Elevated CO₂ concentration increases maize growth under water deficit or soil salinity but with a higher risk of hydraulic failure

Abstract

Climate change presents a challenge for plants to acclimate their water relations under changing environmental conditions, and may increase the risks of hydraulic failure under stress. In this study, maize plants were acclimated to two different CO₂ concentrations ([CO₂]; 400 ppm and 700 ppm) while under either water stress (WS) or soil salinity (SS) treatments, and their growth and hydraulic traits were examined in detail. Both WS and SS inhibited growth and had significant impacts on hydraulic traits. In particular, the water potential at 50% loss of stem hydraulic conductance (P₅₀) decreased by 1 MPa in both treatments at 400 ppm. When subjected to elevated [CO₂], the plants under both WS and SS showed improved growth by 7–23%. Elevated [CO₂] also significantly increased xylem vulnerability (measured as loss of conductivity with decreasing xylem pressure), resulting in smaller hydraulic safety margins. According to the plant desiccation model, the critical desiccation degree (time×vapor pressure deficit) that the plants could tolerate under drought was reduced by 43–64% under elevated [CO₂]. In addition, sensitivity analysis showed that P₅₀ was the most important trait in determining the critical desiccation degree. Thus, our results demonstrated that whilst elevated [CO₂] benefited plant growth under WS or SS, it also interfered with hydraulic acclimation, thereby potentially placing the plants at a higher risk of hydraulic failure and increased mortality.

Acclimation, elevated CO₂ concentration, growth, hydraulic risk, maize, plant desiccation model, soil salinity, water deficit, xylem, Zea mays

Introduction

It is widely known that water shortages and soil salinity severely restrict agricultural production globally (FAO, 2020). The atmospheric CO₂ concentration ([CO₂]) has risen from ~280 ppm to the current level of ~416 ppm since the Industrial Revolution, and it is predicted to exceed 1000 ppm at the end of this century unless there are strict controls on CO₂ emissions (IPCC, 2014). Elevated [CO₂] has been hypothesized to ameliorate the impact of future drought (Kang et al., 2002; Li et al., 2018) and salinity (Pérez-López et al., 2009; Zaghdoud et al., 2013) on crops because it leads to improved water use efficiency. However, owing to the complex nature of plant responses to different stresses and elevated [CO₂], the important implications of rising [CO₂] for integrated plant–water dynamics and drought tolerance largely remain unresolved (Becklin et al., 2017).

Hydraulic failure is one of the major mechanisms in explaining plant responses to drought and increase in mortality (McDowell et al., 2008; Choat et al., 2018). Xylem conduits function under negative pressure, and they are therefore faced with the threat of cavitation (Sperry and Tyree, 1988), which disrupts water transport. Previous studies have shown that cavitation events commonly occur in crops (Tyree et al., 1986; Gleason et al., 2017). The water potential that leads to a 50% loss of hydraulic conductivity (P₅₀) is considered an important marker for drought resistance (Brodribb, 2009; Choat et al., 2012), while the potential that leads to an 88% loss (P₈₈) is commonly regarded as the critical level of hydraulic failure that leads to plant death in angiosperms (Kursar et al., 2009; Urli et al., 2013). In stressful environments, many species acclimate by reducing their xylem vulnerability to cavitation (Holste et al., 2006; Stiller, 2009; Awad et al., 2010; Plavcová and Hacke, 2012; Way et al., 2013; Hacke, 2014; Cardoso et al., 2018; Cary et al., 2020). However, there are limited numbers of studies on the acclimation of xylem vulnerability in stems of crops under elevated [CO₂], and interestingly contrasting results have been found among species. For example, xylem vulnerability decreases in sunflower (Helianthus annuus; Rico et al., 2013) but either increases or does not change in maize (Zea mays; Liu et al., 2020). It is important to note that there is an interaction between [CO₂] and drought. High [CO₂] leads to stomatal closure and consequently to higher plant water status (Hao et al., 2018; Li et al., 2019), which is likely to interfere with xylem acclimation. Consequently, information on hydraulic acclimation in response to the combination of elevated [CO₂] and various stresses is critical for understanding the likely resilience of crops under future climate change.

As well as the resistance of the xylem to embolism, the desiccation process is also influenced by the leaf minimal conductance and the interior water storage of the plant (Blackman et al., 2016). When plants experience drought, water uptake by the roots is limited or even stopped entirely. Hydraulic capacitance and water released from tissues can buffer the increased xylem tension and therefore delay or avoid hydraulic failure (Meinzer et al., 2009; Salomon et al., 2020). Under such conditions, transpiration is greatly reduced to the minimum leaf conductance (g_min), which represents leakage from the cuticle and the closed stomata (Duursma et al., 2019). Previous studies have indicated that plants growing under stress usually have leaves with lower g_min to reduce unnecessary water loss (Fanourakis et al., 2013; Duursma et al., 2019), but we lack an understanding of the acclimation response of g_min at elevated [CO₂].

In addition, we also lack knowledge of the acclimations of water storage to changing environments in herbaceous crops. The scale of this reserve of water depends on hydraulic capacitance and plant size (Blackman et al., 2019). The hydraulic capacitance is usually defined as the mass of water that can be released per unit of water potential per unit of tissue volume, dry mass, or area (Tyree and Ewers, 1991). Previous studies on the acclimation of crop hydraulic capacitance are limited, whilst those that have focused on tree species have shown that acclimation is species specific. Thus, under drought or salt stress the leaf hydraulic capacitance can significantly increase (Pistacia lentiscus, Álvarez et al., 2018; Robinia pseudoacacia, Zhang et al., 2019), decrease (Ziziphus species, Clifford et al., 2002; Atriplex portulacoides, Benzarti et al., 2014; Phragmites karka, Shoukat et al., 2018), or show no significant difference (Arbutus unedo, Nadal et al., 2020). Studies on the hydraulic capacitance of stems and roots under stresses remain rare (but see Barnard et al., 2011; Salomon et al., 2020), and different tissues within a species can have different degrees of acclimation to stress (Kandiko et al., 1980). Studies performed at elevated [CO₂] have shown significantly reduced leaf hydraulic capacitance in Pinus (Domec et al., 2009) but increased stem hydraulic capacitance in aspen trees in during early-season drought (Lauriks et al., 2021). In short, more information on water reserves is needed to improve our ability to understand and anticipate drought-induced risk in crops (Martinez-Vilalta et al., 2019).

Perhaps because irrigation reduces the impact of environmental stresses, less attention has been given to crop hydraulics, especially to their acclimation responses under water stress, salt stress, or elevated [CO₂]. Most previous studies have focused on trees and other woody species (McDowell et al., 2019), and these have shown large differences in hydraulic acclimation to the environment (Hacke et al., 2017). Given that water shortages and serious soil salinization are two general features in arid and semiarid agricultural regions (Yu et al., 2021), and that rainfed agricultural regions that lack irrigation facilities also face severe drought risks (Ai et al., 2020), there is an urgent need to gain a comprehensive knowledge of the hydraulic properties of herbaceous crops and thereby systematically evaluate the potential risks that they face.

In the current study, we used maize grown under controlled conditions to evaluate the impacts of abiotic stress (drought or salinity) on plant hydraulics when combined with elevated [CO₂]. We hypothesized first that plants that are exposed to moderate drought or salinity conditions would modify their hydraulic traits (acclimation) to shift to lower conductance and hence reduced hydraulic risk, and second that elevated [CO₂] would reduce the degree of acclimation due to stomatal closure and the consequent improvement of plant water status, and this would result in increased potential hydraulic risks. Our aim was to study the effects of [CO₂] combined with drought or salinity on a large set of hydraulic traits, and specifically those that are important for sustaining plant survival during long-term droughts.

