Photosynthetic Physiology of Blue, Green, and Red Light: Light Intensity Effects and Underlying Mechanisms

Red and blue light are traditionally believed to have a higher quantum yield of CO2 assimilation (QY, moles of CO2 assimilated per mole of photons) than green light, because green light is absorbed less efficiently. However, because of its lower absorptance, green light can penetrate deeper and excite chlorophyll deeper in leaves. We hypothesized that, at high photosynthetic photon flux density (PPFD), green light may achieve higher QY and net CO2 assimilation rate (An) than red or blue light, because of its more uniform absorption throughtout leaves. To test the interactive effects of PPFD and light spectrum on photosynthesis, we measured leaf An of “Green Tower” lettuce (Lactuca sativa) under red, blue, and green light, and combinations of those at PPFDs from 30 to 1,300 μmol⋅m–2⋅s–1. The electron transport rates (J) and the maximum Rubisco carboxylation rate (Vc,max) at low (200 μmol⋅m–2⋅s–1) and high PPFD (1,000 μmol⋅m–2⋅s–1) were estimated from photosynthetic CO2 response curves. Both QYm,inc (maximum QY on incident PPFD basis) and J at low PPFD were higher under red light than under blue and green light. Factoring in light absorption, QYm,abs (the maximum QY on absorbed PPFD basis) under green and red light were both higher than under blue light, indicating that the low QYm,inc under green light was due to lower absorptance, while absorbed blue photons were used inherently least efficiently. At high PPFD, the QYinc [gross CO2 assimilation (Ag)/incident PPFD] and J under red and green light were similar, and higher than under blue light, confirming our hypothesis. Vc,max may not limit photosynthesis at a PPFD of 200 μmol m–2 s–1 and was largely unaffected by light spectrum at 1,000 μmol⋅m–2⋅s–1. Ag and J under different spectra were positively correlated, suggesting that the interactive effect between light spectrum and PPFD on photosynthesis was due to effects on J. No interaction between the three colors of light was detected. In summary, at low PPFD, green light had the lowest photosynthetic efficiency because of its low absorptance. Contrary, at high PPFD, QYinc under green light was among the highest, likely resulting from more uniform distribution of green light in leaves.


INTRODUCTION
The photosynthetic activity of light is wavelength dependent. Based on McCree's work (McCree, 1971(McCree, , 1972, photosynthetically active radiation is typically defined as light with a wavelength range from 400 to 700 nm. Light with a wavelength shorter than 400 nm or longer than 700 nm was considered as unimportant for photosynthesis, due to its low quantum yield of CO 2 assimilation, when applied as a single waveband (Figure 1). Within the 400-700 nm range, McCree (1971) showed that light in the red region (600-700 nm) resulted in the highest quantum yield of CO 2 assimilation of plants. Light in the green region (500-600 nm) generally resulted in a slightly higher quantum yield than light in the blue region (400-500 nm) (Figure 1; McCree, 1971). The low absorptance of green light is partly responsible for its low quantum yield of CO 2 assimilation. Within the visible spectrum, green leaves have the highest absorptance in the blue region, followed by red. Green light is least absorbed by green leaves, which gives leaves their green appearance (McCree, 1971;Zhen et al., 2019).
Since red and blue light are absorbed more strongly by photosynthetic pigments than green light, they are predominantly absorbed by the top few cell layers, while green light can penetrate deeper into leaf tissues (Nishio, 2000;Vogelmann and Evans, 2002;Terashima et al., 2009;Brodersen and Vogelmann, 2010), thus giving it the potential to excite photosystems in deeper cell layers. Leaf photosynthesis may benefit from the more uniform light distribution throughout a leaf under green light. Absorption of photons by chloroplasts near the adaxial surface may induce heat dissipation of excess excitation energy in those chloroplasts, while chloroplasts deeper into the leaf receive little excitation energy (Sun et al., 1998;Nishio, 2000). Blue and red photons, therefore, may be used less efficiently and are more likely to be dissipated as heat than green photons.
FIGURE 1 | The normalized action spectrum of the maximum quantum yield of CO 2 assimilation for narrow wavebands of light from ultra-violet to far-red wavelengths (McCree, 1971). Redrawn using data from Sager et al. (1988).
The misconception that red and blue light are used more efficiently by plants than green light still occasionally appears (Singh et al., 2015), often citing McCree's action spectrum or the poor absorption of green light by chlorophyll extracts. The limitations of McCree's action spectrum were explained in his original paper: the quantum yield was measured under low photosynthetic photon flux density (PPFD), using narrow waveband light, and expressed on an incident light basis (McCree, 1971), but these limitations are sometimes ignored. The importance of green light for photosynthesis has been well established in more recent studies (Sun et al., 1998;Nishio, 2000;Terashima et al., 2009;Hogewoning et al., 2012;Smith et al., 2017).
From those studies, one trend has emerged that has not received much attention: there is an interactive effect of light quality and intensity on photosynthesis (Sun et al., 1998;Evans and Vogelmann, 2003;Terashima et al., 2009). At low PPFD, green light has the lowest QY inc (quantum yield of CO 2 assimilation on incident light basis) because of its low absorptance; at high PPFD, on the other hand, red and blue light have a lower QY inc than green light, because of their high absorptance by photosynthetic pigments, which shifts much of the light absorption closer to the upper leaf surface. This reduces both the quantum yield of CO 2 assimilation in cells in the upper part of a leaf and light availability in the bottom part of a leaf.
The interactive effect between light quality and intensity was illustrated in an elegant study that quantified the differential quantum yield, or the increase in leaf CO 2 assimilation per unit of additional light (Terashima et al., 2009). The differential quantum yield was measured by adding red or green light to a background illumination of white light of different intensities. At low background white light levels, the differential quantum yield of red light was higher than that of green light, due to the low absorptance of green light. But as the background light level increased, the differential quantum yield of green light decreased more slowly than that of red light, and was eventually higher than that of red light (Terashima et al., 2009). The red light was absorbed efficiently by the chloroplasts in the upper part of leaves. With a high background level of white light, those chloroplasts already received a large amount of excitation energy from white light and up-regulated non-photochemical quenching (NPQ) to dissipate excess excitation energy as heat, causing the additional red light to be used inefficiently. Green light, on the other hand, was able to reach the chloroplasts deeper in the mesophyll and excited those chloroplasts that received relatively little excitation energy from white light. Therefore, with high background white light intensity, additional green light increased leaf photosynthesis more efficiently than red light (Terashima et al., 2009).
In this paper, we present a comprehensive study to explore potential interactive effect of light intensity and light quality on C 3 photosynthesis and underlying processes. We quantified the photosynthetic response of plants to blue, green, and red light over a wide PPFD range to better describe how light intensity and waveband interact. In addition, we examined potential interactions among blue, green, and red light, using light with different ratios and intensities of the three narrow waveband lights. To get a better understanding of the biochemical reasons for the effects of light spectrum and intensity on CO 2 assimilation, we constructed assimilationinternal leaf CO 2 (C i ) response curves (A/C i curves) under blue, green, and red light, as well as combinations of the three narrow waveband lights at both high and low PPFD. We hypothesized that effects of different light spectra would be reflected in the electron transport rate (J) required to regenerate consumed ribulose 1,5-bisphosphate (RuBP), rather than the maximum carboxylation rate of ribulose-1,5bisphosphate carboxylase/oxygenase (Rubisco) (V c,max ).

