Stomatal responses of C3, C3-C4 and C4Flaveria species to light and intercellular CO2 concentration: implications for the evolution of stomatal behaviour

Authors


Correspondence (current address): Travis Huxman, Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721-0088, USA. E-mail: huxman@email.arizona.edu

ABSTRACT

Stomatal function mediates physiological trade-offs associated with maintaining a favourable H2O balance in leaf tissues while acquiring CO2 as a photosynthetic substrate. The C3 and C4 species appear to have different patterns of stomatal response to changing light conditions, and variation in this behaviour may have played a role in the functional diversification of the different photosynthetic pathways. In the current study, we used gain analysis theory to characterize the stomatal conductance response to light intensity in nine different C3, C4 and C3-C4 intermediate species Flaveria species. The response of stomatal conductance (gs) to a change in light intensity represents both a direct (related to a change in incident light intensity, I) and indirect (related to a change in intercellular CO2 concentration, Ci) response. The slope of the line relating the change in gs to Ci was steeper in C4 species, compared with C3 species, with C3-C4 species having an intermediate response. This response reflects the greater relative contribution of the indirect versus direct component of the gs versus I response in the C4 species. The C3-C4 species, Flaveria floridana, exhibited a C4-like response whereas the C3-C4 species, Flaveria sonorensis and Flaveria chloraefolia, exhibited C3-like responses, similar to their hypothesized position along the evolutionary trajectory of the development of C4 photosynthesis. There was a positive correlation between the relative contribution of the indirect component of the gs versus I response and water use efficiency when evaluated across all species. Assuming that the C3-C4 intermediate species reflect an evolutionary progression from fully expressed C3 ancestors, the results of the current study demonstrate an increase in the contribution of the indirect component of the gs versus I response as taxa evolve toward the C4 extreme. The greater relative contribution of the indirect component of the stomatal response occurs through both increases in the indirect stomatal components and through decreases in the direct. Increases in the magnitude of the indirect component may be related to the maintenance of higher water use efficiencies in the intermediate evolutionary stages, before the appearance of fully integrated C4 photosynthesis.

INTRODUCTION

Stomatal function mediates one of the most important physiological trade-offs in vascular plants, maintaining a favourable H2O balance in leaf tissues while acquiring CO2 as a photosynthetic substrate (Cowan 1982; Zeiger, Farquhar & Cowan 1987). In controlling the rates of CO2 uptake and H2O loss, the stomatal guard cells must be responsive to a number of environmental variables, such as light (quantity and quality), intercellular CO2 partial pressure (Ci), humidity, soil water potential, and temperature (Zeiger 1983). A number of studies have examined specific aspects of stomatal response to these environmental factors, with an effort to discuss them in terms of optimization theory (Farquhar, Dubbe & Raschke 1978; Dubbe, Farquhar & Raschke 1978; Williams 1983; Morison & Gifford 1983). Recently, studies have focused on cellular and molecular mechanisms related to the transduction of these environmental variables into a stomatal response (Outlaw et al. 1996; Webb & Hetherington 1997). There is still a need to understand the nature of interactions among these variables in an intact system, including how they have been shaped by evolutionary trends toward optimization and how patterns of variation may have contributed to divergence in plant lineages.

The fact that stomata respond to changes in incident light intensity has been known since the mid-1800s (von Mohl 1856). Three mechanisms have been described as causes of the stomatal light response. First, the guard cells appear to possess light receptor pigments that directly detect changes in light intensity, and adjust stomatal conductance accordingly (Zeiger 1983) (a ‘direct’ response, see Wong, Cowan & Farquhar 1978). Second, the guard cells may indirectly sense a change in light intensity through a change in the intercellular CO2 partial pressure (Ci) that is, in turn, caused by a change in photosynthetic rate of the mesophyll tissues (an ‘indirect’ response, see Wong et al. 1978). This latter response is the so-called ‘CO2 feedback loop’, which affects both stomatal conductance and assimilation rate. The mechanism for this response is as yet unknown, but suspected to be related to the state of the apoplastic malate pool that may lead to stomatal response through activation of plasma membrane anion channels (Hedrich & Marten 1993; Hedrich et al. 1994). Third, the guard cells may indirectly respond to changes in photosynthesis rate through a transmissible chemical signal that is transported from the mesophyll cells to the guard cells (Wong et al. 1978) (a second possible ‘indirect’ response, although one that has not received much attention).