Materials and methods

Plant material and growth conditions

The experiments were carried out in two different years, 2018 and 2020. In 2018, we tested the physiological effects of water deficit or soil salinity combined with three levels of [CO₂], namely 400, 700, and 1000 ppm. On the basis of the results obtained, in 2020 we performed a similar experiment with only 400 ppm and 700 ppm [CO₂], but with a more comprehensive characterization of identified critical hydraulic traits. Consequently, the results from 2020 are presented in here the main text whilst the results from 2018 are provided as supplementary data to support the conclusions made in 2020.

In both years, the experiment was carried out at the Shiyanghe experimental station of the China Agricultural University (N 37°52ʹ, E 102°50ʹ, altitude 1581 m). In 2018, the experiment ran from October 2018 to January 2019, and in 2020 it ran from August to December. In both years, seedlings of maize (Zea mays, cultivar ‘Qiangsheng 51’) were grown in 15 litre pots in a greenhouse at ambient [CO₂], and thinned to only one plant per pot at the four-leaf stage. Each pot was filled with local sandy loam of bulk density 1.45 g cm^–3 and field capacity (θ_f) of 0.248 cm³ cm^–3, with fine sand spread at the bottom to act as a filtration layer. We applied 0.86 g urea, 0.11 g KH₂PO₄, and 0.65 g Ca(H₂PO₄)₂ to each pot four times during the growth period. The plants suffered no stress during the early stages of growth before the sixth leaf entirely developed.

When the plants started to elongate (sixth leaf entirely developed, vegetative stage) at the sixth (2018) or fifth (2020) week after sowing, they were moved into a natural-light phytotron consisting of separate rooms (see Li et al., 2018 for details) and the treatments commenced (see Supplementary Table S1 for timelines of the treatments and measurements). In 2018, the target [CO₂] of three rooms were set to 400, 700, and 1000 ppm, and the actual [CO₂] recorded in each were 413.8 ± 1.1, 685.8 ± 1.3, and 977.8 ± 1.9 ppm, respectively (mean± SE; Supplementary Fig. S1). In the 400 ppm and 700 ppm rooms, the well-watered (WW1) treatment was set as θ_f=65–100% and the water-stress (WS) treatment was set as θ_f=50–70%, corresponding to a moderate drought. In the 400, 700, and 1000 ppm rooms, the well-watered (WW2) treatment was set as θ_f=65–100% and 0% NaCl, and the soil salinity (SS) treatment was set as θ_f=75–100% and 0.3% NaCl soil salt content (SSC, g salt g^-1 dry soil), corresponding to a mild salinization stress. The plants for the salinity×[CO₂] experiment were sown 10 d later than the drought×[CO₂] experiment. In 2020, the target [CO₂] of two rooms in the phytotron were set to 400 ppm and two rooms were set to 700 ppm, and the actual mean [CO₂] values recorded were 408.9 ± 0.4 and 691.3 ± 1.4 ppm, respectively (Supplementary Fig. S1). Under each [CO₂], the following treatments were set: well-watered (WW), θ_f=65–100-65%; water stress (WS), θ_f=45–65% (moderate drought); and soil salinity (SS), θ_f=75–100%, 0.15% NaCl. For the NaCl treatments either 15 g (2018) or 7.5 g (2020) of NaCl was dissolved in 800 ml water and applied to each pot, repeated three times in 10 d at the beginning of the treatment to reach the desired SSC values. The SSC treatment in 2020 was reduced to 0.15% because 0.3% treatment in 2018 seriously inhibited plant growth and the stems were too short to measure their hydraulic vulnerability. In all the rooms, the photoperiod was 12/12 h (07.00–19.00 h) and the temperature was set to 27/18 °C, with relative humidity of 60/80% (Supplementary Fig. S1). The changes in soil water content (v/v) were recorded by weighing the pots every 1–5 d, depending on the weather and growth stage (Supplementary Fig. S2).

In 2018, 100 plants in total were used with 10 plants per treatment, and in 2020 there were 108 plants in total with 18 plants per treatment, with nine in each room at the same [CO₂]. Each plant was considered as a biological replicate in this study.

Leaf gas exchange

Leaf gas exchange was measured after irrigation on clear days using a LI-6400 photosynthetic system (LI-COR Inc.). A transparent leaf chamber was used for the measurements with a flow rate of 500 μmol s^–1, and the temperature, humidity, and [CO₂] inside the chamber were not controlled (see Supplementary Table S2). Measurements in 2018 were taken at the 10th week after sowing, and at the 8th, 10th, and 11th week in 2020 (all during the vegetative stage). Three plants from each treatment were random selected and the third leaf from the top was used for measurement. The net photosynthetic rate (P_n, μmol m^–2 s^–1), stomatal conductance (g_s, mol m^–2 s^–1), and transpiration rate (T_r, mmol m^–2 s^–1) were measured at 2 h intervals from 08.00 h to 18.00 h and the mean values were determined. Intrinsic water use efficiency (iWUE, mmol CO₂ mol^–1 H₂O) was calculated as the ratio of P_n to T_r (for a full list of abbreviations see Table 1).

Table 1.

List of abbreviations used in the text, including units of parameters

Abbreviation	Description	Units
Aleaf	Total leaf area per plant	m2 plant–1
[CO2]	Carbon dioxide concentration	ppm
Cpost	Hydraulic capacitance after Ψtlp	MPa–1 or mmol m–2 MPa–1
Cpre	Hydraulic capacitance before Ψtlp	MPa–1 or mmol m–2 MPa–1
DW	Dry weight	g plant–1
gmin	The minimum conductance of the full expanded leaf	mmol m–2 s–1
gmin_roll	The minimum conductance of the maximum rolled leaf	mmol m–2 s–1
gs	Leaf stomatal conductance	mmol m–2 s–1
HSM50/HSM88	The hydraulic safety margin from the water potential of stomatal closure to P50/P88	MPa
iWUE	Intrinsic water use efficiency	mmol CO2 mol–1 H2O
kleaf	Leaf hydraulic conductance	mmol m–2 MPa–1 s–1
ks	Stem-specific hydraulic conductance	mg mm–1 kPa–1 s–1
P12/P50/P88	The water potential that induces 12%/50%/88% loss of hydraulic conductivity	MPa
PLC	Percentage loss of conductivity	%
Pn	Leaf net photosynthetic rate	μmol m–2 s–1
SS	Salt stress treatment	–
SWC	Saturated water content	g H2O g–1 DW
(t/b)2	A ratio to characterize cell wall reinforcement	–
tcrit	The time to the critical point of dehydration	kPa h
Transpirationmin	Minimum water loss rate per plant	g s–1 plant–1
Tr	Leaf transpiration rate	mmol m–2 s–1
VPD	Vapor pressure deficit	kPa
Vw	Interior available water storage	g plant–1
WS	Water stress treatment	–
WW	Well-watered treatment	–
Ψmd	Leaf water potential at midday	MPa
Ψom	Leaf osmotic potential	MPa
Ψpd	Leaf water potential at pre-dawn	MPa
Ψtlp	The water potential at turgor loss point	MPa
Ψtp	Leaf turgor pressure	MPa

