Leaf hydraulic vulnerability triggers the decline in stomatal and mesophyll conductance during drought in rice

Leaf hydraulic conductance plays a role in stomatal closure during soil drought, and reduction in CO₂ diffusion is a strong driver of the photosynthetic decline during drought.

Abstract

Understanding the physiological responses of crops to drought is important for ensuring sustained crop productivity under climate change, which is expected to exacerbate the frequency and intensity of periods of drought. Drought responses involve multiple traits, and the correlations between these traits are poorly understood. Using a variety of techniques, we estimated the changes in gas exchange, leaf hydraulic conductance, and leaf turgor in rice (Oryza sativa) in response to both short- and long-term soil drought. We performed a photosynthetic limitation analysis to quantify the contributions of each limiting factor to the resultant overall decrease in photosynthesis during drought. Biomass, leaf area, and leaf width significantly decreased during the 2-week drought treatment, but leaf mass per area and leaf vein density increased. Light-saturated photosynthetic rate declined dramatically during soil drought, mainly due to the decrease in stomatal conductance (g_s) and mesophyll conductance (g_m). Stomatal modeling suggested that the decline in leaf hydraulic conductance explained most of the decrease in stomatal closure during the drought treatment, and may also trigger the drought-related decrease of stomatal conductance and mesophyll conductance. The results of this study provide insight into the regulation of carbon assimilation under drought conditions.

Introduction

Plant productivity is significantly impacted by drought events, which are expected to occur more intensely and frequently as global climate change continues (Trenberth et al., 2013). To develop new approaches to improve crop production under future conditions of water limitation, the responses of several physiological processes, including photosynthesis, plant hydraulic conductivity, and cell turgor pressure, have been widely documented (Flexas et al., 2002; Grassi & Magnani, 2005; Flexas et al., 2009; Galle et al., 2011; Cano et al., 2013; Galmés et al., 2013; Rodriguez-Dominguez et al., 2016; Gleason et al., 2017; Martínez-Vilalta & Garcia-Forner, 2017); however, the correlations among these physiological traits have not been fully evaluated under drought conditions.

In C₃ plants, the light-saturated leaf photosynthetic rate (A) is limited by stomatal conductance (g_s), mesophyll conductance to CO₂ (g_m), and/or the photosynthetic biochemistry related to either carboxylation velocity, V_cmax, or the maximum electron transport rate set by photochemical and Calvin cycle activities, J_max (Tosens et al., 2012; Tomás et al., 2013; Tosens et al., 2016; Veromann-Jürgenson et al., 2017). Grassi and Magnani (2005) developed a method to estimate the partial contribution of each limiting factor to the overall reduction of photosynthesis; this approach has since been applied to many species under a variety of environmental stresses (Flexas et al., 2009; Galle et al., 2009; Galle et al., 2011; Galmés et al., 2013; Wang et al., 2018). Although the limiting effects of g_s, g_m, and photosynthetic biochemistry on A are dependent on the species, A has been suggested to be first inhibited by a decrease in g_s and g_m under drought conditions, with the biochemical inhibition occurring later, under more severe drought stress conditions (Grassi & Magnani, 2005; Flexas et al., 2009; Galle et al., 2009; Galle et al., 2011; Galmés et al., 2013; Galmés et al., 2017). However, the contribution of each limiting factor to A under drought conditions, especially dynamic drought conditions, is unknown for rice (Oryza sativa), despite its status as one of the most important cereal crops in the world.

When plants are exposed to drought, their stomata close, preventing a decline in leaf water potential (ψ_leaf) and thereby ensuring that the water demand in leaves does not exceed the safe threshold of the hydraulic system (Scoffoni et al., 2017b); however, the mechanisms underlying stomatal closure in response to soil drought are poorly understood. Both hormonal (Dodd, 2005) and leaf turgor (Sperry et al., 2002; Buckley, 2005; Brodribb & Cochard, 2009; Rodriguez-Dominguez et al., 2016) signals have been proposed to explain stomatal closure in angiosperms during drought conditions. The hormonal hypothesis suggests that stomatal closure in the leaves is principally driven by hormonal signals, especially abscisic acid (ABA) produced de novo in the leaf (Holbrook et al., 2002; McAdam et al., 2016; Zhang et al., 2018). The leaf turgor hypothesis proposes that the decline in g_s during soil drought is caused by change in leaf turgor. Recently, a serial study (McAdam & Brodribb, 2016; McAdam et al., 2016) tried to link these two hypotheses by demonstrating that, in response to low relative humidity, ABA is rapidly synthesized de novo and accumulates in the leaf once the leaf turgor declines in angiosperms. By contrast, a recent theoretical analysis suggested that ABA accumulation in dehydrated leaves is associated with a decline in cell volume, rather than a loss of turgor pressure (Sack et al., 2018).

