Measurements were taken on Zea mays and Miscanthus × giganteus. Leaves were collected from >1 ha plots of a high-yielding modern Z. mays hybrid as described previously (Pignon et al., 2017) on the University of Illinois South Farms (40°02′N, 88°14′W, 216 m above sea level), and leaves of M. × giganteus (“Illinois” clone) from 4 ha plots on University of Illinois Energy Farm (40°07′N, 5 88°21′W, 228 m above sea level) as described previously (Joo et al., 2017). Soils at these sites are deep Drummer/Flanagan series (a fine silty, mixed, mesic Typic Endoaquoll) with high organic matter typical of the central Illinois region of the Corn Belt (Smith et al., 2013). Both plots were rainfed. The M. × giganteus plots were 9 years old, with a stem density of about 100 tillers m^–2; these plots were unfertilized. Z. mays was sown in early May at a density of 75,000 seeds ha^–1. Prior to planting, 140 kg [N] ha^–1 was applied, in line with regional production practice.

Measurements were taken between July 26 and August 06 of 2018. Leaves were cut pre-dawn at the base, then the base was submerged in water and re-cut to prevent air blockage in the xylem as described in Pignon et al. (2017). Removing leaves from plants in this way has been shown not to bias photosynthetic measurements (Leakey et al., 2006). Leaves were then brought back to the laboratory, where they remained in low light until measurement. This procedure avoided any photoinhibition or transient water stress that could develop differentially in shade and sun leaves over a day.

Leaves were sampled from two canopy positions (upper and lower) and two plot positions, centre and the south edge. For each plant sampled, two leaves were collected; an upper canopy leaf, defined as the youngest fully expanded leaf, indicated by a fully emerged ligule, and a lower canopy leaf; the seventh counting down from the first fully emerged leaf. This ensured that within a species and canopy position, leaves from the plot centre and edge were of the same age. The lower canopy leaves in the plot centre were strongly shaded, whereas lower canopy leaves at the plot’s edge were not. The south edge of the plot was chosen since on clear sky days these leaves were exposed to sunlight for 12 h per day.

Portable photosynthetic gas exchange systems (LI 6400 and LI 6400-40 modulated fluorescence chamber head; LI-COR, Inc., Lincoln, NE, United States) were used to measure CO₂ and water vapor exchange on a 2 cm² area of each leaf, along with modulated chlorophyll fluorescence, in the system’s controlled environment leaf cuvette. Air temperature was controlled at a constant 25.0°C, chamber [CO₂] at 400 ppm, and water vapour pressure deficit at 1.6–2.4 kPa.

The measurement sequence began with estimation of maximum dark-adapted quantum yield of PSII photochemistry (F_v/F_m). A photosynthetic light response curve was generated as follows: integrated LEDs emitted uniform light consisting of 10% blue (465 nm wavelength) and 90% red (635 nm wavelength) across the leaf surface. In order to limit photoinhibition caused by sudden exposure to saturating light on enclosure in the cuvette, leaves were first subjected to a photosynthetic photon flux density (PPFD) of 100 μmol m^–2 s^–1 for 5 min, and subsequently exposed to 2000 μmol m^–2 s^–1 for 30–60 min until A reached a steady-state. PPFD was then decreased from 2000 in steps to 1500, 1000, 500, 200, 180, 160, 140, 120, 100, 80, 60, 40, 20, and 0 μmol m^–2 s^–1. Each PPFD step lasted 5–10 min to allow A to reach a steady state before measuring. Steady-state gas-exchange was recorded at each level of PPFD and used to calculate A (von Caemmerer and Farquhar, 1981). Modulated fluorescence measurements were made at each level of PPFD to determine the operating quantum yield of PSII (ϕ_PSII) using a multiphase flash protocol (Loriaux et al., 2013). In turn, ϕ_PSII was used to calculate the rate of linear electron flux through PSII (J), using measured values for leaf fractional absorptance of photosynthetically active photon flux (α, described below) and assuming a photon partitioning factor of 0.4 for PSII vs. PSI, i.e., accounting for increased photon partitioning to PSI to produce ATP through cyclic electron flux (Yin and Struik, 2012; Ver Sagun et al., 2019). Each A-PPFD response curve was fit to a four-parameter non-rectangular hyperbola using PROC NLIN (SAS v9.4, SAS Institute, Cary, NC, United States), which produced an asymptote, taken to represent light-saturated A (A_sat), and a Y-intercept, taken to represent dark respiration (R_d). The third parameter described light-limited A and the fourth parameter described the inflexion between light-limited and light-saturated A with increasing PPFD.