The development of a quantitative theory for separating the direct and indirect responses of stomatal conductance to light intensity has provided a sound basis for partitioning their contributions to the overall stomatal response (Cowan 1977; Wong et al. 1978). Substantial interspecific variability has been observed in the relative magnitudes of the direct and indirect responses. In Zea mays, a C4 monocot, studies have demonstrated a particularly strong contribution of the indirect response, observed as a high sensitivity of the stomatal conductance to changes in Ci (Sharkey & Raschke 1981). Similar results have been observed for the C4 species Amaranthus powellii (Dubbe et al. 1978) and Amaranthus hypochondriacus (Ramos & Hall 1982). In several of these studies, the high gain of the CO2 feedback loop resulted in the stomatal response to changes in light intensity which were dominated by the indirect response. In three separate comparative studies, C4 species exhibited greater sensitivity of stomatal conductance to changes in Ci than did C3 species (Dubbe et al. 1978; Sharkey & Raschke 1981; Ramos & Hall 1982). These results have prompted the hypothesis that in general, the indirect component of the stomatal response to light intensity, involving the CO2 feedback loop, will be greater in C4 species than in C3 species. However, in one study, the stomatal sensitivities to changes in Ci were shown to be similar in two C3 and two C4 grasses (Morison & Gifford 1983). Although the latter study did not specifically address the contribution of the CO2 feedback loop to the overall stomatal response to light, the results do complicate any attempts to generalize about the role that stomatal behaviour may have played in the functional diversification of photosynthetic pathways.

An examination of stomatal responses to light and CO2 in closely related species with different photosynthetic pathways seems warranted. This approach would allow us to separate differences related to phylogenetic history from those related to photosynthetic processes (Monson 1996). Such an effort would accomplish two goals, including development of a better understanding of the factors that regulate plant response to the environment, and identifying potential steps in the evolution of plants with different photosynthetic pathways. In the current study, we used gain analysis theory (Wong et al. 1978) to quantify the direct and indirect portions of the stomatal conductance response to light intensity in nine different plant species, including several C3, C4 and C3-C4 intermediate species in the genus Flaveria. The study provides us with a phylogenetically constrained comparison of stomatal response in plants with different photosynthetic pathways, which can be used to provide unique insight into the evolution of C4 photosynthesis (Ehleringer & Monson 1993; Monson 1996).

MATERIALS AND METHODS

Plant material

Plants of the following species were established in April 2000 from cuttings, 12 cm in length, of the same clones used in previous studies of C3-C4 intermediate photosynthesis in the genus Flaveria (see Ku et al. 1983; Monson et al. 1986) including; Flaveria cronquistii (C3), F. robusta (C3), F. pringlei (C3), F. sonorensis (C3-C4), F. chloraefolia (C3-C4), F. floridana (C3-C4), F. trinervia (C4), F. bidentis (C4), and F. vaginata (C4). These plants were grown for approximately a year in the glasshouse at the University of Colorado at Boulder with supplemental lighting provided from November 2000 to March 2001. The supplemental lighting extended the photoperiod to 12 h by providing 175 µmol  m−2 s−1 photon flux density (PPFD) (measured with a quantum sensor at the leaf level) in equal time blocks prior to sunrise and following sunset. During days of full sun, maximum light intensity in the glasshouse was 1300 µmol m−2  s−1. All plants were initiated in 2 L pots, with a peat : sand :  vermiculite : perlite (2 : 1 : 1 : 1) potting mix. Above-ground structures were cut back and plants were re-potted into 5 L pots in November 2000. Gas exchange sampling occurred on fully expanded leaves in the months of February and March 2001. The plants were watered regularly and fertilized once per week with full-strength Hoagland's solution (Hoagland & Arnon 1950).

Gas exchange measurements

Measurements of CO2 and H2O exchange were made on intact leaves using a portable open-flow gas exchange system (Li-6400; Li-Cor Inc., Lincoln, NE, USA). The Li-6400 has a controllable red–blue light source and a CO2-injection system to manipulate CO2 concentration in the cuvette. The measurements were conducted on one leaf of each species over a period of five hours between 1000 and 1500 h. Incident PPFD and ambient CO2 concentration were varied, but temperature was maintained constant at 30 °C and leaf-to-air vapour pressure deficit was maintained between 2·0 and 2·5 kPa.

During an initial 15 min period of acclimation in the gas exchange system for each leaf, the CO2 concentration was maintained at 340 µmol mol−2 CO2 with a light intensity of 2000 µmol m−2 s−1. Once steady-state gas exchange properties were observed, a PPFD response curve was performed with the following PPFD values: 2000, 1500, 1200, 1000, 800, 500, 300, 200, 100 and 50 µmol m−2 s−1. Leaves were allowed to acclimate to each PPFD level for a period of at least 4 min and then only after steady-state gas exchange properties were observed and logged, was the PPFD level in the cuvette changed.

