillering phase except week 9, when the flowering heads began to emerge. During measurements, the leaf chamber was controlled at a CO₂ mole fraction of 380 µmol mol^-1, a leaf temperature of 25°C and irradiance of 1500 µmol m^-2 s^-1. The vapour pressure deficit inside the leaf chamber varied between 0.6 and 2.3 kPa, depending on g _sw. Mesophyll conductance was estimated by combining leaf gas exchange measurements with measurements of carbon isotope discrimination using a tunable-diode laser absorption spectrometer (TDLAS, model TGA100A, Campbell Scientific, Inc., Logan, UT, USA) as described by Barbour et al. (2010) and Jahan et al. (2021).

When assessing the recovery of plants from the water stress, leaves were placed side-by-side in a 12 cm² (2×6) leaf chamber (Li6400-11) attached to a LI-6400XT portable photosynthesis system (LI-COR, Lincoln, NE, USA) fitted with a red-green-blue light set to mimic the red-blue light source used in the first experiment (Li6400-18 RGB light source). The chamber environmental conditions were similar for all physiological measurements and represented light-saturated photosynthesis. Leaves were labelled based on age of their full expansion: young leaf is the newly fully expanded leaf (1 day old), mature leaf is one week older than young leaf (8 days old) and old leaf is one week older than mature leaf (15 days old), and not by their position or tiller number.

The value of g _m was calculated from the difference between predicted discrimination assuming infinite mesophyll conductance (Δ_i) and measured discrimination (Δ_obs), as described by Jahan et al. (2021) using equations developed by Evans et al. (1986) and Barbour et al. (2010), and including a ternary effect as described by Farquhar and Cernusak (2012). Average δ¹³C during the light period for a 7-day running average (δ¹³C _CO2 ) was measured using a stable isotope cavity ring down laser (G11101-i, Picarro CA, USA). We used the average ambient growth δ¹³ C_CO2 values (-13.2 ‰) in the calculation of g _m. The Picarro laser was calibrated as described by Thurgood et al. (2014). Here, values of fractionation factors were used in the calculation of g _m as described by Jahan et al. (2014); the fractionation associated with carboxylation was assumed to be 29‰, the fractionation during dissolution and diffusion through water was assumed to be 1.8‰, the fractionation associated with photorespiration was assumed to be 16.2‰, the fractionation occurring during diffusion through the leaf boundary layer was assumed to be 2.9‰ and the rate of day respiration was taken to be 0.8 µmol m^-2 s^-1.

Leaves measured for gas exchange were analysed using microscopy to assess anatomical features as a function of leaf age and drought treatment. Leaf samples approximately 5 mm × 2 mm were cut for sectioning, avoiding major veins. Sections were fixed in a solution of 2.5% glutaraldehyde, 3% paraformaldehyde and 0.1 M phosphate buffer (pH 7.2) for 18 h period under vacuum. Samples were then fixed in 2% osmium tetroxide for 2 h, then further fixed using 2% uranyl acetate for 30 min. Samples were dehydrated using an ethanol series of 15, 30, 50, 70, 90, 95 and 100% and then embedded in LR-white resin series of 20, 40, 60, 80, 100% (London Resin Company, London, UK). Semi-thin leaf cross-sections of 0.5 µm for light microscope were cut using an ultra-microtome with glass tip blades, then sections were stained with toluidine blue.

Sections were viewed in bright field with a Leica DM6000B upright light microscope with phase contrast at 400× magnification. We used IMAGE J software (National Institute of Health, Bethesda, MD, USA) for analysis of images. Images were used to determine: (a) leaf thickness, (b) mesophyll thickness between the two epidermal layers (t _mes), (c) the surface area of mesophyll cells exposed to the intercellular airspaces (S _mes), (d) the surface area of chloroplast exposed to intercellular airspace (S _c) and (e) chloroplast number per unit leaf area (Z, in number per m²). S _mes, S _c and Z were calculated as described by Evans et al. (1994).

where L_mes is the total length of mesophyll cells facing the intercellular airspace, W is the width of the section measured and F is the curvature correction factor (1.25) by Thain (1983). Then,

where, L _c is the length of chloroplast exposed to the intercellular airspace.

where, z is the total number of chloroplasts in the section measured. The volume fraction of inter-cellular airspace (fias) was calculated as described by Xiong et al. (2016),

where S _ias is the total cross-sectional intercellular airspace area.