Leaf water potential, osmotic potential, and turgor

Leaf water potentials were measured the day after irrigation on a clear day at the 11th week after sowing in 2018 or the 10th week in 2020. Five (2018) or six (2020) plants in each treatment were randomly selected, and a fully emerged leaf from the top of each plant was cut at pre-dawn (06.00 h) and another at midday (13.00 h) to measure the water potential (Ψ_pd and Ψ_md, respectively, MPa). A dew-point water potential meter (WP4C; Decagon Device Inc., USA) and a pressure chamber (SKPM 1400; Skye Ltd., UK) were used to measure the leaf water potential in 2018 and 2020, respectively. In 2020, a sample from each leaf was then frozen in liquid nitrogen and used to measure the osmotic concentration (C, mmol kg^–1) with a vapor-pressure osmometer (VAPRO5600; Wescor Inc., USA). The pre-dawn and midday osmotic potentials (Ψ_{om_pd} and Ψ_{om_md}, respectively, MPa) were then calculated using the Van’t Hoff equation:

$${\displaystyle \begin{array}{l}{\displaystyle {\mathrm{\Psi }}_{\mathrm{o}\mathrm{m}}=-\mathrm{i}\times C\times \mathrm{R}\times T}\end{array}}$$

(1)

where i is the dissociation coefficient (=1), R is the ideal gas constant (=8.314 J mol^–1 k^–1), and T is the absolute temperature (in K). The turgor pressures at pre-dawn and midday (Ψ_{p_pd} and Ψ_{p_md}, respectively, MPa) were then obtained by subtracting the osmotic potential from the water potential, neglecting the apoplastic water fraction and assuming that the bulk leaf osmotic concentration was similar to the osmotic concentration of the cells (Barzilai et al., 2021).

Minimum leaf conductance and leaf rolling

In 2020, the minimum leaf conductance (g_min) was measured at the 10th week after sowing. Four plants in each treatment were randomly selected and exposed to four consecutive days without irrigation (these plants were then no longer used for other measurements). All the pots were extremely dry, and the leaves showed a maximal amount of rolling (leaf rolling did not increase further after excision; Supplementary Table S3). Two mature leaves near the ear were cut to measure the minimum conductance of the maximally rolled leaf (g_{min_roll}, mmol m^–2 s^–1). The plants were then re-watered until their leaves expanded again, and one mature leaf near the ear was cut to measure the minimum conductance of the fully expanded leaf (g_min, mmol m^–2 s^–1). The loss in mass of the detached leaves was measured to calculate g_min and g_{min_roll} (Anfodillo et al., 2002; Duursma et al., 2019). First, plastic tape was applied at the cut end of the leaf to prevent additional evaporation. Given the initial unstable conductance due to gradual stomatal closure (Anfodillo et al., 2002; Duursma et al., 2019), the leaves were left in the dark for 3 h before measurement. The leaf weight (W, g) was then recorded every hour for at least 7 h until the value of W decreased linearly. During the measurements, the leaves were kept in a dark room at 20 °C and 50% relative humidity, with a fan was blowing air across the surface. The value of g_min (or g_{min_roll}) was calculated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle g_{\mathrm{m}\mathrm{i}\mathrm{n}}=\frac{-\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}\left(W/\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}\right)\times p_{\mathrm{a}}}{a_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}\times VPD\times m_{\mathrm{H}2\mathrm{O}}}}\end{array}}$$

(2)

where (W versus time) refers to the linear part of the slope (g s^–1); a_leaf is the area of the leaf (m², one side); VPD is the vapor pressure deficit (kPa) according to the temperature and the relative humidity; p_a is the atmospheric pressure (101.6 kPa), and m_H2O is the molecular mass of H₂O (18 g mol^–1).

We defined a ‘rolling rate’ to represent the degree of rolling of the leaves, on the assumption that the change of boundary layer conductance is the dominant factor leading to the different measured results between g_min and g_{min_roll}:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{R}\mathrm{o}\mathrm{l}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}=1-\frac{g_{\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{\_}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{l}}}{\mathrm{A}\mathrm{V}\mathrm{G}(g\mathrm{m}\mathrm{i}\mathrm{n})}}\end{array}}$$

(3)

where AVG(g_min) is the average value of the eight replications of g_min in each treatment.

Pressure–volume curves

In 2020, seven plants in each treatment were randomly selected for determination of leaf, stem, and root pressure–volume (P–V) curves at the 13th–14th weeks after sowing. The samples taken were the leaf above the ear, the upper stem (4–7 mm in diameter, between about the 3th and 5th leaves from the top of the plant), and a fine root in the middle part of the pot (<3 mm in diameter, ~10 cm in length, 0.5–2 g fresh weight). Since the roots were delicate and branched, great care was necessary during the sampling as well as during the measurements. All the sampled tissues were recut under water and rehydrated with the cut end submerged in water for 2 h (root and stem) or 3 h (leaf) hours to reach saturation (greater than –0.2 MPa).

The P–V curves were constructed by progressively drying the tissues on a laboratory bench while measuring the water potential and mass at intervals (Tyree and Hammel, 1972). The SKPM 1400 pressure chamber was used to measure the water potential. The potential at the turgor-loss point (Ψ_tlp, MPa), the hydraulic capacitance before and after Ψ_tlp (C_pre and C_post, respectively, MPa^–1), and other parameters were calculated according the methods of Sack and Pasquet-Kok (2011). The leaf hydraulic capacitance per unit leaf area (mmol m⁻² MPa⁻¹) was also calculated for the subsequent determination of leaf hydraulic conductance.

Hydraulic conductance and vulnerability

Leaf hydraulic conductance

In 2020, leaf hydraulic conductance (k_leaf, mmol m^–2 MPa^–1 s^–1) was measured at midday (13.00 h) on a clear day (the day after irrigation) at the 13th week after sowing (reproductive stage) using the rehydration kinetics method (Brodribb and Holbrook, 2003). Seven plants in each treatment were randomly selected, and the shoot containing the second, third, and fourth leaves from the top of the plant was cut. The samples were enclosed in plastic bags and placed in the dark for an equilibrium period of 2 h. The water potentials of the second and fourth leaves were measured and averaged to determine the initial water potential (Ψ₀, MPa). The third leaf was then cut under water and rehydrated for 20–30 s, after which the water potential after rehydration (Ψ_f, MPa) was measured. k_leaf was calculated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle k_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}=C\times \ln \left({\mathrm{\Psi }}_0/{\mathrm{\Psi }}_{\mathrm{f}}\right)/t_{\mathrm{r}}}\end{array}}$$

(4)

where C is the hydraulic capacitance per leaf area determined from the P–V curves and t_r is the rehydration time.

Stem hydraulic conductance and vulnerability

In 2020, stem xylem hydraulic conductance and vulnerability were measured at the 14th–15th weeks after sowing. Seven plants in each treatment were randomly selected and were watered daily to saturation for 5 d in advance of the measurements to recover the hydraulic conductance to its maximum. The pots of the soil salinity treatment were then drenched to remove the salt. Stem segments near nodes 4–5 (below the ear) were harvested before dawn and then recut to 27.5 cm in length under water. The stem maximum specific hydraulic conductance (k_s, mg mm^–1 kPa^–1 s^–1) was measured using the gravity method (Sperry et al., 1988). The static centrifuge method was applied to produce stem xylem vulnerability curves (originally described by Alder et al., 1997; see Liu et al., 2020 for detailed procedures). Briefly, the segment was secured in the rotor and spun on the centrifuge (H2050R-1; Xiangyi, China); after spinning to induce the desired negative pressure at the center of the stem for 3 min, the stem was removed, and the hydraulic conductance was measured again. The process was repeated at progressively higher centrifuge speeds corresponding to an increase of 0.5 MPa each time until more than 90% of the k_s value had been lost. The percentage loss of conductivity (PLC) with decreasing xylem pressure (P_x) was fitted using the following sigmoidal equation (Pammenter and Vander Willigen, 1998; Hochberg et al., 2019):

$${\displaystyle \begin{array}{l}{\displaystyle PLC=\frac{100}{1+\mathrm{e}\mathrm{x}\mathrm{p}[a\times (P_{\mathrm{x}}-P_{50})]}}\end{array}}$$

(5)

where P₅₀ (MPa) is the xylem water potential at 50% loss of hydraulic conductivity, and a is the slope parameter of the curve at P₅₀. The water potentials at 12% and 88% loss of hydraulic conductivity (P₁₂ and P₈₈, respectively, MPa) were calculated from the fitted curves.

