Fluctuating Light Takes Crop Photosynthesis on a Rollercoaster Ride^,

Abstract

Crops are regularly exposed to frequent irradiance fluctuations, which decrease their integrated CO₂ assimilation and affect their phenotype.

The environment of the natural world in which plants live and have evolved and within which photosynthesis operates is one characterized by change. The time scales over which change occurs can range from seconds (or less) all the way to the geological scale. All of these changes are relevant for understanding plants and the vegetation they create. In this update review, we will focus on how photosynthesis responds to fluctuations in irradiance with time constants up to the range of tens of minutes. Photosynthesis is a highly regulated process, in which photochemistry as well as the electron and proton transport processes leading to the formation of ATP and reducing power (reduced ferredoxin and NADPH) need to be coordinated with the activity of metabolic processes (Foyer and Harbinson, 1994). Light, temperature, the supply of the predominant substrate for photosynthetic metabolism (CO₂), and the demand for the products of photosynthetic metabolism are all factors that are involved in short-term alterations of steady-state photosynthetic activity. The coordinated regulation of metabolism with the formation of the metabolic driving forces of ATP and reducing power is subject to various constraints that limit the freedom of response of the system. Of these constraints, the most prominent are the need to limit the rate of formation of active oxygen species by limiting the lifetime of excited states of chlorophyll a and the potential of the driving forces for electron transport (Foyer and Harbinson, 1994; Foyer et al., 2012; Rutherford et al., 2012; Murchie and Harbinson, 2014; Liu and Last, 2017), limiting the decrease of lumen pH to avoid damaging the oxygen evolving complex of PSII (Krieger and Weis, 1993), and adjusting stomatal conductance (g_s) to optimize photosynthetic water-use efficiency (Lawson and Blatt, 2014).

The processes that regulate electron and proton transport, enzyme activation, and CO₂ diffusion into the chloroplast under steady-state conditions also react in a dynamic and highly concerted manner to changes in irradiance, balancing between light use and photoprotection. This overview of the physiological control underlying dynamic photosynthesis is specific to the C₃ photosynthetic pathway. Much less is known about the dynamic regulation of the C₄ and Crassulacean acid metabolism (CAM) pathways, though given their C₃ heritage, we expect that they share much of the regulation of C₃ photosynthesis. We note here that in comparison to C₃ plants, some C₄ species, including maize (Zea mays), show a very slow photosynthetic induction after an irradiance increase (Furbank and Walker, 1985; Chen et al., 2013) and that this phenomenon deserves further attention.

If we grant that the regulation of photosynthesis at steady-state is in some way optimal and represents an ideal balance between light-use efficiency and photoprotection and an ideal balance between CO₂ diffusion into the leaf with the loss of water vapor from the leaf, then significance to photosynthesis under a fluctuating irradiance is the loss of optimal regulation. The faster the response to change, the less is the loss of efficiency, whether that be in terms of water use efficiency (WUE) or light use efficiency.

Since its birth one hundred years ago (Osterhout and Haas, 1918), research on the dynamics of photosynthesis and the limitations it produces in a fluctuating irradiance has come a long way (Box 1). While it has been apparent for some time that sunflecks occur in all kinds of canopies (e.g. Pearcy et al., 1990), research on sunfleck photosynthesis was until recently driven by its importance for forest understory shrubs and trees. The ecophysiological importance of sunflecks, photosynthetic responses, and plant growth focused on the importance of these responses for understory plants growing in shade (Pearcy et al., 1996; Way and Pearcy, 2012). Attention has more recently shifted to crop stands grown in full sunlight and the fact that the slow response of photosynthesis to sunflecks is a limitation to crop growth in the field (e.g. Lawson et al., 2012; Carmo-Silva et al., 2015). The importance of improved photosynthesis as a route to improving crop yields (Ort et al., 2015) has given new impetus into better understanding the physiology and the genetics of photosynthetic responses to fluctuating light and improving upon them (e.g. Kromdijk et al., 2016).

FLUCTUATING IRRADIANCE IN CANOPIES

Sunflecks

Most studies have focused on irradiance fluctuations at the bottom of canopies or in forest understories. In these situations, a shade environment with little diurnal variation prevails, and most incoming irradiance arrives due to transmission and scattering by leaves higher up in the canopy. Also, gaps in the canopy, which move in response to wind, allow brief but significant increases in irradiance (Pearcy, 1990). Smith and Berry (2013) proposed a detailed classification of these fluctuations, resulting in the terms sunfleck (<8 min and peak irradiance lower than above-canopy irradiance), sun patch (>8 min), sun gap (>60 min), and clearing (>120 min).