A decrease in g_m in response to soil drought was also observed in many previous studies, although the mechanisms for this decrease are unclear (Flexas et al., 2002; Grassi & Magnani, 2005; Warren, 2008; Galle et al., 2009; Cano et al., 2013; Théroux-Rancourt et al., 2014). Many studies have demonstrated the parallel responses of g_s and g_m to environmental changes (see review in Flexas et al., 2012). The physiological basis of this relationship is largely unknown; however, recent studies in plant hydraulics suggest that leaf hydraulic conductance (K_leaf) mediates the covariation of g_s and g_m (Flexas et al., 2013; Xiong et al., 2015b; Xiong et al., 2017; Xiong et al., 2018). The liquid water transport pathways in the mesophyll are partially shared with the CO₂ diffusion pathways; hence, a functional linkage between g_m and K_leaf has been suggested. Similarly, g_s and K_leaf may be coupled because of the common stomatal pathway for the exchange of water and CO₂ between the leaf and the atmosphere. Correlations between K_leaf and g_s or g_m have been observed in many species and genotypes (Brodribb et al., 2005; Brodribb et al., 2007; Flexas et al., 2013; Théroux-Rancourt et al., 2015; Xiong et al., 2018). Nonetheless, it is unclear whether a coordinated regulation of g_s, g_m, and K_leaf occurs under varied environmental conditions, for instance, during water stress. Indeed, K_leaf declines rapidly between full turgor and the turgor loss point and even more strongly during extreme dehydration (reviewed in Scoffoni et al., 2017b). The response of K_leaf to dehydration has been suggested to arise mainly due to the vulnerability of tissues outside the xylem, such as mesophyll (Scoffoni et al., 2017a), the major tissue where water transport and CO₂ diffusion may share a common pathway (Xiong et al., 2017). Théroux-Rancourt et al. (2014) observed that K_leaf and g_s decreased as the soil water potential declined, but that g_m decreased only after g_s was <0.15 mol m⁻² s⁻¹ in poplars (Populus sp.). Revealing the regulatory patterns of these traits in response to drought is necessary for enhancing our understanding of plant responses to water limitation (Scoffoni et al., 2017b).

In this study, we estimated gas exchange, K_leaf, and leaf turgor in response to both short- and long-term soil drought in two rice genotypes to reveal the correlations between and sequences of changes in these traits during the response to drought stress. The objectives of this study were (i) to reveal the dynamic limiting effects of g_s, g_m, and the photosynthetic biochemistry on A during drought in rice; and (ii) to clarify the vulnerabilities of A, g_s, g_m, and K_leaf and their relationships under drought conditions.

Materials and methods

Plant materials and growth conditions

Two ‘Super’ hybrid rice cultivars, Yangliangyou 6 (YLY6) and Chaoyou 1000 (CY1000), were used in this study. YLY6 is a widely used reference cultivar in promotion trials of newly developed ‘Super’ varieties in China, while CY1000 is a recently developed ‘Super’ variety with high yield and wide adaptation characteristics. Seeds were germinated and grown in a nursery for 2 weeks, and the seedlings were then transplanted into 11 litre plastic pots containing 10 kg of soil, at a density of three plants per pot. Before transplantation, 7.0 g of compound fertilizer (N:P₂O₅:K₂O=16:16:16%; Batian Ecological Engineering Limited, Shenzhen, China) was mixed into the soil of each pot. For each genotype, 60 randomly arranged pots of seedlings were grown, and pots were watered daily before the drought experiment began. Seven weeks after transplantation, 10 pots of each genotype were subjected to long-term water deficiency stress by maintaining a relative soil water content of ~75% for 2 weeks (Fig. 1A).

Fig. 1.

(A) Soil relative water content of the well-watered (WW) and water-deficient (WD) treatments. (B) Predawn leaf water potential (ψ_predawn) and leaf osmotic potential (ψ_osmotic) of two rice genotypes after a 2-week drought treatment. See Supplementary Table S1 for definitions of the parameters. (C) Responses of leaf traits to 2 weeks of drought. The response was calculated as ln(X_WD/X_WW), where X_WD and X_WW were mean values of trait X under the WD and WW treatments, respectively.

Gas exchange and chlorophyll fluorescence measurements

To avoid the effects of fluctuations in outdoor air temperatures, light intensity, and humidity (see Supplementary Fig. S1 at JXB online) on gas exchange, each measurement was taken between 09.00 h and 16.00 h in an environmentally controlled growth chamber (Model GR48; Conviron, Controlled Environments Limited, Winnipeg, MB, Canada), with an air temperature of 25 °C, a relative air humidity of 70%, and a photosynthetic photon flux density (PPFD) of 600 μmol m⁻² s⁻¹. The night before the gas exchange measurements, the second fully expanded leaf of each plant was covered with both a plastic sheet and aluminum foil to estimate the stem water potential (ψ_stem) of the plant. After acclimating the plants overnight in the growth chamber, gas exchange measurements were carried out on the uppermost newly and fully expanded leaves, using a LI-6400 portable photosynthesis system equipped with a LI-6400–40 chamber (LI-COR Inc., Lincoln, NE, USA). In the leaf chamber, the PPFD was maintained at 1500 μmol m⁻² s⁻¹, the leaf-to-air vapor pressure deficit (VPD) was 1.5–2.0 kPa, and the CO₂ concentration was adjusted to 400 μmol mol⁻¹ using a CO₂ mixer. The block temperature during the measurements was set to 25 °C. After stabilization to a steady state, the gas exchange parameters, steady-state fluorescence (F_s), and maximum fluorescence (F′_m) were recorded. The actual photochemical efficiency of photosystem II (Φ_PSII) was calculated as follows:

$${\Phi }_{\text{PSII}}\text{=}\frac{(F_{\text{m}}^{`}-F_{\text{s}})}{F_{\text{m}}^{`}}$$ (1)

The electron transport rates (J_f) were computed as follows:

$$J_{\text{f}}={\Phi }_{\text{PSII}}\cdot \text{PPFD}\cdot \alpha \cdot \beta$$ (2)

where α is the leaf absorbance and β represents the distribution of electrons between photosystem I and photosystem II. After the gas exchange measurement, both ψ_stem and the leaf water potential (ψ_leaf) were determined using a pressure chamber (PMS Instrument Company, Albany, OR, USA) after equilibrating for at least 30 min.

To estimate α and β, light response curves for both well-watered and water-stressed plants were measured. The gas exchange system was switched to a low O₂ concentration (<2%) by injecting pure N₂, and simultaneous measurements of the light response curves and chlorophyll fluorescence were performed. During the measurements, the chamber conditions were the same as those described above, except that a gradient of PPFD values was used: 2000, 1500, 1200, 1000, 800, 600, 400, 200, 100, and 0 µmol m⁻² s⁻¹. After reaching a steady state, the parameters of gas exchange and chlorophyll fluorescence were simultaneously recorded. The slope of the relationship between Φ_PSII and 4Φ_CO2 (the quantum efficiency of CO₂ uptake) was considered to represent the value of α·β (Valentini et al., 1995). As there were no differences in the α·β values between the control and water-stressed leaves, the average value for all genotypes was used.

