The photosynthetic response of C3 and C4 bioenergy grass species to fluctuating light

Bioenergy grass species are a renewable energy source, but their productivity has not been fully realized. Improving photosynthetic efficiency has been proposed as a mechanism to increase the productivity of bioenergy grass species. Fluctuating light, experienced by all field grown crops, is known to reduce photosynthetic efficiency. This experiment aimed to evaluate the photosynthetic performance of both C3 and C4 bioenergy grass species under steady state and fluctuating light conditions by examining leaf gas exchange. The fluctuating light regime used here decreased carbon assimilation across all species when compared to expected steady state values. Overall, C4 species assimilated more carbon than C3 species during the fluctuating light regime, with both photosynthetic types assimilating about 16% less carbon than expected based on steady state measurements. Little diversity was observed in response to fluctuating light among C3 species, and photorespiration partially contributed to the rapid decreases in net photosynthetic rates during high to low light transitions. In C4 species, differences among the four NADP‐ME species were apparent. Diversity observed among C4 species in this experiment provides evidence that photosynthetic efficiency in response to fluctuating light may be targeted to increase C4 bioenergy grass productivity.

within a canopy, wind, passing clouds, and changes in sun angle (Slattery et al., 2018;Vialet-Chabrand et al., 2017). If sufficient variation in photosynthetic efficiency during fluctuating light exists across different bioenergy crops, then these traits can be targeted by future research to improve yields. Because of their high productivity, many grass species are currently being investigated for their utility as bioenergy crops (Jablonowski & Schrey, 2021); however, little is known about how photosynthetic efficiency during fluctuating light varies across these species.
Photosynthesis, the process of using light energy to assimilate CO 2 , is sensitive to changes in incident light. Changes in light intensity can be rapid, but photosynthetic rates adjust slower, which ultimately affects crop productivity (Slattery et al., 2018). When plants are transitioned from low to high or high to low light intensities, the initial changes in the rate of CO 2 assimilation (A net ) could be related to the processes of electron transport, buildup of metabolite pools, enzyme activities, photoprotection, or stomatal conductance (Kirschbaum & Pearcy, 1988;Sassentath-Cole & Pearcy, 1992, 1994Way & Pearcy, 2012;Yamori et al., 2012). These limitations likely vary among species and even among cultivars (Acevedo-Siaca et al., 2020;Pignon et al., 2021).
Bioenergy grasses also include both C 3 and C 4 photosynthetic types, adding another source of variation. Many of the most productive species like Miscanthus × giganteus and switchgrass use C 4 photosynthesis, while other species like giant reed use C 3 photosynthesis. C 4 species mainly differ from C 3 species by operating a CO 2 concentrating mechanism (CCM) achieved by the C 4 cycle. The CCM increases the CO 2 concentration around the enzyme Rubisco. Rubisco serves as the entry point of carbon into C 3 cycle by catalyzing the reaction of CO 2 with RuBP (Ribulose-1,5-bisphosphate). Both C 3 and C 4 species use the C 3 cycle to produce the chemical energy for cellular respiration as part of the process of photosynthesis. While the addition of the C 4 cycle comes with additional costs in the forms of ATP and reducing equivalents (e.g., NADPH) it has benefits. C 4 species typically display greater water and nitrogen use efficiency than C 3 species (Ghannoum et al., 2010). But how do they compare during fluctuating light? Slattery et al. (2018) reviewed the impacts of fluctuating light on crop performance and apart from highlighting the previously listed possible limitations, presented contrasting hypotheses for how C 3 and C 4 species may compare under fluctuating light: C 4 species could be more negatively impacted by fluctuating light because increased complexity of the C 4 system results in incoordination between metabolic pathways leading to futile cycling of metabolites, or C 4 species are less negatively impacted by fluctuating light because the added complexity of the C 4 system increases flexibility in the production and consumption of ATP, NADPH, and other redox equivalents (Stitt & Zhu, 2014). Given that Slattery et al. (2018) provided contrasting ideas on the subject, it can be inferred that there is currently no consensus on how the photosynthetic efficiency of C 3 and C 4 species compare during fluctuating light. Adding yet another source of variation among bioenergy grass species is the C 4 subtype (NADP-ME, NAD-ME, and PEPCK), which has also been shown to affect photosynthetic responses to fluctuating light (Laisk & Edwards, 1997).
Here, we examine the changes in photosynthetic leaf gas exchange parameters over time as the leaf transitions between high and low light intensities. The objectives of this study were to (1) quantify photosynthetic efficiency in major bioenergy grasses under both steady state and fluctuating light and (2) contrast C 3 and C 4 performance under fluctuating light. Six C 3 species and six C 4 species were included. Of the six C 4 species, four were NADP-ME, one was NAD-ME, and one was PEPCK. The experiments presented here will help guide future research on increasing bioenergy grass productivity through altering photosynthetic efficiency.