Ultrathin cross- and paradermal sections of 70 nm were prepared by an ultramicrotome for use in the scanning transmission electron microscope (STEM). The sections for STEM were stained with lead citrate and mounted on glass with Eukitt (Electron Microscopy Sciences, Hatfield, PA, USA). Cell wall thickness (t _cw), and chloroplast to outer cell wall distance were measured from STEM micrographs by using IMAGE J software.

The effects of leaf age on physiological measures were assessed using two-factor analysis of variance (ANOVA); means were compared using Fisher’s unprotected least significant difference tests. The associations between g _m and A, between g _m and g _sc, between S_c and g _m, and between t_cw and g _m were assessed using linear regression and correlation. GenStat 14^th edition (VSN International Ltd, Hemel Hempstead, UK) was used for statistical analyses, and in all tests, results were considered statistically significant when P< 0.05.

Photosynthetic rate (A), stomatal conductance to CO₂ (g _sc) and mesophyll conductance (g _m) significantly (P< 0.01) decreased as leaves aged ( Figures 2A–C ) regardless of water treatments. For leaf 6, A decreased by 78% (rate of 0.98 µmol m^–2 s^–1 day^-1), g _sc by 73% (0.008 mol m^-2 s^-1 day^-1) and g _m by 87% (0.031 mol m^-2 s^-1 bar^-1 day^-1) from week 6 to week 9. The linear trend in A over the study period differed significantly (P< 0.05) among leaf numbers: for leaf 5, the rate of decline was -8.78 ± 4.31 per week, while leaf 6 declined by -4.89 ± 1.34 per week, and leaf 7 declined by -2.06 ± 4.31 per week ( Figure 2A ). This linear trend was also significant (P< 0.05) for g _sc ( Figure 2B ), whereas the linear trend was not significant for g _m and A/g _sw ( Figures 2C, D ).When leaves of different ages were compared, A, g _sc and g _m for the oldest leaves (22 day-old) were significantly lower than younger leaves (15, 8 or 1 day-old) ( Figures 3A–C ). There were no significant leaf age effects for leaf-intrinsic water-use efficiency (A/g _sw) ( Figure 2D ), or between different leaves of different ages ( Figure 3D ).

Photosynthetic rate (A), stomatal conductance to CO₂ (g _sc) and mesophyll conductance (g _m) of the youngest fully expanded leaves decreased significantly (P< 0.001) from week 5 to week 9, in well-watered plants ( Figures 4A–C ). In well-watered plants, A, g _sc and g _m of the youngest fully expanded leaf initially increased with rising plant age up to week 7, and then decreased rapidly with increasing plant age. Leaf-intrinsic water-use efficiency (A/g _sw) was not affected by plant age for well-watered plants from week 5 to week 9 ( Figure 4D ).

The water-stress treatment applied in the second experiment resulted in significantly (P = 0.05) more negative midday leaf water potentials (Ψ _L) for water-stressed compared to well-watered plants (averages of -1.4 and -1.1 MPa across all leaf ages for water-stressed and well-watered plants, respectively) ( Supplementary Table 1 ).

Only stomatal conductance to CO₂ (g _sc) was significantly higher (P = 0.03) in well-watered plants compared to water-stressed plants, when individual leaves were considered ( Figure 2B ). These changes resulted in higher A/g _sw in drought plants compared to irrigated plants. However, the slope of the decline in photosynthetic rate (A) with leaf age did not significantly differ between well-watered and water-stressed plants ( Figure 2A ). Overall, mesophyll conductance (g _m) was slightly, but not significantly, higher in well-watered compared to water-stressed leaves when individual leaves were compared up to week 8 ( Figure 2C ).

When we consider leaves at different ages measured on the same day, water availability did not affect photosynthetic rate for 1 day- and 8 day-old leaves, whereas A was significantly (P = 0.05) higher for water-stressed plants compared to well-watered plants, for 15 and 22 day-old leaves ( Figure 3A ). However, the rate of decline in photosynthetic rate (A) with different leaf positions varied significantly (P = 0.03) for well-watered plants only ( Figure 3A ). Well-watered plants showed significantly higher g _sc compared to water-stressed leaves, only for 1 day- and 8 day-old leaves, whereas for the oldest leaf measured at day 22, water-stressed plants had higher g _sc than well-watered leaves at the same age. There was no water effect on g _m for any leaf ages except 15 day-old leaves ( Figure 3C ). Water-stressed plants had significantly higher A/g _sw compared to well-watered leaves for all leaf ages ( Figure 3D ). Moreover, the rate of decline in A/g _sw with different leaf positions varied significantly (P< 0.03) for both well-watered and water-stressed plants ( Figure 3D )

Considering the youngest fully expanded leaf over 5 weeks of measurement, water-stress decreased A and g _sc (P = 0.01), but the proportional decrease in A was less than the decrease in g _sc, resulting in higher values of A/g _sw (P< 0.001) for water-stressed plants compared to well-watered plants. Leaf-intrinsic water-use efficiency increased with increasing plant age (P = 0.001) only for water stressed plant ( Figure 4D ). Water-stress did not significantly affect g _m for the youngest fully expanded leaf at any period.

