The convergent quantum yield hypothesis (CQY) assumes that thermodynamics and natural selection jointly limit variation in the maximum energetic efficiency of photosynthesis in low light under otherwise specified conditions (e.g. temperature and CO2 concentration). A literature survey of photosynthetic quantum yield (ϕ) studies in terrestrial plants from C3, C4, and CAM photosynthetic types was conducted to test the CQY hypothesis. Broad variation in ϕ values from C3 plants could partially be explained by accounting for whether the measuring conditions were permissive or restrictive for photorespiration. Assimilatory quotients (AQ), calculated from the CO2 ϕ:O2 ϕ ratios, indicated that 49% and 29% of absorbed light energy was allocated to carbon fixation and photorespiration in C3 plants, respectively. The unexplained remainder (22%) may represent diversion to various other energy-demanding processes (e.g. starch synthesis, nitrogen assimilation). Individual and cumulative effects of these other processes on photosynthetic efficiency are poorly quantified. In C4 plants, little variation in ϕ values was observed, consistent with the fact that C4 plants exhibit little photorespiration. As before, AQ values indicate that 22% of absorbed light energy cannot be accounted for by carbon fixation in C4 plants. Among all three photosynthetic types, the ϕ of photosynthesis in CAM plants is the least studied, appears to be highly variable, and may present the greatest challenge to the CQY hypothesis. The high amount of energy diverted to processes other than carbon fixation in C3 and C4 plants and the poor characterization of photosynthetic efficiency in CAM plants are significant deficiencies in our otherwise robust understanding of the energetics of terrestrial photoautotrophy.
The quantum yield of photosynthesis (ϕ) is a definitive measure of the energetic efficiency of photoautotrophy. The quantum yield for any defined light-dependent process is the rate at which that defined event occurs relative to the rate of photon absorption by the system. As such, the quantum yield is a measure of the efficiency with which absorbed light produces a particular effect. Figure 1 shows the results of simultaneous assessments of net carbon fixation, the instantaneous quantum yield of photosystem II electron transport, and the non-productive dissipation of absorbed light as heat (non-photochemical quenching; NPQ) from an intact C3 leaf over a range of absorbed photon flux densities. The dashed line in Fig. 1A is fit to the linear portion of the light response of carbon assimilation. This line describes the rate of net carbon gain as a function of absorbed light for a leaf operating at maximum realized photosynthetic efficiency. The slope of this line is the conventional measure of the maximum ϕ of photosynthesis. This simple measure of maximum photosynthetic efficiency should not to be confused with ‘instantaneous’ measures of photosynthetic efficiency such as the quantum yield of photosystem II (PSII) activity (Fig. 1B) derived from chlorophyll fluorescence measurements (Maxwell and Johnson, 2000; Lichtenthaler et al., 2007). In low-light, where carbon fixation operates more or less at maximum photosynthetic efficiency, Fig. 1 shows that the instantaneous quantum yield of PSII activity remains high and the non-productive dissipation of absorbed light is modest. However, as carbon fixation becomes increasingly light-saturated, the instantaneous quantum yield of PSII activity declines linearly and heat dissipation (NPQ) increases linearly with increasing light availability. Thus, instantaneous measures of photosynthetic efficiency vary with light availability, whereas the maximum ϕ is an intrinsic characteristic of the photosynthetic tissue of interest. The use of the term ‘quantum yield’ here will be reserved for this maximum ϕ as derived from the initial slope of the light response curve of leaf photosynthesis.
As recently as half a century ago, the actual value of the maximum ϕ was unresolved and proposed estimates for the uppermost efficiency of photosynthesis were a source of contention (Emerson and Lewis, 1941; Warburg, 1958; reviewed in Govindjee, 1999). Today it is known that for each O2 produced and each CO2 fixed into triose phosphates, the minimum assimilatory power requirement is 3 ATP and 2 NADPH (Hill and Bendall, 1960; Jagendorf and Uribe, 1996; Edwards and Walker, 1983; Seelert et al., 2000; Blankenship, 2002). If this alone was all that mattered it would translate into a maximum ϕ of approximately 0.111 mol CO2 fixed (or O2 produced) per mol absorbed photons for C3 photosynthesis. This value serves as an important ‘upper limit’ on photosynthetic efficiency. However, as of 25 years ago it was clear that this potential ϕ value could only be achieved in intact C3 photosynthetic tissues under a restricted set of artificial laboratory conditions and that realized ϕ values under ambient conditions in C3 plants fell well below this upper limit (Ehleringer and Björkman, 1977; Bjorkman and Demmig, 1987). The discrepancy between the potential and the realized maximum ϕ indicates that other processes, such as photorespiration or starch synthesis or nitrogen assimilation, substantially reduce the maximum energetic efficiency of carbon gain.
With the upper ceiling on photosynthetic efficiency firmly set by fundamental physiochemical limitations, one may wonder if a lower threshold on photosynthetic efficiency also exists. It is axiomatic that plant survival and productivity depend absolutely upon light-driven carbon gain. To the extent that the green tissues on an individual plant operate under light-limited conditions, knowing about ϕ can inform us about ultimate restrictions on plant productivity, and ecological and evolutionary success (Monteith, 1972; Ort and Baker, 1988; Long et al., 2006). The assumption that most green cells in most plants are light-limited much of the time predicts that natural selection will continually and decisively maintain plant populations with foliar photosynthetic efficiencies at or near the maximum achievable ϕ for any given environmental setting. Being sandwiched between fitness and physics, we have come to expect that the realized ϕ values in different C3 species under non-stressful conditions should be maximal and invariant. This expectation may be referred to as the ‘convergent quantum yield’ (CQY) hypothesis. Indeed, it is the fact that this simple trait is believed to be invariant and serves as a clear interface between fundamental physiochemical and evolutionary constraints on plant performance that has made ϕ an important target of research in plant ecophysiology (Osmond et al., 1999). Similarly, although de-emphasized here, the ample literature on photoinhibition may be interpreted as a search for causes and consequences of reductions in ϕ (Ewart, 1894; Skillman and Osmond, 1998; Demmig-Adams et al., 2006).