| Plant materials and growth conditions
Seven bioenergy grass species were selected: Miscanthus × giganteus (hereafter M. × giganteus), sugarcane, switchgrass, big bluestem, prairie cordgrass, giant reed, and reed canarygrass. Of these species, only giant reed and reed canarygrass are C 3 . The C 3 species, tall wheatgrass and tall fescue, were also included as they are considered potential bioenergy crops. Wheat, tobacco, and maize were included given that they are commonly measured for leaf gas exchange. Altogether, 12 species were analyzed, including 6 C 3 and 6 C 4 species. The common name, scientific name, abbreviation, and photosynthetic type of these 12 species are shown in Table 1. M. × giganteus were collected from the University of Illinois Energy Farm (Lee et al., 2019).
Two separate growth conditions were used in this study, greenhouse and field. For greenhouse experiments, seeds of each species were individually planted at a depth of 3 mm in a propagation tray liner (Nursery Supplies Inc) using Berger BM7 (Berger) as the growing medium, with the exception of M. × giganteus, which was propagated by rhizomes, and sugarcane and giant reed which were propagated by nodes. After 4 weeks, seedlings were transplanted into pots (30.16 cm diameter × 27.94 cm deep, Nursery Supplies Inc.). All plants were fertilized with granulated fertilizer (Osmocote Plus 13/13/13, The Scotts Company LLC), water-soluble nutrient solution (Peter's Excel 15-5-15, Everris NA Inc), and iron chelate supplement (Sprint 330, BASF Corp.) once every 4 weeks. The greenhouse temperature was kept around 27°C (day) and 16°C (night). A 14-h day length was maintained with high-pressure sodium lamps providing an additional 400 μmol m −2 s −1 photosynthetic photon flux density (PPFD) at canopy level above ambient when necessary. Pots were arranged in a randomized complete block design with four replications. Plants were measured for 2 days at 12 weeks after planting and again for 4 days at 14 weeks after planting.
For the field grown plants, only six species were planted at the University of Illinois Energy Farm: M. × giganteus, switchgrass, big bluestem, prairie cordgrass, tall fescue, and reed canarygrass ( Table 1). The field experimental design was a randomized complete block design with four replications, and blocks were separated by alleys (1.5 m). Four plants of each species were planted in the individual plots (0.9 × 2.7 m). Before field planting in this experiment, plant seeds or rhizomes were transplanted into pots (12.06 cm diameter × 11.74 cm deep, Nursery Supplies Inc.) containing Berger BM7 (Berger) as the growing medium. Plants were grown in a greenhouse for 8 weeks and transplanted by hand in May 2020. Plants were measured for gas exchange 4 weeks after transplanting in field.

| Steady state gas exchange measurements
Light response and CO 2 response curves were measured on the youngest fully expanded leaves using a portable infrared gas exchange system (LI-6800, LI-COR Inc.). Leaves were placed in the leaf chamber at 1500 µmol m −2 s −1 PPFD (LI-6800-01A, LI-COR Inc) at 90% red (635 nm wavelength) and 10% blue (465 nm wavelength). Block temperature was 30°C, flow rate was 500 µmol s −1 , and relative humidity was 60%. Photosynthetic CO 2 response (A/C i ) was measured by varying the CO 2 reference concentration in the following sequence: 400,300,200,150,100,75,0,400,400,400,600,800,1000,1200,1400, and 400 µmol mol −1 . Light response (A/Q abs ) was measured on the same leaf following 15-30 min to allow photosynthesis to reach steady state after increasing the light intensity to 2000 μmol m −2 s −1 . For A/Q abs curves, the CO 2 sample was maintained at 400 µmol mol −1 (~40 Pa) and light intensity was varied as follows : 2000, 1600, 1200, 900, 750, 600, 500, 400, 300, 200, 120, 60, and 20 µmol m −2 s −1 . An additional A/C i curve was performed at a light intensity of 100 μmol m −2 s −1 on the same leaf allowing 15-30 min for the leaf to reach steady state. Chamber settings and CO 2 concentrations matched the initial A/C i curve measured at PPFD of 1500 μmol m −2 s −1 . The reference and sample infrared gas analyzers (IRGAs) were matched at every measurement point. For field experiments, measurements were identical except no A/C i curves were measured at 100 µmol m −2 s −1 PPFD.

| Fluctuating light gas exchange measurements
Photosynthetic responses to fluctuating light were measured on the same leaf following steady state measurements. The leaf chamber was set to 1500 µmol m −2 s −1 PPFD. The block T A B L E 1 Species examined in this study, photosynthetic pathways, and the experiments species were used in are shown below. The experiments presented are fluctuating light gas exchange in greenhouse grown plants (GH) and field grown plants (Field), the flow rate test of the gas exchange system (FT), and measurements at 2% oxygen (O 2 ) temperature was 30°C, flow rate was 500 µmol s −1 , the reference CO 2 was set to 400 μmol mol −1 , and H 2 O reference was fixed to give an approximate sample RH of 60%. The leaf was allowed 15-30 min at these conditions until photosynthetic rates reached a steady state. A fluctuating light program was written as follows: 1500 µmol m −2 s −1 PPFD for 4 min, 100 µmol m −2 s −1 PPFD for 2 min, repeated three additional times, and ending with 1500 µmol m −2 s −1 for 4 min. Gas exchange data were recorded every 5 s during the 28-min program using default averaging time of 4 s. The IRGAs were matched prior to starting the fluctuating light measurements and were not matched during program to avoid interferences with the 5 s data sampling interval.

