High mesophyll conductance in the high-yielding rice cultivar Takanari quantified with the combined gas exchange and chlorophyll fluorescence measurements under free-air CO2 enrichment

ABSTRACT An effective strategy for increasing crop production is increasing the rate of photosynthesis. In this study, we conducted gas exchange and chlorophyll fluorescence measurements for a high-yielding rice cultivar, Takanari, to identify the leaf physiological properties that contribute to high capacity for photosynthesis of the uppermost leaves before (panicle initiation stage) and after heading (grain-filling stage) in the Tsukuba free-air CO2 enrichment (FACE) facility. The higher photosynthesis rate of Takanari compared with that of the commonly cultivated cultivar, Koshihikari, was mainly attributed to the greater stomatal conductance for CO2 (gsc) at the panicle initiation stage and to the greater mesophyll conductance (gm) at the grain-filling stage in both current and elevated atmospheric CO2 concentrations [CO2]. Takanari had a higher level of leaf nitrogen content (Nl) compared with Koshihikari at the grain-filling stage, which led to greater gm and maximum carboxylation rate (Vc,max), but Nl alone did not explain the variations of gm within the variety. A clear correlation was found between Vc,max and Nl. Calculating Vc,max taking gm into consideration removed the artifact of Vc,max25 in relation to Nl that was observed when gm was assumed to be infinite. Our results emphasize the need to separate the roles of Vc,max and gm to accurately understand the ecophysiological processes that control leaf photosynthesis in Takanari. Graphical Abstract


Introduction
In the face of the world's growing population, global crop production needs to be increased. Increasing the rate of photosynthesis is one strategy to increase crop production (Hubbart, Peng, Horton, Chen & Murchie, 2007;Long, Zhu, Naidu & Ort, 2006;Mann, 1999;Murchie, Pinto & Horton, 2009;Ort et al., 2015). Investigations have been conducted to understand the ecophysiological traits of available crop cultivars that support high rates of photosynthesis in order to maximize the use of available genetic resources.
High mesophyll conductance (g m ) supports a high rate of leaf photosynthesis. However, due to the difficulty of accurately quantifying g m , the effect of mesophyll conductance is often neglected in terrestrial ecosystem models that investigate crop photosynthesis at a canopy scale (Ikawa et al., 2018). Researchers often use maximum carboxylation rate (V c,max ) based on intercellular CO 2 concentration (C i ) instead of on actual CO 2 concentration at the carboxylation site (C c ), assuming that g m is infinite and, therefore, C c is equal to C i . Such an assumption leads to an underestimation of V c,max (apparent V c,max ) and overestimation of C c . However, a growing number of studies have suggested that the impact of g m on photosynthesis is not trivial, highlighting the need to disentangle the roles of V c,max and g m in photosynthesis (Adachi et al., 2013;Knauer et al., 2019;Lauteri, Haworth, Serraj, Monteverdi & Centritto, 2014). Furthermore, Sun et al. (2014) demonstrated that photosynthesis models based on the apparent V c,max assuming infinite g m are accurate only within the limited conditions in which the A-C i curve was observed.
Takanari (Oryza sativa L. cv. Takanari) is known for its high grain yield (Imbe et al., 2004). The Tsukuba rice freeair CO 2 enrichment (FACE) experiment revealed that both high sink (i.e. large panicles) and source (i.e. high carbon supply) capacities of Takanari (Chen et al., 2014;Ikawa et al., 2018;Nakano et al., 2017) contribute to the greater grain yield with less loss in the quality in elevated atmospheric CO 2 concentration [CO 2 ] compared to a commonly grown cultivar, Koshihikari (Hasegawa et al., , 2019Zhang et al., 2013Zhang et al., , 2015. Based on a canopy scale model, Ikawa et al. (2018) attributed Takanari's high canopy photosynthesis to greater stomatal conductance (g sc for CO 2 ) and a better nitrogen allocation compared to Koshihikari. However, their model did not explicitly consider the effect of g m . Chen et al. (2014) reported that the high leaf photosynthesis of Takanari was supported mainly by high g m in later growth stage. However, the method to quantify g m used in Chen et al. (2014) depends on the curve-fitting method, which is not necessarily recommendable if any alternative method is available (Pons et al., 2009).