The mesophyll conductance of CO₂ (g_m) was calculated based on the variable J method described by Harley et al. (1992), as follows:

$$g_{\text{m}}=A/\text{(}{C}_{\text{i}}-\frac{\Gamma *\left(J_{\text{f}}+\text{8}\left(A+R_{\text{d}}\right)\right)}{J_{\text{f}}-\text{4}\left(A+R_{\text{d}}\right)}\text{)}$$ (3)

where Γ* represents the CO₂ compensation point in the absence of respiration, R_d is the day respiration rate, which was assumed to be half of the dark respiration rate (R_dark), C_i represents the intercellular CO₂ concentration, which was determined from an estimation of the cuticular conductance (see below) in this study, and C_c is the CO₂ concentration in the chloroplast. Г* is related to the Rubisco specific factor (S_C/O), which is relatively conserved at a given temperature. In the present study, the rice S_C/O at 25 °C was obtained from Hermida-Carrera et al. (2016).

Cuticular conductance and C_i calibration

The method of Sack and Scoffoni (2011) was used to estimate the minimum leaf conductance (g_cut). The leaves were scanned using a Canon EOS M50 (Canon Inc., Tokyo, Japan) to calculate their area, and then dried in a room with an air temperature of 25.0 °C and a light intensity of <5 µmol m⁻² s⁻¹. Leaves were weighed every 10 min over ~300 min using a digital balance (Sartorius BP 2215, Gottingen, Germany). The cuticular transpiration rate was determined from the regression of the change in leaf mass over time. Temperature and humidity sensors (HOB; H21-002; Onset Computer Corporation, Bourne, MA, USA) were placed next to the samples, and the air temperature and relative humidity were recorded at the beginning of each weighing cycle to determine the VPD. The value of g_cut was calculated as the transpiration rate divided by the VPD.

It is a widely accepted norm that water vapor diffusing through stomata can be used to calculate the C_i; however, the calculations assume an identical gas phase path for CO₂ and water vapor, which does not hold under drought conditions. As stomata close, the cuticle becomes the dominant path of water vapor diffusion (Boyer et al., 1997; Boyer, 2015a; Hanson et al., 2016). Indeed, it has been suggested that C_i could potentially be overestimated, as the cuticular conductance is far greater for water than for CO₂ (Boyer, 2015b). Thus, in this study, we recalculated C_i to take g_cut into account, as follows:

$$C_{\text{i}}=\frac{\text{(}{g}_{\text{sc}}\text{-}{E}_s/2\text{)}\cdot C_{as}-A}{g_{\text{sc}}\text{+}{E}_s/2}$$ (4)

$$g_{\text{sc}}=\text{(}{g}_{\text{lw}}\text{-}{g}_{\text{cut}}\text{)}/1.6$$ (5)

$$E_{\text{s}}=E_{\text{l}}-g_{\text{cut}}(W_{\text{l}}-W_a)$$ (6)

where C_as is the CO₂ concentration at the leaf surface (400 µmol mol⁻¹), g_sc is the true stomatal conductance to CO₂, E_s is stomatal transpiration, g_lw is the total leaf conductance to water, E_l is leaf transpiration, and W_l and W_a are the water vapor values inside and outside the leaf, respectively.

Hydraulic vulnerability

Three methods were used to estimate K_leaf: the standard evaporative flux method (EFM), the rehydration kinetic method (RKM), and the gas exchange-based EFM method. The EFM was calculated following the methods outlined by Scoffoni et al. (2012) and Xiong et al. (2017). The rice tillers were bench-dried, and then the initial leaf water potential (ψ₀) was measured in the neighboring leaves. The dehydrated leaves were excised from the tillers under water and connected to a tube system, which was connected to a reservoir of degassed water situated on a high-precision digital balance (NBL 84e, Adam Equipment Inc., Oxford, UK). The balance logs data to a computer every 10 s. Once the water flow rate was stable (~30 min), the water flow rate (E) into the leaves under favorable conditions (on a large box fan with PPFD>1000 µmol m⁻² s⁻¹) was recorded, along with the leaf temperature. Next, the leaf area (LA) and final leaf water potential (ψ_final) were measured. The K_leaf-EFM was calculated as follows:

$$K_{\text{leaf-EFM}}=\frac{E}{\text{LA}\cdot {\text{(0-ψ}}_{\text{final}}\text{)}}$$ (7)

The RKM was calculated following the method outlined by Blackman and Brodribb (2011). The ψ₀, leaf temperature, initial maximum rehydration flow of water into leaves (I), and LA were measured in a similar manner to the EFM method described above, except that the leaves were covered with moist paper and were not exposed to light, in order to prevent transpiration during the I measurement. I was calculated by fitting an exponential curve through the first 10 s of the flow data and extrapolating back to the initial point of leaf excision. K_leaf-RKM was calculated as follows:

$$K_{\text{leaf-RKM}}=\frac{I}{\text{LA}\cdot {\text{ψ}}_0}$$ (8)

We also measured K_leaf using the transpiration rate (T_r) values from the gas exchange measurement and ψ_stem, and ψ_leaf after the gas exchange. For this method, K_leaf-licor was calculated as follows:

$$K_{\text{leaf-licor}}=\frac{T_{\text{r}}}{{\text{ψ}}_{\text{stem}}-{\text{ψ}}_{\text{leaf}}}$$ (9)

To construct the vulnerability curve, K_leaf was plotted against the lowest ψ_leaf (i.e. ψ₀ in RKM; ψ₀ or ψ_final in EFM, and ψ_leaf in the gas exchange method). Because the viscosity of water is temperature dependent, the K_leaf values in this study were standardized to their corresponding value at 25 °C (Scoffoni et al., 2012).