| Fluctuating light flow test
The gas exchange system used cannot provide instantaneous measurements of leaf gas exchange. From the manufacturer's application note (https://www.licor.com/env/ suppo rt/LI-6800/topic s/chamb er-custo m-note.html): where C t is the chamber concentration at time t, C e is the concentration entering the chamber, C o is the initial chamber concentration, f is the flow rate, and V is the chamber volume. The chamber volume of LI-6800-01A is 87.3 cm 3 (personal communication with manufacturer). We derived the time required to reach 95% (t 95 ) of the new concentration as: such that flow and volume determine the time required to reach the new concentration. As volume of the chamber is constant, four flow rates were tested: 500, 700, 900, and 1100 μmol s −1 . The calculated time required to reach 95% of the new chamber concentrations was 21, 15, 11, and 9 s, respectively. These calculated equilibration times were longer than our 5 s logging interval and 4 s averaging time. While these calculations were for instantaneous changes in concentrations, we do not expect leaf fluxes of CO 2 and H 2 O to be instantaneous, but we do want the equilibration time to be faster than the changes in leaf flux. Therefore, we tested the effect of flow rate on fluctuating light measurements. The same starting conditions as listed above were used. After a leaf achieved steady state, it was exposed to 1500 µmol m −2 s −1 PPFD for 4 min, 100 µmol m −2 s −1 PPFD for 2 min, then returned to 1500 µmol m −2 s −1 PPFD for 5 min. The leaf was allowed 15-30 min between each flow rate to return to steady state before starting a new flow rate.
Measurements were made 2 weeks after the initial gas exchange measurements at 14 weeks after planting.

| 2% oxygen test
Atmospheric oxygen concentrations are known to affect the net CO 2 assimilation rates (A net ) of leaves. All the above measurements were conducted at 21% O 2 . To test the effect of oxygen on A net response to fluctuating light, plants were measured at 2% O 2 . Two C 3 species, giant reed and reed canarygrass, and two C 4 species, M. × giganteus and switchgrass, were measured. The above methodology was used for both steady state and fluctuating light measurements except that the air being provided to the leaf came from a 2% O 2 gas cylinder balanced in N 2 (Airgas USA) connected to the LI-6800 following manufacturer's specifications. Measurements were made 2 weeks after the initial gas exchange measurements at 14 weeks after planting.

| Leaf spectral qualities
Following gas exchange measurements, on the same leaves, leaf absorbance was measured using an integrating sphere (Spectroclip-JAZ-TR, Ocean Optics). Leaf absorbance (L A ) was calculated following: where L I is the incident radiation, L T is the transmitted radiation, and L R is the reflected radiation (400-700 nm).
The L A was used to calculate the amount of incident light that was absorbed for the A/Q abs curves. A SPAD 502 Plus Chlorophyll Meter was also used to characterize the greenness of leaves (Konica Minolta).

| A/C i curve analysis
A/C i curves were modeled using the following equation for a non-rectangular hyperbola: from Bellasio et al. (2016). The observed values of A net and the intercellular CO 2 partial pressure (C i ) were calculated by the gas exchange system. The carboxylation efficiency (CE) is the initial slope of the A/C i response. The CO 2 compensation point (Γ) is the C i value where A net is equal to zero. The term A max is the CO 2 saturated rate of A net . The curvature (1) factor (ω) is a unitless value ranging between 0 and 1. Model fits were performed in Excel (Microsoft) using the solver add-in to minimize the sum of the differences squared between the observed and modeled values of A net at a given C i , by changing the parameters CE, Γ, A sat , and ω. The repeated points at a reference CO 2 of 400 μmol mol −1 were excluded from model fits, only the first measurement was used. The same model was fit to the A/C i data collected at 100 µmol m −2 s −1 PPFD.

| A/Q abs curve analysis
A/Q abs curves were modeled using the following equation for a non-rectangular hyperbola: from Bellasio et al. (2016) but modified to include respiration (R) and absorbed (Q abs ) rather than incident PPFD (Q in ). Absorbed PPFD was calculated as: The parameters A net and Q in were output by the gas exchange system. The conversion efficiency of converting PPFD into assimilated CO 2 (Φ CO 2 ) is the initial slope of the A/Q abs response. The respiration rate (R) is the y-intercept of the function when PPFD is equal to zero. The term A sat is the PPFD saturated rate of A net . The curvature factor (θ) is a unitless value ranging between 0 and 1. Model fits were performed in Excel (Microsoft) using the solver add-in to minimize the sum of the differences squared between the observed and modeled values of A net at a given PPFD, by changing the parameters Φ CO 2 , R, A sat , and θ.