In addition to the length of the fluctuation, classifying a fluctuation as a sunfleck depends on the irradiance increasing above a specific threshold during the fluctuation. Often, fixed thresholds are used, but their values vary greatly (60–300 µmol m⁻² s⁻¹; Pearcy, 1983, 1990; Tang et al., 1988; Roden and Pearcy, 1993; Barradas et al., 1998; Naumburg and Ellsworth, 2002). Thresholds may be adjusted depending on canopy structure, position within the canopy where measurements are taken, and angle of measurement (Pearcy, 1990; Barradas et al., 1998). An alternative approach is to use the fraction of irradiance transmitted by the canopy instead of absolute irradiance to calculate the threshold (Barradas et al., 1998). However, this approach requires an additional measurement of irradiance above the canopy.

Short-lived sunflecks with low peak irradiance are particularly abundant in the lower layers of canopies and forest understories. Pearcy et al. (1990) reported that 79% of sunflecks were ≤1.6 s long in a soybean (Glycine max) canopy, and the same distribution was reported for aspen (Populus tremuloides; Roden and Pearcy, 1993). Peressotti et al. (2001) reported that most sunflecks in wheat (Triticum aestivum), maize, and sunflower (Helianthus annuus) were ≤1 s long. Most sunflecks in bean (Phaseolus vulgaris) and rice (Oryza sativa) canopies were ≤5.0 s long (Barradas et al., 1998; Nishimura et al., 1998). These results agree with our measurements in durum wheat (Triticum durum) and white mustard (Sinapis alba; Fig. 1).

Figure 1.

Sunflecks in two crop canopies. A, Irradiance fluctuations above and below a durum wheat and white mustard canopy, logged at 1 s resolution. B, Fraction of the total number of sunflecks as a function of sunfleck duration; calculations based on data displayed in A. Photosynthetically active irradiance (PAR; 400–700 nm) was logged using two LI-190R quantum sensors (Li-Cor Biosciences) and an LI-1400 (Li-Cor) data logger. Data were recorded 10 cm above the ground for measurements below canopies and just above canopies for 6 h (11:00–17:00) on two consecutive days (May 26 and 27, 2017) in Wageningen, The Netherlands (51.97 °N, 5.67 °E, 12 m above sea level). The two days were cloudless with average wind speeds of 3.5 m s⁻¹ and 4.2 m s⁻¹, respectively. In the absence of sunflecks, the irradiance measured below the canopy was 2.4% and 3.7% of above-canopy PAR for white mustard and wheat, respectively, indicating full canopy closure. To detect sunflecks, a baseline was constructed by interpolating PAR values in the absence of sunflecks and defining a sunfleck as the absolute change in PAR with respect to the baseline >10 µmol m⁻² s⁻¹ (this was larger than the measurement error).

Canopy structure is assumed to affect sunfleck distribution (Pearcy, 1990), but this has so far only been systematically tested by Peressotti et al. (2001), who compared sunflecks in different crop canopies and found only small differences between wheat, maize, and sunflower. Our data, on the other hand, revealed bigger differences between crops despite similar meteorological conditions (Fig. 1): in durum wheat, 2,606 sunflecks (83% of total irradiance) were detected within 6 h, while only 213 (22%) were observed in white mustard (Fig. 1A). In white mustard, sunflecks tended to be shorter and weaker, though for both crops most sunflecks were <5 s long (Fig. 1B). For most sunflecks, the average irradiance increase was <350 µmol m⁻² s⁻¹, and peak irradiance was always below the irradiance measured above the canopy (Fig. 1A). However, a large proportion of short sunflecks may not always contribute much to integrated irradiance, partly because of their short duration and partly because of their low peak irradiance (Pearcy, 1990). For example, in a soybean canopy, the peak irradiance in sunflecks less than 1.6 s long was two to three times less than that of longer sunflecks, and contributed only 6.7% of the total irradiance, while sunflecks lasting up to 10 s contributed only 33% of the total irradiance (Pearcy et al., 1990).