Recent advancements in gas exchange measurement systems enable a wide range of users to quantify g m more reliably and easily. Using a newly-released photosynthesis system (LI-6800, LICOR, USA), this study aims to redefine the physiological traits of Takanari and determine why Takanari has a high rate of leaf photosynthesis. We set three hypotheses: (1) Takanari has greater V c,max , g m , and g sc than Koshihikari both in current and elevated [CO 2 ]; (2) the difference in g m between [CO 2 ] treatment and variety is greater than that of either V c,max or g sc with respect to their effects on leaf photosynthesis; and (3) Takanari's high V c,max and g m are explained by high leaf nitrogen contents. The rationale for the first two hypotheses is based on the results of the study by Chen et al. (2014). The third hypothesis is based on the fact that the uppermost leaves of Takanari have a high nitrogen content (Taylaran, Adachi, Ookawa, Usuda & Hirasawa, 2011), and also on recent studies reporting that the V c,max and g m of rice leaves are correlated to leaf nitrogen content (Cai et al., 2018).
To test these hypotheses, the objectives of this study are: (1) to calculate the V c,max , g m , and g sc using combined gas exchange and chlorophyll Table 1. Measurement date, time, block, measurement order, and meteorological conditions during the combined gas exchange and fluorescence measurements in 2017. The order of the measurements was randomized for each variety (KH: Koshihikari and TN: Takanari) and for CO 2 treatment [A-CO 2 : a control plot with current (CO 2 ;~390 μmol mol −1 ) and E-CO 2 : elevated (CO 2 ;~590 μmol mol −1 )]. Air temperature (T a ) and relative humidity (Rh) were measured at 2 m above the ground and wind speed (U) was measured at 2.5 m above the ground. PPFD (photosynthetic photon flux density) is the value on the horizontal plane. (2) to conduct a sensitivity analysis of each parameter using the Farquhar, von Caemmerer, and Berry leaf photosynthesis model (Farquhar, von Caemmerer & Berry, 1980); and (3) to measure leaf nitrogen content and compare it with V c,max and g m .

Site descriptions
Leaf-level gas exchange measurements were conducted in the Tsukuba FACE experimental facility in Tsukubamirai, Ibaraki prefecture, Japan (35°58′ N, 139°6 0′ E, 10 m.a.s.l.). The experimental facility includes four FACE rings (hereafter referred to as E-CO 2 plots), with a diameter of 17 m and four control blocks (A-CO 2 plots) (Nakamura et al., 2012). Rice seedlings of both Koshihikari and Takanari were transplanted in experimental plots on May 24-25 in 2017. Heading occured about 70 and 75 days after the transplanting for Koshihikari and Takanari, respectively (Hasegawa et al., 2019). Measurements were made at panicle initiation (July 11-15) stage and at grain-filling stage (August 13-25) in 2017 (Table 1).

