Summary
Cuticle conductance (gcw) can bias calculations of intercellular CO2 concentration inside the leaf (Ci) when stomatal conductance (gsw) is small.
We examined how the light induction of photosynthesis impacts calculations by directly measuring Ci along with standard gas exchange in sunflower and tobacco leaves.
When photosynthesis was induced from dark to saturating light (1200 μmol m−2 s−1 PAR) the calculated Ci was significantly larger than measured Ci and the difference decreased as gsw increased. This difference could lead to over-estimation of rubisco deactivation by limited CO2 supply during early induction of photosynthesis. However, only small differences in Ci were observed during the induction from shade (50 μmol m−2 s−1 PAR) because gsw was sufficiently large. The induction from dark also allowed robust estimations of gcw when combined with direct Ci measurements. These gcw estimates succeeded in correcting the calculation, suggesting that the cuticle was the major source of error.
Despite a technical restriction to amphi-stomatous leaves, the presented technique has a potential to provide insights into the cuticle conductance on intact stomatous leaf surfaces.
Introduction
Since Moss and Rawlins (1963) first calculated intercellular CO2 concentration inside leaves (Ci) from the gas exchange of single leaves, the technique has become standard practice in gas exchange studies. The calculations became gained utility when Farquhar et al. (1980) developed a photosynthesis model that uses Ci to estimate substrate availability for ribulose-1,5-bisphosphate (RuBP) carboxylase oxygenase (rubisco). Based on a simple principle that rubisco carboxylation is limited by either of the two substrates, RuBP or CO2, the model predicts gas exchange of leaves as well as extrapolates underlying photosynthetic biochemistry (von Caemmerer, 2013). Accordingly, the calculations are the foundation of modern understanding of leaf carbon and water exchange.
However, calculations of Ci have been questioned because uncertainties exist in assumptions (Hanson et al., 2016). Ci is calculated from the relationship between water vapor exiting and CO2 entering the leaf by assuming both gasses only diffuse through stomata (Moss & Rawlins, 1963). In fact, water also diffuses through the cuticle. By including the cuticular transpiration (or cuticle conductance: gcw) in stomatal transpiration (or stomatal conductance: gsw), the calculation can overestimate the CO2 that has entered through stomata. Correction for cuticle effects is difficult as standard gas exchange measurements cannot distinguish cuticular and stomatal transpirations. Therefore, gcw is generally measured in isolated cuticles (Kerstiens, 1996; Schuster et al., 2017). The calculations also assume lateral uniformity in physicochemical properties over the leaf surface. Any heterogeneity, such as non-uniform stomatal apertures (Terashima, 1992; West et al., 2005), potentially bias the calculations as well (Terashima et al., 1988; Buckley, 1997; Meyer & Genty, 1998).
The accuracy of Ci can be validated by directly measuring Ci along with normal gas exchange measurements (Sharkey et al., 1982; Boyer & Kawamitsu, 2011; Tominaga & Kawamitsu, 2015a). In previous studies, amphi-stomatous leaves were put between an open gas exchange system and a closed system. The CO2 concentration inside the closed system is measured after it equilibrates with the Ci while gas exchange is measured on the other side. Indeed, Sharkey et al. (1982) found that the measured Ci was in close agreement to the value calculated on the other side when stomatal conductance was sufficiently large. Using a similar system, however, the over-estimation of calculated Ci has been demonstrated by a large discrepancy from the measured Ci when gsw is small (Boyer, 2015a; Tominaga & Kawamitsu, 2015b; Tominaga et al., 2018).
Ci is critical for understanding how photosynthesis approaches a steady state during light induction. Within the first minute after a shift from low to high irradiance photosynthesis can be limited by the slow regeneration of RuBP while the Calvin Benson Cycle enzymes are activated (Sassenrath-Cole & Pearcy, 1992). Much of the rest of induction may be (co)-limited by stomatal opening and activation of rubisco. Upon illumination Ci is depleted because stomatal opening is generally slower than activation of photosynthetic enzymes, thereby limiting rubisco carboxylation (Kirshbaum & Pearcy, 1988; Pearcy et al., 1996; Lawson and Blatt, 2014; McAusland et al., 2016; Dean et al., 2018). Meanwhile, rubisco is activated by sequential binding of CO2 and Mg2+ to the active site (Portis, 2003). This process is inhibited by a range of sugar-phosphates bound to the active site in darkness, and promoted by rubisco activase which releases these ligands in light. The reduction of Ci also hinders this activation process, thereby limiting rubisco carboxylation indirectly (Mott & Woodrow, 1993). Therefore, initial gsw at the onset of light induction as well as the rapidity of stomatal opening have a large impact on gas exchange kinetics (Wachendorf & Küppers, 2017; Vialet-Chabrand et al., 2017). Both rapid stomatal regulation (Wang et al., 2014; Lawson & Blatt, 2014; McAusland et al., 2016) and rubisco activation (Yamori et al., 2012; Carmo-Silva & Salvucci, 2013; Soleh et al., 2016; Taylor and Long, 2017) can play significant roles and are novel targets for improving photosynthesis.