Interestingly, an earlier literature survey demonstrated that there was surprisingly broad scatter in this putatively conserved trait in terrestrial C3 plants even under non-stressful conditions (Singsaas et al., 2001). This comparative literature survey approach provided an opening for Singsaas et al. (2001) to demonstrate how measurement methods and data analysis methods can make a large difference in the calculated ϕ estimates for C3 photosynthesis. For example, not accounting for light effects on mitochondrial respiration (i.e. the Kok effect) leads to erroneous overestimates of the ϕ of photosynthesis (Sharp et al., 1984; Singaas et al., 2001). Reassuringly, these authors found that, in many cases, the observed spread in measured C3 ϕ values is attributable to analytical errors and is not a result of biological variation for this trait.
In this review, the literature survey approach used by Singass et al. (2001) is employed to examine the range of variation in ϕ values in terrestrial plants in toto and broken down by mode of photosynthesis [C3, C4, and Crassulacean acid metabolism (CAM)]. This approach is used as a means of identifying areas where our understanding of photosynthetic quantum yield biology is sound and areas where our knowledge is still incomplete. The surveyed literature was restricted to a large set of authoritative studies where an accurate determination of ϕ from leaves of one or more terrestrial plant species was a central objective of the study and where the described methods and analysis techniques were judged to be sound, thus presumably avoiding the sources of error identified by Singsaas et al. (2001). Although the selected set of studies analysed here is undoubtedly not exhaustive, it does include many of the important ϕ studies made on intact leaf tissue and is large enough to allow informative generalizations to be made. It will become evident that most ecophysiological studies on ϕ variation have concentrated only on C3 and C4 plants and that much of the observed variation can be understood in terms of trade-offs between these two well-studied modes of photosynthesis. By contrast, a consensus on the causes and range of variation in ϕ among CAM plants has not been achieved. In addition, there is little quantitative data available on how other metabolic processes (e.g. nitrogen assimilation) affect the efficiency of photosynthetic carbon gain. The high amount of energy diverted to processes other than carbon fixation in C3 and C4 plants and the poor characterization of photosynthetic efficiency in CAM plants are significant deficiencies in our understanding of the energetics of terrestrial photoautotrophy.
Observed photosynthetic quantum yield values for terrestrial C3 plants
The distribution of values from a set of authoritative ϕ studies on various terrestrial plants, including representatives of all three photosynthetic pathways, is presented in Fig. 2. These data were collected on healthy, non-stressed plants over, in most cases, a narrow range of temperatures (25–30 °C). This plot for 218 observations exhibits broad and bimodally distributed variation (coefficient of variation (CV)=33%). Both the high variance and bimodality may, on first consideration, appear to be at odds with the CQY hypothesis. The broad scatter is not simply a consequence of lumping the three different photosynthetic types into a common dataset. These data, when pared down to include only C3 species (Fig. 3), gives essentially the same result (N=127, CV=33%). However, there are other important factors besides photosynthetic pathway differences that are subsumed within Fig. 2 that can affect measured ϕ values and can explain much of the variation among the C3 plants shown in Fig. 3.
The effect of photorespiration on quantum yield in C3 plants
About half of the C3 ϕ data summarized in Fig. 3 were measured with leaves under artificial atmospheres that were either enriched in CO2 or depleted in O2 down to about 2%. These data constitute the right side of the histogram in Fig. 3 (i.e. both white and hatched bars). The remaining data on the left side of the histogram (black bars) are from leaves in ambient atmospheres (21% O2 and ∼0.03% CO2). The relative concentrations of O2 and CO2 affect the maximum measurable efficiency of C3 photosynthesis because Rubisco, the key enzyme of global photoautotrophy, possesses both a carboxylase and an oxygenase activity (Bowes et al., 1971). The carboxylation and oxygenation activities of Rubisco are competitive and so depend upon the relative concentrations of these gases as well as the active site specificity for CO2 relative to O2 (Sc/o). Carboxylation of RuBP by Rubisco leads to the productive gain of carbon via the photosynthetic carbon reduction (PCR) cycle which is used to support plant growth. Oxygenation of RuBP by Rubisco yields a two-carbon compound (phosphoglycolate) which can only be partially utilized for growth and only after first being diverted through an alternate and extensive series of reactions distributed across several cellular organelles (chloroplast, peroxisome, and mitochondria). This extensive and energetically costly series of reactions is referred to as the photorespiratory carbon oxidation (PCO) cycle (Fig. 4).
The co-ordinated regulation of organelle numbers and their positions, the synthesis and regulation of PCO metabolic machinery, the overall flux of photosynthetic products through this additional cycle, and the actual photorespiratory oxidation of previously reduced carbon increases the total energetic requirements of carbon fixation in C3 plants under ambient atmospheres (Heldt, 2004). This reduced energetic efficiency is clearly illustrated when comparing C3 ϕ values measured under ambient conditions against those under so-called non-photorespiratory conditions (Fig. 3). The average ϕ of C3 photosynthesis under ambient atmospheres at moderate temperatures is 0.052 mol mol−1 (N=61, CV=6%; black bars in Fig. 3; see also Table 1). The average C3 ϕ measured under conditions that inhibit PCO cycle activity (i.e. reduced O2 or elevated CO2), extracted from the data in Fig. 3, is 0.096 mol mol−1 (N=66, CV=15%; white and hatched bars in Fig. 3). This approaches the theoretical maximum ϕ value of 0.111 mol mol−1 assumed for C3 photosynthesis. Distinguishing between photosynthetic efficiencies measured with and without photorespiration largely explains the bimodal distribution of C3 ϕ values in Fig. 3.
|Photosynthetic pathway||Number of observations||Quantum yield(mol mol−1)||Coefficient of variation (%)||Referencesa|
|C3||61||0.052±0.003||6||1, 3, 5, 7, 9, 11, 12, 13, 14, 15|
|C4||56||0.057±0.006||11||1, 3, 5, 7, 9, 11, 12, 13|
|CAM||6||0.033±0.017||52||2, 4, 6, 8, 10|
|Photosynthetic pathway||Number of observations||Quantum yield(mol mol−1)||Coefficient of variation (%)||Referencesa|
|C3||61||0.052±0.003||6||1, 3, 5, 7, 9, 11, 12, 13, 14, 15|
|C4||56||0.057±0.006||11||1, 3, 5, 7, 9, 11, 12, 13|
|CAM||6||0.033±0.017||52||2, 4, 6, 8, 10|
All measurements included in this summary were made under ambient concentrations of atmospheric CO2 and O2 (i.e. photorespiratory conditions) in non-stressed photosynthetic tissues and, in most cases, at temperatures between 25 °C and 30 °C.