After gas-exchange measurements were completed, absorptance (α) was measured using an integrating sphere and associated spectrometer (Jaz-Spectroclip-TR, Ocean Optics, Largo, FL, United States) and operating software (Spectrasuite, Ocean Optics). α was weighted for 10% blue (465 nm wavelength) and 90% red (635 nm wavelength) incident light to match illumination in the gas-exchange chamber.

The maximum quantum yield of CO₂ assimilation on an incident light basis (ϕCO_{2 max, app}) was calculated from the slope of the linear regression of A against PPFD from 40 to 140 μmol m^–2 s^–1 using PROC GLM (SAS v9.4) (Yin et al., 2014; Pignon et al., 2017). This interval was chosen to account for the Kok effect where respiration increases at very low light levels (PPFD < 40 μmol m^–2 s^–1), and to avoid high light levels (PPFD > 140 μmol m^–2 s^–1) where A is no longer strictly light-limited causing deviations from the linear relationship of A and PPFD. The maximum quantum yield of CO₂ assimilation on an absorbed light basis (ϕCO_{2 max, abs}) was given by ϕCO_{2 max, app} /α. Finally, the maximum quantum yield of CO₂ assimilation on an absorbed light basis and corrected for concurrent changes in ϕPSII (ϕCO_{2 max, abs PSII}) was calculated as in Yin et al. (2014). To test for alternative electron sinks to photosynthetic carbon metabolism, the slope of A vs. J was calculated for PPFD between 40 and 140 μmol m^–2 s^–1 using linear regression (SAS v9.4). The slope of this relationship gives the mol CO₂ assimilated per mol electrons in linear electron flux (1/k) (Baker, 2008). Here k is the mol electrons through linear electron flux required for photosynthesis to fix one mol CO₂. 1/k is an indicator of alternative energy sinks, where any reduction in 1/k is assumed to result from alternative energy sinks, including utilization of ATP and NADPH in processes other than photosynthetic carbon metabolism.

Data were analysed by ANOVA using PROC GLM (SAS v9.4), testing for the fixed effect of species (S effect: Z. mays vs. M. x giganteus), the fixed effect of canopy position (C effect: upper vs. lower canopy), and the fixed effect of plot position (P effect: centre vs. edge), along with all two-way interactions (S x P, S x C, P x C). This model was used to test for significant (p = 0.05 threshold) and marginally significant (p = 0.1 threshold) differences in the following traits: ϕCO_{2 max, app}, ϕCO_{2 max, abs}, ϕCO_{2 max, abs PSII}, 1/k, α, A_sat, R_d, and F_v/F_m. Homogeneity of group variances was tested by Levene’s at p = 0.05 threshold in PROC GLM (SAS v9.4). Normality of Studentized residual distribution was tested by Shapiro–Wilk at p = 0.01 threshold in PROC UNIVARIATE (SAS v9.4). Replication was n = 8–16 in different traits.

In a self-shading crop canopy, the optimal response to shade would be to maintain or increase ϕCO_{2 max, abs} and increase α in order to maximize photosynthesis in light limited conditions. This would increase the linear slope of the response of A to PPFD at low light. This response is observed in shade adapted C₃ plants and in C₃ cereal crops (Givnish, 1988; Beyschlag et al., 1990; Hoyaux et al., 2008). However, the two C₄ crops Z. mays and M. x giganteus studied here, show decreased ϕCO_{2 max, abs} in the lower canopy at the plot centre, but not at the plot edge. This suggests that loss of ϕCO_{2 max, abs} in shade leaves was not due to leaf age, since leaf age was equivalent across plot positions for each species and canopy position. Understanding the basis for this maladaptive response in photosynthetic efficiency is important, as it costs an estimated 10% of potential canopy CO₂ assimilation in the field (Pignon et al., 2017).

If not age, then some environmental factor must trigger the decline in ϕCO_{2 max, abs} of these shaded leaves. The most obvious environmental change between the top and bottom of the canopy is the light environment, with lower leaves receiving less light and an altered spectral distribution, depleted of red and blue and enriched in far-red wavelengths (Sattin et al., 1994). The hypothesis that self-shading is the primary cause for the loss of ϕCO_{2 max, abs} in shade leaves of these C₄ NADP-ME crops is supported by the following observations: (1) when comparing both studied species, the loss of ϕCO_{2 max, abs} in shade leaves at the plot centre was more pronounced in M. x giganteus, which produces a denser canopy with considerably more self-shading than Z. mays. Profiles of canopy light interception in field stands of both species show that the lowest photosynthetically active leaves of Z. mays receive as much as twice the incident PPFD compared to equivalent leaves in M. x giganteus (Pignon et al., 2017). (2) In a previous study comparing two field-grown sugarcane varieties with high and low self-shading, photosynthetic light response curves measured at the top and bottom of the canopy produced contrasting results in the response of A to PPFD at low PPFD (<500 μmol m^–2 s^–1) (Marchiori et al., 2014). In the low self-shading variety, A at low PPFD was slightly greater at the bottom than at the top of the canopy, while the opposite was seen in the high self-shading variety. These studies implemented shade acclimation under realistic field conditions, which produce different results than artificial shading including altered spectral light composition and increased incidence of sun and shade flecks (Pearcy, 1990; Bellasio and Griffiths, 2014; Yabiku and Ueno, 2019).