Combined gas exchange and chlorophyll fluorescence measurements
The uppermost expanded leaves were targeted for gas exchange and chlorophyll fluorescence measurements with the LI-6800. The heading date varied among plants, particularly in Takanari, and flag leaves with a fullyemerged panicle were selected at the grain-filling stage. Leaf samples were exposed to photosynthetic photon flux density (PPFD) at 1,500 μmol m −2 s −1 for 15-20 min until the steady-state condition was achieved under [CO 2 ] of 390 μmol mol −1 in A-CO 2 and 590 μmol mol −1 in E-CO 2 . When it was overcast, samples were first exposed to PPFD at 1,000 μmol m −2 s −1 for several minutes and PPFD was increased to 1,500 μmol m −2 s −1 . The fractions of red and blue lights were set at 0.9 and 0.1, respectively. Gas exchange data including net leaf photosynthesis rate (A n ) and g sc were then logged at the steady-state condition. Once the steady-state condition was achieved, [CO 2 ] in the reference cell was decreased and was kept at 200 µmol mol −1 for several minutes. Subsequently, [CO 2 ] was further decreased to 150, 100, 55, and 20 µmol mol −1 , and data were logged within 150 s at each CO 2 level. The chamber temperature and humidity were set at 30°C-35°C and 60%, respectively. Other chamber conditions were: fan speed, 10,000 rpm; flow rate to the sample cell, 500 μmol air s −1 ; and overpressure, 0.2 kPa. Chlorophyll fluorescence measurements were conducted at the same time as the gas exchange measurements with a multiphase flash fluorometer. Genty, Briantais and Baker (1989) first reported that the electron transport rate (J) based on the fluorometry (J F ) was linearly related to the quantum yield of photosystem II. On the basis of their theory, J F was estimated as follows: where Q is PPFD and F m ′ and F s are maximum and steady-state chlorophyll fluorescence under illumination, respectively. We assumed that leaf light absorptance (α l = 0.843) and the fraction of electron distributed to photosystem II (β = 0.5) are constant. The multiphase flash (MPF) method was used to estimate F m ′ at an infinitely high saturation light (Loriaux et al., 2013). A saturation light of 8,000 μmol m −2 s −1 with 25% attenuation for 0.3 s was used at each of the three phases. The reasonability of the saturation light was examined in the fluorescence data during the MPF measurements (0.9 s). For measurements with low CO 2 , the fluorescence of Koshihikari seemed to decline continuously during the third phase of MPF, but we considered that the impact on J F was minimal ( Figure S1 in Supplementary Material). The MPF method increased J F by 20% ( Figure S1 in Supplementary Material), and both J and J F were nearly identical when measured in the laboratory under saturated light levels (data not shown). We therefore assumed actual J equals J F based on our protocol at least under high PPFD conditions.

Calculation of V c,max and g m
The variable J method was used for calculating g m from the gas exchange and chlorophyll fluorescence measurements (Harley, Loreto, Di Marco & Sharkey, 1992) using the following equation: where Γ* is CO 2 compensation point at C c and R d is daytime mitochondrial respiration, which was determined to be half of the dark respiration rate for Takanari and Koshihikari measured in both A-CO 2 and E-CO 2 in the study by Noguchi et al. (2018). The calculation of g m based on the variable J method is sensitive to Γ* and R d (Centritto, Lauteri, Monteverdi & Serraj, 2009;Pons et al., 2009). We independently performed the so-called Laisk's method for plant samples within the experimental plot between the two stages (August 4-7) for Koshihikari and Takanari and found that CO 2 compensation point based on C i (C i *) did not statistically differ between varieties (n = 7). Adachi et al. (2013) also reported that Γ* did not differ between rice varieties. Instead of estimating Γ* from measured C i *, we decided to estimate Γ* following Bernacchi, Portis, Nakano, von Caemmerer and Long (2002) because of the following reasons: The temperature function of g m used in this study (Scafaro, von Caemmerer, Evans & Atwell, 2011) is also based on Bernacchi et al. (2002) and the amount of our data was too limited to identify a temperature response of Γ*.
Our pilot measurements in the laboratory for rice plants under saturated light levels showed that the values of g m estimated by our protocol generally agreed with the online isotope method described in Evans and von Caemmerer (2013) using the isotope measurement system employed in Nishida, Kodama, Yonemura and Hanba (2015)  ] = 21%) was tuned at 30 μmol mol −1 , which is relatively lower than the values estimated from the temperature function. Appropriate parameterization for Γ* awaits further investigation.
Maximum carboxylation rate (V c,max ) was calculated by fitting the FvCB model (Farquhar et al., 1980) to A n and C c for [CO 2 ] ranging from 20-200 µmol mol −1 using the lsqcurvefit function of MATLAB (MATLAB R2015b, MathWorks, USA). CO 2 concentration at the carboxylation site (C c ) was estimated as follows: Michaelis constant for CO 2 and O 2 (K c and K o , respectively) required for the calculation of V c,max were based on Bernacchi et al. (2002). The values of V c,max and g m at 25°C (V c,max25 and g m25 ) were estimated based on the temperature function, as described in Bernacchi, Pimentel and Long (2003) and Scafaro et al. (2011), respectively. We used the same unit for both g sc and g m to compare their magnitude, although the physically more appropriate unit of g m is molar flux per pressure because g m in liquid phase is defined as the ratio of a net photosynthesis rate and the gradient of partial pressure of CO 2 (Sharkey, Bernacchi, Farquhar & Singsaas, 2007). The LI-6800 generally outputs stomatal conductance for H 2 O (g sw ), and g sw was assumed to be 1.6 times greater than g sc .