Because the initial gsw tends to be small in low irradiance or darkness the Ci may be miscalculated during photosynthetic induction. Here we examine how light induction experiments impact the standard calculations by directly measuring Ci along with gas exchange in leaves of sunflower (Heliunthus annuus L.) and tobacco (Nicotiana tabacum L.). Our results show when cuticle conductance matters and how it affects data interpretation. Together, we propose a novel technique to estimate intact cuticle conductance using the dual gas exchange system.
Materials and methods
Plant material
Tobacco (N. tabacum L. cv. Samsun) and sunflower (H. annuus L. cv. Mammoth Russian) plants, 9–11 weeks old and 5–7 weeks old respectively, were used. The tobacco plants were grown in 2.5-L plastic pots, and the sunflower plants were grown in 3.8-L plastic pots during December 2018 to April 2019 under ambient CO2 at 24.2/19.9 °C day/night temperature; each pot contained a soil mixture (Metro-Mix® 360; Sun Gro Horticulture, Agawam, MA, USA) in a greenhouse located at the University of New Mexico [35.08°N, 106.62°W, 1587 m.a.s.l]. The plants were automatically watered three times each day and the pots were manually saturated with nutrient solutions containing 357 ppm N (Peters Professional® 20-20-20 General Purpose; ICL Specialty Fertilizers, Summerville, SC, USA) and 195 ppm chelated Fe (Southern Agricultural Insecticides, Inc., Rubonia, FL, USA) twice and once weekly, respectively. Daylength in the greenhouse was extended to 15 h with LED lamps (VRPRx PLUS; Fluence Bioengineering, Inc., Austin, TX, USA) to prevent early flowering. All experiments with these plants were conducted with fully expanded upper leaves.
Direct measurement system incorporated into LI-6800
The direct Ci measurement system has previously been incorporated into a commercial open gas exchange system LI-6400 (LI-COR, Lincoln, NE, USA) and reported in detail elsewhere (Tominaga & Kawamitsu, 2015a). We applied a similar system for a new LI-6800. Instead of customizing a chamber, the manufactured LI-6800 chambers could be slightly modified to include the direct system (Fig. S1). In the present experiments, a large 6 x 6 cm chamber (6800-13; LI-COR) was used to maximize measurement precision, but the following system is applicable to the other chambers (e.g., Fluorometer chamber; 6800-01A; LI-COR). The chamber bottom-half was detached from the manifold and rotated 180°. Then, the flexible bellows originally attached to the bottom chamber were replaced with a solid adapter. The manifold to the bottom chamber was closed by a blanking plate with a rubber plate in between so that air only flows through the top chamber (i.e., open system). All experimental leaves were large enough to cover the entire chamber area, so the top and bottom chambers were divided by the leaf. Two miniature impeller pumps were connected in parallel to the adapter (inlet and outlet) of the rotated bottom chamber, and smoothly circulated the air to an infrared gas analyzer (IRGA; LI-7000; LI-COR) in a loop. When a sample leaf closed the loop, a water jacketed glass tube (condenser) maintained the dew point slightly lower than the room temperature (checked by the LI-7000), preventing condensation in the IRGA. In the closed loop, CO2 equilibrated with that inside the stomatal pores (Ci(m)) was continuously measured while gas exchange through the opposite side of same leaf area was measured with the open system, which automatically calculated Ci (Ci(c)). Accordingly, both Ci(m) and Ci(c) were obtained simultaneously for the same leaf area. We configured LI-7000 to the differential mode by leading the air to the reference cell from the LI-6800 reference port (located at the backside of the sensor head) at a constant flow rate of 400 μmol s−1. The values of CO2 and water vapor concentrations in the reference air measured by the LI-6800 was exported to the LI-7000 whereas the sample CO2 concentration measured by the LI-7000 (i.e. Ci(m)) was exported to and recorded by the LI-6800 along with other gas exchange parameters. By sharing the reference gas and data between the two systems, measurements were accurate and in sync. Plumbing of the closed loop was made mostly with copper tubing to avoid diffusion leaks. The advanced polymer gaskets (6568-512; LI-COR) and a large leaf area to edge ratio minimized the diffusion leaks such that any leaks were too small to be detected for the empty chamber in our experimental conditions, and no corrections were performed.