In 2018, either four plants from each of the [CO₂]×water treatments or eight plants from each of the [CO₂]×salt treatments were randomly selected for measurement of k_s at the 15th week after sowing, as described above. Different to the steps in 2020, initial embolisms were removed by flushing at 100 kPa before the k_s measurement.

After the hydraulic measurements, the middle part of the stem segment was sampled for paraffin cross-sectioning to observe the xylem anatomy (Supplementary Fig. S3). Twenty vascular bundles in each stem cross-section were randomly selected and imaged using a digital optical microscope (BA210; Motic, China). The xylem conduit span (b, μm) and wall thickness (t, μm) were measured using the ImageJ software, the value of (t/b)² was calculated to characterize the conduit structural strength (Hacke and Sperry, 2001).

Plant growth

Prior to the hydraulic measurements, the height (cm) of the sampled plants was measured and the total leaf area (A_leaf, m²) was determined. The leaf area was calculated by summing the leaf lengths multiplied by the maximum leaf widths multiplied by a conversion factor of 0.74 (Li et al., 2008). The plants were in the early stage of reproductive growth, and their vegetative growth had stopped.

In 2020, the plants sampled for hydraulic measurements were also used to measure the dry matter. All the measured samples were retained and together with the remainder of the plants they were oven-dried at 80 °C for 48 h. The dry weight per plant was divided into three parts: leaf (DW_l, g), stem (DW_s, g), and root (DW_r, g). The spikes and ears were grouped in DW_s.

Modeling and statistical analysis

The t_critmodel

Using the t_crit model developed by Blackman et al. (2016), we calculated the plant interior available water storage (V_w, g plant^–1) and the critical time to a hydraulic threshold (t_crit, kPa h). Different to the original model, the water storages of roots, stems, and leaves were distinguished, and the effect of leaf rolling under stress was also considered. Therefore, the equations were modified to fit our research:

$${\displaystyle \begin{array}{l}{\displaystyle t_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}=\frac{V_{\mathrm{w}}}{Transpiratio{n}_{\mathrm{m}\mathrm{i}\mathrm{n}}}=\frac{V_{\mathrm{w}\mathrm{\_}\mathrm{l}}+V_{\mathrm{w}\mathrm{\_}\mathrm{s}}+V_{\mathrm{w}\mathrm{\_}\mathrm{r}}}{Transpiratio{n}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}\end{array}}$$

(6)

where the Transpiration_min (g s^–1 plant^–1) is the minimum water loss rate per plant, and is calculated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle Transpiratio{n}_{\mathrm{m}\mathrm{i}\mathrm{n}}=g_{\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{\_}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{l}}\times A_{\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{f}}\times \frac{VPD}{p_{\mathrm{a}}}\times m_{\mathrm{H}2\mathrm{O}}}\end{array}}$$

(7)

where VPD is the vapor pressure deficit (kPa), p_a is the atmospheric pressure (taken as 101.3 kPa), and m_H2O is the molecular mass of H₂O (18 g mol^–1). Since the value of VPD varies all the time and has a great impact on the results, t_crit is an index to represent the critical desiccation degree considering both time and VPD variation (Blackman et al., 2019).

The V_w of the plant consists of V_{w_l}, V_{w_s}, and V_{w_r}, representing the water storage up to the hydraulic threshold in the leaves, stems, and roots, respectively. They were modified and calculated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle V_{\mathrm{w}\mathrm{\_}\mathrm{l}}=C_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{l}}\times ({\mathrm{\Psi }}_{\mathrm{g}\mathrm{s}0}-{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}})\times D{W}_{\mathrm{l}}\times SW{C}_{\mathrm{l}}}\end{array}}$$

(8A)

$${\displaystyle \begin{array}{l}{\displaystyle V_{\mathrm{w}\mathrm{\_}\mathrm{s}}=\{\begin{array}{l}{C}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{\_}\mathrm{s}}\times HSM\times D{W}_{\mathrm{s}}\times SW{C}_{\mathrm{s}},\mathrm{i}\mathrm{f}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{s}}\ge {\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}[C_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{\_}\mathrm{s}}\times ({\mathrm{\Psi }}_{\mathrm{g}\mathrm{s}0}-{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{s}})+C_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{s}}\times ({\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{s}}-{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}})]\times D{W}_{\mathrm{s}}\times SW{C}_{\mathrm{s}},\mathrm{i}\mathrm{f}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{s}}<{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}\end{array}}\end{array}}$$

(8B)

$${\displaystyle \begin{array}{l}{\displaystyle V_{\mathrm{w}\mathrm{\_}\mathrm{r}}=\{\begin{array}{l}{C}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{\_}\mathrm{r}}\times HSM\times D{W}_{\mathrm{r}}\times SW{C}_{\mathrm{r}},\mathrm{i}\mathrm{f}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{r}}\ge {\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}[C_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{\_}\mathrm{r}}\times ({\mathrm{\Psi }}_{\mathrm{g}\mathrm{s}0}-{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{r}})+{\mathrm{C}}_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{r}}\times ({\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{r}}-{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}})]\times D{W}_{\mathrm{r}}\times SW{C}_{\mathrm{r}},\mathrm{i}\mathrm{f}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{r}}<{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}\end{array}}\end{array}}$$

(8C)

where HSM is the hydraulic safety margin (MPa), the water potential difference between stomatal closure (Ψ_gs0), and the stem xylem hydraulic thresholds at P₅₀ or P₈₈ (HSM₅₀ or HSM₈₈, respectively).

Since we did not measure the water potential of stomatal closure, the value Ψ_gs0 of the plants under 400 ppm [CO₂] in the WW treatment was empirically set to –1.5 MPa (Cochard, 2002) as the beginning of HSM. In the other treatments, the value of Ψ_gs0 was set to –1.5 MPa plus the difference in leaf Ψ_tlp compared to the 400 ppm+WW plants, as previous studies have indicated that stomatal closure is correlated with the loss of turgor in maize and other species (Mustardy et al., 1982; Yoon and Richter, 1990; Brodribb et al., 2003; Rodriguez-Dominguez et al., 2016). It was assumed that the water potentials in the leaf, stem, and roots were equal after stomatal closure.