Leaf Absorptance, Transmittance, and Reflectance
Leaf absorptance was determined using a method similar to that of Zhen et al. (2019). Three plants were randomly selected. A newly expanded leaf from each plant was illuminated with a broad-spectrum halogen bulb (70W; Sylvania, Wilmington, MA, United States) for leaf transmittance measurement. Transmittance was measured with a spectroradiometer (SS-110, Apogee, Logan, UT, United States). The halogen light spectrum was taken as reference measurement with the spectroradiometer placed directly under the halogen bulb in a dark room. Then, a lettuce leaf was placed between the halogen bulb and spectroradiometer, with its adaxial side facing the halogen bulb and transmitted light was measured. Leaf transmittance was then calculated on 1 nm resolution. Light reflectance of the leaves was measured using a spectrometer with a leaf clip (UniSpec, PP systems, Amesbury, MA, United States). Light absorptance was calculated as 1 − reflectance − transmittance. We verified that this method results in similar absorptance spectra as the use of an integrating sphere. Absorptance of each of the nine light spectra used in this study were calculated from the overall leaf absorptance spectrum and the spectra of the red, green, and blue LEDs.

Leaf Photosynthesis Measurements
All gas exchange measurements were made with a leaf gas exchange system (CIRAS-3, PP Systems). Light was provided by the LEDs built into the chlorophyll fluorescence module (CFM-3, PP Systems). This module has dimmable LED arrays of different colors, with peaks at 653 nm [red, full width at half maximum (FWHM) of 17 nm], 523 nm (green, FWHM of 36 nm), and 446 nm (blue, FWHM of 16 nm). Nine different combinations of red, green, and blue light were used in this study (Table 1). Throughout the measurements, the environmental conditions inside the cuvette were controlled by the leaf gas exchange system. Leaf temperature was 23.0 ± 0.1 • C, CO 2 concentration was 400.5 ± 4.1 µmol·mol −1 , and the VPD of air in the leaf cuvette was 1.8 ± 0.3 kPa (mean ± SD).