| Fluctuating light analysis
The observed A net value was reported as A obs . The expected A net value, that is, if the leaf could instantaneously reach steady state (A exp ), was determined using Equations (5 and 6) with Q in for each 5 s data interval. The expected A net value, if stomatal and boundary layer conductance were infinite (i.e., C i = C a , where C a is the atmospheric CO 2 partial pressure measured by the gas exchange system) and the leaf could reach steady state instantaneously (A * C a ), was calculated using Equation (4) for the appropriate light level (i.e., A/C i parameters for 1500 or 100 µmol m −2 s −1 ) and C a at each 5 s data interval. The expected A net value, based on observed C i , if the leaf could reach steady state instantaneously (A * C i ), was calculated using Equation (4) for the appropriate light level and C i at each 5 s data interval. To estimate the carbon lost due to fluctuating photosynthetic rates, A obs was subtracted from A exp at each time point. To estimate the amount of carbon lost due to stomatal limitation and fluctuating photosynthetic rates, A * C i was subtracted from A * C a similar to Kaiser et al. (2017). To estimate the amount of carbon lost due to non-stomatal limitation and fluctuating photosynthetic rates, A obs was subtracted from A * C i similar to Kaiser et al. (2017).
For the high to low light transitions (2 min), low to high light transitions (4 min), and both periods together (6 min), the amount of carbon assimilated (C obs , , respectively) during the period multiplied by the sampling interval (i.e., 5 s) resulting in units of mmol m −2 . The four repeated events were treated as technical replicates. For the flow test and 2% O 2 test, only the first 40 s were calculated for C obs . Values were normalized by dividing the observed value of A net at any given time by the average A net value for the 30 s prior to the first light change (A initial ).

| Statistical analysis
Experimental design was a randomized complete block design with four replications. Normal distribution and equality of the variances were evaluated using the UNIVARIATE procedure in SAS (SAS institute). If data were not normally distributed, log transformation was performed. Data that met assumptions were analyzed in a mixed-model analysis of variance using PROC MIXED and GLIMMIX procedures in SAS. All statistical significances were determined using Tukey's range test at α = 0.05. Datasets of 2% oxygen test were analyzed by a pairwise comparison using SAS at α = 0.05 (SAS institute).

| Photosynthetic performance during steady state
For greenhouse grown plants, net CO 2 assimilation response to intercellular CO 2 partial pressure (A/C i ) was conducted at high light (1500 µmol m −2 s −1 PPFD) and low light on all 12 species at 21% O 2 (100 µmol m −2 s −1 PPFD; Figure 1; Figure S1). For modeled A/C i parameters at high and low light, C 4 species had higher CE, lower CO 2 compensation point (Γ), and lower CO 2 saturated net CO 2 assimilation rates (A max ) compared to C 3 species as expected (Figure 1; Table 2). The fitting of the nonrectangular hyperbola model often resulted in a value of 0 Pa for Γ in C 4 species; therefore, the current methodology may not be capable of discerning differences in C 4 compensation points among species. Field plants had similar A/C i responses as greenhouse plants ( Figure S1). At 2% O 2 , C 4 species showed no notable changes in their A/C i response, the C 3 species had lower Γ and higher CE compared to measurements at 21% O 2 as expected ( Figure S1).
For greenhouse grown plants, net CO 2 assimilation response to absorbed light (A/Q abs ) was measured on all 12 species at an atmospheric CO 2 partial pressure of 40 Pa and 21% O 2 (Figure 1; Figure S2; Table 2). On average, C 4 species showed higher light saturated rates of net CO 2 assimilation (A sat ) and respiration rates (R) compared to C 3 species (Table 2). Leaf spectral characteristics were similar among all species, with only sugarcane and tall fescue having significantly higher light absorbance than tobacco (Table S1). Field measurements of A/Q abs were similar to greenhouse measurements ( Figure S2). At 2% O 2 , C 3 species had higher Φ CO 2 and A sat compared to measurements at 21% O 2 as expected ( Figure S2).