Significant positive relationships (P< 0.001) were found between g _m and A, and between g _m and g _sc in both well-watered and water-stressed conditions. The relationship between g _m and A was stronger (well-watered: g _m = 0.028A – 0.078, r = 0.78, P< 0.001; water-stressed: g _m = 0.027A – 0.095, r = 0.58, P< 0.001) than between g _m and g _sc (well-watered: g _m = 1.90g _sc + 0.027, r = 0.69, P< 0.001; water-stressed: g _m = 1.59 g _sc + 0.19, r = 0.45, P< 0.001). However, there was no significant water availability effect (P > 0.05) on the association between g _m and A, or between g _m and g _sc ( Supplementary Figure S1 ).

Changes were observed in leaf structure as leaves aged, with young leaves having the highest surface area of chloroplast exposed to intercellular airspace (S _c) (P< 0.001), followed by mature leaves, and then old leaves ( Table 1 ). The ratio between S _c and the surface area of mesophyll cells exposed to the intercellular airspaces (S _mes) was significantly higher for young leaves compared to old leaves (P< 0.001), whereas there was no age effect on S _mes. Mesophyll thickness was marginally (but not significantly, P = 0.08) higher for well-watered compared to water-stressed plants, whereas S _c/S _mes was higher for water-stressed plants compared to well-watered plants (but not statistically significant, P = 0.06) ( Table 1 ). S _c declined more quickly with increasing age for well-watered leaves compared with that for water-stressed leaves ( Figure 5 ). Furthermore, the number of chloroplasts per unit leaf surface area was higher for water-stressed plants than for well-watered plants but did not change with leaf age. Taken together, these results suggest that water limitation had subtle effects on chloroplast number, size and distribution, and on the changes in these anatomical properties as leaves aged. Moreover, there was no water or leaf age effect on f _ias ( Supplementary Table 2 ).

There was a significant (P = 0.01) relationship between t _cw and different leaf stages; young leaves had significantly lower t _cw compared to old leaves regardless of the water treatment ( Figures 6 , 7 ). There was a significant positive relationship between S _c and g _m (P = 0.01) ( Figure 8A ) whereas, the relationship between t _cw and g _m was not significant (P = 0.22) ( Figure 8B ) for three different leaf stages at two water treatments.

After one day of re-watering, leaves were less water stressed as Ψ _L rose from -1.38 MPa to -1.10 MPa. Gas exchange parameters were measured on the same leaves for water-stress plants and one week later following re-watering. As expected, g _sc significantly increased after re-watering ( Figure 9B ), whereas A and A/g_sw did not change significantly following re-watering ( Figures 9A, D ). There was a significant interaction effect between re-watering and leaf age in g _m (P = 0.05) where g _m decreased for young leaves and increased for mature leaves ( Figure 9C ).

It is widely observed that stomatal conductance (g _sc) and photosynthetic rate decline as leaves aged and includes crops such as rice (Gu et al. (2012)) and maize (Azizian and Sepaskhah (2014)) and other herbaceous plants (e.g. Arabidopsis (Flexas et al. (2007)). Other examples include an alpine orchid (Zhang et al. (2008)) and woody species such as pine (Warren (2006)) and oak (Niinemets et al. (2005)). The decline in A has sometimes been attributed to a decline in mesophyll conductance (g_m ) in oak (Niinemets et al. (2005), and mustard (Monti et al. (2009)) which in some instances was due to changes in leaf anatomy including tobacco (Clarke et al., 2021), maple (Hanba et al. (2001) and poplar (Tosens et al. (2012)). However, other studies in pine report that g _m limitations to A did not increase with leaf age (Warren (2006)) and in southern beech, g _m itself was not significantly lower in mature compared to young leaves (Whitehead et al. (2011)). We found that A declined by 1.2 µmol m^–2 s^–1 day^-1, and that g _m declined by 0.035 mol m^-2 s^-1 bar^-1 day^-1, as leaves matured. Evans and Vellen (1996) also report a decline in A and g _m as wheat leaves aged but did not provide actual ages of the leaves so a direct comparison is not possible. We have previously observed declines in A and g _m of 1 µmol m^–2 s^–1 day^-1 and 0.05 mol m^-2 s^-1 bar^-1 day^-1, respectively, over one week for newly expanded leaves in the wheat cultivar ‘Scout’, but no significant decline in A or g _m over same the period for the cultivar ‘Cranbrook’ (Jahan et al., 2014). Clarke et al. (2021) observed that A and g _m decreased by 68% and 50% respectively among different leaf positions over 3 weeks. Similarly, we also observed that A and g _m decreased by 64% and 61% respectively among different leaves over the same 3-week period (from Figure 3 ).