Light induction
Plants were taken in the morning from the greenhouse to the laboratory. There, whole plants were acclimated in the dark for at least 1 h. Leaves were then clamped and illuminated by the chamber equipped with a large light source (6800-03; LI-COR). We set three levels of irradiances with white LEDs: 50, 200, and 1200 μmol m−2 s−1 photosynthetically active radiation (PAR) with the spectrum shown in Fig. S2. Measurements were automatically logged every 10 s until reaching a steady state. In all irradiances, full induction usually took less than 1.5 h and 2 h for tobacco and sunflower leaves, respectively. In addition to the induction from the dark, leaves were also subjected to saturating light (1200 μmol m−2 s−1 PAR) after being acclimated in shade (50 μmol m−2 s−1 PAR). Throughout the experiments, incoming air with a flow rate of 700 μmol s−1, air temperature of 25 °C, water vapor of 10 mmol mol−1 and CO2 concentration of 400 μmol mol−1 were held constant to avoid transient artefacts. Leaf temperature increased instantaneously upon illumination and was decreased throughout the induction by cooling from transpiration as stomata opened. Leaf temperature ranges varied among inductions: in tobacco, 24.5 ± 0.2□25.1 ± 0.1 °C (50 PAR), 24.4 ± 0.1□25.3 ± 0.1 °C (200 PAR), and 26.0 ± 0.1□27.4 ± 0.2°C (1200 PAR), and in sunflower, 24.0 ± 0.1□24.7 ± 0.1 °C (50 PAR), 24.1 ± 0.2□25.3 ± 0.2°C (200 PAR), and 25.8 ± 0.6 26.9 ± 0.4°C (1200 PAR), respectively.
Calculations
The intercellular CO2 concentration inside the leaf (Ci) was calculated from the gas exchange measurements on the adaxial leaf side, according to von Caemmerer and Farquhar (1981) as: where Cb is the CO2 concentration out of the leaf in the boundary layer (mol mol−1), A and E are the CO2 flux (mol CO2 m−2 s−1) and water vapor flux (mol H2O m−2 s−1), respectively, gsc is the stomatal conductance of CO2 (mol CO2 m−2 s−1). Given that Cb, A, and E are measured variables, gsc is the critical parameter that determines validity of Ci(c) (though measurement precision of Cb, A, and E affect the calculation). Then, gsc would be expressed as: where gsw is the stomatal conductance of water vapor (mol H2O m−2 s−1) and 1.6 is the ratio of diffusivities of CO2 and water vapor in air. Then, gsw was calculated as: where Wi and Wb are the water vapor concentrations inside the leaf (mol mol−1) and out of the leaf in the boundary layer (mol mol−1), respectively. The Wi was assumed to be saturated at the leaf temperature. However, the gsw includes cuticle conductance of water vapor (gcw) as: where gsw’ is the strict sense of stomatal conductance of water vapor, which is only based on water vapor passing through the stomatal pore, and is in parallel with cuticle conductance (Jarvis, 1971).
Instead of estimating from gsw (Eq. (3)), we also calculated gsc from the directly measured Ci (Ci(m)) (Boyer and Kawamitsu, 2011) as: where gsc’ was directly calculated from the measured CO2 gradients across the adaxial leaf surface.
According to the manufacturer’s calibration, boundary layer conductance of CO2 (1.5 CO2 mol m−2 s−1) and that of water vapor (2.1 mol H2O m−2 s−1) were estimated from the constant fan speed of 14,000 rpm, and used for calculating the Cb and Wb, respectively. A summary of parameters referred to within the text is shown with accompanying units in Table 1.
Stomatal kinetics
To get a sense of ‘rapidity’ of stomatal openings and increase of gsw in the leaves used here, we estimated kinetic parameters for the temporal response of gsw according to a sigmoidal model (McAusland et al., 2016; Vialet-Chabrand et al., 2017): where G0 and Gsteady are the gsw (mmol H2O m−2 s−1) at the time, t (min), for the beginning of light induction (t = 0) and at the steady state, respectively, ki (min) is the time constants for the increase of gsw and λ is the initial lag time (min). The parameter values were estimated with using a curve-fitting routine in Microsoft Excel provided elsewhere (Supplemental File S1 in Vialet-Chabrand et al., 2017). The maximum slope of gsw increase (SImax, mmol m−2 s−2) was calculated based on the previously described parameters as:
Our estimates only reflect the adaxial stomata while they usually represent both adaxial and abaxial stomata in amphi-stomatous leaves (e.g., McAusland et al., 2016).