Leaf nitrogen content measurements
Leaf samples used for gas exchange and chlorophyll fluorescence measurements were sampled and dried at 80°C. Specific leaf nitrogen concentration (N l ) was calculated from nitrogen concentrations measured using an NC analyzer (Sumigraph NC-22, SCAS Ltd., Japan) and the weight of the dry samples.

Effects of leaf parameters on RuBP-saturated photosynthesis rate
To quantitatively compare the effects of each leaf parameter (V c,max , g m , and g sc ) on leaf photosynthesis, the rate of RuBP-saturated photosynthesis (A c ) (Farquhar et al., 1980) as a proxy of leaf photosynthesis was calculated by changing one of the parameters under the same [CO 2 ] of 390 μmol mol −1 and in similar environmental conditions as when the measurements were taken (air and leaf temperature: 30°C; relative humidity: 60%; PPFD: 1,500 μmol m −2 s −1 ; and leaf boundary conductance for H 2 O (g bw ): 3 mol m −2 s −1 ). The detailed calculation procedure is explained in Appendix A.

field season data
To supplement our results with information from another field season, A-C i curves obtained at the grain-filling stage in 2016 and that were used in the study by Ikawa et al. (2018) were analyzed to obtain V c,max and g m based on the curve-fitting method (Sharkey et al., 2007). Note that Ikawa et al. (2018) in their study calculated apparent V c,max neglecting g m . The Excel tool provided in the study by Sharkey et al. (2007) was modified so that respiration rate was fixed at the same rate as that of the grain-filling stage data in 2017, and the temperature function for g m of Oryza sativa reported in Scafaro et al. (2011) was introduced. Further details about the measurements can be found in the study by Ikawa et al. (2018).

Statistical analysis
A split-plot approach was used with variety effect nested within [CO 2 ] and replicate blocks (n = 4). An R script used in the study by Mauritz et al. (2017) and Ikawa et al. (2018) was adapted for the analysis. Measurements were taken twice at two of the blocks on different days at the grain-filling stage, and the data were treated as subsets of each block. The overall variation of the simulated A c was smaller than the variation of other parameters, and the variety effect and CO 2 effect were tested separately.

Results
Takanari had a higher A n than Koshihikari under both CO 2 treatments and in both panicle initiating and grain-filling stages (Figure 1(a,b) and Table 2). E-CO 2 increased A n for both varieties at both growth stages, but the increase for Koshihikari was not statistically significant at the grainfilling stage. At the grain-filling stage, the A n of Takanari responded to CO 2 more strongly (30%) than did that of Koshihikari (2%), and the effect of CO 2 treatment and variety difference in A n was significant.
There was not a clear varietal difference in V c,max 25 at the panicle initiation stage, but Takanari had a  −1 (a, b), maximum carboxylation rate at 25°C (V c,max25 ) (c, d), mesophyll conductance at 25°C (g m25 ) (e, f), and stomatal conductance for CO 2 (g sc ) (g, h) measured for Koshihikari (KH) and Takanari (TN) in current [CO 2 ] (A-CO 2 ) and elevated [CO 2 ] (E-CO 2 ) at the panicle initiation stage and grain-filling stage in 2017. Data from 2016 based on the curve-fitting method are included for V c,max25 and g m25 . Measurements in 2017 for A n , g m25 , and g sc were conducted by setting the chamber [CO 2 ] at 390 μmol mol −1 for plants grown in A-CO 2 and 590 μmol mol −1 for those grown in E-CO 2 . Four measurement blocks were treated as replicates (n = 4), and any subset samples were averaged within the block.