(1) Ehleringer and Björkman, 1977; (2) Nobel, 1977; (3) Ku and Edwards, 1978; (4) Nobel and Hartsock, 1978; (5) Robichaux and Pearcy, 1980; (6) Spalding et al. 1980; (7) Monson et al. 1982; (8) Nobel, 1982; (9) Ehleringer and Pearcy, 1983; (10) Nobel and Hartsock, 1983; (11) Osborne and Garret, 1983; (12) Monson et al., 1986; (13) Monson et al., 1987; (14) Long et al., 1993; (15) Singaas et al., 2001.
Observed variation in C3 ϕ under ambient atmospheres would be even greater if measurements over a broader range of temperatures were included in this analysis. This is because photorespiration in C3 plants varies with temperature (Ehleringer and Björkman, 1977). Temperature-dependent changes in the kinetic properties of Rubisco result in a drop in Sc/o with rising temperatures, favouring increased RuBP oxygenation at higher temperatures (Jordan and Ogren, 1984). In addition, the solubility of CO2 in water decreases more than the solubility of O2 in water for each incremental increase in temperature. Consequently, the relative concentration of CO2 to O2 in the vicinity of the active site of Rubisco changes with temperature, favouring RuBP oxygenation more at increasing temperatures (Ku and Edwards, 1978). Together these factors explain the marked drop in the C3 ϕ under warm compared with cool temperatures (Ehleringer and Björkman, 1977). The effect of temperature on the relative RuBP oxygenation and carboxylation rates, which is reflected in the temperature effect on the C3 ϕ, appears to have far-reaching ecological and evolutionary implications (Ehleringer et al., 1997; Sage and Kubien, 2003).
Over the narrow range of temperatures used in the studies summarized in this review, the Sc/o of Rubisco in C3 plants is typically quite high (>80) with little variation observed among different C3 species (Jordan and Ogren, 1983; Tcherkez et al., 2006). A high Sc/o value indicates that the competitive interaction between CO2 and O2 at the Rubisco active site is strongly biased in favour of carboxylation. This suggests that there has been selection to minimize Rubisco oxygenase activity in C3 plants as might be predicted from the CQY hypothesis. The strong kinetic bias for carboxylation notwithstanding, in the contemporary atmosphere (21% O2 and 0.04% CO2), photorespiration remains as an unavoidable drain on photosynthetic efficiency in C3 plants.
Given the inevitability of photorespiration, it is not surprising that plants exhibit various means by which they minimize and/or exploit PCO activity for other essential metabolic processes such as avoiding photo-oxidative damage (Powles and Osmond, 1978) or facilitating foliar nitrate assimilation (Rachmilevitch et al., 2004). Despite the fact that plants are able to make the best of a bad reaction, RuBP oxygenation imposes a substantial penalty on the overall efficiency of carbon gain in C3 plants. But ultimately, it is the realized ϕ in C3 plants (i.e. the maximum efficiency under ambient atmospheric conditions) that is relevant to the discussion of their ecological and evolutionary success in the field. Accordingly, the variation in ϕ among C3 plants, when measured over a narrow temperature range under ambient atmospheric conditions, turns out to be quite narrow and unimodal (Table 1; black bars in Fig. 3). This is consistent with the CQY hypothesis.
Other sinks for assimilatory power
Some examples of chloroplast reactions that can divert assimilatory power away from the PCR/PCO cycles are shown in Fig. 4. These include the synthesis of carbohydrates (starch and/or sucrose) from photosynthetically derived triose phosphates, the photoassimilation of inorganic nitrogen (e.g. chloroplastic nitrite reduction and ammonia assimilation) and the water–water cycle [i.e. photoreduction of O2 (the Mehler reaction) and the subsequent reductive scavenging of reactive oxygen species]. Additional cellular processes not shown that may also depend upon photogenerated ATP and reductant [collectively, reduced Ferredoxin (Fd) and NADPH] include inorganic sulphur assimilation, cytosolic nitrate reduction, reductive poising of redox metabolites (e.g. thioredioxin, ascorbic acid, glutathione, tocopherol), and fatty acid synthesis (Harwood and Russell, 1984; Noctor and Foyer, 1998; Saito, 2004; Noctor, 2006). Furthermore, differential engagement of photosynthetic processes that regulate the ATP:reductant production ratio can also affect the efficiency of photosynthesis. For example, photosystem I can use absorbed light to drive cyclic-photophosphorylation without the involvement of photosystem II activity, thereby contributing to ATP synthesis without producing either O2 or NADPH. Increased diversion of absorbed light to cyclic-photophosphorylation and away from linear electron transport will lower the ϕ of O2 evolution while increasing the ATP:NADP production ratio (Allen, 2003). Clearly many metabolic processes can alter the availability of assimilatory power for joint PCR/PCO activities and so may be expected to lower the overall efficiency of photosynthesis below the theoretical maximum ϕ. These other reactions will be referred to en masse as ancillary processes. Evidence for the proposal that a substantial portion of assimilatory power is directed to these ancillary processes emerges upon consideration of whether the summarized ϕ values were measured as CO2 uptake or O2 production. This distinction was subsumed within Fig. 2. However, as Fig. 3 indicates, this methodological distinction clearly explains some of the observed variation in ϕ among C3 plants.