Rubisco activation kinetics
The activation of rubisco would limit photosynthesis within approximately the initial 10 min of the increase in PAR (Woodrow & Mott, 1989,1992). This phase can be found as a linear portion of a semilogarithmic plot of normalized photosynthesis to a constant Ci(A*) versus time (t) (Woodrow & Mott, 1989), expressed as: where a and b are the slope and intercept of the regression for this linear portion, is the A* in the final steady state. The relaxation time for the rubisco phase (τ) is then defined from the slope a as:
In the induction experiments from dark to saturating light, we estimated τ with using Ci(c) or Ci(m), according to Woodrow and Mott (1989). A was normalized to a Ci of 250 μmol mol−1 by assuming a linear relationship between A and Ci, that passed through the CO2 compensation point of 50 μmol mol−1. The maximum A* was used as during each induction. a was determined by linear regression of the data for 3 min (i.e., 18 points).
Stomatal density
Nail polish was spread at three points on each upper and lower side of ten leaves. The dried polish was striped with clear tape and attached to a glass slide (a replica of the epidermis). Microphotographs (0.60 mm2) of the replicates were taken with a digital camera (AxioCam HRc; Carl Zeiss, Göttingen, Germany) attached to a microscope (Axioskop 2 Mot Plus; Carl Zeiss, Göttingen, Germany). Stomata were counted in each replicate and averaged to calculate stomatal density for each side of the leaves.
Intercellular CO2 conductance and gradients
Vertical gradients of CO2 exist in a leaf because CO2 has to further diffuse through the mesophyll airspace once inside the stomata (Parkhurst et al. 1988). In our measurement system, Ci gets lower towards abaxial side because CO2 is only supplied from the adaxial side and more CO2 is consumed as it diffuses deeper into the leaf (Boyer and Kawamitsu, 2011). In one-dimensional model, the CO2 gradient is expressed as: where gias is the intercellular conductance of CO2 (mol CO2 m−2 s−1). We determined gias anatomically from the fraction of mesophyll volume occupied by intercellular airspace (fias) and the effective diffusion path length (ΔLias) (Syvertsen et al., 1995) as: where ζ is the tortuosity factor and Da is the diffusion coefficient for CO2 in the gas phase (1.51 10−5 m2 s−1). We used mesophyll thickness as ΔLias and a constant ζ of 1.57 (Tosens et al., 2012). The fias and the mesophyll thickness between the two epidermal layers were measured from microscopic observations of leaf sections with the same equipment described above.
Statistics
The number of replications and the statistical tests, if applicable, are presented in figure legends for each experiment. The results are given as means with SDs unless otherwise indicated.
Results
Leaf anatomy and intercellular conductance to CO2
Sunflower leaves had more stomata than tobacco leaves while their stomatal ratios were similar (Table 2). For adaxial sides on which gas exchange was measured, the mean stomatal density was 49 ± 4 mm−2 in tobacco and 118 ± 9 mm−2 in sunflower, respectively. In both species, adaxial side had fewer stomata than abaxial side with the mean stomatal ratio (adaxial/abaxial) of 0.6-0.7. Intercellular conductance of CO2 (gias) was determined from anatomical properties of leaf sections (Table 2). Sunflower leaves trended toward having a smaller fraction of airspace (0.27 ± 0.01) than tobacco leaves (0.35 ± 0.05) whereas both leaves had a comparable mesophyll thickness of around 200 μm. There were no significant differences in the mean gias between tobacco (617 ± 87 mmol CO2 m−2 s−1) and sunflower (538 ± 135 mmol CO2 m−2 s−1).