Partitioning contributions

The numerical integration method that is commonly used in stomatal and photosynthetic models (Grassi and Magnani, 2005; Buckley and Diaz-Espejo, 2015; Rodriguez-Dominguez et al., 2016) was applied in the t_crit model to partition changes in V_w or t_crit under the different treatments into contributions from different variables. The exact differential of a function parses an infinitesimal change in the function into contributions from each variable. Applied to V_{w_l} in Equation 8A, this gives:

$${\displaystyle \begin{array}{lll}{\displaystyle \mathrm{d}{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}=}& \frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{C}_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{l}}}\mathrm{d}{C}_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }D{W}_{\mathrm{l}}}\mathrm{d}D{W}_{\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }SW{C}_{\mathrm{l}}}\mathrm{d}SW{C}_{\mathrm{l}}& +\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{l}}}\mathrm{d}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}}\mathrm{d}{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}\end{array}}$$

(9)

Numerically integrating Equation 9 between a 400 ppm+WW point and a 400 ppm+WS point gives finite partial changes in V_{w_l} due to each variable, which add up to the total change in V_{w_l}:

$${\displaystyle \begin{array}{l}{\displaystyle {\displaystyle \underset{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{{\displaystyle \stackrel{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{\int }}}}\mathrm{d}{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}={\displaystyle \underset{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{{\displaystyle \stackrel{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{S}}{\int }}}}\left(\begin{array}{l}\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{C}_{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{\_}\mathrm{l}}}\mathrm{d}{C}_{\mathrm{p}\mathrm{o}\mathrm{s}{\mathrm{t}}_{\mathrm{l}}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }D{W}_{\mathrm{l}}}\mathrm{d}D{W}_{\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }SW{C}_{\mathrm{l}}}\mathrm{d}SW{C}_{\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{l}}}\mathrm{d}{\mathrm{\Psi }}_{\mathrm{t}\mathrm{l}\mathrm{p}\mathrm{\_}\mathrm{l}}+\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}}\mathrm{d}{\mathrm{\Psi }}_{\mathrm{c}\mathrm{r}\mathrm{i}\mathrm{t}}\end{array}\right)}\end{array}}$$

(10A)

$${\displaystyle \begin{array}{l}{\displaystyle {\displaystyle \underset{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{{\displaystyle \stackrel{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{S}}{\int }}}}\mathrm{d}{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}={\displaystyle \sum _j}\left({\displaystyle \underset{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{{\displaystyle \stackrel{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{S}}{\int }}}}\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{x}_{\mathrm{j}}}\mathrm{d}{x}_{\mathrm{j}}\right)={\displaystyle \sum _j}{p}_{\mathrm{j}}}\end{array}}$$

(10B)

where x_j is C_post, DW_l, SWC_l, Ψ_{tlp_l}, or Ψ_crit, and p_j is the partial change in V_{w_l} due to x_j, i.e. the contribution.

The integral in Equation 10B can be approximated as follows:

$${\displaystyle \begin{array}{l}{\displaystyle \begin{array}{llll}& {\displaystyle {\displaystyle \underset{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{W}}{{\displaystyle \stackrel{400\mathrm{p}\mathrm{p}\mathrm{m}+\mathrm{W}\mathrm{S}}{\int }}}}\frac{\mathrm{\partial }{V}_{\mathrm{w}\mathrm{\_}\mathrm{l}}}{\mathrm{\partial }{x}_j}\mathrm{d}{x}_j\approx }& {\displaystyle \sum _{k=0}^{n-1}}\left[\frac{V_{\mathrm{w}\mathrm{\_}\mathrm{l}}(x_{j,k+1},x_{i\ne j,k})-V_{\mathrm{w}\mathrm{\_}\mathrm{l}}(x_{j,k},x_{i\ne j,k})}{x_{j,k+1}-x_{j,k}}\right]\times (x_{j,k+1}-x_{j,k})& ={\displaystyle \sum _{k=0}^{n-1}}[V_{\mathrm{w}\mathrm{\_}\mathrm{l}}(x_{j,k+1},x_{i\ne j,k})-V_{\mathrm{w}\mathrm{\_}\mathrm{l}}(x_{j,k},x_{i\ne j,k})]\end{array}}\end{array}}$$

(11)

where the index k is one of n equivalents between the 400 ppm+WW and 400 ppm+WS points, and the functional notation V_{w_l} (… ) is Equation 8A, calculated by the information in parentheses. According to the considerations of Rodriguez-Dominguez et al. (2016) and Buckley and Diaz-Espejo (2015), step n was set as 1000. The numerical error in our results was less than 0.3%, and thus negligible.

Sensitivity analysis

Since the desiccation model contained 16 variables and any variations could result in different outputs, a sensitivity analysis using ±5% of the three traits that contributed the most was performed to determine the confidence in the quantitative calculations of the modelling.

Significance and correlation analysis

We used SPSS 17.0 for the statistical analysis. Both one-way and two-way ANOVA followed by Duncan’s test were carried out on plant growth, leaf gas exchange, plant water status, and the hydraulic data to determine significant differences between treatments and the effects of [CO₂], drought stress, salt stress, and their interactions (CO₂×drought and CO₂×salt). SigmaPlot 12.5 was used to examine the correlations and prepare the figures. Regression analyses were performed with mean values to test the correlations between parameters by linear models.

Results

Plant growth and leaf gas exchange

Maize plant growth and leaf gas exchange were sensitive to both water and salt stress. At 400 ppm [CO₂], the WS and SS treatments significantly decreased plant height and A_leaf by 25–37%, while DW, P_n, g_s, and T_r decreased by 57–75% (Fig. 1; Supplementary Table S4). The 700 ppm [CO₂] treatment partially ameliorated the effects of WS and SS, especially on P_n, which was 83% and 95% higher, respectively, when compared with 400 ppm [CO₂]. Surprisingly, these increases in P_n at 700 ppm [CO₂] were accompanied by only small (7–23%) increases in plant height, A_leaf, and DW. Regardless, elevated [CO₂] increased plant growth under water deficit and soil salinity. The data from 2018 also supported this conclusion (Supplementary Table S5). Both WS and SS had limited effects on iWUE, but the higher P_n and lower g_s at 700 ppm [CO₂] resulted in a 51–104% increase in iWUE when compared with that at 400 ppm [CO₂].

Fig. 1.

Dry weight of maize plants and its distribution in the leaves, stems, and roots under different [CO₂], drought and salinity treatments. WW, well-watered; WS, water stress; SS, soil salinity. The data represent means, n=6–7. The error bar is the SE of the total plant DW. Different letters indicate significant differences in total DW as determined using one-way ANOVA followed by Duncan’s test (P<0.05). Two-way ANOVA of treatment effects: *P<0.05, ***P<0.001, ns, not significant.

Leaf water potential, osmotic potential, and turgor

The leaf water status declined under both water and salt stress. At 400 ppm [CO₂] the WS and SS treatments decreased Ψ_pd by 0.07–0.19 MPa and decreased Ψ_md by 0.30–0.34 MPa (Fig. 2). The leaf osmotic potential decreased from approximately –1.05 MPa at pre-dawn to approximately –1.32 MPa at midday, but no significant differences were found between treatments. In addition, WS and SS at 400 ppm [CO₂] significantly decreased leaf turgor pressure to zero at midday, but had no significant effect at pre-dawn. At 700 ppm [CO₂], the Ψ_md of WW, WS, and SS plants significantly increased by 0.19, 0.18, and 0.27 MPa, respectively. The leaf turgor pressure at midday of WS and SS plants was also partially recovered at 700 ppm [CO₂]. Overall, elevated [CO₂] mitigated the negative effects of soil drought and salinity on leaf water status. Similar results were also found in 2018 (Supplementary Table S5).

Fig. 2.