Anatomical and structural changes are observed during and after leaf development. When fully mature Populus tremula leaves are aging (based on a leaf plastochron index, LPI, of 6 to 12), leaf thickness (t _leaf), mesophyll thickness (t _mes), S _mes and fraction of internal airspace (f_{i as}) remained constant (Tosens et al., 2012), while S _c, S _c/S _mes, thickness of chloroplasts (t _chl) and length of chloroplasts (L _chl) began to decrease (Tosens et al., 2012). In the current study, we also observed that S _c and S _c/S _mes decreased with increasing leaf age, but leaf age did not affect f _ias, and that this was due to a decline in the size of individual chloroplasts for young leaves compared to old leaves. Mesophyll conductance and S _c were strongly positively correlated in wheat leaves. Similarly, Evans (2020) also observed in his review paper that for any given species, g _m varies in direct proportion to S _c.

Clearly, S _c is partly responsible for the reduction in g _m observed in the current experiment. However, the age-related reduction in S _c is proportionally much smaller than in g _m, so other anatomical and metabolic factors must also play a role. In general, a negative relationship was found between cell wall thickness (t _cw) and g _m because of the longer pathway for CO₂ diffusion from the intercellular airspace to the plasma membrane (Peguero-Pina et al., 2012; Tosens et al., 2012; Giuliani et al., 2013; Peguero-Pina et al., 2015). In this experiment, old leaves have thicker t _cw compared to young or mature leaves, and overall g _m is not associated with t _cw; this indicates that across the leaf stages, cell wall thickness is responsible for age-related reduction in g _m for wheat cultivar ‘Tasman’. Evans (2020) reviewed g _m in C4 leaves and suggested that the effective porosity of cell walls decrease with increasing cell wall thickness. Hence, changes in the porosity of cell walls with varying leaf age need to be considered to identify the association of g _m and t _cw at different leaf ages. In addition to anatomical and structural factors, biochemical factors should be considered in association with g _m and t _cw at different leaf ages. For example, Warren (2006) observed that the down-regulated activity of Rubisco with age was responsible for the variation in g _m with leaf age for Pinus pinaster, and for wheat, Rubisco activation decreases in mature leaves compared to old leaves (Sharwood et al., 2016).

Mesophyll conductance can be divided into three components: conductance through intercellular airspaces (g _ias), through cell walls (g _w) and through the liquid phase inside cells (g _liq) (Flexas et al., 2008). In this experiment, gas-phase conductance from sub-stomatal cavities to the outer surface of cell walls (g _ias) was calculated as described by Tosens et al. (2012); we observed no leaf age effect on g _ias. Apart from anatomical changes, enzyme activity also typically declines as leaves age, including carbonic anhydrase, and aquaporins. Carbonic anhydrase (CA) catalyses the reversible interconversion of CO₂ with HCO₃ ^-, and occurs in multiple forms in the chloroplast, cytosol, mitochondria and plasma membrane (Fabre et al., 2007; Ogeíe et al., 2018). The influence of CA activity on photosynthesis in C₃ plants is small (Price et al., 1994; Williams et al., 1996), and the effects of CA on g _m have not been widely reported. Gillon and Yakir (2000) measured both g _m and total leaf CA activity in tobacco and soy, and reported no relationship between them. However, on a latitudinal genotype transect in Populus trichocarpa, northern genotypes exhibited higher A as a function of higher g _m and elevated CA activity (Momayyezi and Guy, 2017). However, it is well known that CA activity declines as leaves age (Novichkova et al., 2006), and therefore the decline in g _m as leaves age may be in part due to the decline in CA activity.