Pressure–volume curves

Four pressure–volume curves per genotype were conducted with well-watered plants, to estimate their osmotic potential at full turgor (π₀) and at the turgor loss point (π_tlp), as well as their modulus of elasticity (ɛ) (Sack et al., 2003; Scoffoni et al., 2011). Leaves were sampled from well-watered plants and rehydrated overnight before desiccation. Briefly, the leaf weight and ψ_leaf were measured at least 10 times over the desiccation period until ψ_leaf dropped to –3.0 MPa. Finally, the leaves were dried at 70 °C for 2 days and their dry mass was measured.

Osmotic potential measurements

The fully expanded young leaves of well-watered and water-stressed plants were sampled in the morning. The leaf samples were immersed in liquid nitrogen and then stored at –80 °C. The osmotic potentials of these leaves were measured using a vapor pressure osmometer (VAPRO 5520; Wescor Inc., Logan, UT, USA).

Leaf vein density

The newly developed and fully expanded leaves of both well-watered and water-stressed plants were chemically cleared in 15% NaOH (w/v) and then bleached following the standard protocol for rice (Xiong et al., 2015b; Xiong et al., 2017). The cleared leaves were stained with safranin and fast green in ethanol. After being rinsed in water, the leaves were scanned using a Canon EOS M50 (Canon Inc., Tokyo, Japan) to enable quantification of their area and major vein lengths. To measure the minor veins, a light microscope (U-TVO.5XC; Olympus, Tokyo, Japan) with a 5× objective was used to observe the leaves, and photographs were taken of the top, middle, and bottom of each leaf. LA and vein length were manually measured using ImageJ (Wayne Rasband/NIH, Bethesda, MD, USA). The total vein density (VLA), major vein density (VLA_major, including the midrib and large veins), and minor vein density (VLA_minor) were estimated.

Biomass and leaf area

Four plants per treatment were sampled after 2 weeks of drought treatment and were separated into stems and leaves. The LA was measured using a LA meter (LI-3100; LI-COR Inc., Lincoln, NE, USA). The samples were dried to a constant weight at 80 °C and their biomass was recorded.

Photosynthetic limitation analysis

A limitation analysis is a helpful tool for quantifying the effects of stress on various factors affecting A (Grassi & Magnani, 2005; Buckley & Diaz-Espejo, 2015). The relative photosynthetic limitations, including the relative stomatal (l_s), mesophyll (l_m), and biochemical (l_b) limiting effects, were modeled as previously described by Grassi and Magnani (2005):

$$l_{\text{s}}=\frac{g_t\text{/}{g}_{\text{sc}}\cdot \partial A\text{/}\partial C_{\text{c}}}{g_t+\partial A\text{/}\partial C_{\text{c}}}$$ (10)

$$l_{\text{m}}=\frac{g_t\text{/}{g}_{\text{m}}\cdot \partial A\text{/}\partial C_{\text{c}}}{g_t+\partial A\text{/}\partial C_{\text{c}}}$$ (11)

$$l_{\text{b}}=\frac{g_t}{g_t+\partial A\text{/}\partial C_{\text{c}}}$$ (12)

where g_t is the total conductance, which is calculated as:

$$g_{\text{t}}=\frac{1}{\frac{1}{g_{\text{sc}}}+\frac{1}{g_{\text{m}}}}$$ (13)

To assess the impact of ψ_leaf change on photosynthesis, the limiting effects were linked to overall changes in A:

$$\frac{dA}{A}=\text{LS+LM+LB}=\frac{d{g}_{\text{sc}}}{g_{\text{sc}}}{l}_{\text{s}}+\frac{d{g}_{\text{m}}}{g_{\text{m}}}{l}_{\text{m}}+\frac{d{J}_{\text{f}}}{J_{\text{f}}}{l}_{\text{b}}$$ (14)

where LS, LM, and LB are the reduction fractional limitations in A caused by a reduction in stomatal conductance, mesophyll conductance, and biochemistry, respectively. In the current study, the fitted photosynthetic parameters at ψ_leaf = –0.3 MPa were used as reference values. Thus,

$$\frac{dx}{x}\cong \frac{x_{0.3}-x}{x_{0.3}}$$ (15)

where x represents the fitted g_s, g_m, or J_f, (see Supplementary Table S1 for definitions of these and other mathematical parameters used in this paper), and x_0.3 represents the x value at ψ_leaf = –0.3 MPa.

Quantification of the contributions of hydraulic and hormonal signals to stomatal closure

The g_s model, originally presented by Buckley et al. (2003) and subsequently modified by Rodriguez-Dominguez et al. (2016), was used to examine the contributions of hydraulic and hormonal signals to the decline in g_s under drought. In this model, g_s is expressed as:

$$g_{\text{s}}=\frac{na{K}_{\text{leaf}}({\psi }_{stem}+\pi )}{K_{\text{leaf}}+na\Delta w}$$ (16)

where π is bulk leaf osmotic pressure, Δw is the leaf-to-air water vapor mole fraction gradient, n represents the effect of hormonal signals on the sensitivity of guard cell osmotic pressure to leaf turgor, and a represents the relative adenosine triphosphate concentration. In this study, K_leaf, ψ_stem, Δw, and π were measured, a was simulated, and n was fitted. The π value was measured using a WP4C water potential meter (Decagon, Pullman, WA, USA).