Sunflecks can also be caused by the penumbra effect (Smith et al., 1989), a “soft shadow” that occurs when a light source is partially blocked. In canopies, a penumbra is produced by small canopy elements that partially obscure the solar disc as viewed from a lower leaf. When combined with rapid leaf movements, the penumbra causes sunflecks on leaves that are otherwise shaded. Due to the penumbra effect, it was estimated that a gap in a canopy must have an angular size greater than 0.5° in order for the sunfleck to reach full solar irradiance (Pearcy, 1990). The frequent, short sunflecks discussed above are probably caused by penumbra (Smith and Berry, 2013) and contribute to a substantial fraction of total irradiance in forest understories (Pearcy, 1990).

Due to wind-induced movements, the structure of canopies is not static. Wind has two effects: (1) movement of the whole plant or “swaying” (de Langre, 2008; Tadrist et al., 2014; Burgess et al., 2016) and (2) fluttering of single leaves, especially in trees (Roden and Pearcy, 1993; Roden, 2003; de Langre, 2008). Plant swaying alters the spatial distribution of canopy gaps and the exposure of leaves to these gaps, adding sunflecks and shadeflecks to the baseline irradiance that would occur in the absence of wind. Fluttering allows individual leaves to have a more uniform diurnal distribution of absorbed irradiance and to maintain a high photosynthetic induction state (Roden, 2003). Fluttering further increases the number of sunflecks at the bottom of the canopy (Roden and Pearcy, 1993). Leaves flutter at a wide frequency range (1–100 Hz; Roden and Pearcy, 1993; Roden, 2003; de Langre, 2008), whereas plant swaying occurs at 0.1 to 10 Hz (de Langre, 2008; Burgess et al., 2016). Wind thus introduces rapid irradiance fluctuations in the entire canopy. Without wind, sunflecks and shadeflecks can still be caused by gaps in the canopy structure and by penumbra, but high wind speeds have been correlated with increasing irradiance fluctuations (Tang et al., 1988).

Shadeflecks

As long as the total irradiance intercepted by a canopy remains the same, the existence of sunflecks necessitates the existence of shadeflecks (i.e. transient excursions below a baseline that is the average irradiance (Pearcy, 1990; Pearcy et al., 1990; Barradas et al., 1998; Lawson et al., 2012). It is important to distinguish between sunflecks and shadeflecks, as the dynamic responses of photosynthesis are different for increasing and decreasing irradiance and involve different potentially limiting processes (see below). A shadefleck should not be seen as a “period between sunflecks,” but rather as a brief period of low irradiance with respect to a baseline of intermediate or high irradiance, which tends to occur in the top and middle layers of a canopy. A special type of shadefleck is a cloudfleck (Box 2; Knapp and Smith, 1988).

THE REGULATION OF PHOTOSYNTHESIS IN FLUCTUATING IRRADIANCE

Responses and Regulation of Electron and Proton Transport

The shorter term physiological responses of photosynthesis begin with light-driven redox state and pH changes occurring within and close to the thylakoid membranes. Photochemistry, the primary chemical event of photosynthesis, provides the redox driving forces for electron and proton transport, which result in the feed-forward activation of metabolic processes that produce CO₂ assimilation. Metabolism, when limiting, will down-regulate electron transport via feed-back mechanisms. This balance between feed-forward and feed-back regulation is at the heart of photosynthetic regulation, including responses to changing irradiance.