Turgor pressure, water potential, and osmotic potential of maize leaves under different [CO₂], drought, and salinity treatments. WW, well-watered; WS, water stress; SS, soil salinity. [CO₂] was either 400 ppm or 700ppm. Turgor (Ψ_tp) is shown as histograms, water potential (Ψ) is shown as circles, and osmotic potential (Ψ_om) is shown as squares. (A) Values at pre-dawn (pd) and (B) values at midday (md). Data are means (±SE), n=6. Different uppercase and lowercase letters indicate significant differences among water potentials and turgor pressure, respectively, as determined using one-way ANOVA followed by Duncan’s test (P<0.05). Two-way ANOVA of treatment effects: **P<0.01, ***P<0.001, ns, not significant.

g_min and leaf rolling

A lower g_min and a higher degree of leaf rolling are beneficial for delaying the drying process of plants. At 400 ppm [CO₂], the value of g_min significantly decreased from 2.10 mmol m^–2 s^–1 in WW plants to 1.70 mmol m^–2 s^–1 in WS plants and 1.75 mmol m^–2 s^–1 under SS plants, and there was no significant effect of elevated [CO₂] (Supplementary Fig. S4A). The degree of rolling of the leaves varied under the different treatments. At 400 ppm [CO₂], WS and SS significantly reduced the rolling rate of maize leaves compared with WW, but the effect was not significant at 700 ppm [CO₂] (Supplementary Fig. S4C). The maximally rolled leaves had a significantly lower minimum conductance (g_{min_roll} compared with g_min), with an mean value of 1.21 mmol m^–2 s^–1, but no significant differences were found among the treatments (Supplementary Fig. S4B), mainly due to the differences in the degree of rolling.

Hydraulic conductance and vulnerability

Similar to the results for plant growth and leaf gas exchange, the water transport capacity of the maize xylem was also sensitive to soil drought and salinity, and elevated [CO₂] also showed mitigative effects (Table 2). At 400 ppm [CO₂], the value of k_s was significantly decreased by 53% under the WS treatment and 56% under the SS treatment compared with WW. However, k_leaf declined to a greater extent, decreasing by 83% under the WS treatment and 80% under the SS treatment. Elevated [CO₂] alleviated the negative impact of the SS treatments on k_s and the negative impact of the WS treatment on k_leaf.

Table 2.

Maize leaf and stem hydraulic conductances under the different treatments

[CO2] (ppm)	Soil treatment	ks (mg mm–1 kPa–1 s–1)	kleaf (mmol m–2 MPa–1 s–1)
400	WW	0.94 ± 0.03a	29.3 ± 2.7a
	WS	0.44 ± 0.03c	5.0 ± 0.7c
	SS	0.41 ± 0.02c	5.9 ± 0.7c
700	WW	0.98 ± 0.036a	29.1 ± 5.1a
	WS	0.50 ± 0.04c	12.0 ± 1.6b
	SS	0.63 ± 0.05b	8.1 ± 1.0bc
Two-way ANOVA	CO2	**	ns
	Water	***	***
	Salt	***	***
	CO2 × Water	ns	*
	CO2 × Salt	*	ns

WW, well-watered; WS, water stress; SS, soil salinity; k_s, stem-specific hydraulic conductance; k_leaf, leaf hydraulic conductance. Data are means (±SE): n=7. Different letters indicate significant differences among means as determined by one-way ANOVA followed by Duncan’s test (P<0.05). Two-way ANOVA of treatment effects: * P<0.05, **P<0.01, ***P<0.001; ns, not significant.

In 2020, we calculated the values of P₁₂, P₅₀, and P₈₈ (Supplementary Table S6) to evaluate the water transport vulnerability according to PLC curves (Fig. 3). At 400 ppm [CO₂], the value of P₅₀ significantly decreased from –1.89 (±0.09) MPa for WW plants to –2.91 (±0.10) MPa for WS plants and to –2.91 (±0.08) MPa for SS plants. P₁₂ and P₈₈ also decreased by ~1 MPa under the stressed treatments. As a consequence, the values of HSM were increased by ~1 MPa under the WS and SS treatments. At 700 ppm [CO₂], P₁₂, P₅₀, and P₈₈ of the WS and SS plants significantly increased by ~0.70 MPa, resulting in smaller HSMs. PLC curves were not obtained in 2018, but calculated values of (t/b)², a trait frequently correlated with P₅₀ (Hacke and Sperry, 2001; Supplementary Fig. S5B, R²=0.91, P=0.003), supported the results from 2020. At 400 ppm [CO₂], (t/b)² increased by 83% under the WS treatment and by 89% under the SS treatment compared with the controls (Supplementary Fig. S5A). In addition, in both years the value of (t/b)² was consistently lower at 700 ppm [CO₂] than at 400 ppm [CO₂] in the WS and SS plants.

Fig. 3.

Percentage loss of maize stem xylem conductivity with decreasing water potential of under (A) well-watered (WW), (B) water stress (WS), and (C) salt stress (SS) treatments grown under 400 ppm or 700 ppm [CO₂]. The plots show the original data points, the fitted curves, and the 95% confidence bands (shaded areas). The data were obtained from n=6–7 biological replicates.

Water transport capacities displayed coordinated acclimation to treatments with plant growth and leaf gas exchange (Fig. 4). Both k_s (R²=0.83, P=0.012) and k_leaf (R²=0.85, P=0.009) showed significant correlations with P_n across treatments, and even stronger correlations with DW (k_s, R²=0.97, P<0.001; k_leaf, R²=0.98, P<0.001). In addition, both P₅₀ (Fig. 5A; R²=0.82, P=0.012) and P₈₈ (Fig. 5B; R²=0.81, P=0.015) were closely correlated with k_s across the different treatments.

Fig. 4.

Correlations between water transport efficiency of the stems and leaves and plant photosynthesis and growth under the different treatments. WW, well-watered; WS, water stress; SS, salt stress; [CO₂] was either 400 ppm or 700ppm. (A, B), Net photosynthetic rate (P_n) versus (A) stem specific hydraulic conductance (k_s) and (B) leaf hydraulic conductance (k_leaf). (C, D) Dry weight (DW) versus (C) k_s and (D) k_leaf. Data are means (±SE), n=6–7 biological replicates. Significance of R² values: *P<0.05, **P<0.01, ***P<0.001.

Fig. 5.

Trade-offs between water transport efficiency and hydraulic safety in maize stems under the different treatments. WW, well-watered; WS, water stress; SS, salt stress; [CO₂] was either 400 ppm or 700ppm. (A) The water potential at 50% loss of stem hydraulic conductance (P₅₀) versus stem specific hydraulic conductance (k_s) and (B) the water potential at 88% loss of stem hydraulic conductance (P₈₈) versus k_s. Data are means (±SE), n=6–7 biological replicates. Significance of R² values: *P<0.05.

P–V curve traits

According to the P–V curves, the leaves, stems, and roots of maize had different levels of Ψ_tlp, hydraulic capacitance, and SWC (Supplementary Table S7), all of which are critical inputs for the V_w and t_crit simulations. Leaves had the highest values of Ψ_tlp, ranging from –1.06 to –1.19 MPa across all treatments, while values in the stems ranged from –1.62 to –2.40 MPa, and in the roots from –1.74 to –2.09 MPa. In addition, roots had the highest SWC and hydraulic capacitance. Moreover, the roots, stems, and leaves showed tissue-specific responses to water stress and salt stress. The roots were the most insensitive organ, as the WS and SS treatments had no significant effects on their Ψ_tlp and hydraulic capacitance. Nevertheless, we found that root hydraulic capacitance was reduced to a certain extent under the WS and SS treatments, and the large heterogeneity of the roots might have accounted for these responses not being found significant. The WS treatment significantly reduced the C_post of leaves and increased the Ψ_tlp of stems and the SWC of stems and leaves, while the SS treatment significantly reduced the C_post of leaves and the C_pre of stems and increased the SWC of stems and leaves. Except for alleviating the impacts of the WS and SS treatments on the SWC of leaves, 700 ppm [CO₂] had limited effects on these traits in all tissues.

The t_crit model

The drought simulation performed using the t_crit model showed that the V_w and t_crit of maize were improved when the plants were acclimated to water or salinity stress, indicating a lower risk of hydraulic failure and mortality under severe drought, but the acclimation was greatly reduced under elevated [CO₂].