Statistical analysis

Regressions were fitted with a linear model, and regression lines are shown when P<0.05. The correlations between the leaf functional traits (K_leaf, A, g_s, g_m, and J_f) and ψ_leaf were tested by four functions described in a previous study (Scoffoni et al., 2012): a linear function (K_leaf = aψ_leaf+b), a sigmoidal function ($K_{\text{leaf}}=\frac{a}{\text{1+}{e}^{\text{-(}\frac{{\text{ψ}}_{\text{leaf}}-x_0}{b}\text{)}}}$), a logistic function ($K_{\text{leaf}}=\frac{a}{\text{1+(}\frac{{\text{ψ}}_{\text{leaf}}}{x_0}{\text{)}}^b}$), and an exponential function ($K_{\text{leaf}}=y_0+a\cdot e^{\text{-}b\cdot {\text{ψ}}_{\text{leaf}}}$). The functions were compared using the Akaike Information Criterion (AIC) corrected for low n. The function with the lowest AIC value was chosen as the maximum likelihood function. The differences of K_leaf vulnerability among genotypes and methods were compared using a two-sample Kolmogorov–Smirnov test. All of the analyses were performed in the program R (R Core Team, 2018).

Results

Effects of 2 weeks of water stress on plant performance

The 2-week drought treatment led to a significant reduction in rice biomass (by 17.5% in CY1000 and 20.9% in YLY6; Fig. 1). Drought stress significantly increased LMA (by 13.8% in CY1000 and 15.5% in YLY6) and VLA (by 25.1% in CY1000 and 5.7% in YLY6), but decreased LA (by 14.1% in CY1000 and 15.0% in YLY6) and leaf width (LW) (by 10.8% in CY1000 and 9.3% in YLY6). VLA_major increased by 11.33% in CY1000 under water stress, but no changes were observed in YLY6; more pronounced increases in VLA_minor were observed for both genotypes (28.4% increase in CY1000 and 6.8% in YLY6).

Drought stress significantly decreased the gas exchange and leaf hydraulic traits in both rice genotypes, with more pronounced effects in YLY6 than CY1000 (Fig. 1). Overall, water stress decreased A, g_s, g_m, and K_leaf by 28.4%, 43.0%, 19.6%, and 50.2%, respectively. The leaf osmotic potential (ψ_osmotic) increased by 30.2% in CY1000 and 21.0% in YLY6 following the 2-week drought treatment. In addition, water stress decreased g_cut in CY1000, but not in YLY6 (Fig. 1).

Leaf hydraulic and photosynthetic dynamics during short-term drought

The gas exchange and leaf hydraulic traits of rice were sensitive to short-term drought (Table 1; Figs 2 and 3). A, g_s, and g_m declined exponentially with decreasing ψ_leaf, with very similar patterns observed for the two genotypes (Fig. 2). Overall, the maximum A, g_s, and g_m values were 19.30 µmol CO₂ m⁻² s⁻¹, 0.31 mol H₂O m⁻² s⁻¹, and 0.14 mol CO₂ m⁻² s⁻¹, respectively, with slightly higher values in YLY6 than CY1000. To quantify the sensitivity of the gas exchange traits to leaf drying, we estimated the leaf water potential values at 50% and 80% loss of function (P₅₀ and P₈₀, respectively; Table 1; Fig. 2). The P₅₀ values for A, g_s, and g_m were –0.99 MPa, –0.93 MPa, and –1.03 MPa, respectively; however, the P₅₀ of J_f was –1.99 MPa, which was lower than that of A.

Table 1.

Pressure–volume, gas exchange, and leaf hydraulic vulnerability parameters of rice

Trait	CY1000	YLY6	Mean
SWC (g g−1)	2.31 ± 0.05	2.22 ± 0.06	2.27
π0 (MPa)	–0.67 ± 0.14	–0.83 ± 0.11	–0.75
πtlp (MPa)	–1.13 ± 0.12	–1.19 ± 0.06	–1.16
ɛ (MPa)	8.41 ± 2.10	6.96 ± 0.79	7.69
K max-RKM (mmol m−2 s−1 MPa−1)	8.02	9.25	8.47
K max-EFM (mmol m−2 s−1 MPa−1)	12.2	11.73	11.9
K max-licor (mmol m−2 s−1 MPa−1)	15.4	15.47	15.5
A max (µmol CO2 m−2 s−1)	18.6	20.23	19.3
g smax (mol H2O m−2 s−1)	0.29	0.33	0.31
g mmax (mol CO2 m−2 s−1)	0.13	0.15	0.14
P50-RKM (MPa)	–0.82	–0.85	–0.84
P50-EFM (MPa)	–0.91	–0.82	–0.87
P50-licor (MPa)	–0.64	–0.64	–0.64
P50-A (MPa)	–0.93	–1.01	–0.99
P50-gs (MPa)	–0.91	–0.94	–0.93
P50-gm (MPa)	–0.99	–1.05	–1.03
P50-Jf (MPa)	–1.99	–1.98	–1.99

See Supplementary Table S1 for definitions of the parameters.

Fig. 2.

Response of the gas exchange parameters to decreasing (A–D) leaf water potentials (ψ_leaf) and (E–H) stem water potentials (ψ_stem). The vertical solid and dotted lines indicate the water potential at 50% and 80% loss of function, respectively. The triangle represents the turgor loss point (Table 1). Fitted lines are the best-fit functions selected using maximum likelihood.

Fig. 3.

Vulnerability of leaf hydraulic conductance (K_leaf) estimated by (A) the standard evaporative flux method (EFM), (B) the rehydration kinetic method, and (C) the gas exchange based EFM method. The vertical solid and dotted lines indicate the water potentials at 50% and 80% loss of K_leaf, respectively. The triangle represents the turgor loss point (Table 1). Fitted lines are the best-fit functions selected using maximum likelihood.