The relative engagement of these various ancillary processes compared to carbon assimilation will affect the assimilatory quotient (AQ, the ratio of net CO2 uptake to net O2 evolution) when photorespiration is controlled for (Rachmilevitch et al., 2004). Under conditions where all the assimilatory power is being used for carbon fixation, AQ=1.0. Under conditions where assimilatory power is diverted to processes other than carbon fixation, AQ <1.0. As shown in Fig. 3, ϕ measurements made in the absence of photorespiration used two different methods which, when viewed together, suggest an AQ <1.0. Non-photorespiratory C3 ϕ observations using closed Clark-type oxygen electrode (CTOE) systems to analyse leaf O2 production in atmospheres containing ∼5% CO2 and 21% O2 had an average ϕ of 0.106 mol mol−1 (N=39, CV=6%, hatched bars in Fig. 3). This is remarkably close to the theoretical maximum ϕ value of 0.111 mol mol−1. Conversely, non-photorespiratory C3 ϕ observations made using open-flow infrared gas analysis (IRGA) systems to quantify leaf CO2 uptake in atmospheres containing ambient CO2 and ∼2% O2 had an average ϕ of 0.083 mol mol−1 (N=28, CV=12%, white bars in Fig. 3). This is well below the predicted ϕ value of 0.111 mol mol−1. From these data it is not possible to determine whether the discrepancy between the two photorespiration-free ϕ datasets is due to different methodologies or due to real biological phenomena. But, if it is assumed that this difference is not a methodological artefact, these data give an AQ of 0.78. Unfortunately, no simultaneous quantum yield measurements of CO2 and O2 fluxes are available to confirm this estimate. However, consistent with this value of 0.78, AQ values of approximately 0.80 were reported for tomato leaves under non-photorespiratory conditions (2% O2) over a broad range of brighter light levels (Searles and Bloom, 2003). An AQ of 0.78 measured under non-photorespiratory conditions suggests that, at maximum photosynthetic efficiency, about 22% of assimilatory power is allocated to these various ancillary processes. Comparing the ϕ of CO2 fixation under ambient conditions (0.052) to the ϕ of O2 production under non-photorespiratory conditions (0.106) gives an AQ of 0.49. This suggests that the maximum efficiency of net CO2 fixation, after diverting assimilatory power to these ancillary processes and photorespiration, is only about 49% of the maximum efficiency of net O2 evolution. The remaining 29% (after accounting for the 22% diverted to ancillary processes and the 49% dedicated to carbon fixation) may be taken as an estimate of the energetic costs of the PCO cycle. This latter estimate agrees well with numerous other published estimates of the relative energy requirement of photorespiration (Edwards and Walker, 1983; Ogren, 1994; Heldt, 2004).
Terrestrial plants are strict autotrophs and photosynthetic carbon acquisition is an absolute prerequisite to carbon allocation to all other plant metabolic processes. Consequently, it is surprising that, under light-limiting conditions, as much as 22% of absorbed light energy in C3 plants appears to be diverted away from carbon fixation and its unavoidable partner, photorespiration. Of course, essential constitutive energy-demanding processes that are intimately linked to photosynthetic activity (e.g. starch synthesis) would be expected to require assimilatory power in direct proportion to that being used in carbon fixation, even in low light. At maximum efficiency, the assimilatory requirement for the fixation of six CO2 molecules to produce a single hexose is 18 ATP and 12 NADPH. Polymerizing one hexose molecule to starch in the chloroplast requires one additional ATP, raising the overall assimilatory requirement to 19 ATP and 12 NADPH (Heldt, 2004). This represents, at most, a 6% reduction in the ϕ. Cytosolic sucrose synthesis should have an even smaller effect on photosynthetic efficiency if the necessary ATP equivalents are derived from mitochondrial respiration. Thus, starch and sucrose synthesis only explain a small portion of the apparent 22% reduction in ϕ. The remaining fractional reduction in ϕ (∼16%) must be due to other ancillary processes. However, unlike starch and sucrose synthesis, these other ancillary processes tend to be facultatively active, depending upon plant environmental and developmental status. For example, chloroplast fatty acid production cannot be a substantial drain on assimilatory power in mature photosynthetic tissues where growth rates and glycerolipid synthesis rates are low. In addition, these other ancillary processes are not as tightly coupled to photosynthesis as carbohydrate synthesis is. For example, the assimilation of nitrate can take place in either roots or leaves. Root nitrate assimilation relies upon energy derived from mitochondrial respiration and therefore cannot act as a drain on foliar assimilatory power. In order to maximize the efficient use of energy for carbon gain, light-dependent regulatory mechanisms might be expected to be in place that preferentially allocate assimilatory power to carbon gain and away from these facultative ancillary processes when light is limiting. Accordingly, only at brighter light levels would there be assimilatory power directed to these other reactions.
Among various ancillary processes other than starch/sucrose synthesis, foliar nitrate assimilation has received considerable attention as a potential competitive sink for photogenerated ATP and reductant. Nitrate photoassimilation is a four-step process beginning with nitrate reduction to nitrite in the cytosol, nitrite reduction to ammonia in the chloroplast (Fig. 4), and the assimilation of ammonia first to glutamine and then to glutamate in the chloroplast (Fig. 4). Robinson (1988) was unable to detect an inhibitory effect of CO2 assimilation on nitrite reduction even in very low light in isolated spinach chloroplsts. Likewise, induction of nitrite reduction activity did not inhibit light-limited CO2 fixation (Robinson, 1988). This was interpreted to mean that carbon fixation and nitrite reduction do not actually compete for assimilatory power even in low light and that nitrite assimilation does not affect the maximum efficiency of carbon fixation. This argues against the necessity of light-dependent regulation of assimilatory power to these two processes. On the other hand, light regulation is implicated in both the expression of genes encoding nitrate reductase and the activity of nitrate reductase (Heldt, 2004). Rachmilevitch et al. (2004) also point out that the relative affinities of chloroplast enzymes for reduced ferredoxin (Fd) may provide a means for regulated flux of reductant to carbon fixation over nitrite reduction and ammonia assimilation (Fig. 4). For example, the relative affinity of Fd-NADP reductase for Fd is about six times that of nitrite reductase. Consequently, Fd-dependent nitrite reduction should proceed only if the Fd concentration in the chloroplast exceeds that needed for maintaining NADPH at sufficient levels for carbon assimilation (Rachmilevich et al., 2004). Thus, at the levels of gene expression, enzyme activity, and enzyme kinetics, there appear to be regulatory mechanisms in place that regulate the allocation of assimilatory power preferentially to PCR (and PCO) activity and away from foliar nitrate assimilation in low light. Comparative studies of the assimilatory quotient under bright light in tomato and wheat leaves from plants fertilized with nitrate or ammonia support this expectation, but wild-type Arabidopsis thaliana did not conform to this expectation (Searles and Bloom, 2003; Rachmilevitch et al., 2004). More research is needed to achieve a consensus on whether or not N assimilation reactions compete directly with photosynthesis for assimilatory power under light-limiting conditions.