greater V c,max25 than did Koshihikari at the grain-filling stage, both in 2016 and 2017 (Figure 1(c,d) and Table 2). The downregulation of V c,max25 under E-CO 2 was also evident at the grain-filling stage, although in 2016 it was not statistically significant. The varietal differences in g sc and g m25 were more apparent than in the case of V c,max25 at both panicle initiation stage and grain-filling stage (Figure 1(e,h) and Table 2). E- Table 2. Statistical results (p values) of the mixed-effects model for variety (Koshihikari and Takanari) and CO 2 effects with respect to net leaf photosynthesis rate (A n ), maximum carboxylation at 25°C (V c,max25 ), mesophyll conductance at 25°C (g m25 ), stomatal conductance for CO 2 (g sc ), and the sensitivity of the three parameters to RuBP-saturated photosynthesis rate (A c ). * p < 0.05, ** p < 0.01, *** p < 0.001. (2017) Figure 2. Simulated RuBP-saturated photosynthesis rate (A c ) of Koshihikari (KH) grown in current CO 2 (A-CO 2 ), and A c with only one of the physiological parameters changed to that of KH grown in elevated CO 2 (E-CO 2 ) and that of Takanari (TN) grown in A-CO 2 and E-CO 2 . Physiological parameters included maximum carboxylation rate at 25°C (V c,max25 ) (a, b), mesophyll conductance at 25°C (g m25 ) (c, d), and stomatal conductance for CO 2 (g sc ) (e, f). The simulation was conducted with air and leaf temperature at 30°C, relative humidity at 60%, PPFD at 1,500 μmol m −2 s −1 , leaf boundary conductance for H 2 O at 3 mol m −2 s −1 , and [CO 2 ] at 390 μmol mol −1.
CO 2 also decreased both g m25 and g sc at the grainfilling stage. The magnitude of g sc was always greater than that of g m , and their ratio was greater in A-CO 2 (~1.5) than it was in E-CO 2 (~1.3) at the panicle initiation stage, and greater in Koshihikari (~1.7) than in Takanari (~1.2) at the grain-filling stage. The varietal difference in g sc had a significant impact on the calculated A c at the panicle initiation stage, and the varietal difference in g m had a significant effect on A c at the grain-filling stage ( Figure 2 and Table 2). Although varietal differences in V c,max25 and g sc were found at the grain-filling stage, they did not lead to a statistically significant difference in A c between the varieties (Table 2).
A clear correlation (R 2 = 0.60) was found between V c,max25 and N l , and the correlation was much clearer when V c,max25 was calculated taking g m into consideration rather than assuming g m to be infinite (i.e. apparent V c,max25 ) (Figure 3(a,b)). Analysis of covariance (ANCOVA) indicated that the relationship between V c,max25 and N l was not different between Koshihikari and Takanari, but that the relationship between apparent V c,max25 and N l significantly differed between varieties (p < 0.0005). Apparent V c,max25 of each variety was also smaller than that estimated from N l based on the data from 2016 and on the relationship reported in Ikawa et al. (2018) (Figure 3(b)). Interestingly, the relationship between V c,max25 and N l was comparable between 2016 and 2017 (Figure 3(a)), despite using different methods and instruments to estimate V c,max25 . For the data in 2016, Ikawa et al. (2018) used a LI-6400 photosynthesis system, but we found no significant difference in the measured apparent V c,max25 between LI-6400 and LI-6800 ( Figure S2 in Supplementary Material).
A correlation was also found between N l and g m25 (R 2 = 0.37), but the relationship within the variety was obscure (R 2 = 0.20 for Koshihikari and R 2 = 0.24 for Takanari). Takanari had greater N l than did Koshihikari at the grain-filling stage. At the panicle initiation stage, E-CO 2 also decreased N l for both varieties, but the decrease was not statistically significant (p = 0.1).