K _leaf vulnerability curves were determined using three independent methods (Fig. 3). Although the curves of the two genotypes were indistinguishable when estimated with the same method (Supplementary Table S2), the curves estimated using the three methods were different (Supplementary Table S3). The maximum K_leaf from the gas exchange based EFM method was 15.5 mmol m⁻² s⁻¹ MPa⁻¹, almost twice as high as the 8.5 mmol m⁻² s⁻¹ MPa⁻¹ estimated in the RKM method (Table 1). The P₅₀ values of K_leaf were –0.84 MPa, –0.87 MPa, and –0.64 MPa for the RKM, EFM, and gas exchange based EFM methods, respectively. Moreover, the pressure–volume traits were similar in the two rice genotypes (Table 1), with average values for π₀, π_tlp, and ɛ of –0.75 MPa, –1.16 MPa, and 7.69 MPa, respectively.

Photosynthetic limitation analysis

Both g_s (r²=0.78; P<0.001) and g_m (r²=0.69; P<0.001) were tightly correlated with A during the drought treatment (Fig. 4A, B). Close correlations were also observed between K_leaf and g_s (r²=0.39; P<0.001), and between K_leaf and g_m (r²=0.31; P<0.001). The impacts of drought on the relative stomatal (l_s), mesophyll (l_m), and biochemical (l_b) limitations are shown in Fig. 5A. g_m was found to be the major limiting factor for photosynthesis in rice, as l_m contributed more than 40% of the relative limitation at any level of ψ_leaf. As ψ_leaf decreased in response to soil drought, both l_s and l_m increased; however, l_b declined dramatically. Diffusion processes appear to have a prominent role in limiting photosynthesis during soil drying (Fig. 5B), with diffusion through stomata (LS) and mesophyll (LM) having the greatest effect. Overall, the diffusion limitation (LM+LS) reached ~50% at a ψ_leaf of –1.0 MPa, while the contribution of biochemistry (LB) to the limitation of photosynthesis was very small.

Fig. 4.

(A, B) Relationships between light-saturated photosynthetic rate (A) and stomatal conductance (g_s) or mesophyll conductance to CO₂ (g_m). (C, D) Relationships between gas exchange based EFM estimation of leaf hydraulic conductance (K_leaf) and g_s or g_m.

Fig. 5.

Effects of leaf water potential (ψ_leaf) on (A) the distribution of the relative limits on photosynthesis caused by stomatal diffusion (l_s), mesophyll diffusion (l_m), and biochemistry (l_b), and (B) the overall ψ_leaf-dependent reduction in photosynthesis due to stomatal diffusion (LB), mesophyll diffusion (LM), and photosynthetic biochemistry (LB).

We fitted the stomatal model to our data and then partitioned the observed declines in g_s into contributions from each variable in the model (Fig. 6). The turgor-independent parameter, n, declined dramatically with leaf dehydration, with a 4-fold decrease as the leaf water potential decreased from 0 to –1.5 MPa during soil drought. The a parameter was quite stable during leaf dehydration; however, the leaf osmotic pressure, π, increased exponentially under drought.

Fig. 6.

Responses of variables in the stomatal model to changes in leaf water potential (ψ_leaf). (A) n, a turgor-independent parameter representing the effects of hormonal signals on the sensitivity of guard cell osmotic pressure to leaf turgor. (B) a, the relative concentration of adenosine triphosphate. (C) π, leaf osmotic pressure. (D) The parameters normalized by their values at a ψ_leaf of 0 MPa. K, gas exchange based EFM estimation of K_leaf, as in Fig. 3C.

Discussion

In the present study, A declined under water stress, which resulted in a significant decrease in biomass accumulation. Photosynthesis in C₃ plants such as rice is limited by g_s, g_m, and/or the biochemistry of photosynthesis itself, including the enzymes and metabolites involved in the process as well as components of the thylakoid electron transport chain (Flexas, 2016). Our analysis showed that the total relative limitation of photosynthesis by g_s and g_m was greater than 80% in rice when the ψ_leaf dropped to –1.0 MPa following soil drying (Fig. 5A). The results of the present study, as well as those of previous studies (Flexas et al., 2002; Galle et al., 2011; Galmés et al., 2013), highlight a major role for CO₂ diffusion in limiting A under conditions of water stress.

The decrease in g_s can be largely explained by K_leaf vulnerability under drought

We found that the g_s of rice declined with decreases in the stem (ψ_stem) and leaf (ψ_leaf) water potentials under drought conditions (Fig. 2). This decrease in g_s during soil drought has been widely studied, although the mechanisms for this response remain unclear. Two major mechanisms that regulate stomatal closure under drought conditions have been suggested to involve hydraulic (Sperry et al., 2002; Buckley, 2005; Brodribb & Cochard, 2009; Rodriguez-Dominguez et al., 2016) or hormonal (Dodd, 2005; McAdam et al., 2016) processes. Although hormonal signals were not measured in the present study, we quantified the responses to the hormonal and hydraulic signals of drought by modeling them (Fig. 6). Consistent with the findings of Rodriguez-Dominguez et al. (2016), we demonstrated that stomatal closure during drought can be largely explained by hydraulic signals, although hormonal signals also play a role in decreasing g_s. Nevertheless, a recent study suggested that the drought-induced decline in K_leaf in isohydric grapevine (Vitis vinifera) genotypes is regulated by ABA accumulation (Coupel-Ledru et al., 2017); thus, ABA could directly or indirectly regulate stomatal closure by decreasing K_leaf. Future studies are required to clarify the direct and indirect impacts of ABA on stomatal closure under soil drought conditions.