Light induction from dark to saturating irradiance
When dark-adapted leaves were clamped with the chamber under saturating irradiance (1200 PAR), both assimilation and conductance increased slowly (Fig. 1a,b). Meanwhile, CO2 concentration in the closed loop was rapidly consumed by photosynthesis until reaching an equilibrium with the airspace inside the leaf (Ci(m) in Fig. 1c). Until then, the Ci(m) was higher than the actual Ci and the gsc’ derived from the Ci(m) was incorrect (shown with grey background in Fig. 1). The equilibrium state was indicated by the parallel changes in the gsw and gsc’ as both changes should commonly reflect the stomatal behavior (inset in conductance of Fig. 1b). The equilibrium was generally expected to occur within 5 min after clamping on the leaf in both species. The Ci(c) was significantly larger than the Ci(m), and the difference became largest when the Ci(m) was around the minimum (Fig. 1c). The Ci(m) decreased to 96 ± 15 μmol mol−1 in tobacco and 76 ± 11 μmol mol−1 in sunflower, whereas the minimum Ci(c) was 137 ± 19 μmol mol−1 and 150 ± 10 μmol mol−1, respectively. Therefore, rubisco underwent much lower Ci than was predicted from the calculation. Because rubisco activity can be inferred from A relative to Ci within the early induction phase, these differences in the minimum Ci would predict a different activation status of rubisco. Based on the Ci(m) or Ci(c), we estimated the relaxation time (τ) primarily limited by the rate of rubisco activation as illustrated in Fig. S3. The τ estimated from the Ci(c) (5.0 ± 2.7 min and 7.1 ± 2.8 min in tobacco and sunflower, respectively) were significantly longer than those estimated from the Ci(m) (2.0 ± 1.0 min and 3.3 ± 2.0 min, respectively) (Fig. 2). The difference between the Ci(c) and Ci(m) diminished as the conductance increased (Fig. 1c).
Light induction from shade to saturating irradiances
In light induction experiments, leaves are often pre-acclimated to shade, rather than to the darkness. We also looked at the induction from shade (50 PAR) (Fig. 3). While the induction from dark to shade similarly resulted in a large discrepancy between the Ci(c) and Ci(m) as dark to saturating light, the subsequent induction from shade to saturating irradiance caused less discrepancies (Fig. 3c). However, tobacco Ci(c) and Ci(m) remained different at saturating light (insets of Figs. 1c & 2c). A large difference between the minimum Ci(c) and Ci(m) was also evident in the induction from the dark to 200 PAR (data not shown). Therefore, the calculations primarily deviated from the direct measurements only in the induction from the darkness. For those dark-adapted leaves, the initial gsw right after the induction was as small as 9.4 ± 6.6 mmol H2O m−2 s−1 and 9.4 ± 6.8 mmol H2O m−2 s−1 for gsw in tobacco and sunflower, respectively. On the other hand, when the photosynthesis reached a steady-state in the shade, gsw increased to as much as 61 ± 20 mmol m−2 s−1 in tobacco, and 190 ± 56 mmol m−2 s−1 in sunflower. These results are consistent with the previous direct measurements showing the over-estimation of Ci(c) mostly when gsw is small (Tominaga & Kawamitsu, 2015b, Boyer, 2015a; Tominaga et al., 2018), and these initial gsw in the shade-acclimated leaves must have been sufficiently large to avoid over-estimations.
Kinetics of stomatal openings
Kinetic parameters for temporal response of gsw during the light induction were summarized in Table 3. In general, increase of gsw was sigmoidal in tobacco whereas it was multiple stepwise in sunflower (Figs. 1b & 3b). As a result, the smooth sigmoidal function (Eq. 6) was less fit in sunflower showing the higher normalized root-mean-square error (NRMSE). Time constant of stomatal openings (ki) was shorter in the induction from dark to shade (0 to 50 PAR), suggesting that stomata reached steady-state more quickly with partial openings. On the other hand, the lower SImax indicated the slower increase of gsw in the shade. In the induction from dark to saturating light (0 to 1200 PAR), average SImax was higher in sunflower (0.139 mmol m−2 s−2) than in tobacco (0.047 mmol m−2 s−2) despite the similar k (approx. 20 min) because of the larger steady-state gsw with the higher stomatal density (Table 2). These SImax values were within or close to the high end of reported values for C3 forbs displaying the similar growth form and guard cell type (McAusland et al., 2016). Hence, it follows that over-estimations of Ci(c) in the above experiments occurred in leaves with moderate to rapid gsw responses. Shade acclimation (50 to 1200 PAR) neither shortened the ki nor increased the SImax (Table 3).