| Photosynthetic performance during fluctuating light
Photosynthetic response to fluctuating light varied among the 12 species of greenhouse grown plants measured at 21% O 2 (Figure 2a; Table 3). It was expected based on A/Q abs curves that C 4 species would have similar carbon assimilation at low light and higher carbon assimilation at high light compared to C 3 species ( Figure  2b; Table 3). During high to low light transitions, carbon assimilation was higher than expected due to slow decreases in photosynthetic rates. On average, A net of C 4 species decreased slower during high to low light transitions compared to A net of C 3 species (Figures 2a and 3; Figure S3). As a result, observed carbon assimilation (C obs ) during high to low light transitions was higher on average in C 4 compared to C 3 species (Table  3). C 4 species assimilated 118% more carbon than expected, compared to only a 34% increase in C 3 species (Table 3, C exp − C obs ). Neither C 3 nor C 4 species experienced an overall stomatal (C *  Table 3).
During low to high light transitions, all species assimilated less carbon than expected (Figure 2; Figure  S3; Table 3). On average, A net of C 4 species increased similarly during low to high light transitions compared to A net of C 3 species (Figures 2a and 3; Figure S3). The observed carbon assimilated during low to high light transitions was greater for C 4 compared to C 3 species because C 4 species had higher A net at high light compared to C 3 species as was predicted from A/Q abs curves (Figure 2a,b). Both C 3 and C 4 species assimilated about 20% less carbon than expected (Table 3). During low to high light transitions C 4 species experienced less stomatal limitation (C *  Table 3).
Overall, including both high to low and low to high light transitions, both C 3 and C 4 species assimilated less carbon than expected, losing more carbon from low to high light transitions than was gained from high to low light transitions (Table 3, C exp − C obs ). C 3 species had a highly uniform response to the fluctuating light regime (Figure 3m), while C 4 species were highly variable (Figure 3n,o). The NADP-ME subtypes had similar shapes F I G U R E 1 Steady state response of net CO 2 assimilation (A net ). (a) A net response to intercellular CO 2 partial pressure (C i ) at high (1500 μmol m −2 s −1 ) and low (100 μmol m −2 s −1 ) light. (b) A net response to absorbed photosynthetic photon flux density (Q abs ) at atmospheric CO 2 concentration of ~40 Pa. Six C 4 (blue lines) and six C 3 (red lines) species are shown. Lines are the mean of four replicates (n) except for wheat where n = 3, species as indicated in legend but varied in magnitude (Figure 3n). The PEPCK and NAD-ME species showed the most distinctive responses; however, only a single species was measured for each, so it remains to be seen the extent of variability in these subtypes (Figure 3o).
Possible limitations in our methodology could be due to holding H 2 O constant. This was done to avoid artifacts that could obscure the true response of the leaf. However, as a result, vapor pressure deficit (VPD) varied during the fluctuating light regime. In this T A B L E 2 Parameter means for steady state model fits are shown with ± SE. For A/C i curves, CO 2 saturated rate of A net (A max ), the carboxylation efficiency (CE), CO 2 compensation point (Γ), and the curvature factor (ω) were statistically compared within group (C 3 or C 4 ), high and low light were separated. For A/Q abs response, the light saturated rate of A net (A sat ), the light conversion efficiency for CO 2 assimilation (Φ CO 2 ), respiration rate (R), and the curvature factor (θ) were statistically compared within group (C 3 or C 4 ). Lower case letters indicate significant differences with group at α = 0.05. Group without letters were not significantly different, except for C 4 θ, which failed to meet normality assumptions of the statistical test. Species values are the mean of four replicates except for TA, where n = 3 experiment, VPD experienced by C 3 and C 4 species was not drastically different with mean values of 1.60 and 1.74 kPa, respectively. However, VPD may be a confounding factor that needs further experimentation to disentangle from carbon assimilation responses to fluctuating light.

| Flow test for measuring photosynthetic response to fluctuating light
Historically, measurements of A net during non-steady state were conducted using in-house built gas exchange systems with high response times (Laisk & Edwards, 1997;Ruuska et al., 1998). To test the utility of the LI-6800 system for the measurements presented here, four flow rates (500, 700, 900, and 1100 µmol s −1 ) were tested to determine whether response times of the system were fast enough to capture the rapid changes in leaf gas exchange. Higher flow rates did reveal faster changes in A net during fluctuating light (Figure 3). During high to low light transitions, dips in A net became more pronounced for prairie cordgrass, switchgrass, and all C 3 species as flow rate increased (Figure 3e-l). Comparisons among species remained consistent regardless of flow rate (i.e., the general shape of A net response was captured at the lowest flow rate tested; Figure S4).

| Field test for measuring photosynthetic response to fluctuating light
In greenhouses, environmental conditions are controlled. To test that observed A net responses to fluctuating light were not artifacts of greenhouse growth conditions, we also measured our fluctuating light regime on field grown plants. Overall, photosynthetic response from field plants was similar to that of greenhouse grown plants ( Figure S3). The delayed or biphasic increase in A net of NADP-ME species observed during low to high light transitions in greenhouse plants was also apparent in field grown big bluestem and M. × giganteus ( Figure S3).  (Figure 4). A net decreased faster for the two C 3 species at 21% O 2 compared to 2% O 2 with minimal or no change observed in the C 4 species. The carbon assimilated during the first 40 s after the high to low light transition was significantly lower at 21% than at 2% O 2 in the C 3 species (Figure 4e). Our objective was to quantify natural variation in photosynthetic efficiency of bioenergy grass species. As fluctuating light is a known limitation of photosynthetic efficiency (Slattery et al., 2018), we measured both steady and non-steady state conditions. Steady state measurements separated C 3 and C 4 species as expected; however, T A B L E 3 The carbon assimilated during high to low light, low to high light, or both transitions together (total) are shown for observed (C obs ) and derived values. C exp is the expected carbon assimilation calculated from A/Q abs curves, C * C a is the expected carbon assimilation calculated from A/C i curves assuming infinite stomatal conductance and steady state, and C * C i is the expected carbon assimilation calculated from A/C i curves using the observed C i value during fluctuating light. The term C exp − C obs indicates the loss of carbon due to non-steady state,