The mechanisms of K_leaf decline during dehydration are still largely unknown. K_leaf consists of at least two components, the conductance within the xylem (K_x) and the conductance through tissues outside the xylem (K_ox); therefore, the decline of K_leaf during dehydration could potentially be caused by changes in either or both of these factors. Increases in xylem tension during dehydration can cause air bubbles to form in the xylem conduit pit (Brodribb et al., 2016a; Brodribb et al., 2016b; Skelton et al., 2017), which decrease K_x. When the tension in the xylem conduits exceeds the biomechanical resistance of the cell wall, the conduits collapse (Cochard et al., 2004; Brodribb & Holbrook, 2005; Bouche et al., 2016). K_x vulnerability cannot always fully explain the observed decline in K_leaf; for instance, K_leaf can decline early, at high ψ_leaf, before an embolism has been observed (Brodribb & Holbrook, 2006; Scoffoni et al., 2012; Sack et al., 2016). Indeed, some direct insights have challenged the major role for K_x vulnerability in driving K_leaf decline (Trifiló et al., 2016; Scoffoni et al., 2017a). In this study, we did not separate the contributions of K_x and K_ox to the decline in K_leaf during drought; however, Stiller et al. (2003) reported that the P₅₀ of K_x in rice is approximately –2.0 MPa, which is far lower than that of the K_leaf observed here (Table 1), indicating that the decrease in K_leaf in rice during drought might be more closely connected to K_ox vulnerability. Water movement outside the xylem is complex and dynamic, involving apoplastic, symplastic, and transmembrane liquid flow paths and vapor diffusion within the intercellular airspaces (Buckley, 2015; Buckley et al., 2015; Buckley et al., 2017). During leaf dehydration, cells may be less well connected to each other owing to changes in their shape and size caused by leaf shrinkage. Indeed, the initial slope of the vulnerability curve, before the turgor loss point, has been suggested to be more related to decreases in K_ox than K_x (Scoffoni et al., 2014; Hernandez-Santana et al., 2016). In this study, K_leaf decreased sharply before π_tlp, suggesting that K_ox vulnerability played a major role in K_leaf decline. Moreover, the membrane permeability of the bundle sheath and mesophyll tissues has been suggested to influence K_leaf, and this effect may be related to the activities of aquaporins (Laur & Hacke, 2014; Sade et al., 2014b).

Currently, all methods for estimating K_leaf have limitations (both common and specific to each method) that require consideration when interpreting data (Flexas et al., 2013). As shown in Fig. 3, the K_leaf vulnerability curve of rice is method dependent, despite the similar values observed between genotypes for any given method. The K_leaf vulnerability curve produced using EFM has a similar shape to the one generated using RKM (see the equations in Fig. 3 and statistics in Supplementary Table S2); however, the EFM method shows a much higher maximum K_leaf (K_max; Table 1) value. Considering that the RKM measurement was performed in darkness, the difference may have been caused by light-dependent aquaporin activation (Cochard et al., 2007; Scoffoni et al., 2008). Indeed, we previously observed that rice K_leaf measured using EFM was strongly affected by light (Xiong et al., 2018). In addition, the shape of the K_leaf vulnerability curve generated using the gas exchange based EFM method clearly differed from those produced using the other two methods, especially at high ψ_leaf values (>–0.5 MPa). One possible reason for these high K_leaf values at high ψ_leaf could be the imprecise method used to measure ψ_stem. Although the leaves were wrapped and equilibrated overnight in this study, ψ_stem is technically challenging to measure precisely using pressure chambers in leaves that are close to full hydration.

Responses of g_m to short-term soil drought

As reported for many species (Grassi & Magnani, 2005; Flexas et al., 2006a; Warren, 2008; Flexas et al., 2009; Galle et al., 2009; Cano et al., 2013), we observed that g_m in rice decreased with soil drought. Methodological problems exist in all currently available estimation techniques for g_m (Tholen et al., 2012; Gu & Sun, 2014). One of the challenges for measuring g_m under drought conditions (low g_s) is the accurate estimation of C_i, because of the increasing relative contribution of g_cut to the overall leaf conductance, since the cuticular conductance for water is far greater than that for CO₂. In this study, we carefully ruled out the effects of g_cut on C_i. As it was not possible to estimate g_m under non-photorespiratory conditions using the variable J method, the effects of mitochondrial recycling of CO₂ on g_m were not estimated here; however, the 3-fold decrease in g_m observed in this study is unlikely to have been caused by (photo)respiration alone. We therefore assume that the decrease in g_m values observed during drought was mostly due to the decline of g_mper se.

The causes of the decrease in g_m during leaf dehydration are largely unknown, although g_m has been confirmed to be tightly correlated with mesophyll structure, membrane permeability, and the function of enzymes in the cytoplasm and chloroplast stroma (Flexas et al., 2008; Evans et al., 2009; Xiong et al., 2015a; Xiong et al., 2017). The two most important structural traits related to g_m are the cell wall thickness and the area of the chloroplast surface facing the intercellular airspace per unit leaf area (S_c; Evans et al., 2009; Tosens et al., 2012; Tomás et al., 2013; Tosens et al., 2016; Xiong et al., 2017). S_c is related to the mesophyll cells themselves, as well as to the shape of chloroplasts and the light-dependent arrangement of chloroplasts (Tholen et al., 2008). During leaf dehydration, the chloroplasts may move to reduce photodamage to the photosystems, and thus potentially change the values of S_c (Tholen et al., 2008). As for water transport outside the xylem, the decline of membrane permeability, mediated by aquaporins, is suggested to correspond with the decrease in g_m (Flexas et al., 2006b; Perez-Martin et al., 2014; Sade et al., 2014a). The change in cell wall properties might be one of the reasons for the decline in g_m under drought, as water stress usually introduces changes in the bulk elastic modulus of the cell wall (Brodribb & Holbrook, 2003; Saito & Terashima, 2004; Guyot et al., 2012), involving alteration of its biochemical composition and/or thickness. In addition, as shown in this study and previously (Théroux-Rancourt et al., 2014; Théroux-Rancourt et al., 2015), the K_leaf, A, g_s, and g_m vulnerability curves are almost always described as containing large measurement noise and/or high variability. Although estimation biases are inherently associated with all of the currently available techniques used to estimate K_leaf and g_m, the large number of different leaves used to construct the curves may be responsible for the major sources of variability. Hence, developing new methods to construct vulnerability curves based on a single leaf is perhaps one way to reduce the estimation variability in the future.