Photosynthesis -Light Response Curves
To explore photosynthetic efficiency of light with different spectra, we constructed light response curves for lettuce plants using each light spectrum. Lettuce plants were exposed to 10 PPFD levels ranging from 30 to 1,300 µmol·m −2 ·s −1 (30,60,90,120,200,350,500,700, 1,000, and 1,300 µmol·m −2 ·s −1 ) in ascending orders for light response curves. Photosynthetic measurements were taken on 40-66 days old lettuce plants.
Lettuce plants were taken out of the growth chamber and dark-adapted for 30 min. Starting from the lowest PPFD, one newly expanded leaf was exposed to all nine spectra. Net CO 2 assimilation rate (A n ) of the leaf was measured using the leaf gas exchange system. Under each light spectrum, three A n readings were recorded at 10 s intervals after readings were stable (about 4-20 min depending on PPFD after changing PPFD and spectrum). The three A n readings were averaged for analysis. After A n measurements under all nine light spectra were taken, the leaf was exposed to the next PPFD level and A n measurements were taken with the light spectra in the same order, until measurements were completed at all PPFD levels. Throughout the light response curves, C i decreased with increasing PPFD, from 396 ± 10 µmol·mol −1 at a PPFD of 30 µmol·m −2 ·s −1 to 242 ± 44 µmol·mol −1 at a PPFD of 1,300 µmol·m −2 ·s −1 . To account for the potential effect of plants and the order of the spectra on assimilation rates, the order of the different spectra was re-randomized for each light response curve, using a Latin square design with plant and spectrum as the blocking factors. Data were collected on nine different plants. Regression curves (exponential rise to maximum) were fitted to the data for each light spectrum and replication (plant): where R d is the dark respiration rate, QY m,inc is the maximum quantum yield of CO 2 assimilation (initial slope of light response curve, mol of CO 2 fixed per mol of incident photons) and A g,max is the light-saturated gross assimilation rate. The A n,max is the light-saturated net assimilation rate and was calculated as A n,max = A g,max − R d . The maximum quantum yield of CO 2 assimilation was also calculated on absorbed light basis as QY m,abs = QY m,inc light absorptance . The instantaneous quantum yield of CO 2 assimilation based on incident PPFD (QY inc ) was calculated as A g PPFD for each PPFD at which A n was measured, where the gross CO 2 assimilation rate (A g ) was calculated as A g = A n + R d . To account for differences in absorptance among the different light spectra, the quantum yield of CO 2 assimilation was also calculated based on absorbed light base, as QY abs = A g PPFD × light absorptance , where light absorptance is the absorptance of lettuce leaves for each specific light spectrum. The differential QY, the increase in assimilation rate per unit of additional incident PPFD, was calculated as the derivative of Eq. 1:

Photosynthesis -Internal CO 2 Response (A/C i ) Curves
To explore the underlying physiological mechanisms of assimilation responses to different light spectra, we constructed A/C i curves. Typically, A/C i curves are collected under saturating PPFD. We collected A/C i curves at two PPFDs (200 and 1,000 µmol·m −2 ·s −1 ) to explore interactive effects of light spectrum and PPFD on the assimilation rate. At a PPFD of 200 µmol·m −2 ·s −1 , red light has the highest A n and green light the lowest A n , while at PPFD of 1,000 µmol·m −2 ·s −1 , red and green light resulted in the highest A n and blue light in the lowest A n .
We used the rapid A/C i response (RACiR) technique that greatly accelerates the process of constructing A/C i curves (Stinziano et al., 2017). We used a Latin square design, similar to the light response curves. A/C i curves were measured under the same nine spectra used for the light response curves. Nine lettuce plants were used as replicates. For each A/C i curve, CO 2 concentration in the leaf cuvette started from 0 µmol·mol −1 , steadily ramping to 1,200 µmol·mol −1 over 6 min. A reference measurement was also taken at the beginning of each replication with an empty cuvette to correct for the reaction time of the leaf gas exchange system. Post-ramp data processing was used to calculate the real A and C i with the spreadsheet provided by PP systems, which yielded the actual A/C i curves with C i range of about 100-950 µmol mol −1 . Throughout the data collection, leaf temperature was 24.4 ± 1.3 • C and VPD in the cuvette was 1.4 ± 0.2 kPa. FIGURE 2 | Light absorptance, reflectance, and transmittance spectrum of a newly fully expanded "Green Towers" lettuce leaf.
Curve fitting for A/C i curves was done by minimizing the residual sum of squares, following the protocol developed by Sharkey et al. (2007). Among our nine replicates, four plants did not show clear Rubisco limitations at low PPFD and for those plants Rubisco limitation (V c,max ) was not included in the model (Sharkey et al., 2007). We therefore report V c,max values for high PPFD only. The J was determined for all light spectra at both PPFDs. We therefore report V c,max was determined for all light spectra only at high PPFD. The quantum yield of electron transport [QY(J)] was calculated on both incident and absorbed PPFD basis as QY(J) inc = J PPFD and QY(J) abs = QY(J) inc light absorptance , respectively. We did not estimate triose phosphate utilization, because the A/C i curves often did not show a clear plateau.

Data Analysis
The QY m,inc , QY m,abs , and A g,max were analyzed with ANOVA to determine the effects of light spectrum using SAS (SAS University Edition; SAS Institute, Cary, NC, United States). A n , QY inc , and QY abs at each PPFD level and V c,max and J estimated from A/C i curves were similarly analyzed with ANOVA using SAS. A n at different PPFD levels were analyzed with regression analysis to detect interactive effect of blue, green, and red light on leaf assimilation rates using the fractions of red, blue, and green light as explanatory variables (JMP Pro 15, SAS Institute).