Discussion
Our results partly support the hypothesis that Takanari has greater V c,max , g m , and g sc than does Koshihikari both in current and elevated [CO 2 ]. Both g m25 and g sc were greater in Takanari than in Koshihikari, but V c,max25 did not differ between varieties at the panicle initiation stage. There was no clear varietal difference in V c,max25 at the panicle initiation stage and only a possible difference (p = 0.06) was detected at the grain-filling stage. This result might be related to N l . Takanari had a greater N l than did Koshihikari at the grain-filling stage, but N l did not differ significantly at the panicle initiation stage (data not shown). In the same FACE experiment, Muryono et al. (2017) reported that N l in the top canopy leaves was even lower in Takanari than Figure 3. Maximum carboxylation rate at 25°C (V c,max25 ) in relation to leaf nitrogen content (N l ) (a); V c,max25 with mesophyll conductance (g m ) assumed to be infinite (apparent V c,max25 ) in relation to N l (b); g m at 25°C (g m25 ) in relation to N l (c). ANCOVA (analysis of covariance) indicates that the relationship between V c,max25 (or g m25 ) and N l was different for Koshihikari and Takanari in (b) and (c). No difference was found between CO 2 treatments (i.e. A-CO 2 and E-CO 2 ). it was in Koshihikari at the panicle initiation stage. Generally, Takanari begins to show a high level of N l in the top canopy leaves after heading (e.g. Taylaran et al., 2011), and it is likely that the V c,max of Takanari also follows the seasonal trend of N l . Chen et al. (2014) reported in their study that the smaller mesophyll conductance limitation (MCL) of Takanari resulted in a higher rate of photosynthesis than that of Koshihikari, and that the varietal difference in MCL was greater than that of stomatal conductance limitation (SCL), particularly at later growth stages. Our results at the grain-filling stage agree with the results of the study by Chen et al. (2014) and support our second hypothesis: the difference in g m between [CO 2 ] treatment and variety is greater than that of either V c,max or g sc with respect to their effects on leaf photosynthesis. At the panicle initiation stage, however, our results suggested that the varietal difference in SCL was greater than it was in MCL. The small varietal difference in g m at the panicle initiation stage might be due to the small varietal difference in N l as compared with the grain-filling stage.
The third hypothesis: Takanari's high V c,max and g m are explained by high leaf nitrogen contents, was also partly supported by our results. The greater N l of Takanari correlated with the greater V c,max25 and g m25 as compared with Koshihikari at the grain-filling stage. A high N l is considered to indicate an anatomical advantage for g m by increasing the effective surface of chloroplast (Cai et al., 2018). However, the correlation between N l and g m25 was not as strong as the correlation between N l and V c,max25 , and the relation was not clear within varieties, suggesting that other factors influence g m25 . Responses to environmental changes (Flexas et al., 2007;Mizokami, Noguchi, Kojima, Sakakibara & Terashima, 2015), anatomical structure (Adachi et al., 2013;Terashima, Miyazawa & Hanba, 2001), and aquaporins that possibly control not only H 2 O transport but also the transport of other materials including CO 2 (Ferrio et al., 2012;Miyazawa, Yoshimura, Shinzaki, Maeshima & Miyake, 2008) could all influence g m .
Our results indicate that g m is significant for determining leaf photosynthesis; that its importance changes with the growth stage; and that its variation is not simply explained by N l . Furthermore, apparent V c,max25 given the same N l was estimated to be higher in Takanari than Koshihikari, and was generally higher in 2016 measurements than in 2017 measurements (Figure 3(b)); however, such bias in V c,max25 was no longer observed when g m was considered in the calculation. These results emphasize the importance of separately considering the roles of V c,max and g m for accurately understanding the ecophysiological processes that control leaf photosynthesis.
The magnitudes of V c,max25 and g m25 and their relation to N l were comparable between 2016 and 2017, even though different instruments (LI-6400 in 2016 and LI-6800 in 2017) and different methods (curve-fitting method in 2016 and variable J method in 2017) were used (Figures 1 and 3). When the values of g m reported in the study by Chen et al. (2014) were converted to 25°C according to the temperature function of Scafaro et al. (2011), they ranged roughly from 0.1 to 0.15 mol m −2 s −1 for Koshihikari and 0.15 to 0.2 mol m −2 s −1 for Takanari in the mid grain-filling stage in 2012 and the heading stage in 2013. These values are comparable to those of our present study (Figure 1). Our results suggest that the curve-fitting method would be useful to recompute V c,max and g m based on A-C i curves obtained in the past as long as careful comparisons between the measurement techniques are made.