K _leaf vulnerability as a potential trigger for decline in g_s and g_m

Correlations between A, g_s, g_m, and K_leaf have been widely observed in many species, in part because of the common pathways for CO₂ diffusion and water transport within leaves, as well as between the leaf and atmosphere (Flexas et al., 2013; Théroux-Rancourt et al., 2015; Xiong et al., 2015b; Xiong et al., 2018). To determine whether these traits truly influence each other, these correlations would need to be observed for plants grown in the same conditions and measured under variable environmental conditions. As shown in Fig. 4, rice grown in the same environment and exposed to short-term changes in soil water content displayed a positive correlation between K_leaf and both g_s and g_m. Positive correlations between g_s and K_leaf across short-term environmental changes have been observed in many species (Théroux-Rancourt et al., 2015; Gleason et al., 2017; Xiong et al., 2018); however, the positive correlation observed between g_m and K_leaf contradicts the findings of Loucos et al. (2017), who found no correlation between these traits in cotton (Gossypium sp.) measured under different light intensities. Although different species were used, the reason for this discrepancy is unclear. However, our results do support previous observations in grapevine (Vitis sp.) and poplar (Populus sp.) subjected to short-term soil drought (Ferrio et al., 2012; Théroux-Rancourt et al., 2014).

One of the novel findings of this study is the role of K_leaf vulnerability in triggering the decrease in g_s and g_m. In general, changes in A, g_s, and g_m in response to ψ_leaf were similar to the K_leaf vulnerability curves in rice; however, the P₅₀ of K_leaf was higher than for g_s and g_m (Figs 2 and 3; Table 1). Our observations in rice disprove the previously proposed hypothesis, which suggested that stomata close early to reduce xylem tension and thus prevent plant hydraulic dysfunction (Cochard et al., 2004; Brodribb & Holbrook, 2006; Hochberg et al., 2017). As discussed above, K_leaf vulnerability might be largely determined by the non-xylem water movement pathways, and thus be influenced directly by the hydraulic effects that also trigger stomatal closure (Brodribb & Holbrook, 2003; Guyot et al., 2012). Indeed, a recent study showed that stomatal closure under drought is induced by hydraulic signals but maintained by ABA (Tombesi et al., 2015).Changes in hormone levels and/or leaf structural properties potentially decrease g_m. Interestingly, accumulation of ABA during drought conditions has been reported to decrease g_m significantly (Mizokami et al., 2015); moreover, a slight increase in the leaf ABA level is enough to decrease g_s, but decreases in g_m require higher leaf levels of ABA (Mizokami et al., 2015). The observation that g_s is more sensitive to drought than g_m may relate to the accumulation of ABA in leaves.

Modification of leaf anatomy facilitates the acclimation of leaf physiology to long-term drought

The acclimation of leaf anatomy and physiology to long-term drought was found to be coordinated in rice (Fig. 1). The LA and LW of the two rice genotypes displayed coordinated acclimation to drought, with the decrease in LA largely resulting from the narrowing of the leaf. Interestingly, a previous study found that grass species with naturally narrow leaves have high physiological drought tolerance (Craine et al., 2012). The decrease in LW is also associated with an increase in leaf vein density, which could result from the declining LW and/or increasing vein numbers. For instance, a perfectly coordinated acclimation of vein density and LW would suggest that vein spacing is determined passively by differences in leaf expansion. In this study, the major vein (VLA_major) and minor vein densities(VLA_minor) increased to different degrees under drought, suggesting that the increased leaf vein density is regulated both passively and actively in rice (Fig. 1). Moreover, the genotype-dependent differences in leaf vein density changes under drought may underpin the different drought tolerances of the two genotypes studied. The acclimation of the physiological traits to long-term drought was genotype dependent, providing further evidence that modification of leaf vein density facilitates the physiological acclimation to drought in rice. Higher leaf vein densities in drought-acclimated leaves have a higher hydraulic capacity, and thus assimilate higher quantities of carbon. Vein density is closely related to K_leaf because greater vein densities, especially of the minor veins, are associated with higher K_ox and K_x values (Buckley et al., 2015). In the present study, the responses of g_m and g_cut were also genotype dependent, suggesting that the mesophyll and epidermal tissues are also responsive to physiological acclimation in rice. However, we could not evaluate the effects of drought-induced anatomical and physiological changes on drought tolerance capacity because we did not construct pressure–volume curves and K_leaf vulnerability curves after drought treatment. Future research should focus on the effects of anatomical and physiological changes on drought tolerance.

In conclusion, these results provide new evidence that K_leaf and gas exchange are coordinated under drought conditions. Photosynthesis under drought conditions is primarily limited by g_s and g_m, and the decreased g_s was mainly determined by the decline in K_leaf, although it was also related to drought-induced hormonal signals. The decreased g_m and K_leaf are likely related to the changes in leaf anatomy and membrane permeability caused by drought.

Supplementary data

Supplementary data are available at JXB online.

Table S1. List of mathematical parameters and their units of measurement.

Table S2. Two-sample Kolmogorov-Smirnov test results in comparing K_leaf vulnerability of two rice genotypes.

Table S3. Two-sample Kolmogorov-Smirnov test results comparing K_leaf vulnerability methods.

Fig. S1. Climate information during the experiment (2017).

Author contributions

DX planned and designed the research; XW and TD performed the experiments; DX, XW, and TD analyzed the data and wrote the manuscript; all authors revised the manuscript.