Leaf Absorptance
A representative spectrum of light absorptance, reflectance and transmittance of a newly fully expanded lettuce leaf is shown in Figure 2. In the blue region, 400-500 nm, the absorptance by "Green Towers" lettuce leaves was high and fairly constant, averaging 91.6%. The leaf absorptance decreased as the wavelength increased from 500 to 551 nm where the absorptance minimum was 69.8%. Absorptance increased again at longer wavelengths, with a second peak at 666 nm (92.6%). Above 675 nm, the absorptance decreased steadily to <5% at 747 nm (Figure 2). The absorptance spectrum of our lettuce leaves is similar to what McCree (1971) obtained for growth chambergrown lettuce, with the exception of slightly higher absorptance in the green part of the spectrum in our lettuce plants. Using this spectrum, the absorptance of the blue, green, and red LED lights were calculated to be 93.2 ± 1.0%, 81.1 ± 1.9% and 91.6 ± 1.1%, respectively. Absorptance of all nine spectra was calculated based on their ratios of red, green, and blue light ( Table 2).

Light Quality and Intensity Effects on Photosynthetic Parameters
Light response curves of lettuce under all nine spectra are shown in Figure 3, with regression coefficients in Supplementary Table 1. It is worth noting that a few plants showed photoinhibition under 100B (decrease in A n with PPFD > 1,000 µmol·m −2 ·s −1 ). Those data were excluded in curve fitting for light response curves to better estimate asymptotes. Photoinhibition was not observed under other spectra.
The QY m,inc of lettuce plants was 22 and 27% higher under red light (74.3 mmol·mol −1 ) than under either 100G (60.8 mmol·mol −1 ) or 100B (58.4 mmol·mol −1 ), respectively ( Figure 4A and Supplementary Table 1). Spectra with a high fraction of red light (64% or more) resulted in a high QY m,inc (Figure 4A), while 80G20R resulted in an intermediate QY m,inc ( Figure 4A). To determine whether differences in QY m,inc were due to differences in absorptance or in the ability of plants to use the absorbed photons for CO 2 assimilation, we also calculated QY m,abs . On an absorbed light basis, 100B light still resulted in the lowest QY m,abs (62.7 mmol·mol −1 ) and red light resulted in the highest QY m,abs (81.1 mmol·mol −1 ) among narrow waveband lights ( Figure 4B). Green light resulted in a QY m,abs (74.9 mmol·mol −1 ) similar to that under red light, but significantly higher than that of blue light ( Figure 4B). We did not find any interactions (synergism or antagonism) between lights of different colors, with all physiological responses under See Figure 2 for the leaf absorptance spectrum. *See spectral composition in Table 1.
mixed spectra being similar to the weighted average of responses under single colors. Thus, for the rest of the results we focus on the three narrow waveband spectra. Among the three narrow waveband lights, 100G resulted in the highest A g,max (20.0 µmol·m −2 ·s −1 ), followed by red (18.9 µmol·m −2 ·s −1 ), and blue light (17.0 µmol·m −2 ·s −1 ) (Figure 5 and Supplementary Table 1). As with QY m,inc and QY m,abs , combining two or three colors of light resulted in an A g,max similar to the weighted averages of individual light colors.
QY inc initially increased with increasing PPFD and peaked at 90-200 µmol·m −2 ·s −1 , then decreased at higher PPFDs ( Figure 6A). The QY inc under 100R was higher than under either green or blue light at low PPFD (≤300 µmol·m −2 ·s −1 ). Although 100G resulted in lower QY inc than 100B at low PPFD (≤300 µmol·m −2 ·s −1 ), the decrease in QY inc under 100G with increasing PPFD was slower than that with 100B or 100R. Above 500 µmol m −2 s −1 , the QY inc with 100G was comparable to the QY inc with 100R, and higher than with 100B ( Figure 6A). The QY abs with 100R was higher than that with either 100G or 100B at PPFDs from 60 to 120 µmol·m −2 ·s −1 (p < 0.05). The QY abs with 100G was similar to 100B at low PPFD, but decreased slower than that with either 100R or 100B as PPFD increased. At PPFD ≥ 500 µmol·m −2 ·s −1 , QY abs was lowest under 100B among the three monochromatic lights (p < 0.05) (Figure 6B).