These findings are important in light of recent efforts to develop crops with a more even vertical light distribution, where either more erect (Perez et al., 2018; San et al., 2018) or more transparent (Slattery et al., 2016; Walker et al., 2018) leaves allow more light to filter to the bottom of the canopy, ultimately increasing canopy photosynthesis (Zhu et al., 2010). The benefits of this type of canopy manipulation could be 2-fold in NADP-ME C₄ crops such as Z. mays, M. x giganteus, sugarcane or sorghum, providing both increased light to drive more photosynthesis and minimizing the loss of photosynthetic efficiency in lower canopy leaves.

Apart from light, temperature is one other important change in microclimate between canopy and plot positions, as shaded leaves can be expected to be cooler. However, temperature is a less likely candidate than light to explain the lost photosynthetic efficiency of shaded leaves seen at the plot centre in the present study. In C₃ plants, ϕCO_{2 max, abs} is highly temperature-sensitive, primarily due to increased photorespiration at high temperatures (Ehleringer and Bjorkman, 1977; Long and Spence, 2013). In contrast, due to the C₄ cycle’s suppression of photorespiration, ϕCO_{2 max, abs} has been found to be constant with temperature from 15 to 40°C in C₄ species such as Atriplex rosea (Ehleringer and Bjorkman, 1977) and Alloteropsis semialata (Osborne et al., 2008). Although loss of ϕCO_{2 max, abs} has been observed in NADP-ME C₄ grasses such as Z. mays due to photodamage during long-term exposure to a combination of high light and cool temperatures (<15°C) (Long and Spence, 2013), this is unlikely to have occurred in the warm summer months during which the present study took place, with maximum daily air temperatures ranging from 19.5 to 33°C at the time measurements were taken. Indeed, since the lower canopy leaves on the exposed southern edge of the stands were exposed to higher light intensities than the shaded lower leaves in the centre of the stands, the expectation would be of a lower ϕCO_{2 max, abs} due to photodamage in the exposed lower canopy leaves, yet the opposite was found.

Under limiting light, reduced α in lower canopy leaves would limit the amount of incident light made available for use within the leaf, and would result in reduced maximum quantum yield on an incident light basis (i.e., ϕCO_{2 max, app}). The fact that α increased in lower canopy leaves shows that in fact their light absorption was improved, not impaired. This pattern in α, along with R_d and A_sat, matches established mechanisms of acclimation to low light, as shade leaves: (1) reduce R_d, (2) remobilize N away from photosynthetic enzymes and toward chlorophyll to improve α under limiting light, and (3) translocate N to the upper canopy so sun leaves can increase photosynthetic enzyme content and improve A_sat (Boardman, 1977; Chen et al., 2014; Niinemets, 2016b; D’Odorico et al., 2018).

Because of the difference in light availability between sun and shade leaves, shade leaves benefit from partitioning relatively more N toward chlorophyll, compared to sun leaves that partition much more N toward photosynthetic enzymes. Therefore, while shade leaves typically reallocate the N stored in their photosynthetic enzymes and decrease total N content, this primarily results in a loss of A_sat, while the apparent maximum quantum yield (ϕCO_{2 max, app}) rises due to increased chlorophyll and, in turn, increased α. The unusual feature in this study is that ϕCO_{2 max, app} falls despite an increase in α – hence our use of the term maladaptive to describe the response of studied shade leaves to low light. Also, as ϕCO_{2 max, abs} is measured on an absorbed light basis and derived from the initial linear slope of the light response curve, it is by definition where A is strictly light-limited, ruling out any limitation by N or protein amounts which primarily affect A_sat (Hikosaka and Terashima, 1995). In fact, the maximum quantum yield of CO₂ assimilation corrected for chlorophyll content was equivalent in N-stressed and control maize plants (Lu and Zhang, 2000).

Efficient energy transfer at PSII is essential to power photosynthesis under limiting light. F_v/F_m is an effective probe to determine whether damage to PSII has occurred (Baker, 2008). However, the <1% loss of F_v/F_m observed here in lower canopy leaves cannot explain the much more substantial losses in ϕCO_{2 max,abs}.