In a leaf initially subject to a subsaturating irradiance, a sudden increase in irradiance results in an increase in the rate of photochemistry and then an increase in the rate of linear electron flow (LEF) from water to ferredoxin within milliseconds. For every electron passing along the LEF, three protons are translocated from the stroma into the thylakoid lumen, which changes the electric (Δψ) and pH (ΔpH) gradients across the thylakoid membrane. Together, Δψ and ΔpH constitute the proton motive force (pmf). The pmf is further modulated by cyclic electron flux around PSI (Strand et al., 2015; Shikanai and Yamamoto, 2017) and alternative noncyclic electron flux (Asada, 2000; Bloom et al., 2002), making the pmf more flexible to changing metabolic demands for ATP and NADPH (Kramer and Evans, 2011) and adjustments in lumen pH resulting in regulatory responses of thylakoid electron transport and nonphotochemical quenching (NPQ). The acidification of the lumen upon increases in irradiance partially drives the fastest component of NPQ (Fig. 2), qE. This form of NPQ acts to reduce the lifetime of excited singlet states of chlorophyll a (¹chl*) in PSII. When the rate of PSII excitation and ¹chl* formation exceeds the potential for photochemical dissipation of ¹chl* via electron transport (e.g. during irradiance increases), the lifetime of ¹chl* in PSII tends to increase, potentially increasing the rate of formation of triplet chlorophylls in the PSII pigment bed and reaction center, resulting in the formation of reactive singlet oxygen (Müller et al., 2001). Up-regulating NPQ activity counteracts the tendency for increased ¹chl* lifetime and moderates the increase in singlet oxygen formation (Müller et al., 2001). The protein PsbS senses the low pH in the lumen (Li et al., 2000, 2002) and may mediate conformational changes in trimeric light harvesting complex II (LHCII) antenna complexes that allow the light harvesting complex (LHC) to more efficiently dissipate excitons formed in PSII as heat (Ruban, 2016). The presence of the carotenoid zeaxanthin further amplifies qE (Niyogi et al., 1998). Zeaxanthin is formed from violaxanthin via antheraxanthin by the enzyme violaxanthin de-epoxidase upon acidification of the thylakoid lumen and is reconverted to violaxanthin as lumen pH increases (Demmig-Adams, 1990).

Figure 2.

Schematic depiction of dynamic reactions of leaf photosynthetic processes to irradiance fluctuations. The leaf is initially adapted to shade (50 µmol m⁻² s⁻¹), then exposed to strong irradiance (1,000 µmol m⁻² s⁻¹) for 60 min, after which it is shaded again for 35 min. Displayed are net photosynthesis rate (A; red line, continuous), stomatal conductance (g_s; blue line, dots), substomatal CO₂ partial pressure (C_i; green line, long dashes), nonphotochemical quenching (NPQ; gray line, short dashes), and the electron transport efficiency of PSII (Φ_PSII; black line, long dashes and dots). These values are representative of Arabidopsis Col-0, grown in climate chambers at a constant irradiance of 170 µmol m⁻² s⁻¹.

Since after drops in irradiance NPQ relaxes only slowly (Fig. 2), LEF is transiently limited by an overprotected and quenched PSII, potentially limiting photosynthesis (Zhu et al., 2004). In Arabidopsis (Arabidopsis thaliana), the ΔpH component of the pmf was increased in plants overexpressing K⁺ efflux antiporter proteins, accelerating NPQ induction and relaxation kinetics and diminishing transient reductions in LEF and CO₂ assimilation upon transitions from high to low irradiance (Armbruster et al., 2014). In tobacco (Nicotiana tabacum), the simultaneous overexpression of PsbS, violaxanthin de-epoxidase and zeaxanthin epoxidase increased the rate of NPQ relaxation, which subsequently increased growth in the field by 14% to 20% (Kromdijk et al., 2016). These results prove that slow NPQ relaxation is an important limitation in naturally fluctuating irradiance. Further, the results of Kromdijk et al. (2016) are a powerful testament to the fact that irradiance fluctuations strongly diminish growth in the field; they provide a glimpse into growth accelerations that would be possible if the rate constants of other processes responding to fluctuating irradiance were enhanced.

Chloroplast Movement

Another potential limitation to electron transport under fluctuating irradiance is the movement of chloroplasts in response to blue irradiance. At high blue irradiance, chloroplasts move toward the anticlinal walls of the mesophyll cells, while at low blue irradiance, they move to the periclinal walls (Haupt and Scheuerlein, 1990), resulting in decreases and increases of absorptance, respectively (Gorton et al., 2003; Williams et al., 2003; Tholen et al., 2008; Loreto et al., 2009). In leaves of some species, chloroplast movements can change irradiance absorptance by >10%, although in other species the effect is <1% (Davis et al., 2011). The reduction in absorptance in high irradiance has a photoprotective effect, and significant reductions in photoinhibition have been demonstrated for Arabidopsis (Kasahara et al., 2002; Davis and Hangarter, 2012). Furthermore, chloroplast movements alter the area of chloroplasts exposed to the intercellular spaces, changing mesophyll conductance (g_m). Importantly, chloroplasts move within minutes (Brugnoli and Björkman, 1992; Dutta et al., 2015; Łabuz et al., 2015), so the effects of their movement on absorptance and g_m (Box 3) should be relevant under naturally fluctuating irradiance. In particular, slow chloroplast movement toward the low irradiance position (time constants of 6–12 min; Davis and Hangarter, 2012; Łabuz et al., 2015), which lead to increased absorptance, would transiently decrease absorptance after drops in irradiance, thus limiting electron transport and photosynthesis (i.e. similar to the effect of slow qE relaxation, see above). However, experimental evidence of this possible limitation is currently lacking.