Aquaporins influence the diffusion of CO₂ through membranes and may regulate g _m (Hanba et al., 2004; Flexas et al., 2012; Uehlein et al., 2012; Ermakova et al., 2021). Terashima and Ono (2002) investigated the involvement of aquaporins in CO₂ diffusion across the plasma membrane of Vicia faba and observed that g _m decreased by 40% and 30% when the leaflets were fed with the aquaporin inhibitor 0.3 mM and 1.2 mM HgCl₂, respectively. Evans et al. (2009) reviewed the role of aquaporins in CO₂ diffusion conductance in plants and suggested that aquaporins are responsible for 25% of lipid-phase diffusion conductance. Moreover, Boron et al. (2011) also suggested that aquaporins play a key role in membrane permeability. A wide range of aquaporins are known in plant leaves, with more than 30 isoforms identified in higher plants, and they constitute the most abundant plasma membrane and tonoplast proteins (reviewed by Maurel et al., 2008). The expression of different aquaporins occurs during leaf development (Wei et al., 2007; Prodo and Maurel, 2013), so aquaporin expression and activity may contribute to age-related decline in g _m although this is yet to be demonstrated. Therefore, high membrane permeability may explain the lack of an association between g _m and cell wall thickness.

Further studies on membrane permeability and the porosity of cell wall are needed to examine the effects of leaf anatomical characteristics and metabolic factors (aquaporins) for different aged leaves on g _m in wheat plants. Obtaining a thorough understanding of the association of g _m with anatomical traits is important to interpret photosynthetic traits, which will lead to a better understanding of how to manipulate g _m to improve plant performance and increase yield in stressful environments.

A review by Flexas et al. (2007) demonstrated that a reduction in g _m in response to drought is commonly observed. For example, in Populus tremula, water stress reduced the exposed chloroplast to leaf area ratio (S _c/S) and increased cell wall thickness, and these changes combined to reduce g _m (Tosens et al., 2012). However, reduced g _m in response to drought is not always observed (Flexas et al., 2012), or is transient (Flexas et al., 2009; Galle et al., 2009). In the current experiment, one week after the water treatment started (week 6), water-deficit plants had slightly lower g _m (0.53 mol m^-2 s^-1 bar^-1) than irrigated plants (0.58 mol m^-2 s^-1 bar^-1) averaged across all leaf ages. Three weeks later, these same leaves of water-deficit plants (now 15- and 22-day-old leaves) had higher g _m and A than their well-watered counterparts. That is, the reduction in g _m and A as leaves aged was less pronounced in water-deficit plants than well-watered, perhaps reflecting a slowing of leaf development under water-deficit.

Wheat plants re-watered after a water stress event showed an increase in g _sc and different g _m recovery rates for different aged leaves, which indicates that photosynthetic machinery is preserved with no irreversible damage. Water stress has been implicated in reducing photosynthetic biochemistry such as maximum photochemical efficiency (F_v/F_m) and the efficiency of electron transfer through PSII (Φ_PSII), however, rewatering for 14 days restored Fv/Fm and Φ_PSII to pre-water deficit conditions (Wang et al., 2018). Mesophyll conductance (g _m) was somewhat reversible when water stress was released in sugar beet (Monti et al., 2006) while Galle et al. (2009) observed a rapid recovery of g _m in tobacco from water stress in Spring compared with Summer. Peguero-Pina et al. (2018) observed complete recovery of g _m in holm oak (Quercus ilex L.) after re-watering. Furthermore, it has been showed that g _m recovered more quickly in Eucalyptus dumosa than in Eucalyptus pauciflora because aquaporins were deactivated in E. pauciflora associated with higher oxidative stress (Cano et al., 2014). In this study, rates of g _m recovery for different leaf ages were not equal when comparing water-deficit plants and plants after re-watering. That is, young leaves did not recover to unstressed values after re-watering, while in mature leaves g _m increased after re-watering. There is no clear explanation for the differences in rates of recovery after rewatering, but these differences may relate to: a) the conditions under which the leaves formed (young leaves formed under water stress, while mature leaves did not); b) differences in leaf anatomy including maturity of stomata development (the degree of recovery of stomatal conductance was inversely related to leaf age); or c) redistribution of resources from old leaves to younger leaves.

We consider the applied water stress as moderate because when measured in high light and moderate vapour pressure deficit, stomatal conductance to water vapour (g _sw) was higher than 0.16 mol H₂O m^-2 s^-1 and the average leaf water potential (Ψ _L) was -1.38 MPa ( Supplementary Table 2 .). Similarly, Shrestha et al. (2018) also observed similar stomatal conductance to water vapour (g _sw) for three chickpea genotypes and described them as moderately water stressed. Leaf age, and/or duration of water stress can alter the recovery rate of leaf physiological parameters; e.g. Warren et al. (2011) found that long-term water stress delayed the recovery rate of A in Eucalyptus. Genotypic variation was also found to affect the recovery rate of A between two Eucalyptus species (Cano et al., 2014).