Estimation and correction of cuticle conductance
The standard calculations use the diffusivity ratio of water vapor to CO2, which is a constant value of 1.6 in air (Sharkey & Farquhar, 1982; LI-COR, 2019) while assuming that stomata are the dominant path for both gasses (Eq. (2)). Note that Massman (1998) reported uncertainty (maximum deviations after removing outliers) of ± 7% and ± 5% for the respective diffusivities, indicating ± 12% uncertainty (± 0.19) in the mean ratio of 1.58. We tested this assumption that gsw = 1.6gsc’ by comparing the gsw with the gsc’ during the initial light induction from the dark (gsc’<20 mmol CO2 m−2 s−1) (Fig. 4). The gsw was linearly related to the gsc’, indicating a tight stomatal regulation on the gsw in this early induction phase. However, the gsw was consistently larger than the assumption of gsw = 1.6gsc’ (broken lines in Fig. 4). Given little CO2 transport through the cuticle (Boyer, 2015a), a zero gsc’ would be obtained with complete stomatal closure. Thus, the gsw with complete stomatal closure can be estimated by extrapolating the regression line to the intersection with the y-axis. This intersection is consistently above the origin (Fig. 4), indicating that water transport continues even if stomata close completely. We considered this residual gsw as cuticle conductance for water vapor (gcw).
To see the impact of cuticle conductance, we corrected Ci(c) by calculating the strict sense of stomatal conductance to water vapor (gsw’), as described in Eq.(4), and substituting it for the gsw. The initial extrapolated gsw’ right after the induction from dark was 6.8 ± 6.5 mmol H2O m−2 s−1 and 6.1 ± 6.4 mmol H2O m−2 s−1 in tobacco and sunflower, respectively, that are 63 ± 21% and 56 ± 22% of the initial gsw. The corrected Ci(c) (Ci(c),cut) followed the Ci(m) more closely than the Ci(c) and the discrepancy became much smaller (Fig. 5a,b), suggesting that the discrepancy in the early induction was mostly attributable to the contribution of cuticular water loss. In contrast, the gcw marginally corrected the Ci(c) as the gsw increased, and there remained the discrepancy even when the gsw was maximum, especially in tobacco (insets of Fig. 5a,b).
The gcw did not show a particular trend in response to the irradiance, with the mean across all irradiances of 2.65 ± 1.10 mmol m−2 s−1 in tobacco and 3.27 ± 1.93 mmol m−2 s−1 in sunflower, respectively (Fig. 6a). Because of the strict linearity of the regressions (Fig. 4), the standard error of each gcw estimate (i.e., y-intercept) was less than 0.1 mmol m−2 s−1 in both species, that is much smaller than the variations among leaves (Fig. 6a). The slope of the regression was mostly within or slightly above the uncertainty interval of 1.58, except for the inductions in shade (Fig. 6b). The slower photosynthesis in the shade (Table S1) would delay the equilibrium in the closed system, and perhaps affected the slope. The slope value was greater in sunflower than in tobacco across the irradiances (Fig. 6b).
Effect of intercellular CO2 gradients
Because intercellular conductance is finite, vertical gradients of Ci must be present in a leaf. In our measurement system, the Ci(m) should be the lowest Ci as the CO2 is only supplied from the opposite side, and the CO2 gradients would be attributable to the residual difference of Ci(c)cut from Ci(m). Assuming that the Ci(c) represents the value just inside the stomatal pore or ‘stomatal cavity’ on the adaxial stomata (Sharkey et al., 1982; Parkhurst et al; 1988; Buckley et al., 2017), we subtracted the gradients estimated with the mean gias (Table 2) according to Eq. (10) from the Ci(c),cut. The gradient reduced the Ci(c) when the gsw was relatively large (Ci(c),cut + grad in Fig. 7a,b). The gradients slightly overcorrected the Ci(c) in tobacco (insets of Fig. 7a,b), and these mismatches were adjustable by approximately doubling the gias (data not shown). In sunflower the Ci(c) was already close to or even lower than the Ci(m) by several-ppm when the stomata were open (Figs. 1c & 2c). Then, the gradients greatly overcorrected the Ci(c) (Fig. 7a,b), and the gias had to be unrealistically large—more than 10X larger than the anatomical gias (>5000 mmol m−2 s−1)—in order to match the Ci(c),cut+grad with the Ci(m). Therefore, the over-correction would rather indicate that Ci(c) was indicated deeper inside the leaf than the adaxial stomatal cavity and/or that Ci(c) was underestimated.