Enzyme Activation and Metabolite Turnover

The activity of several key enzymes in the Calvin Benson cycle (CBC) is regulated in an irradiance-dependent manner, much of which depends on the thioredoxin (TRX) system (Geigenberger et al., 2017). There is a multitude of TRX types and isoforms. For example, Arabidopsis chloroplasts contain 10 different TRX isoforms (Michelet et al., 2013). Chloroplastic TRXs may be reduced by ferredoxin-dependent or NADPH-dependent thioredoxin reductases (Nikkanen et al., 2016; Thormählen et al., 2017). In the chloroplast, f-type TRXs control the activation state of Fru-1,6-bisphosphatase (FBPase), sedoheptulose-1,7-bisphosphatase (SBPase), and Rubisco activase (Rca; Michelet et al., 2013; Naranjo et al., 2016). While oxidized FBPase maintains a basal activity of 20% to 30%, the oxidized form of SBPase is completely inactive (Michelet et al., 2013). In Pisum sativum, the activities of phosphoribulokinase (PRK) and glyceraldehyde-3-phosphate dehydrogenase are controlled by the redox-regulated protein CP12, which binds the enzymes together in low irradiance and thereby inactivates them even if they are reduced (i.e. active; Howard et al., 2008). However, this type of regulation by CP12 is not universal, as in several species, the complex formed by CP12, glyceraldehyde-3-phosphate dehydrogenase, and PRK was mostly absent in darkness or the enzymes existed both in the bound and free form (Howard et al., 2011). Apart from the action of CP12, PRK activity is also regulated by TRX m and f (Schürmann and Buchanan, 2008).

Within the first minute after a switch from low to high irradiance, SBPase, FBPase, and PRK are believed to limit photosynthesis via the slow regeneration of ribulose-1,5-bisphosphate (RuBP; Sassenrath-Cole and Pearcy, 1992, 1994; Sassenrath-Cole et al., 1994; Pearcy et al., 1996). These enzymes activate and deactivate quickly, with time constants (τ) of ∼1 to 3 min for activation and ∼2 to 4 min for deactivation (Supplemental Table S1). Compared to limitation by either Rubisco or g_s (see below), which often (co)-limit photosynthetic induction for 10 to 60 min, the limitation due to activation of SBPase, FBPase, and PRK appears negligible but is relatively understudied. Due to their relatively quick deactivation in low irradiance, it may be that in the field the activation states of these enzymes are a stronger limitation of CO₂ assimilation than Rubisco or g_s (Pearcy et al., 1996), as the majority of sunflecks in canopies are short and narrowly spaced (see above). More research into this potentially large limitation is needed, e.g. by using plants with increased concentrations of CBC enzymes (e.g. Simkin et al., 2015), as well as “always-active” FBPase and PRK (Nikkanen et al., 2016).

The dependence of the activation state of Rubisco upon irradiance resembles that of an irradiance response curve of photosynthesis (Lan et al., 1992). In low irradiance, 30% to 50% of the total pool of Rubisco is active (Pearcy, 1988; Lan et al., 1992; Carmo-Silva and Salvucci, 2013). The remainder is activated with a τ of 3 to 5 min after switching to high irradiance (Pearcy, 1988; Woodrow and Mott, 1989; Kaiser et al., 2016; Taylor and Long, 2017). Activation of Rubisco active sites requires the binding of Mg²⁺ and CO₂ to form a catalytically competent (carbamylated) enzyme, after which RuBP and another CO₂ or O₂ molecule have to bind for either carboxylation or oxygenation to occur (Tcherkez, 2013). Rubisco activates more quickly at higher CO₂ partial pressures, both in folio (Kaiser et al., 2017) and in vitro (Woodrow et al., 1996), a phenomenon that is not well understood and whose kinetics cannot be explained by carbamylation.