Another of the ancillary processes that has received considerable interest is the water–water cycle (Fig. 4). Direct photoreduction of O2 at photosystem I to the superoxide radical (), followed by the enzymatic reduction of superoxide ultimately back to H2O, is widely believed to be an important alternate sink for photosynthetic reducing power in high light but not at low light (Asada, 1999; Ort and Baker, 2002). However, the evidence in support of this proposal is modest and equivocal (Badger et al., 2000). Moreover, the means of regulation for differences in relative activity of the water–water cycle at low compared to high light is not well understood. The limited and conflicting evidence for the possibility of light-dependent regulation of just two of the many ancillary energy-demanding processes (i.e. N assimilation and the water–water cycle) indicate that, at this point, the unexpected suggestion that approximately 20–25% of photosynthetic assimilatory power is allocated to reactions other than PCR and PCO cycles under light-limiting conditions cannot be rejected. Because of the far-reaching implications of C3 photosynthetic efficiency for our understanding of plant biology, agricultural and ecological production, and the global carbon cycle, this area of research deserves greater attention. Importantly, modern integrative approaches incorporating microarray analyses and metabolite profiling of the expression of key genes and key metabolites in these various interacting ancillary pathways, coupled with biochemical models and innovative gas exchange studies should allow us to quantify better the effect of these assorted ancillary processes on carbon gain efficiency.
Observed photosynthetic quantum yield values for terrestrial C4 and CAM plants
The carbon-concentrating mechanisms of CAM and C4 photosynthesis are widely believed to have evolved as means for minimizing the energetically wasteful process of photorespiration. Interestingly, data summarized by Tcherkez et al. (2006) indicate that the Sc/o of Rubisco in C4 plants is more variable than that in C3 plants and that the mean value is significantly lower in C4 plants than in C3 plants. Comparable data for Rubisco from multiple CAM species are not available. However, the C3/C4 contrast in Rubisco kinetic properties is consistent with the idea that photorespiration-avoidance is the raison d’être for C4. But, as discussed in this section, avoidance of photorespiration by means of the C4 cycle also comes with a penalty for photosynthetic efficiency. As will be seen, the comparative role of photorespiration avoidance in the evolution and energetics of CAM is less clear.
The distribution of values from a set of authoritative ϕ studies in terrestrial C4 plants is presented in Fig. 5A. All data were collected on healthy, non-stressed plants over a narrow range of temperatures (25–30 °C). This plot encompasses measurements made under both artificial non-photorespiratory atmospheres and ambient atmospheric conditions. Among data collected under non-photorespiratory conditions are included measurements of both the ϕ of CO2 uptake and the ϕ of O2 evolution. Little variation in observed ϕ values in C4 plants exists regardless of the methodological approach employed (Fig. 5A). The average ϕ of C4 photosynthesis, pooled for all measurement methods and conditions, is 0.058 mol mol−1 (N=74, CV=12%). This difference in total ϕ variation compared with that for the C3 plants reflects the fact that C4 plants normally exhibit greatly reduced rates of photorespiration. Indeed, a t test indicated that the average C4 ϕ of CO2 uptake measured under non-photorespiratory conditions (0.054 mol mol−1, N=18, CV=15%) is not significantly different from the C4 ϕ of CO2 uptake under ambient conditions (0.057 mol mol−1, N=56, CV=10%; Table 1). This narrow distribution is even more remarkable given that the dataset includes different taxonomic and biochemical groups of C4 plants.
Systematic differences in C4 photosynthetic efficiencies are known to exist between monocots and dicots, among members of the different C4 subtypes, and among plants with different propensities for bundle sheath leakage (Pearcy and Ehleringer, 1984; Ogle, 2003; Kubásek et al., 2007). Bundle sheath leakage of CO2 in particular may be the biggest source of variation in photosynthetic efficiency among C4 plants. Leakage of previously fixed CO2 out of the bundle sheath cells reduces photosynthetic efficiency by raising the ATP requirement for the C4 cycle. Bundle sheath leakage of CO2 as a portion of carbon fixed by PEP Carboxylase (PEPCase) in C4 plants is greater in low-light than in high-light (Henderson et al., 1992; Meinzer and Zhu, 1998; Kubásek et al., 2007; Tazoe et al., 2008). Tazoe et al. (2008) examined the effect of bundle sheath leakage on ϕ estimates in Amaranthus edulis, a dicotyledonous NAD-malic enzyme C4 plant. Simultaneous measurements of carbon assimilation and carbon isotope discrimination over a range of low-light levels demonstrated that CO2 leakage increases proportionally with decreasing irradiance in the linear portion of the light response curve. This implies that conventional ϕ measurements (i.e. from the slope of the linear portion of the light response curve of net CO2 uptake) in this C4 plant may actually overestimate the maximum efficiency of C4 carboxylation by as much as 10% (Tazoe et al., 2008). This is particularly interesting because ϕ values of net CO2 uptake for NAD-malic enzyme dicots, like A. edulis, are about 10% lower than other C4 subtype dicots and all C4 monocots when assessed through conventional light-response curve methods (Ehleringer and Pearcy, 1983). It will be interesting to see if the strict light dependence of bundle sheath leakage observed in A. edulis generalizes to other C4 taxonomic groups and biochemical subtypes. The possibility of differential rates of bundle sheath leakage notwithstanding, it is difficult to imagine a better argument in support of the CQY hypothesis than the narrow distribution of C4 ϕ values observed across biochemically, anatomically, and taxonomically distinct groups even when assessed under distinct measurement conditions (Fig. 5A).