Discussion
In this study, the intercellular CO2 concentration inside leaves was directly measured as well as calculated from the standard gas exchange measurement. A commercially available open system was minimally modified to include the direct system, and simultaneous measurements on the same leaf area allowed a close comparison. The standard calculations for the gas exchange measurements were validated under a square-wave light pattern that is often employed for induction studies. Over-estimations of Ci were observed in the dark-adapted leaves whose stomata started out closed. This raises a potential problem for interpreting gas exchange kinetics during non-steady state photosynthesis because Ci is an essential indicator of dynamic limitations (Kirshbaum & Pearcy, 1988; Pearcy et al., 1996). In our experiments, over-estimation of Ci was profound most at around its minimum during the early induction phase (Fig. 1c). Within this timescale photosynthesis is primarily limited by the rate at which rubisco is converted from an inactive to an active form (Woodrow & Mott, 1989, 1992). We found that the over-estimation of the minimum Ci delayed the relaxation period of the rubisco phase (Fig. 2). This suggests that rubisco activation status derived from A vs Ci relationship is potentially underestimated while the rubisco actually experiences lower CO2 concentrations than the calculation indicates. In other words, one would be fooled as if rubisco is decarbamylated at the higher Ci(c) than the actual Ci. Our study demonstrated that the cuticular water loss was responsible for the calculation error and should be taken into account in the induction measurements.
Importantly, the gcw hardly affected the induction calculation when stomata were partially opened in the shade prior to the full induction (cf. Ci(c) & Ci(c),cut in Fig. 5b). In the induction from shade in Alocasia macrorrhiza (giant taro), Kirshbaum and Pearcy (1988) also realized that the A/Ci slope (at the initial 1 min) declined when the initial gsw was suppressed under high CO2. While stomata would partially open in fairly low irradiance under the current atmospheric CO2 level (Valladares et al., 1997), the initial gsw varies greatly among species in both shade (McAusland et al., 2016; Deans et al., 2018) and darkness (Wachendorf & Küppers, 2017). The initial gsw is confined in and likely scaled to the maximum gsw among taxa in vascular plants (McAusland et al., 2016, Deans et al., 2019). Because greater stomatal conductance appears to evolve through the vascular plant lineages (Flanks & Beerling, 2009; McElwain et al., 2016), ferns and gymnosperms may have smaller initial gsw than angiosperms. Together, ‘sluggish’ stomatal openings could exacerbate the calculation by prolonging the cuticle effect. In general, graminoids with dumbbell-shaped guard cells exhibit much faster light response of gsw than the other vascular plants with elliptic guard cells (Franks & Farquhar, 2007; Vico et al., 2011; McAusland et al., 2016). In the elliptic guard cells, both stomatal opening and gsw increase are slower with lower maximum gsw (McAusland et al., 2016), and gymnosperm opens stomata more slowly than angiosperm (Vico et al., 2011). Besides genetic variations, water stress decreases the initial gsw (Allen & Pearcy, 2000; Wachendorf & Küppers, 2017) and so drought conditions should be prone to the error.
Cuticle conductance also varies among leaves in a species (Fig. 6a) and potentially more among species (Kerstiens, 1996; Schuster et al., 2017). Furthermore, intact cuticle conductance may change with leaf water status (Boyer et al., 1997; Burghardt & Riederer, 2003; Boyer, 2015b). Therefore, a correction seems problematic unless gcw for each intact leaf is known. Indeed, the average gcw could result in over-correction of Ci in our experiments (data not shown). Although, in principle, the cuticle conductance can be easily included in the calculations for Ci (Ci(c),cut), accurate measurement of gcw is essential for a reliable correction.
Estimation and correction of cuticle conductance
Due to technical difficulty in the measurement of cuticular transpiration on a stomatous leaf surface/cuticle (e.g., Šantrůček et al., 2004), cuticle conductance has been studied exclusively in hypostomatous species with using the astomasous leaf surface/cuticle (Kerstiens, 1996; Schuster et al., 2017). Kirshbaum and Pearcy (1988) first estimated the gcw on the stomatous leaf surface from the above induction measurements in giant taro. They determined the gcw that could match the two initial A/Ci slopes observed with a low and high gsw in each leaf, and reported the mean of 2.7 ± 0.3 mmol m−2 s−1 (± SE, n=7). As they noted, however, the method assumes that the activation of rubisco is independent from the initial gsw, which would not hold (Mott & Woodrow, 1993). In the previous experiments using similar dual gas-exchange systems (Boyer 2015a, Tominaga & Kawamitsu, 2015b), gcw or cuticular transpiration was estimated from a deviation of the calculation from the direct measurement. Unlike the Kirshbaum and Pearcy method, these estimations can be made from a single measurement without assuming the rubisco activation status, thereby increasing the throughput. Nevertheless, these methods still assumed that gcw is solely responsible for their measurement discrepancies, and potentially neglected other factors, such as intercellular CO2 gradients. In contrast, our intercept method extrapolates gcw at virtually complete stomatal closure when the CO2 gradient becomes negligibly small with A≈0 (Fig. 4). Furthermore, the linear relationship allowed a simple extrapolation for a correction that improved the agreement between calculated and measured values (Fig. 5).