Several types of sugar phosphates can bind to Rubisco catalytic sites and block their complete activation (Bracher et al., 2017). Removal of these inhibitors requires the action of Rca (Salvucci et al., 1985), whose activity depends on thioredoxin and ATP. Rca light-activates with a τ of ∼4 min in spinach (Spinacia oleracea; Lan et al., 1992). In Arabidopsis, Rca is present in two isoforms, of which the larger, α-isoform is redox regulated and the smaller, β-isoform is regulated by the α-isoform (Zhang and Portis, 1999; Zhang et al., 2002). In transgenic plants only containing the β-isoform, photosynthetic induction after a transition from low to high irradiance was faster than in the wild type, as Rca activity was constitutively high and independent of irradiance (Carmo-Silva and Salvucci, 2013; Kaiser et al., 2016). Modifying the composition of Rca (Prins et al., 2016) or its concentration, either transgenically (Yamori et al., 2012) or through classical breeding (Martínez-Barajas et al., 1997), might enhance photosynthesis and growth in fluctuating irradiance (Carmo-Silva et al., 2015).

After the fixation of CO₂ into RuBP, the triose phosphates may be transported out of the chloroplast and converted into sugars, after which the phosphate is transported back into the chloroplast and recycled via the chloroplast ATPase and the CBC (Stitt et al., 2010). The enzyme Suc phosphate synthase can transiently limit photosynthesis after a transition from low to high irradiance, but this has so far only been shown in elevated CO₂ (Stitt and Grosse, 1988). After decreases in irradiance, pools of CBC intermediates can transiently enhance photosynthesis (“postillumination CO₂ fixation”), while the turnover of Gly in the photorespiratory pathway may be visible as a transient decrease in photosynthesis (“postillumination CO₂ burst”). After very short (≤1 s) sunflecks, postillumination CO₂ fixation enhances total sunfleck carbon gain greatly, such that the CO₂ fixed directly after a sunfleck exceeds the CO₂ fixed during the sunfleck (Pons and Pearcy, 1992). The negative effect of postillumination CO₂ fixation on the carbon balance of a sunfleck seems less pronounced in comparison (Leakey et al., 2002). For more details on both phenomena, see Kaiser et al. (2015).

CO₂ Diffusion into the Chloroplast

Diffusion of CO₂ to the site of carboxylation is mediated by g_s and g_m. Stomata tend to decrease their aperture in low irradiance, when evaporative demand and demand for CO₂ diffusion are small. Vast differences exist between species (15- to 25-fold) for steady-state g_s in low and high irradiance (e.g. McAusland et al., 2016), for rates of stomatal opening after irradiance increases (τ = 4–29 min) and for rates of stomatal closure after irradiance decreases (τ = 6–18 min; Vico et al., 2011). Often, initial g_s after a switch from low to high irradiance is small enough, and stomatal opening is slow enough (Fig. 2), to transiently limit photosynthesis (McAusland et al., 2016; Wachendorf and Küppers, 2017). Manipulating g_s to respond more quickly to irradiance could greatly enhance photosynthesis and WUE in fluctuating irradiance (Lawson and Blatt, 2014; Vialet-Chabrand et al., 2017b). Mesophyll conductance will further affect the CO₂ available for photosynthesis (Tholen et al., 2012; Yin and Struik, 2017), and steady-state g_m affects CO₂ diffusion as strongly as does g_s (Flexas et al., 2008, 2012). Mesophyll conductance may be variable under fluctuating irradiance (Campany et al., 2016), as some of the processes determining g_m are flexible (Price et al., 1994; Flexas et al., 2006; Uehlein et al., 2008; Otto et al., 2010; Kaldenhoff, 2012). The possibility that transient g_m changes limit photosynthesis in fluctuating irradiance is discussed in Box 3.