Interactive Effect of Light Spectrum and PPFD on Photosynthesis
There was an interactive effect of light spectrum and PPFD on photosynthetic properties of lettuce. Under low light conditions (≤200 µmol·m −2 ·s −1 ), the QY inc of lettuce leaves under green light was lowest among blue, green, and red light (Figure 6A), due to its lower absorptance by lettuce leaves. After accounting for absorptance, green photons were used at similar efficiency as blue photons, while red photons were used most efficiently ( Figure 6B). The QY m,abs under green and red light were higher than under blue light (Figure 4B). At high PPFD, green and red light had similar quantum yield, higher than that of blue light, both on an absorbed and incident light basis ( Figure 6A).
Multiple factors contributed to the interactive effect of light spectrum and PPFD on quantum yield and photosynthesis.
Light Absorptance and Non-Photosynthetic Pigments Determine Assimilation at Low PPFD QY m,inc with blue and green light was lower than with red light (Figure 4A), consistent with McCree's action spectrum (McCree, 1971). But when taking leaf absorptance into account, QY m,abs was similar under green and red light and lower under blue light ( Figure 4B). Similarly, at low PPFD (≤200 µmol·m −2 ·s −1 ), QY inc of lettuce leaves was highest under red, intermediate under blue, and lowest under green light. When accounting for leaf absorptance, QY abs under red light remained highest and QY abs under both green and blue light were similar at low PPFD ( Figure 6A). Consistent with our data, previous studies FIGURE 4 | Maximum quantum yield of CO 2 assimilation of "Green Towers" lettuce based on incident (QY m,inc ) (A) and absorbed light (QY m,abs ) (B) under nine different light spectra. Values are calculated as the initial slope of the light response curves of corresponding light spectra (see Figure 3). Bars with the same letter are not statistically different (p ≤ 0.05). Error bars represent the standard deviation (n = 9). The composition of the nine light spectra is shown in Table 1. also documented that, once absorbed, green light can drive photosynthesis efficiently at low PPFD (Balegh and Biddulph, 1970;McCree, 1971;Evans, 1987;Sun et al., 1998;Nishio, 2000;Terashima et al., 2009;Hogewoning et al., 2012;Vogelmann and Gorton, 2014). For example, the QY m,abs of spinach (Spinacia oleracea) and cabbage (Brassica oleracea L.) was highest under red light, followed by that under green light and lowest with blue light. But on incident light basis, QY m,inc of under green light was lower than under red or blue light (Sun et al., 1998).
Both our data ( Figure 4B) and those of Sun et al. (1998) show that QY m,abs with blue light is lower than that with red and green light, indicating that blue light is used intrinsically less efficiently by lettuce. Blue light, and, to a lesser extent, green light is absorbed not just by chlorophyll, but also by flavonoids and carotenoids (Sun et al., 1998). Those pigments can divert energy away from photochemistry and thus reduce the QY abs under blue light. Flavonoids (e.g., anthocyanins) are primarily located in the vacuole and cannot transfer absorbed light energy to photosynthetic pigments (Sun et al., 1998). Likewise, free carotenoids do not contribute to photochemistry (Hogewoning et al., 2012). Carotenoids in light-harvesting antennae and reaction centers channel light energy to photochemistry, but with lower transfer efficiency than chlorophylls (Croce et al., 2001;de Weerd et al., 2003a,b;Wientjes et al., 2011;Hogewoning et al., 2012). Therefore, absorption of blue light by flavonoids and carotenoids reduces the quantum yield of CO 2 assimilation. Thus, even with the high absorptance of blue light by green leaves, QY m,abs of leaves under blue light was the lowest among FIGURE 5 | Maximum gross assimilation rate (A g,max ) of "Green Towers" lettuce under different light spectra, calculated from the light response curves. Bars with the same letter are not statistically different (p ≤ 0.05). Error bars represent standard deviation (n = 9). The composition of the nine light spectra is shown in Table 1. the three monochromatic lights (Figure 4B). It is likely that the lower QY abs under green light than that under red light was also due to absorption of green light by carotenoids and flavonoids (Hogewoning et al., 2012). At high PPFD, absorption of blue light by flavonoids and carotenoids still occurs, but this is less of a limiting factor for photosynthesis, since light availability is not limiting under high PPFD.

Light Dependence of Respiration and Rubisco Activity May Reduce the Quantum Yield at Low PPFD
At PPFDs below 200 µmol·m −2 ·s −1 , the QY inc and QY abs of lettuce showed an unexpected pattern in response to PPFD (Figure 6). Unlike the quantum yield of PSII, which decreases exponentially with increasing PPFD (Weaver and van Iersel, 2019), QY inc and QY abs increased initially with increasing PPFD (Figure 6). A similar pattern was previously observed by Craver et al. (2020) in petunia (Petunia × hybrida) seedlings. This pattern could result from light-dependent regulation of respiration (Croce et al., 2001), alternative electron sinks such as nitrate reduction (Skillman, 2008;Nunes-Nesi et al., 2010), or Rubisco activity (Campbell and Ogren, 1992;Zhang and Portis, 1999). In our calculations, we assumed that the leaf respiration in the light was the same as R d . However, leaf respiration in the light is lower than in the dark, in a PPFD-dependent manner FIGURE 7 | The differential quantum yield of CO 2 assimilation (differential QY ) of "Green Towers" lettuce under blue, green, and red LED light as a function of the PPFD. The differential QY is the increase in net assimilation per unit additional PPFD and was calculated as the first derivate of the light response curves (Figure 3). The insert shows the differential quantum yield plotted at PPFDs of 1,000-1,300 µmol m −2 s −1 s to better show differences at high PPFD (note the different y-axis scale). (Brooks and Farquhar, 1985;Atkin et al., 1997), which can lead to overestimation of A g with increasing PPFD. When we accounted for this down-regulation of respiration, using the model by Müller et al. (2005) to correct A g , QY inc , and QY abs , we found that depression of respiration by light did not explain the initial increase in QY inc and QY abs we observed (Supplementary  Figure 4). Alternative electron sinks in the chloroplasts that are upregulated in response to light can explain the low QY inc , and QY abs at low PPFD, because they compete with the Calvin cycle for reducing power (ferredoxin/NADPH). Such processes include photorespiration (Krall and Edwards, 1992), nitrate assimilation (Nunes-Nesi et al., 2010), sulfate assimilation (Takahashi et al., 2011) and the Mehler reaction (Badger et al., 2000) and their effect on QY inc , and QY abs would be especially notable under low PPFD (Supplementary Figure 5).
Upregulation of Rubisco activity by Rubisco activase in the light may also have contributed to the increase in QY inc and QY abs at low PPFD (Campbell and Ogren, 1992;Zhang and Portis, 1999). In the dark, 2-carboxy-D-arabinitol-1-phosphate (CA1P) or RuBP binds strongly to the active sites of Rubisco, preventing carboxylation activity. In the light, Rubisco activase releases the inhibitory CA1P or RuBP from the catalytic site of Rubisco, in a light-dependent manner (Campbell and Ogren, 1992;Zhang and Portis, 1999;Parry et al., 2008). At PPFD < 120 µmol·m −2 ·s −1 , low Rubisco activity may have limited photosynthesis.