Few measurements have been made on C4 plants under non-photorespiratory conditions because there is little to be learned in studying the effect of these artificial atmospheres on plants that exhibit little photorespiration. Nevertheless, two intriguing patterns are suggested. First, comparing C3 and C4 data, the C4 ϕ of O2 production (0.069 mol mol−1, N=5, CV=14%) is significantly lower than that of C3 plants (0.106 mol mol−1). This difference is due to the additional energy required to drive the C4 cycle. C4 photosynthesis minimally requires 2ATP per CO2 fixed over and above that of photorespiration-free C3 photosynthesis (Edwards and Walker, 1983). Taking this additional ATP requirement into account (and assuming no CO2 leakage), Ehleringer and Pearcy (1983) calculated that the maximum potential ϕ for C4 plants is 0.067 mol mol−1 (compared with the 0.111 mol mol−1 maximum value in C3 plants). Assuming the additional ATP is made up in cyclic photophosphorylation, this would lower the ϕ of O2 production in C4 plants. The congruence between the predicted and the observed values for the ϕ of O2 production in C4 plants is exceptional. This implies that our overall understanding of photosynthetic efficiency in C4 plants is robust and that other factors (e.g. CO2-leakage, C4 subtype) have only minor effects on steady-state photosynthetic efficiency in low light. Secondly, considering the two sets of measurements in C4 plants made under atmospheres designed to eliminate photorespiration, the C4 ϕ of O2 production in elevated CO2 at 0.069 mol mol−1 is significantly higher than the C4 ϕ of CO2 uptake in reduced O2 at 0.054 mol mol−1. The ratio of these two mean values gives an AQ for C4 photosynthesis of 0.78. Remarkably, this is identical to the AQ estimate in C3 species under the same paired conditions. As before, if it is assumed that these differences are real rather than a methodological artefact, this would again implicate processes other than carbon fixation and photorespiration as a sink for 22% of the absorbed energy in C4 plants operating at maximum photosynthetic efficiency.
Although temperature effects have been intentionally avoided in the ϕ data included in this review, the differential effect of temperature on the ϕ of CO2 uptake in C3 and C4 plants needs to be mentioned. As discussed above, in C3 plants the ratio of Rubisco oxygenase to carboxylase activity increases with temperature. This lowers the C3 ϕ of CO2 uptake from approximately 0.07 mol mol−1 at 10 °C down to approximately 0.04 mol mol−1 at 40 °C (Ehleringer and Björkman, 1977; Ku and Edwards, 1978). The ϕ of CO2 uptake in C4 plants is constant at about 0.06 mol mol−1 across this same temperature range. Thus, the photosynthetic efficiency of C4 plants under ambient atmospheres is less than that of C3 plants at cool temperatures and surpasses them at warmer temperatures. Only at intermediate temperatures of 20–30 °C will C3 and C4 plants have a similar ϕ of CO2 uptake. As it turns out, this ϕ ‘crossover’ temperature has been useful in describing ecological patterns for C3 and C4 plants. For example, the latitude where C4 dominance gives way to C3 dominance can be accurately predicted from the ϕ crossover temperature (Ehleringer et al., 1997; Sage et al., 1999). Similarly, temporal variation in CO2 and temperature have been used to predict how the trade-offs in photosynthetic efficiency between C3 and C4 plants should affect the photosynthetic composition of vegetation over different time periods. There is good agreement between these predictions and available paleoecological data (Ehleringer et al., 1997; Collatz et al., 1998). Sage and Kubien (2003) have argued that it is unlikely that these ecological patterns are actually driven by differences in photosynthetic performance in low light. The strength of the association between ecological distributions and the ϕ crossover temperature must reflect the fact that the ϕ differences between C3 and C4 plants are determined by the relative rates of RuBP oxygenation and carboxylation in these two photosynthetic types. Differences between C3 and C4 plants in the temperature effect on RuBP carboxylation and oxygenation rates will ultimately have a larger effect on the efficiency of carbon fixation under high- rather than low-light conditions. Even so, these ecological observations provide potent circumstantial evidence that there has been and continues to be strong selection for maximizing the realized efficiency of photosynthesis, regardless of pathway. This is precisely what the CQY hypothesis predicts.
Crassulacean acid metabolism
The distribution of values from a set of authoritative ϕ studies in terrestrial CAM plants is presented in Fig. 5B. As before, this plot encompasses measurements made under either artificial non-photorespiratory conditions or ambient atmospheric conditions. It also encompasses several methodological approaches, including, but not limited to, ϕ estimates from IRGA-based CO2 flux measurements and from CTOE-based O2 flux measurements. It is immediately obvious that the photosynthetic ϕ in CAM has received relatively little attention. In order to include as many ϕ estimates for this photosynthetic type as possible, the narrow temperature range requirement that was used for C3 and C4 plants has been relaxed. However, all data included in Fig. 5B were made under ecologically realistic temperature regimes for each of the study species. As before, all measurements were conducted on healthy, non-stressed plants.