Limiting CO₂ diffusion into the chloroplast after a switch from low to high irradiance may transiently limit photosynthesis in two ways: via a transiently low availability of the substrate CO₂ for carboxylation and by decreasing the rate of Rubisco activation (Mott and Woodrow, 1993). While the former limitation is visible through a concomitant increase in A and chloroplast CO₂ partial pressure (C_c) along the steady-state A/C_c relationship (Küppers and Schneider, 1993), the latter can be calculated by log-linearizing CO₂ assimilation after an increase in irradiance, after correcting for changes in C_i (Woodrow and Mott, 1989). The apparent τ for Rubisco activation calculated from gas exchange in folio correlates well with Rubisco activation in vitro (Woodrow and Mott, 1989; Hammond et al., 1998) and with Rca concentrations (Mott and Woodrow, 2000; Yamori et al., 2012). Additionally, Rubisco activation during photosynthetic induction can be approximated by “dynamic A/C_i curves” which are achieved by measuring the rate of photosynthetic induction at several C_i levels and plotting maximum rates of carboxylation (V_cmax) as a function of time (Soleh et al., 2016). It was recently shown that the apparent τ_Rubisco derived from dynamic A/C_i curves was in agreement with values derived using the procedure described by Woodrow and Mott (1989) and Taylor and Long (2017). Apparent τ_Rubisco decreases with increases in C_i (Mott and Woodrow, 1993; Woodrow et al., 1996) and with relative air humidity (Kaiser et al., 2017) during photosynthetic induction. The latter phenomenon was caused by humidity effects in initial g_s, leading to faster depletion of C_c and transiently lower C_c after an increase in irradiance (Kaiser et al., 2017). The mechanism behind this slowing down of Rubisco activation due to lower C_c is as yet unresolved.

PHENOTYPING FOR FASTER PHOTOSYNTHESIS IN FLUCTUATING IRRADIANCE

High-throughput phenotyping for natural variation (including mutant screens, e.g. Cruz et al., 2016) gained importance following the analyses of Lawson et al. (2012), Lawson and Blatt (2014), and Long et al. (2006). These studies highlighted the response times of photosynthesis to changing irradiance as limitations to carbon gain, including the slow response of g_s (Tinoco-Ojanguren and Pearcy, 1993), which can also diminish WUE (Lawson and Blatt, 2014), stressing their value as routes for improving assimilation. Kromdijk et al. (2016) consequently showed that improved relaxation of qE type NPQ improved tobacco yield under field conditions. While they used transgenics, the modifications used—increased amounts of PsbS, violaxanthin de-epoxidase and zeaxanthin epoxidase—could have occurred naturally. In fact, altering gene expression patterns has been a major route to improving the usefulness of plants for agriculture (Swinnen et al., 2016), either through natural variation in the gene pool of natural ancestors or through mutations occurring during domestication. Naturally occurring variation in a trait can be used to analyze the genetic architecture of the trait, and this can be used to increase the efficiency of improving the trait by breeding. Knowing how a trait is genetically determined increases the options for its improvements by breeding beyond those emerging from the physiological or biochemical approaches of the kind used by Kromdijk et al. (2016). Variation for the kinetics of photosynthetic responses to changing irradiance is also another resource for further conventional physiological and biochemical analyses of the regulation and limitations acting on photosynthesis under these conditions.

If variation for a quantitative trait, such as photosynthetic responses, is identified in a genetically diverse population, and the genetic diversity has been mapped by means of e.g. single nucleotide polymorphisms, it is possible to correlate genetic with phenotypic variation (e.g. Harbinson et al., 2012; Rungrat et al., 2016) and to identify the QTL (quantitative trait loci) whose variation correlates with phenotypic variation. Different types of mapping populations can be used for QTL identification: genome-wide association study and linkage mapping using recombinant inbred lines (RIL). These strategies have their own advantages and disadvantages (Bergelson and Roux, 2010; Harbinson et al., 2012; Korte and Farlow, 2013; Rungrat et al., 2016). Once identified, QTL are invaluable as markers for conventional plant breeding approaches and as a starting point for identifying the causal gene for the QTL. It is obviously advantageous to maximize the chances of finding an association by including as much genetic diversity as possible in a mapping population. In the case of crop plants, domestication results in a loss of genetic diversity (Doebley et al., 2006; Shi and Lai, 2015), so there is much to be gained by including field races and wild types in the construction of mapping populations or RILs. The phenotypic data required for QTL mapping requires measurements upon hundreds or thousands of individuals depending on the mapping approach adopted, the precision of the phenotyping procedure compared to the variability of the trait, and the heritability of the trait. In photosynthesis, which even in stable environments can change diurnally, quick measurements are needed (Flood et al., 2016). Measuring this many plants quickly places considerable demands on the design of high-throughput systems. Currently, the measuring technologies that are best suited to automated high-throughput phenotyping of plant photosynthetic traits, including those in unstable irradiance, are chlorophyll fluorescence imaging (Barbagallo et al., 2003; Furbank and Tester, 2011; Harbinson et al., 2012; Rungrat et al., 2016) and thermal imaging for measuring stomatal responses (Jones, 1999; Furbank and Tester, 2011; McAusland et al., 2013). While it is based on fluorescence from PSII, chlorophyll fluorescence allows the measurement of many useful photosynthetic parameters such as the electron transport efficiency of PSII, NPQ, and its components (of which qE is most commonly reported), F_v/F_m, q_P, F_v′/F_m′, and similar parameters (Baker et al., 2007; Furbank and Tester, 2011; Harbinson et al., 2012; Murchie and Harbinson, 2014). Chlorophyll fluorescence procedures are well developed, and the phenomenology and correlations of fluorescence-derived physiological parameters are well understood (e.g. Baker et al., 2007; Baker, 2008; Murchie and Harbinson, 2014). Biomass accumulation can also be used as a measure of plant fitness, and while this is not high-throughput nor specific for a photosynthetic process, it is simple to apply, requires no specific technology, and gives a useful measure of the extent to which a plant can successfully adapt to fluctuating irradiance.