|
indicates the loss of carbon due to stomatal limitation and non-steady state, C * C i − C obs indicates the loss of carbon due to non-stomatal limitations and non-steady state. Lower case letters indicate significant differences between all species at α = 0.05, capital letters indicate differences between the groups (C 3 , C 4 ). Species values are the mean of four replicates except for TA, where n = 3. Standard error is shown giant reed outperformed other C 3 species having a comparable photosynthetic rate to C 4 species at ambient CO 2 partial pressures and high light. Under fluctuating light, all species assimilated less carbon than predicted from steady state measurements, supporting the role of fluctuating light in limiting photosynthetic efficiency. C 3 species showed little diversity in response of A net to light changes ( Figure 3m). The C 4 response of A net to fluctuating light varied but known characteristics of photosynthetic subtypes (i.e., NADP-ME, NAD-ME, PEPCK) were observed (Figure 3n,o; Brown & Gracen, 1972;Downton, 1970;Laisk & Edwards, 1997). Among the four NADP-ME species, the response of A net to fluctuating light was diverse (Figure 3n), suggesting that there is potential for trait improvement to increase photosynthetic efficiency of NADP-ME bioenergy grasses.

| Steady state characteristics of C 3 and C 4 photosynthesis
It is recognized that C 4 species have higher photosynthetic efficiency than C 3 species under photorespiratory conditions such as higher O 2 levels, lower CO 2 levels, higher temperatures, and lower water availability because of a CCM (Pearcy & Ehleringer, 1984). Here, we also observed that C 4 species in our study displayed higher A sat and CE than C 3 species at 21% O 2 (Table 2). Of the C 3 species we observed, giant reed had the highest CE, highest A sat , and lowest Γ (Table 2). At PPFD of 1500 μmol m −2 s −1 , which was used in our fluctuating light regime, giant reed was predicted to have a higher photosynthetic rate than three of the C 4 species in this experiment, based on steady state measurements (Figure 1b). Webster et al. (2016) previously reported the higher photosynthetic capabilities of giant reed. Rossa et al. (1998) found that high photosynthetic rates of giant reed may originate from a lack of light saturation of photosynthesis, which we also observed. In our A/Q abs plot, giant reed showed the least amount of light saturation among C 3 species, followed by tall wheatgrass, with the remaining four C 3 species clustering together ( Figure 1b).

| The effect of fluctuating light on carbon assimilation
Fluctuating light is a certainty for field-grown plants, varying in intensity and duration for many reasons including sun angle, wind, shading within canopies, and cloud movement (Knapp & Smith, 1989;Tanaka et al., 2019). During fluctuating light regime employed here, both C 3 and C 4 species showed excess carbon gain during high to low light transitions, carbon loss during low to high light transitions, for a net loss of carbon when compared to expected values derived from steady state measurements (Table 3). This was observed in greenhouse experiments at 21% O 2 , 2% O 2 , and in field conditions at 21% O 2 . The carbon gain during high to low light transitions did not offset the carbon loss during low to high light transitions. The amount of carbon gain during high to low light compared to carbon loss during low to high light likely depends on duration and intensity of light transitions, which was not tested in our study.

| The effect of fluctuating light on C 3 and C 4 carbon assimilation
The impact of fluctuating light differed between C 3 and C 4 species. For example, in the initial 40 s during high to low light transitions comparing observed carbon assimilation to the expected value for greenhouse grown plants at