When P₅₀ was set as the hydraulic threshold, the V_w increased from 45.9 g plant^–1 in the WW treatment to ~111.2–115.2 g plant^–1 in the WS and SS treatments at 400 ppm [CO₂] (Fig. 6A). However, increasing [CO₂] to 700 ppm greatly reduced V_w by 48–66% for all treatments. In addition, the distribution of V_w among plant organs was markedly affected by the stress treatments. For the WW plants, the leaves were the dominant water storage organ, accounting for 45–60% of the total, while the stem accounted for only 14–16%. At 400 ppm [CO₂], there was a tendency for V_w to shift from the leaves and roots to the stems under the WS and SS treatments.

Fig. 6.

Modeling of the plant interior available water storage (V_w) under the different treatments, based on the difference in water potential between stomatal closure and the point of 50% loss of stem hydraulic conductance, i.e. the hydraulic safety margin HSM₅₀ (see equation 8). WW, well-watered; WS, water stress; SS, salt stress; [CO₂] was either 400 ppm or 700ppm. (A) Distribution of V_w among the leaves, stems, and roots. (B) The available water storage in the plants with increasing desiccation degree (VPD×time). The negative value of the slope of the line represents the minimum water loss rate of the whole plant (Transpiration_min; equation 7), the intercept on the y-axis represents V_w, and the intercept on the x-axis represents the critical desiccation time to a hydraulic threshold (t_crit) for the plants (equation 6).

We simulated the desiccation process of maize and found that the t_crit varied, ranging from a lowest value of 45.7 kPa h at 700 ppm [CO₂] in WW plants to a highest value of 386.5 kPa h at 400 ppm [CO₂] in WS plants (Fig. 6B). Overall, and in accordance with the trends in V_w, at 400 ppm [CO₂], t_crit was largely increased under the WS and SS treatments compared to WW, while at 700 ppm [CO₂] it was largely reduced, by 43–64% across all treatments. P₈₈ is considered a more valid marker for a critical desiccation model, and we present the model output of V_w and t_crit using P₈₈ in Supplementary Fig. S6. Different to the results based on P₅₀, large decreases in V_w and t_crit at 700 ppm [CO₂] were only apparent in the SS plants, and not in the WW and WS plants.

The contribution analysis showed that the largest contribution to t_crit in the acclimated plants was a function of the xylem vulnerability, A_leaf, and DW (Fig. 7; Supplementary Fig. S7). The contributions of traits from the P–V curves were variable and had a limited impact on the final V_w and t_crit. The results of a sensitivity analysis indicated that ±5% of A_leaf or DW had marginal influences on the results of t_crit, while a change of ±5% P₅₀ had the most influence (Supplementary Table S8). We would expect the qualitative results of t_crit found here to be robust.

Fig. 7.

Contributions of individual variables under the different treatments to changes in the critical desiccation time to a hydraulic threshold (t_crit) based on the difference in water potential between stomatal closure and the point of 50% loss of stem hydraulic conductance (P₅₀), i.e. the hydraulic safety margin HSM₅₀ (see equation 8). WW, well-watered; WS, water stress; SS, salt stress; [CO₂] was either 400 ppm or 700ppm. The results are relative to those obtained at 400 ppm in the WW treatment. Subscripts l, s, and r denote leaf, stem, and root, respectively. DW, dry weight; Ψ_tlp, water potential at turgor loss point; C_pre and C_post, hydraulic capacitance before and after Ψ_tlp; SWC, saturated water content g_{min_roll}, the minimum conductance of maximum rolled leaves; A_leaf, total plant leaf area.

Discussion

Overall, our results demonstrated that elevated [CO₂] was beneficial to maize growth under water deficit and under salinity stress, but that it interfered with the hydraulic acclimation of the plants, thus potentially placing them at a higher risk of hydraulic failure and mortality.

The acclimation of maize hydraulic traits to water deficit and soil salinity

The water stress and salt stress treatments significantly decreased the water potential inducing 50% loss of hydraulic conductivity (P₅₀; Supplementary Table S6), as found in previous studies (Stiller, 2009; Awad et al., 2010; Cardoso et al., 2018; Liu et al., 2020). The magnitude of the decrease in P₅₀ (~1.1 MPa) was considerably larger than the decreases in midday water potential, Ψ_md (<0.56 MPa), and leaf water potential at the turgor loss point, Ψ_tlp (<0.04 MPa) (Fig. 2), leading to a significant increase in the hydraulic safety margin (HSM) of the WS and SS plants. The relatively small osmotic adjustment in response to stress (averaging –0.4 MPa in both Ψ_om and Ψ_tlp) was in line with previous findings in maize (Sobrado, 1986; Premachandra et al., 1992; Erdei and Taleisnik, 1993; Jamil et al., 2015). The increase in HSM was in accordance with sunflower acclimation to drought (Cardoso et al., 2018) but different to the adjustments of Ψ_tlp and P₅₀ found in pine stems and grape leaves that result in a constant HSM during dehydration (Sorek et al., 2021; Feng et al., 2023). This suggests that a relatively shallow root system has necessitated that maize (and possibly other small herbaceous plants) has evolved greater phenotypic plasticity in HSM under severe drought. The lower P₅₀ and therefore the higher HSM appears to be crucial for plant acclimation in terms of survival, as P₅₀ contributed the most to the significantly higher time to critical dehydration (t_crit) of the WS and SS plants (Fig. 7).

Acclimation of other traits to stress also contributed to the increase in t_crit. Specifically, the dry mass distribution that significantly determines the interior available water storage, V_w, was changed under stress. Maize growth under water deficit or soil salinity showed a reduction of ~58% in DW (Fig. 1) but only a 29% decrease in A_leaf (Supplementary Table S4), which resulted in a 7–11% higher proportion of leaves (out of the total DW) in the WS and SS plants. This is disadvantageous to plant survival under drought (Holbrook and Sinclair, 1992), as it potentially increases the proportion of evaporative surfaces and decreases the t_crit. Even though the decrease in plant size and changes in morphology under the WS and SS treatments made a large negative contribution to V_w and t_crit, the decreased P₅₀ and hence increased hydraulic safety margin was large enough to compensate for the increase in the leaf proportion (Fig. 7). As a result of these different tissue-specific trait acclimations (especially the hydraulic capacitance), V_w shifted from leaves or roots to the stems (Fig. 6A). Since leaves and roots are considered more vulnerable to embolism (Tsuda and Tyree, 1997; Hochberg et al., 2016; Losso et al., 2019) and are likely to die first, it would make sense to conserve most of the water in the main stem. We speculate that this could be a water storage strategy under stress.

The higher xylem resistance came at a cost. The xylem acclimation process involves the transition into smaller vessels (Liu et al., 2020), resulting in lower hydraulic conductivity in both the leaf and stem (Table 2). The hydraulic conductivity was coordinated with plant growth and leaf gas exchange (Fig. 4), and this trade-off between P₅₀/P₈₈ and k_s confirms that the long-standing safety–efficiency hypothesis (Tyree et al., 1994) is true in maize. In recent years, several studies have suggested that the safety–efficiency trade-off does not exist (Gleason et al., 2016; Liu et al., 2021). However, all of these studies compared different species/genotypes; when a single genotype has been tested under drought or salinity, the trade-off has normally been observed (Holste et al., 2006; Stiller, 2009; Beikircher et al., 2019).