Light Distribution Within Leaves Affects QY at High PPFD
Except for the initial increase at low PPFD, both QY inc and QY abs decreased with increasing PPFD. QY inc decreased slower under green than under red or blue light (Figure 6A). At a PPFD ≥ 500 µmol·m −2 ·s −1 , QY inc under green light was higher than that under blue light ( Figure 6A). Accordingly, A n under blue light was lower than under green and red light at PPFDs above 500 µmol·m −2 ·s −1 (Figure 3A). The lower QY inc under blue light than under green and red light at high PPFD can be explained by disparities in the light distribution within leaves.
Blue and red light were strongly absorbed by lettuce leaves (93.2 and 91.6%, respectively), while green light was absorbed less (81.1%) ( Table 2). Similar low green absorptance was found in sunflower (Helianthus annuus L.), snapdragon (Antirrhínum majus L.) (Brodersen and Vogelmann, 2010), and spinach (Vogelmann and Han, 2000). In leaves of those species, absorption of red and blue light peaked in the upper 20% of leaves, and declined sharply further into the leaf. Absorption of red light decreased slower with increasing depth than that of blue light (Vogelmann and Han, 2000;Brodersen and Vogelmann, 2010). Green light absorption peaked deeper into leaves, and was more evenly distributed throughout leaves, because of low absorption of green light by chlorophyll (Vogelmann and Han, 2000;Brodersen and Vogelmann, 2010). The more even distribution of green light within leaves, as compared to red and blue light, can explain the interactive effects between PPFD and light spectrum on leaf photosynthesis. It was estimated that less than 10% of blue light traveled through the palisade mesophyll and reached the spongy mesophyll in spinach, while about 35% of green light and 25% of red light did so (Vogelmann and Evans, 2002). It was also estimated that chlorophyll in the lowermost chloroplasts of spinach leaves absorbed about 10% of green and <2% of blue light, compared to chlorophyll in the uppermost chloroplasts (Vogelmann and Evans, 2002;Terashima et al., 2009).
The more uniform green light distribution within leaves may be a key contributor to higher leaf level QY inc under high PPFD because less heat dissipation of excess light energy is needed (Nishio, 2000;Terashima et al., 2009). Reaction centers near the adaxial leaf surface receive more excitation energy under blue, and to a lesser extent under red light, than under green light, because of the differences in absorptance. Consequently, under high intensity blue light, NPQ is up-regulated in the chloroplasts near the adaxial leaf surface to dissipate some of the excitation energy (Sun et al., 1998;Nishio, 2000), lowering the QY inc under blue light. Since less green light is absorbed near the adaxial surface, less heat dissipation is required. When incident light increased from 150 to 600 µmol·m −2 ·s −1 , the fraction of whole leaf CO 2 assimilation that occurred in the top half of spinach leaves remained the same under green light (58%), but decreased from 87 to 73% under blue light. This indicates more upregulation of heat dissipation in the top of the leaves under blue, than under green light (Evans and Vogelmann, 2003). On the other hand, the bottom half of the leaves can still utilize the available light with relatively high QY inc , since the amount of light reaching the bottom half is relatively low, even under high PPFD (Nishio, 2000). By channeling more light to the under-utilized bottom part of leaves, leaves could achieve higher QY inc even under high intensity green light. In our study, high QY inc under green light and low QY inc under blue light at high FIGURE 8 | Electron transport rate (J) at PPFDs of 200 (left bars) and 1,000 µmol m −2 s −1 (right bars) (A) and maximum Rubisco carboxylation rate (V c,max ) at a PPFD of 1,000 µmol m −2 s −1 (B) of "Green Towers" lettuce, as estimated from A/C i curves under different light spectra. Bars with the same letter are not statistically different (p ≤ 0.05). Error bars represent the standard deviation (n = 9). The light composition of the nine light spectra is shown in Table 1. PPFD (Figure 6) can be thus explained by the large disparities in the light environment in chloroplasts from the adaxial to the abaxial side of leaves due to differences in leaf absorptance. Similarly, differential QY of lettuce leaves was highest under green light and lower under blue and red light at high PPFD (>300 µmol·m −2 ·s −1 ) (Figure 7), also potentially because of the more uniform distribution of green light and the uneven distribution of blue and red light in leaves.
Along the same line, A n of lettuce leaves was the lowest under blue light at PPFD > 500 µmol·m −2 ·s −1 (Figure 3). Also, A n of lettuce leaves approached light saturation at lower PPFDs under blue and red light, than under green light ( Figure 3A). Under blue, green, and red light, lettuce leaves reached 95% of A n,max at PPFDs of 954, 1,110 and 856 µmol·m −2 ·s −1 , respectively. This can be seen more clearly in the differential QY at high PPFD (Figure 7). At a PPFD of 1,300 µmol·m −2 ·s −1 , green light had a differential QY of 1.09 mmol·mol −1 , while that of red and blue light was only 0.46 and 0.69 mmol·mol −1 , respectively (Figure 7). Green light also resulted in a higher A g,max (22.9 µmol·m −2 ·s −1 ) than red and blue light (21.8 and 19.3 µmol·m −2 ·s −1 , respectively) ( Figure 5). As discussed before, the high A g,max under green light resulted from the more uniform light distribution under green light, allowing deeper cell layers to photosynthesize more. Previous research similarly found that at high PPFD (>500 µmol·m −2 ·s −1 ), A n of both spinach and cabbage were lower under blue light than under white, red and green light (Sun et al., 1998). Overall, under high PPFD, the differences in light distribution throughout a leaf are important to quantum yield and assimilation rate, since it affects NPQ up-regulation (Sun et al., 1998;Nishio, 2000). However, light distribution within a leaf is less important at low than at high PPFD, FIGURE 9 | The correlation between gross CO 2 assimilation rate (A g ) estimated from light response curves and electron transport rate (J) estimated from A/C i curves (A), and between the quantum yield of CO 2 assimilation (QY abs ) and the quantum yield of electron transport on an absorbed light basis [QY(J) abs ] (B), under low PPFD (200 µmol m −2 s −1 ) and high PPFD (1,000 µmol m −2 s −1 ) under nine light spectra averaged over nine "Green Towers" lettuce plants. The color scheme representing the nine spectra is the same as Figure 8.