Elevated CO 2 decreased V c,max , g m , and g sc at the grain-filling stage. Early studies have reported that higher CO 2 decreases g m (e.g. Flexas et al., 2007;Yin et al., 2009), but the validity of the short-term response of g m to CO 2 is debatable, partly because the calculation of g m requires C i as an input Tazoe, von Caemmerer, Badger & Evans, 2009). To examine whether the CO 2 response was the artefact, Mizokami, Noguchi, Kojima, Sakakibara and Terashima (2019) demonstrated that Arabidopsis thaliana with impaired stomatal regulation decreased g m with increasing CO 2 , which supports a hypothesis that the short-term response of g m to CO 2 does occur. It is also known that g m decreases during long-term exposure to high CO 2 (Cai et al., 2018;Mizokami et al., 2019). To investigate whether the decrease of g m in E-CO 2 in the grain-filling stage was due to the direct response to CO 2 or to acclimation effects, we calculated g m25 when [CO 2 ] of the reference cell was set at 150 μmol mol −1 . The paired t-test found no statistically significant difference in g m25 between the growth CO 2 condition and [CO 2 ] at 150 μmol mol −1 . However, g m25 was still lower in the plants grown in E-CO 2 than in those grown in A-CO 2 under the same [CO 2 ] of 150 μmol mol −1 (p < 0.05). We therefore consider that the decrease of g m in E-CO 2 was due to acclimation rather than to a short-term response to CO 2 .
The effects of decreased V c,max , g m , and g sc in E-CO 2 on A c were small and not statistically significant. Cai et al. (2018) also reported that the CO 2 acclimation effect on photosynthesis was small in rice leaves. It should be noted, however, that E-CO 2 increased A n of Takanari much more than that of Koshihikari at the grain-filling stage (Figure 1(b) and Table 2), even though a variety×CO 2 interaction effect was not found in any of the single parameter (Table 2). This implies the combined effects of CO 2 acclimation of physiological parameters were greater in Koshihikari than in Takanari at the grain-filling stage. If CO 2 acclimation did not occur and V c,max , g m , and g sc were assumed to be invariant between CO 2 treatments, then further sensitivity analysis with the FvCB model showed that E-CO 2 benefits A n of Koshihikari slightly more than that of Takanari, which agrees with the theory that high CO 2 benefits individuals with lower conductance (Centritto, Tognetti, Leitgeb, Střelcová & Cohen, 2011).
We consider that the relevance of this study to crop production is limited by the fact that we conducted gas exchange and chlorophyll fluorescence measurements under a similar environment with a high level of PPFD. The performance of the MPF method under different light conditions has yet to be investigated. While rice physiology rapidly changes through the growth (Tatsumi, Kuwabara & Motobayashi, 2019), it was not possible to completely eliminate potential effects of different phenology between Koshihikari and Takanari. Another limitation of our study is that we did not consider the interaction between physiological effects and plants' energy balance (Ikawa et al., 2018;Yoshimoto, Oue & Kobayashi, 2005) that impact the thermal environment and thus photosynthesis and rice production (Usui et al., , 2016.

Conclusions
V c,max , g m , and g sc of the uppermost leaves of the highyielding rice cultivar Takanari were quantified and compared with those in Koshihikari under current and elevated [CO 2 ] before and after heading. The effects of the differences in these physiological parameters between varieties and CO 2 treatments on A c were then quantified, using A c as a proxy of leaf photosynthesis rate. The greater A n of Takanari than of Koshihikari was mainly supported by high g sc at the panicle initiation stage and by high g m at the grain-filling stage. Calculating V c,max25 taking g m into consideration eliminated the artifact of V c,max25 in relation to N l that was observed when g m was assumed to be infinite. Although N l may provide a good estimate of V c,max25 , other factors besides N l play a role in the variations of g m . Our results highlight the importance of considering g m to accurately understand photosynthetic processes and the need to further explore the mechanisms regulating g m . E-CO 2 decreased all three parameters (V c,max , g m , and g sc ) at the grain-filling stage, and the decrease in g m is likely to be due to acclimation to high CO 2 .