The average ϕ of photosynthesis in CAM, pooled for all measurement methods and conditions, is 0.073 mol mol−1 (N=17, CV=48%). The high variation among published CAM ϕ values would seem to refute the CQY hypothesis (Fig. 5B; Table 1). In general, we cannot account for that variation by focusing on differences among methods as was done for C3 plants because so few CAM ϕ studies have been done and they have relied on a broader number of methods, including diel (i.e. 24 h) CO2 flux measurements (Nobel, 1977; Nobel and Hartsock, 1983), nocturnal tissue acidification rates (Nobel and Hartsock, 1978; Nobel, 1982), diurnal malate consumption rates (Spalding et al., 1980), and instantaneous O2 flux rates (Adams et al., 1986; Björkman and Demmig, 1987). Each of these different methods actually measure different, albeit related, metabolic processes and each method has its own set of assumptions and uncertainties. Adding to the uncertainty, most of the methods are not used for C3 or C4 plants and so most CAM data cannot be referenced against these other photosynthetic types. Among these different methods, only the O2 electrode measurements are directly comparable with data available for C3 and C4 plants. The average ϕ of O2 production in CAM plants is 0.10 mol mol−1 (N=10, CV=10%). A t test verified that the efficiency of photosynthetic oxygen production in CAM plants is not different from that of C3 plants. The CAM O2 electrode data account for the entire right side of the histogram in Fig. 5B. This leaves very few data compiled from a disparate set of methods and measurement conditions from which to estimate the ϕ of CO2 fixation in CAM. The average photosynthetic ϕ for the remaining CAM data is 0.033 mol mol−1 (N=6, CV=53%; Table 1). The AQ from these data, if taken at face value, would indicate that at maximum efficiency (i.e. in low light) only 33% of absorbed photoenergy is used for carbon fixation compared to ∼50% in C3 plants and ∼75% in C4 plants. But with the limited number of samples and the variety of methods employed, this estimated ϕ of CO2 fixation in CAM has to be considered of only limited reliability. It is not possible from these data to discern how much of this variation is due to real biological differences in photosynthetic efficiency. A consideration of the photosynthetic efficiency of CAM from a physiological perspective and/or an evolutionary perspective may provide insight into this problem.
Carbon gain physiology of CAM plants is conventionally characterized by the diel pattern of gas exchange which can be divided into four distinct phases (Leegood et al., 1997). Nocturnal PEPCase-mediated uptake and fixation of atmospheric CO2 and vacuolar storage of the resulting malic acid distinguish CAM Phase I. Diurnal deacidification and Rubisco-mediated fixation of endogenous CO2 in the light behind closed stomates distinguish CAM Phase III. Early morning transitional Phase II is characterized by uptake of atmospheric CO2 via open stomates and fixation by some combination of PEPCase/C4-cycle activity and Rubisco/C3-cycle activity. Finally, the late afternoon transitional Phase IV is also characterized by uptake of atmospheric CO2 via open stomates and fixation by some combination of PEPCase/C4-cycle activity and Rubisco/C3-cycle activity.
There are three central aspects to this diel cycle that challenge our ability to predict (and measure) the ϕ of CO2 fixation in CAM, starting with the temporal separation between nocturnal CO2 uptake in Phase I and diurnal photosynthetic CO2 fixation in Phase III. Because Phase III photosynthetic fixation of endogenous CO2 takes place under closed stomata, midday photosynthesis per se cannot be monitored in CAM plants by ‘instantaneous’ IRGA measurements as is routinely done with C3 and C4 plants. This also means that a full accounting of the energetic requirements for CO2 fixed during Phase III must include the ‘hidden’ energy costs of malate production and accumulation from the previous night (Winter and Smith, 1996).
A second set of challenges arises upon more careful consideration of the entire diel CAM cycle with particular emphasis on Phase II and Phase IV. Despite having open stomata during Phases II and IV, strong intercellular resistances to external CO2 diffusion in succulent CAM tissues still result in very low chloroplast CO2 concentrations (Maxwell et al., 1997; Nelson et al., 2005). Limiting CO2 availability at the Rubisco active site will necessarily lower carbon gain efficiency during these daytime transitional phases of CAM. Furthermore, the overall energetics of these transitional phases are not obvious since they can involve varying levels of PEPCase/C4-like activity and Rubisco/C3-like activity. For example, prolonged C4-like activity during Phase II will be expected to come at a greater quantum cost than Phase II activity dominated by Rubisco-mediated fixation (Roberts et al., 1997; Maxwell et al., 1999). In addition, there is potential for sequential co-occurring carboxylation by PEPCase and Rubisco during these transitional phases when both enzymes are active which would permit an energetically wasteful futile cycle (Leegood et al., 1997). Although complementary regulation of the two carboxylases is thought to minimize this possibility (Griffiths et al., 2002; Winter and Holtum, 2002), even modest or transient episodes of futile cycling would markedly lower the overall ϕ of C gain in CAM. In addition, the relative expression of the different phases of CAM is highly variable, responding to both developmental and environmental cues. In the extreme, facultative CAM plants like certain Clusia species or Mesembryanthemum crystallinum can fully switch from predominantly C3 photosynthesis to CAM in response to environmental factors such as water availability (Holtum and Winter, 1982; Cushman and Dodds, 2002; Lüttge, 2006). Constitutive CAM plants also exhibit remarkable plasticity for adjusting the absolute and relative amounts of carbon taken up during the day and night in response to various environmental factors including light availability (Nobel, 1982; Skillman et al., 2005). Strong internal diffusion resistances, the uncertain quantum costs of Phase II and IV, the possibility of futile cycling between the two carboxylases, and the highly variable contribution of different CAM phases to the overall daily carbon gain represents a second set of complications in trying to predict the maximum ϕ of CO2 fixation in CAM from physiological considerations.
The third challenge for predicting the ϕ in this photosynthetic type arises when considering the question of photorespiration during Phase III in CAM tissues. Drawing on the analogy of C4 photosynthesis, it is widely believed that carbon-concentrating via CAM minimizes PCO cycle activity in CAM plants during Phase III (e.g. Nobel, 1991; Black et al., 1996; Winter and Smith, 1996). This would improve the ϕ of CO2-gain in CAM compared with plants that were otherwise expending energy on photorespiration during Phase III. The conclusion of reduced Phase III photorespiration is supported by the early work of Spalding et al. (1979) who found that the intercellular O2:CO2 concentrations for a variety of CAM species during Phase III were approximately one-third that of ambient air. Assuming those intercellular concentrations are indicative of concentrations at the active site of Rubisco, Spalding et al. (1979) concluded that photorespiration should be largely suppressed during CAM Phase III. But more recent work by Maxwell and colleagues indicate that chloroplastic O2:CO2 concentrations are considerably more favourable for Phase III photorespiration than previously believed (Maxwell et al., 1997, 1998). Reviewing these studies and others, Lüttge (2002) concluded that the suppression of photorespiration during CAM phase III is incomplete and that there are likely situations such as severe drought where high rates of Phase III photorespiration may occur in CAM tissues. For now, Phase III photorespiration remains an unknown that further complicates our ability to predict the efficiency of photosynthesis in CAM. As a result of these complications, the realized ϕ of CAM can only be determined empirically by carrying out 24 h gas exchange studies under different limiting light levels. For purposes of assessing the generality of the CQY hypothesis, these 24 h ϕ values in CAM plants would only take on meaning if they were related directly to comparably measured values in C3 and C4 plants. Unfortunately, such comparative 24 h gas exchange data are not available.