While the technologies and procedures for phenotyping and QTL identification are promising, the application of this approach to photosynthesis is still limited, especially in the case of photosynthetic responses to fluctuating irradiance. QTL for qE have been identified using low throughput phenotyping (Jung and Niyogi, 2009). van Rooijen et al. (2017) have identified a gene (YS1) underlying longer-term responses to an irradiance change using a genome-wide association study analysis of an Arabidopsis mapping population (Li et al., 2010). This work demonstrates that phenotyping combined with further genetic analysis can be used for identifying QTLs and genes linked to variation in a photosynthetic trait, opening the door to a new approach to understanding photosynthetic responses to fluctuating irradiance. If a QTL can be found for a trait, such as faster responses to fluctuating light, then by implication there is an association with genetic markers. This association can be used in marker-assisted breeding to accelerate the transfer of the QTL into a genotype that lacks the trait but has otherwise desirable properties.

CONCLUSION

Average rates of photosynthesis decrease under fluctuating irradiance when compared to a constant environment. Whereas part of this decrease is explained by the nonlinear response of photosynthesis to irradiance, further decreases are the result of slow changes in enzyme activities, stomatal conductance, and NPQ. Changes in mesophyll conductance and irradiance absorbance (caused by chloroplast movements) may add to these limitations, but this awaits experimental verification. Whereas much of the earlier research focused on Rubisco activity and dynamic stomatal conductance, recent experimental and modeling studies suggest other processes (and enzymes) to be limiting (Hou et al., 2015; Guo et al., 2016). Therefore, both models and experiments should widen their scope. This requires extending the toolbox of the dynamic photosynthesis experimentalist to include rapid gas exchange systems, chlorophyll fluorescence, and spectroscopic techniques and the design of new measurement protocols and mathematical models to provide the necessary parameters. There is also the realization that the growth environment of plants should approximate that experienced in the field (Poorter et al., 2016). Recent developments of lighting technology (LEDs) enable this. Increasingly, plants are grown under more fluctuating conditions (Külheim et al., 2002; Leakey et al., 2003; Athanasiou et al., 2010; Alter et al., 2012; Vialet-Chabrand et al., 2017a), but the complex nature of natural irradiance fluctuations and the scarcity of measurements in the field mean that to date no standard exists for defining relevant fluctuating growth conditions in the laboratory.

Our review of the literature indicates that the fluctuating regime strongly depends on whether fluctuations are caused by wind and gaps in the canopy (i.e. sunflecks) or by intermittent cloudiness (i.e. cloudflecks; Box 2). Whereas the former consists of fluctuations at the scale of seconds over a low irradiance background, cloudflecks are fluctuations at the scale of minutes over a high irradiance background. Furthermore, the variation across species, canopy structure, and location seems to be small, but further characterization of cloudflecks and sunflecks is needed. Both fluctuating regimes are relevant to crops in the field, but the relative importance of processes limiting photosynthesis could depend on the specific irradiance pattern.

Supplemental Data

The following supplemental materials are available.

• Supplemental Table S1. Time constants of irradiance-dependent activation and deactivation of FBPase, PRK, and SBPase, based on fits to published data.