F I G U R E 3
The effect of flow rate on the response of net CO 2 assimilation (A net ) to fluctuating light. (a-l) The change in A net following the transition from a photosynthetic photon flux density of 1500 to 100 μmol m −2 s −1 at time 0 s and return to high light at 120 s. Line color indicates the flow rate, either 500, 700, 900, or 1100 μmol s −1 . (m) The A net /A initial for the six C 3 species and the average. (n) The A net /A initial for the four NADP-ME subtypes and the C 3 average. (o) The A net /A initial for the NAD-ME and PEPCK subtypes and the C 3 average. Line color indicates species, and only the highest flow rate (1100 μmol s −1 ) is shown. The C 4 species are shown in blue and C 3 species are shown in red. Each line is the average of four replicates (n) except for wheat where n = 3 21% O 2 , C 4 species assimilated more excess carbon during high to low light (0.46 > 0.21 mmol m −2 ), lost more than expected during low to high light (0.67 > 0.46 mmol m −2 ), but lost less overall than expected (0.20 < 0.25 mmol m −2 ) compared to C 3 species (Figure 2, calculations not shown). These calculations, however, were compared to an expected level and did not reflect the actual amount of carbon assimilated. Observation from our highest flow rate (1100 μmol mol −1 ) shows C 4 species assimilated more carbon than C 3 species during the first 40 s of high to low light transitions (0.62 > 0.33 mmol m −2 ) and low to high light transitions (0.86 > 0.71 mmol m −2 ), giving them an overall higher carbon assimilation (1.48 > 1.04 mmol m −2 ). These comparisons depend on timescale. For example, we observed that C 3 species maintained higher A net than C 4 species during the first 15 s following a low to high light transition. This short timescale comparison is consistent with findings from Krall and Pearcy (1993), who demonstrated that maize had lower photosynthetic efficiency at light events lasting <10 s when compared to soybean (Pons & Pearcy, 1992). In general, we found C 4 species decrease carbon assimilation rates slower than C 3 species during high to low light transitions and increase carbon assimilation rates similarly to C 3 species during low to high light transitions. Stitt and Zhu (2014) proposed that large metabolite pools needed to drive diffusion gradients between mesophyll and bundle sheath cells of C 4 species can store or release reducing equivalents and ATP with a larger capacity and longer timescale than what is possible in C 3 species. Modeling presented by Slattery et al. (2018) suggested that the metabolite buffering capacity of C 4 photosynthesis could be capable of sustaining rates of CO 2 assimilation for up to 15 s following a high to low light transition. Indeed, there are many examples of C 4 species maintaining A net after light changes (Krall & Pearcy, 1993;Laisk & Edwards, 1997;Qiao et al., 2020).
We thought it was likely that the faster reduction of C 3 photosynthetic rates during high to low light transitions compared to C 4 species could be affected by photorespiration. Because C 3 species are subjected to atmospheric concentrations of CO 2 , high rates of RuBP oxygenation occur compared to C 4 species. The resulting products of RuBP oxygenation get partially decarboxylated affecting the net CO 2 assimilation rate. The CO 2 release from photorespiration is not instantaneous, possibly affecting the C 3 responses to fluctuating light. Bulley and Tregunna (1971) found that photorespiratory CO 2 release lasts longer than photosynthesis after a sudden decrease in light intensity. Our measurements at 2% O 2 , which should limit photorespiration, resulted in a slower decline in A net during high to low light transitions and more carbon being assimilated. Suggesting a major limitation to carbon assimilation in C 3 species, following a reduction in light intensity, is photorespiratory CO 2 release. We have labeled this event in Figure 3g-l. The amount of photorespiration is in part mediated by stomatal conductance which facilitates CO 2 movement into the leaf (Lawson et al., 2012). In general, stomatal responses to fluctuating light are slower than observed photosynthetic responses (Lawson et al., 2012;McAusland et al., 2016;Tinoco-Ojanguren & Pearcy, 1993). We observed a higher amount of stomatal limitation (C * C a − C * C i ) for C 3 species during both light transitions compared to C 4 species. This is not surprising as C 3 species remain CO 2 limited until C i partial pressures rise above ~60 Pa, whereas C 4 species are not limited at C i values above ~10 Pa, as shown by our A/C i curves.
Our results of higher CO 2 assimilation in C 4 species appear contrary to a report of two C 4 species performing worse than two C 3 species during high to low light transitions  (Kubásek et al., 2013). Kubásek et al. (2013) suggested that C 4 species did worse during fluctuating light compared to C 3 species due to mechanisms involving induction of photosynthesis. In their study, plants were started at 50 μmol m −2 s −1 PPFD, whereas in our fluctuating light regime plants started acclimated to 1500 μmol m −2 s −1 PPFD. These differences highlight the innumerable ways that light can fluctuate in nature, and that our findings may not be applicable to all fluctuating light comparisons of C 3 and C 4 species.