Light Spectrum Affects J, but Not V c,max
We examined the effect of light quality and intensity on J and V c,max (Figure 8). For the light-dependent reactions, the interactive effect between light spectra and PPFD found for CO 2 assimilation and quantum yield was also observed for J (Figure 8A). At low PPFD (200 µmol·m −2 ·s −1 ), green light resulted in the lowest J and red light in the highest J among single waveband spectra. But at a PPFD of 1,000 µmol·m −2 ·s −1 , red and green light resulted in the highest J and blue light in the lowest J (Figure 8A), similar to the differences in A g .
There was no clear evidence of Rubisco limitations to photosynthesis at a PPFD of 200 µmol·m −2 ·s −1 , so the rate of the light-dependent reactions likely limited photosynthesis. This is corroborated by the strong correlation between A g and J at a PPFD of 200 µmol·m −2 ·s −1 . Although Rubisco limitations to photosynthesis were observed at a PPFD of 1,000 µmol·m −2 ·s −1 , there were no meaningful differences in V c,max in response to light spectrum, in contrast to J (Figure 8).
When PPFD increased 5×, from 200 to 1,000 µmol·m −2 ·s −1 , there was only a 1.7 to 2.4× increase in J, indicating a lower QY(J) inc at higher PPFD. This matches the lower QY inc and the asymptotic increase in A n in response to increasing PPFD (Figure 3). The relative increase of J under green light (143%) was greater than that under both blue and red light (73 and 75%, respectively) as PPFD increased. This similarly can be attributed to a more uniform energy distribution of green light among reaction centers throughout a leaf and weaker upregulation of non-photochemical quenching with increasing green light intensity (Sun et al., 1998;Nishio, 2000;Evans and Vogelmann, 2003), as discussed before.
There was a strong correlation between J and A g under the nine light spectra at both PPFD levels ( Figure 9A). QY abs and QY(J) abs are similarly strongly correlated ( Figure 9B). Unlike J, V c,max was largely unaffected by light spectra ( Figure 8B) and was not correlated with A g (data not shown). There was, however, a strong correlation between J and V c,max at a PPFD of 1,000 µmol·m −2 ·s −1 (R 2 = 0.82, Supplementary Figure 3), suggesting that J and V c,max are co-regulated. Similarly, Wullschleger (1993) noted a strong linear relationship between J and V c,max across 109 C 3 species. The ratio between J and V c,max in our study (1.5-2.0) similar to the ratio found by Wullschleger (1993). These results suggest that the interactive effect of light spectra and PPFD resulted from effects on J, which is associated with light energy harvesting by reaction centers, rather than from V c,max .

No Interactive Effects Among Blue, Green, and Red Light
The Emerson enhancement effect describes a synergistic effect between lights of different wavebands (red and far-red) on photosynthesis (Emerson, 1957). McCree (1971) attempted to account for interactions between light with different spectra when developing photosynthetic action spectra and applied low intensity monochromatic lights from 350 to 725 nm with white background light to plants. His results showed no interactive effect between those monochromatic lights and white light (McCree, 1971). We tested different ratios of blue, green, and red light and different PPFDs, and similarly did not find any synergistic or antagonistic effect of different wavebands on any physiological parameters measured or calculated.