1/k, i.e., the ratio of A to the rate of linear electron transport through PSII (J) at low light, is decreased when the energetic compounds NADPH, reduced ferredoxin, and ATP, produced through linear electron flux, are diverted away from photosynthetic carbon metabolism and into other energy-consuming processes (e.g., nitrogen metabolism, Mehler reaction) (Delatorre et al., 1991; Baker, 2008). This is observed as a reduced slope of the linear relationship of A to J. Under limiting light, this will cause a decline in ϕCO_{2 max, abs}. However, in lower canopy leaves, 1/k was greater than at the top of the canopy, implying leaves at the bottom of the canopy actually had fewer, not more, alternative energy sinks to photosynthetic carbon metabolism. In fact, alternative energy sinks overall were minimal: 1/k was always close to the theoretical maximum of 0.25 mol mol^–1, i.e., for each mol CO₂ assimilated, a theoretical minimum of k = 4 mol electron equivalents must be produced through linear electron flux when there are no alternative energy sinks (Baker, 2008).

One possible explanation for loss of ϕCO_{2 max, abs} without reduced 1/k is that lower canopy leaves in the plot centre did have increased alternative sinks, but these were not detected by the leaf fluorescence measurements. One caveat of PSII fluorescence is that the signal is primarily obtained from PSII closest to the leaf surface, with less contribution from PSII deeper in the leaf. Therefore 1/k is obtained from A throughout the entire leaf cross-section, and J obtained from PSII fluorescence at the leaf surface. If alternative energy sinks diverted NADPH and ATP from deeper PSII, this could result in a decrease of ϕCO_{2 max, abs} without an apparent effect to 1/k. Additionally, 1/k only measures the partitioning toward A of NADPH and ATP produced through linear electron flux. ATP can also be produced through cyclic electron flux around PSI, a process which bypasses PSII and produces only ATP (von Caemmerer, 2000). Alternative energy sinks for the ATP produced through cyclic electron flux would not be reflected in 1/k, since 1/k is based on the photochemical efficiency of PSII and not PSI. For instance, shaded leaves of field-grown M. x giganteus show signs of increased leakage of CO₂ from bundle-sheath cells, which should incur additional ATP consumption to power C₄ overcycling of CO₂ (Kromdijk et al., 2008). However, C₄ NADP-ME grasses including Z. mays showed increased photon partitioning to PSI, but no significant change in cyclic electron flux, when grown in the shade (Ver Sagun et al., 2019).

ϕCO_{2 max, abs} measured in non-stressed conditions is typically well conserved across various species (Long et al., 1993). Surprisingly, here Z. mays showed ϕCO_{2 max, abs} 23% greater than M. x giganteus. This may be explained in part by the greater F_v/F_m and 1/k in Z. mays relative to M. x giganteus. In previous measurements on nearby plots of the same species, ϕCO_{2 max, abs} of M. x giganteus and Z. mays were within just 9% of one another (Pignon et al., 2017), suggesting the greater inter-species difference observed here may be an effect of different location or growing season.

These results raise the question of why such productive crops show a maladaptive acclimation to shade. Zea mays in particular is being grown at ever greater densities (Lobell et al., 2014), resulting in increased leaf area indices and self-shading, but these high densities are a recent construct of cultivation. The ancestors of cultivated Z. mays grew largely as isolated plants in semi-arid and nutrient limited environments, such that they would have evolved as plants in which most or all leaves were exposed to full sunlight and shading was rare. Similarly, Miscanthus spp. often occur as single tall clumps, standing above surrounding plants and so too would experience relatively little shading, compared to field production stands. Having evolved as sun plants, there may have been insufficient time for them to adapt to the recent production in dense stands.

Although both species are part of the same C₄ evolutionary clade, modern Z. mays hybrids have been subject to centuries of selection for productivity, which has been particularly intense in the last 50 years, while M. x giganteus is only just emerging as a crop. This may suggest that there is variability that could be selected to overcome this significant Achilles heel in this important group of crops. Z. mays is considered to have diverged in the evolution of the Andropogoneae before divergence of the genera Saccharum, Sorghum and Miscanthus (Kim et al., 2014; Singh et al., 2019). The occurrence of this maladaptation in both Z. mays and M. x giganteus suggests that the major crops sorghum and sugarcane are likely similarly affected. Given that Z. mays accounts for more cereal grain than any other crop globally, overcoming this maladaptation to shade would contribute very significantly toward meeting the 60% increase in food demand anticipated for mid-century (Long et al., 2015; FAO, 2017).