Overall, our measurements and modeling work supported our first hypothesis that water deficit or soil salinity has widespread impacts on maize hydraulic traits, such as hydraulic conductance, vulnerability to embolism, tissue water storage capacity, and g_min. In addition, the total contribution of these acclimations was beneficial to the survival of maize under severe desiccation. However, given that high [CO₂] can increase water use efficiency and thus improve plant water status, we hypothesize that it might also ameliorate the degree of acclimation and thus increase potential hydraulic risks.

Elevated [CO₂] is beneficial to maize growth, especially for plants under WS and SS, but increases hydraulic risk

Elevated [CO₂] ameliorated the stress suffered by the plants, as P_n, plant height, A_leaf, DW, k_s, and k_leaf of the WS and SS plants were significantly increased at 700 ppm [CO₂] (Fig. 1; Supplementary Table S4). Interestingly, the increase in growth (plant height, A_leaf, and DW, by 7–23%) of plants under WS and SS at high [CO₂] was much lower than the increase in photosynthetic rate (60–88%). This kind of mismatch has also been observed in previous studies on maize (Allen et al., 2011; Markelz et al., 2011; Meng et al., 2014; Wijewardana et al., 2016), highlighting that the complex relationship between leaf and plant productivity also depends on the respiration rate (Jiang et al., 2020). In contrast, elevated [CO₂] had limited effects on the assimilation, growth, and water transport efficiency of WW plants. Previous studies pot-grown crop plants have also shown that the benefits of [CO₂] enrichment are more significant under water deficit (Kang et al., 2002; Li et al., 2018) or soil salinity (Pérez-López et al., 2009; Zaghdoud et al., 2013). These results also agree with the findings of free-air CO₂ enrichment experiments that have shown that C₄ crops will not be more productive at elevated [CO₂], except under drought (Leakey et al., 2004; Manderscheid et al., 2014; Ainsworth and Long, 2021). C₄ plants have a much higher internal [CO₂] than C₃ plants at the carboxylation sites in chloroplasts as a result of the C₄ cycle (Taiz and Zeiger, 2010), which means that elevated ambient [CO₂] might have only limited additional effects under WW conditions. The increased water use efficiency does not help to improve plant water status and water transport capacity either as the plants are not under stress. However, given that climate predictions suggest the occurrence of more stress episodes, it seems likely that maize growth and productivity could be improved under future elevated [CO₂].

The higher growth under elevated [CO₂] should also come at a cost. At higher [CO₂], the improved water status (Fig. 2) alleviated the stress suffered under water deficit and soil salinity and thereby reduced the acclimation process to stress. This led to increased plant growth but also to significant increases in vulnerability to embolism and to sharp reductions in the HSM of plants under WS and SS when compared with ambient [CO₂] conditions (Fig. 3; Supplementary Table S6). Furthermore, elevated [CO₂] resulted in higher A_leaf and less leaf-rolling in the WS and SS treatments (Supplementary Table S4; Supplementary Fig. S4). Overall, the t_crit model indicated that the plants grown under 700 ppm [CO₂] were more vulnerable to severe drought (Fig. 6; Supplementary Fig. S6). From the genetic perspective, drought resistance and high yield are contradictory and can hardly be achieved simultaneously (Sun et al., 2023). The trade-off between P₅₀/P₈₈ and k_s and the correlation between plant growth and water supply are the rationales behind our second hypothesis: if plant acclimation to high [CO₂] results in increased xylem conductivity, gas exchange, and plant growth, it comes with an inherent cost, namely greater xylem vulnerability and risk of mortality. It is worth noting that this trade-off between P₅₀/P₈₈ and k_s was weaker under high [CO₂] in this study, and it was not observed in the results of our previous study that had four [CO₂] and two water levels (Liu et al., 2020). A possible explanation is that apart from vessel size, lipid surfactants and pit membranes between the vessels that are associated with xylem vulnerability (Sorek et al., 2021; Levionnois et al., 2022) might have interfered with the trade-off relationship under high [CO₂].

One of the possible criticisms that could be raised with regard our current study is that further dehydration of the soil (on the way to complete desiccation) would enable the plants to acclimate even under elevated [CO₂]. However, the acclimation process, which requires the formation of new xylem, takes many days and might only occur during the vegetative growth period (Sakaguchi and Fukuda, 2008). Specifically, in our experiment, the plants were given 2 months to acclimate to drought or salinity (Supplementary Table S1). Furthermore, under such severe stress, any further growth is likely to be impaired by insufficient turgor (Taiz and Zeiger, 2010). Due to the nature of the soil water retention curve, most water is available under high soil water potential. Accordingly, it is likely that further dehydration from the soil water potential that the plants experienced in our experiment until the stem xylem reached P₅₀ or P₈₈ would not have allowed sufficient time for acclimation. Another potential error in the t_crit model is that for the sake of simplicity, we ignored root water uptake, even though it might still exist even in dry soil (Lobet et al., 2014). If it did exist and the soil water content was partially available to the plants, then the t_crit value could be substantially larger than we have reported. At the same time, previous studies have indicated that crops show a significant decrease in root hydraulic conductance at elevated [CO₂] (Bunce, 1996; Huxman et al., 1999; Zaghdoud et al., 2013; Fang et al., 2019), meaning that if we took the soil water uptake into consideration, our conclusion regarding the greater vulnerability of plants at elevated [CO₂] would only be strengthened.

In conclusion, our results have revealed that elevated [CO₂] improved growth when maize suffered from water shortages or soil salinity. However, the improved water status under elevated [CO₂] interfered with hydraulic acclimation to stress, thereby increasing the threat to plant productivity and survival under severe drought events. It is important to note that this study was conducted on pot-grown plants in a controlled environment, and hence differed from natural conditions. But our results reveal a potential hydraulic threat to one of the world’s most important crops under future elevated [CO₂], and show that there is a need to extend these studies to field conditions.

Supplementary data

The following supplementary data are available at JXB online.

Table S1. The timeline of treatments and measurements during the experiments.

Table S2. Environmental parameters during the gas exchange measurements.

Table S3. Relative changes with time in the projected areas of the leaf samples for measurement of g_{min_roll}.

Table S4. Leaf gas exchange and growth of maize under different treatments in 2020.

Table S5. Leaf gas exchange, water potential, hydraulic conductance, and growth of maize under different treatments in 2018.

Table S6. Related parameters obtained from maize stem xylem embolism vulnerability curves under different treatments.

Table S7. Pressure–volume parameters related to water storage under different treatments.

Table S8. Sensitivity analysis for the effects of ±5% of P₅₀/P₈₈, DW, or A_leaf on t_crit variation under different treatments.

Fig. S1. Variations of temperature, relative humidity and [CO₂] in the phytotron in 2018 and 2020.

Fig. S2. Means and variations of soil water content in the pots in 2018 and 2020.

Fig. S3. Images of maize stem cross-sections.

Fig. S4. The minimum leaf conductance of expanded and rolled maize leaves under different treatments.

Fig. S5. The (t/b)² values of maize stem xylem under different treatments.

Fig. S6. Modeling results for storage water under different treatments based on HSM₈₈.

Fig. S7. Contributions of individual variables to changes in t_crit based on HSM₈₈.

Acknowledgements

JL acknowledges Dr Shabtai Cohen of the Agricultural Research Organization Volcani Center for his inspiration.

Author contributions

JL and SK planned the research; JL and QZ performed the experiments; JL and UH performed the data analysis and led the writing; all the authors contributed to writing the manuscript.

Conflict of interest

The authors declare that they have no conflicts of interest in relation to this work.

Funding

This work was supported by the National Natural Science Fund of China (grant no. 51790534) and the Discipline Innovative Engineering Plan of China (111 Program, grant no. B14002).

Data availability

The data supporting the findings of this study are available from the corresponding author, Shaozhong Kang, upon request.