Importance of Interactions Between PPFD and Light Quality and Its Applications
The interactive effect between PPFD and light quality demonstrates a remarkable adaptation of plants to different light intensities. By not absorbing green light strongly, plants open up a "green window, " as Terashima et al. (2009) called it, to excite chloroplasts deeper into leaves, and thus facilitating CO 2 assimilation throughout the leaf. While red light resulted in relatively high QY inc , QY abs and A n at both high and low PPFD (Figures 3, 6), it is still mainly absorbed in the upper part of leaves (Sun et al., 1998;Brodersen and Vogelmann, 2010). Green light can penetrate deeper into leaves (Brodersen and Vogelmann, 2010) and help plants drive efficient CO 2 assimilation at high PPFD (Figures 3, 5).
Many early photosynthesis studies investigated the absorptance and action spectrum of photosynthesis of green algae, e.g., Haxo and Blinks (1950) or chlorophyll or chloroplasts extracts, e.g., Chen (1952). Extrapolating light absorptance of green algae and suspension of chlorophyll or chloroplast to whole leaves from can lead to an underestimation of absorptance of green light by whole leaves and the belief that green light has little photosynthetic activity (Moss and Loomis, 1952;Smith et al., 2017). Photosynthetic action spectra developed on whole leaves of higher plants, however, have long shown that green light effectively contributes to CO 2 assimilation, although with lower QY inc than red light (Hoover, 1937;McCree, 1971;Inada, 1976;Evans, 1987). The importance of green light for photosynthesis was clearly established in more recent studies, emphasizing its role in more uniformly exciting all chloroplasts, which especially important under high PPFD (Sun et al., 1998;Nishio, 2000;Terashima et al., 2009;Hogewoning et al., 2012;Smith et al., 2017). The idea that red and blue light are more efficient at driving photosynthesis, unfortunately, still lingers, e.g., Singh et al. (2015).
Light-emitting diodes (LEDs) have received wide attention in recent years for use in controlled environment agriculture, as they now have superior efficacy over traditional lighting technologies (Pattison et al., 2018). LEDs can have a narrow spectrum and great controllability. This provides unprecedented opportunities to fine tune light spectra and PPFD to manipulate crop growth and development. Blue and red LEDs have higher efficacy than white and green LEDs (Kusuma et al., 2020). By coincidence, McCree's action spectrum (Figure 1; McCree, 1971) also has peaks in the red and blue region, although the peak in the blue region is substantially lower than the one in the red region. Therefore, red and blue LEDs are sometimes considered optimal for driving photosynthesis. This claim holds true only under low PPFD. Green light plays an important role in photosynthesis, as it helps plants to adapt to different light intensities. The wavelength-dependent absorptance of chlorophylls channels green light deeper into leaves, resulting in more uniform light absorption throughout leaves and providing excitation energy to cells further from the adaxial surface. Under high PPFD, this can increase leaf photosynthesis. Plant evolved under sunlight for hundreds of millions of years, and it seems likely that the relatively low absorptance of green light contributes to the overall photosynthetic efficiency of plants (Nishio, 2000).

CONCLUSION
There was an interactive effect of light spectrum and PPFD on leaf photosynthesis. Under low PPFD, QY inc was lowest under green and highest under red light. The low QY inc under green light at low PPFD was due to low absorptance. In contrast, at high PPFD, green and red light achieved similar QY inc , higher than that of blue light. The strong absorption of blue light by chlorophyll creates a large light gradient from the top to the bottom of leaves. The large amount of excitation energy near the adaxial side of a leaf results in upregulation of nonphotochemical quenching, while chloroplasts near the bottom of a leaf receive little excitation energy under blue light. The more uniform distribution of green light absorption within leaves reduces the need for nonphotochemical quenching near the top of the leaf, while providing more excitation energy to cells near the bottom of the leaf. We also found that the interactive effect of light spectrum and PPFD on photosynthesis was a result of the light-dependent reactions; gross assimilation and J were strongly correlated. We detected no synergistic or antagonistic interactions between blue, green, and red light.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
JL and MI designed the experiment, discussed the data, and revised the manuscript. JL performed the experiment, analyzed data, and prepared the first draft. Both authors contributed to the article and approved the submitted version.  Figure 6) Quantum yield of CO 2 assimilation of "Green Towers" lettuce as a function of incident (QY inc ) (A,C,E,G) and absorbed PPFD (QY abs ) (B,D,F,H) under nine light spectra (see Table 1). Error bars represent standard deviation (n = 9). Figure 7) Differential quantum yield of CO 2 assimilation (differential QY ) of "Green Towers" lettuce under nine light spectra as a function of the PPFD. Inserts show differential QY at PPFDs of 1,000-1,300 µmol·m −2 s −1 s to better show differences at high PPFD (note the different y-axis scale). The composition of the nine light spectra is shown in Table 1. The light spectra in the graphs are (A) 100B, 100G and 100R; (B) 100B, 80B20G, 20B80G and 100G; (C) 100G, 80G20R, 20G80R and 100R; and (D) 20B80R, 16B20G64R and 100G. Figure 6) The correlation between electron transport (J) and maximum Rubisco carboxylation rate (V c,max ) of "Green Towers" lettuce estimated from A/C i curves under PPFD (1000 µmol m −2 s −1 ) under nine light spectra (p < 0.001).