From a physiological perspective, it can only be concluded that it is plausible that the maximum realized efficiency for light-driven carbon gain in CAM is, in fact, highly variable, possibly refuting the CQY hypothesis. From an evolutionary perspective, there are three possible causes for increased variation in a trait worth our consideration; there has been selection for high variability (phenotypic plasticity) for the trait of interest, there is reduced selection on the trait of interest, or the trait of interest is highly correlated with another trait that is under strong selection.
Has there been selection for high phenotypic plasticity in the ϕ of CAM plants? Physiological plasticity seems to be a central feature of CAM. Depending upon environmental and developmental conditions, the metabolic complexities described above may play out differently over each and every 24 h cycle (Dodd et al., 2002; Griffiths et al., 2002; Lüttge, 2006). But, this general pattern of high plasticity in CAM notwithstanding, it is difficult to imagine a situation wherein there would be any advantage to having a low or flexible ϕ per se that is unique to CAM plants. It is unlikely that the putative high variation in ϕ in CAM is a result of selection for greater plasticity in this trait.
Has there been reduced selection for maintaining a maximum realized ϕ in CAM plants? Among desert succulents that thrive in one of the few biomes where light is seldom a limiting factor, one might speculate that relaxed selection for efficient light use has allowed for more variation in this trait. However, there is no evidence among desert C3 plants for ϕ differences. Also, the similar ϕ of O2 production values in CAM and C3 plants argues against there being reduced selection for maintaining overall photosynthetic efficiency in CAM. Moreover, CAM diversity is greatest in the tropics (Gentry and Dodson, 1987; Zotz, 2004) where the efficient use of light in carbon gain should be at a premium (Yoda, 1974; Chazdon and Fetcher, 1984; Graham et al., 2003). It is unlikely that the putative high variation in ϕ in CAM is a result of relaxed selection on this trait.
If the ϕ in CAM is indeed highly variable, it probably represents a correlated response to selection for other traits. The fundamental compromise in carbon gain for all terrestrial vascular plants, regardless of photosynthetic pathway, is with water loss through open stomata. In C3 and C4 plants the balance between carbon gain and water loss can be regulated ‘in real time’ with changes in stomatal conductance. This allows for relatively rapid changes in gas exchange behavior for the optimization of both water use and light use. In CAM, the compromise between water loss and carbon gain is largely regulated through the differential expression of the different phases of CAM. The temporally variable pattern of carbon acquisition means that the effective efficiency of light use in carbon gain among CAM plants only emerges over the course of a 24 h cycle. Perhaps there is a trade-off between the expressed ϕ of 24 h carbon gain and the water use efficiency of 24 h carbon gain in CAM plants which is fundamentally different from that of C3 and C4 plants. Unfortunately, there are no available data from which to evaluate this suggestion.
Summarizing this section, available CAM ϕ data are insufficient for discerning whether or not plants in this photosynthetic pathway conform to predictions of the CQY hypothesis. A consideration of CAM from both a physiological and an evolutionary perspective suggests that this photosynthetic type may indeed prove to have a greater range in photosynthetic efficiencies as a result of the inherent flexibility of CAM which is regulated largely in the service of water conservation. Comparative 24 h gas exchange studies of CAM plants with C3 and C4 plants over a range of low light levels are needed to verify this interesting possibility.
Emerson and Warburg's great debate of 50 years ago was over a 2-fold difference in their respective estimates of the true maximum ϕ of photosynthetic oxygen production. Warburg (1958) interpreted his data to indicate a maximum ϕ of ∼0.25. Emerson estimated a maximum ϕ of ∼0.125 (Emerson and Lewis, 1941). Emerson's estimate has better stood the test of time and most questions regarding the maximum efficiency of photosynthesis seemed to have been settled as of 25 years ago. In particular, seminal studies by Ehleringer and Björkman (1977) and Björkman and Demmig (1987) showed that the maximum potential ϕ of light-driven CO2 fixation and O2 production in intact C3 and C4 leaves agreed well with expectations from the current understanding of the energetic mechanisms of these two different photosynthetic pathways. However, the present analysis of these and several other leaf-level ϕ studies reveals two remaining uncertainties in our understanding of in vivo photosynthetic efficiency. First, estimates presented here suggest that for both C3 and C4 plants, 20–25% of the absorbed energy cannot be accounted for in photosynthesis or photorespiration. Numerous metabolic processes besides the PCR and PCO cycles are known to rely upon photogenerated assimilatory power and, in principle, could explain this surprisingly high level of diverted energy. Unfortunately, the regulation and magnitude of the individual and integrated fluxes through these numerous other pathways are not well quantified, especially in low light where competition for assimilatory power would be the greatest. It would be interesting to know if this same level of energy diversion away from carbon gain also occurs in CAM species but, because of the second identified deficiency in our understanding of ϕ biology, we are a long way from being able to address this question with any confidence. Indeed, the best available estimates for the ϕ in CAM differ by a factor of ten (Fig. 5B), easily dwarfing the kinds of differences over which Emerson and Warburg debated those many years ago. There are clearly unavoidable and perhaps even essential inefficiencies in light-limited carbon gain in all plants, regardless of photosynthetic type. However, evolutionary theory predicts that any avoidable and non-essential reductions in photosynthetic efficiency will be continually winnowed from populations. Without further research, the inefficiencies in photosynthesis characterized in this review will continue to cast long shadows on our greater understanding of the evolutionary energetics of terrestrial photoautotrophy.
This paper is dedicated to Professors Gerry Edwards and Barry Osmond as they approach retirement. Each, in his own way, has long served as a distinguished leader in the study of photosynthesis. I am indebted to Ichiro Terashima for access to the pre-published research from his laboratory. This review has benefited from the thoughtful comments made by Rowan Sage and an anonymous reviewer.