| The effect of fluctuating light on carbon assimilation in C 4 subtypes
Much previous work on the effect of fluctuating light on C 4 species has been done on light to dark transitions (post-illumination), but we observe many similarities to our results presented here. Post-illumination CO 2 burst in C 4 species are characteristic of NAD-ME and PEPCK type plants (Brown & Gracen, 1972;Downton, 1970). The burst is independent of O 2 (Downton, 1970); therefore, it is not a product of photorespiration as is the case for post-illumination CO 2 bursts in C 3 species (Wynn et al., 1973). The hypothesis is that the CO 2 burst results from CO 2 leakage from bundle sheath cells, originating from decarboxylation after the C 3 cycle has stopped and RuBP has been consumed (Downton, 1970). This process is often called over cycling or over pumping, where more CO 2 is released into the bundle sheath than can be used by Rubisco (Furbank et al., 1990;Jenkins et al., 1989;Slattery et al., 2018;von Caemmerer, 2000). We have noted this over pumping event for prairie cordgrass and switchgrass on Figure 3e,f. This explanation also depends on RuBP pool size. If the RuBP pool size is large enough to consume post-illumination CO 2 released from the C 4 cycle, then it will prevent loss of CO 2 from the bundle sheath (Laisk & Edwards, 1997). Because our analysis only included a single NAD-ME and PEPCK species, we do not know how variable this over pumping event might be.
In NADP-ME subtypes, the post-illumination burst is known to be absent (Downton, 1970;Wynn et al., 1973). We also observed gradual decreases in A net , lacking observable CO 2 bursts, during high to low light transitions for the NADP-ME species observed here: maize, big bluestem, M. × giganteus, and sugarcane. This is likely because decarboxylation of malate immediately stops in the dark (Laisk & Edwards, 1997). As ATP production stops, phosphoglycerate kinase in the C 3 cycle no longer produces substrate needed to produce NADP + for malate decarboxylation by NADP-ME, which is located in the bundle sheath chloroplasts (Laisk & Edwards, 1997). During high to low light transitions, this suggests tight coupling of C 4 and C 3 cycles in NADP-ME subtypes, but not NAD-ME or PEPCK subtypes, where the decarboxylase is located outside of bundle sheath chloroplasts (Laisk & Edwards, 1997).
We observed variability among the four NADP-ME species during high to low light transitions. The long persisting CO 2 uptake at levels well above expected during high to low light transitions could be due to conversion of 3-PGA to PEP via phosphoglycerate mutase and enolase. In NADP-ME subtypes, PSII activity is reduced in bundle sheath cells, 3-PGA is shuttled to mesophyll cells where it is converted to triose phosphate in reactions that consume ATP and NADPH, triose phosphate is then transported back to the bundle sheath chloroplast providing the ATP and NADPH equivalents to the Calvin cycle (Arrivault et al., 2017;Stitt & Heldt, 1985). If this 3-PGA shuttle can be utilized to produce PEP, then CO 2 assimilation can continue without ATP needed to convert pyruvate to PEP (Laisk & Edwards, 1997). The long duration of the higher than expected A net values during high to low light transitions may reflect the time it takes to shuttle metabolites from the bundle sheath chloroplast to PEPC in the mesophyll cytoplasm. This process could explain the different amounts of C obs we observed between NADP-ME species. Maize, which had the highest C obs during high to low light transitions, may have larger 3-PGA pool sizes than M. × giganteus, sugarcane, and big bluestem. Because differences were observed within subtype, it suggests that traits are available for selection and improvement related to photosynthetic efficiency during high to low light transitions.
During dark to light transitions, previous work has reported a CO 2 gulp (rapid increase in A net ) in NAD-ME and PEPCK subtypes, resulting from the rapid phosphorylation and conversion of alanine to pyruvate to PEP (Laisk & Edwards, 1997). We did not observe an obvious low to high light transition CO 2 gulp in either prairie cordgrass or switchgrass, but further measurements with additional species of NAD-ME and PEPCK are needed. During NADP-ME transitions from dark to light, a CO 2 burst has been observed (Krall & Pearcy, 1993;Laisk & Edwards, 1997). This is thought to be a result of rapid malate decarboxylation linked to the reduction of large PGA pools. During this initial period, RuBP levels are low and the CO 2 released from malate cannot be fixed and leaks out of the bundle sheath (Laisk & Edwards, 1997). This is another example of over pumping. We observed minimal dips in A net for three of the four NADP-ME species measured here within the first 10 s of high light and labeled them in Figure 3b-d.
Given the small size of the CO 2 burst, the conditions used during our low to high light transition may have facilitated close coupling of RuBP pools with malate and 3-PGA pools, preventing large losses of CO 2 observed in previous studies (Laisk & Edwards, 1997). On the other hand, maize and big bluestem showed a biphasic increase in A net during low to high light transitions, lasting for about 2 min, that was minimal or not consistently observed in M. × giganteus and sugarcane (Figures 2 and 3). This biphasic increase in A net during low to high light transitions accounts for the biggest limitation to photosynthetic efficiency during fluctuating light in maize and big bluestem. Qiao et al. (2020) suggested this biphasic increase in A net was a result of Rubisco deactivation in maize. However, if that were true, we may expect larger CO 2 bursts (i.e., more over pumping) than what we observed in the first 10 s of the low to high light transition. This biphasic response was also observed by Laisk and Edwards (1997), but only at low CO 2 concentrations with no hypothesis put forward. We suggest that it could be due to a reestablishment of large metabolite pools needed for forming a concentration gradient between mesophyll and bundle sheath cells. This could also be due to stomatal closure overshoot depressing photosynthesis; however, non-stomatal limitation was higher than stomatal limitation during this time period suggesting biochemical limitations. It should be noted that our estimations for stomatal and non-stomatal limitations are based on steady state and may not reflect what occurs during fluctuating light. Because the biphasic transition from low to high light was not apparent in all four of the NADP-ME species we observed, it could be targeted by future research to improve photosynthetic efficiency of the low to high light transition of NADP-ME bioenergy grass species.

| CONCLUSION
Understanding natural variation in photosynthetic traits between and among species and cultivars is critical for understanding the regulation of photosynthesis in plants and for providing the necessary knowledge for breeding programs (Flood et al., 2011;Langridge & Fleury, 2011;Tanaka et al., 2019). The diversity we observed in C 4 species response to fluctuating light was remarkable compared to the uniformity of the C 3 response. The different responses of C 4 species during light transitions observed here were related to biochemical subtype of the species and appear to be analogous to previous descriptions of post-illumination measurements in C 4 species. Overall, C 4 species assimilated more carbon than C 3 species for the fluctuating light regime used here, but mismatch between C 3 and C 4 cycles was evident and variable between species providing targets for future research to increase photosynthetic efficiency during fluctuating light in C 4 bioenergy grasses.