The regression slope was slightly larger than 1.58, the modelled diffusivities ratio of CO2 and water vapor in air (Massman, 1998), but mostly within the uncertainty interval (Fig. 6b). A potential explanation for the larger slope is a non-uniform stomatal opening that breaks the assumptions of lateral uniformity over the leaf surface: uniform stomatal aperture, leaf temperature and metabolic capacity, and in the leaf: uniform CO2 distribution (Terashima et al., 1988; Meyer & Genty, 1998; West et al., 2005). Because non-uniformity may be more obvious in heterobaric leaves in which bundle-sheath extensions laterally disconnects the airspace (Terashima, 1992), the larger slope in sunflower than tobacco (Fig. 5b) might be due to the greater heterobaricity (Morison & Lawson, 2007; also noticeable in our microscopic observation). It would be interesting to compare the thermal and fluorescence imaging over the leaf surface (e.g., West et al., 2005) during induction experiments among leaves with a wide range of heterobaricity.
Intercellular CO2 gradients and unsaturation in the airspace
In tobacco, the intercellular CO2 gradient seems to explain the discrepancy between the Ci(c) and Ci(m) when stomata are open and gcw has little impact on the calculation (Fig. 7). The slight over-correction of Ci(c) may be due to under-estimation of the anatomical gias. However, the much larger over-correction of Ci(c) in sunflower would indicate that the Ci(c) was located deeper than the stomatal cavity (i.e., closer to the Ci(m) than we expected). This suggests that internal evaporating surface where humidity is saturated (Wi) occurs deeper than the stomatal cavity, and that gsw includes the diffusion in the intercellular airspace. Also, this notion could explain the discrepancy in the over-correction between the two species. Because closer arrangements of stomata should decrease the lateral diffusion distance of CO2 inside the leaf, the pathlength ratio of water vapor to CO2 per stoma may increase with increasing stomatal density. In sunflower leaves having the 2.4X as many stomata (Table 2), a larger pathlength ratio might cause the deeper Ci.
It is also possible that the over-correction is manifested by under-estimated Ci(c). In calculating the gsw (Eq. 3), Wi, is assumed to be saturated, and if not the gsw is more or less underestimated. We observed a progressive under-estimation of Ci(c) when those tobacco and sunflower leaves were detached and rapidly desiccated (MS#2). There, the relative humidity in leaf airspace was estimated to be less than 80% of saturation in both species as the leaves dried up. Likewise, Cernusak et al. (2018) estimated that the relative humidity could be as low as 77% and 87% in Pinus edulis and Juniperus monosperma even when the leaves opened stomata and actively photosynthesized.
Applications
One could do a direct Ci measurement simply on one leaf side, but the greatest value is pairing with calculated values on the other side for an accurate paired measurement of water fluxes. In that case a correction of the direct measurement is needed for the vertical CO2 gradient especially in leaves with lower gias (e.g., thicker leaves). Anatomical measure of gias may be used for this correction. The dual gas-exchange system allows a robust estimation of cuticle conductance for water vapor on intact stomatous leaf surfaces in amphi-stomatous species, giving an new opportunity to study cuticular physiology in connection with stomatal physiology. The technique, owing to a high throughput (<10 min/ sample), will also allow for exploring a large number of plant species. A disadvantage is a current limitation of the approach to amphi-stomatous leaves. In hypo-stomatous leaves CO2 moves through the astomatous side so slowly that the closed system requires much longer times to reach equilibrium. The applicability also depends on the stomatal behavior. The estimation of gcw would be more accurate as stomatal conductance diminishes, although this is aided by simply conducting induction responses from dark-adapted leaves. In addition to gcw, location and/or unsaturation of Wi deserves more investigations as both have wide implications for gas exchange and water transport in leaves.
Author contributions
JT planned and designed the research. JT and JRS performed experiments and analyzed data. JT and DTH wrote the manuscript.
Acknowledgements
Thanks are due Dr. Roxana Khoshravesh for leaf anatomical measurements. We particularly appreciate thoughtful discussions with Prof. John S. Boyer (University of Missouri). This work was supported by funding to JT through the JSPS KAKENHI 18J00308 at Hiroshima University, and to DTH through the NSF EPSCoR Program under Award # IIA-1301346 and through NSF IOS 1658951 at the University of New Mexico. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. JT is supported by Research Fellowships for Young Scientists from the Japanese Society for the Promotion of Science.