Following the PPFD response curve, the PPFD level in the cuvette was increased to 500 µmol m−2 s−1. After a period of time not less than 10 min and once gas exchange properties returned to similar values as seen in the PPFD response curve for this particular light level, the CO2 concentration in the cuvette was decreased to either 100 µmol mol−1 (for C3 and C3-C4 species) or 75 µmol  mol−1 (for C4 species). The cuvette was maintained at these conditions until steady-state gas-exchange rates were observed and then the CO2 concentration of the cuvette was raised in step-wise fashion to 150, 200, 250, 300, 340, 400, 450, 500 and 600 µmol mol−1 CO2 with at least 4 min of acclimation at each CO2 step. At each point, the next step did not occur until steady-state gas exchange properties were measured.

In a previous study that we report here, we used the laboratory-based gas-exchange system described in (Monson et al. 1986) to enclose entire leaves of some species, and to measure their CO2 and H2O fluxes and stomatal conductance to H2O over a two-day period in response to a broad range of PPFD and ambient CO2 concentrations. The measurement protocol for these studies was chosen to closely approximate those used in previous studies (e.g. Wong et al. 1978; Ramos & Hall 1982). Between 1000 and 1500 h, on the first day of measurement, leaves were exposed to four step-wise changes in PPFD, beginning from the highest value (1500 µmol m−2 s−1) and progressing to the lowest value (100 µmol m−2 s−1). Leaves were maintained at each PPFD for 30 min before recording CO2 and H2O fluxes, as well as leaf temperature, air temperature and relative humidity. The gas-exchange system was programmed to keep leaf temperature at 30 ± 0·5 °C and the dew-point of the ingoing airstream was adjusted to maintain the leaf-to-air water vapour pressure difference at 1·5 kPa. After completing measurements across the entire range of PPFD values, the leaves were returned to the highest PPFD, and observed for verification that CO2 and H2O fluxes returned to within 10% of the starting values. After the first day of measurements, plants were returned to the glasshouse. Between 0900 and 1600 h on the second day, the same leaf was placed in the cuvette, and subjected to measurements of CO2 and H2O fluxes in response to four different ambient CO2 concentrations, at four different PPFD levels. This analysis was initiated by measuring steady-state CO2 and H2O fluxes at 340 µmol CO2 mol−1 air, followed by step-wise decreases to the lowest CO2 concentration, re-establishment of fluxes at 340 µmol mol−1, and step-wise increases to the highest concentration. Leaf temperature and leaf-to-air vapour pressure difference were maintained as described above.

Gain analysis

The overall change in stomatal conductance that is a function of light can be described by the following equation:

dgs/dI = (∂gs/∂I)Ci + (∂gs/∂Ci)I (dCi/dI)(1)

where dgs/dI is taken as the slope of the gs versus I response curve determined within the limit as dI → 0 at any given value of I, and (∂gs/∂Ci)I is taken as the slope of the relationship between gs and Ci assuming constant I, at the chosen I. The term dCi/dI is the slope of the relationship between Ci and I, again at the chosen I. In essence, Eqn 1 states that the overall slope (dgs/dI) is the sum of a partial direct slope (∂gs/∂I)Ci, and a partial indirect slope (∂gs/∂Ci)I. The ratio of (∂gs/∂I)Ci to dgs/dI represents the relative direct stomatal response to I, whereas the ratio of (∂gs/∂Ci)I to dgs/dI represents the relative indirect response of stomatal conductance to I. Each of these are presented as a percentage of the total stomatal response.

In past studies (e.g. Wong et al. 1978; Ramos & Hall 1982), the definition of the response slopes was relaxed to finite intervals of ΔI and ΔCi, and determined over hours to days in the form of gs versus Ci curves measured at several constant I levels. This approach carried the disadvantage of potentially introducing acclimation artifacts due to the many hours a single leaf remained in the gas-exchange cuvette, and precluded the analysis of more than one or two replicates or species because of the long time required for analysis. We have developed an approach to shorten the time required for this analysis. In our approach, as in past studies, we relax the theory to accommodate a measurable response of gs to a finite change in I. Furthermore, we make the simplifying assumption of linearity in the slope (ΔgsI) evaluated across the finite change in I. In terms of finite algebra then, Eqn 1 can be rewritten as:

image(2)

where Term I is taken as the measured linear slope of the gs versus I response curve within a small finite interval of changing I, Term III is taken as the measured linear slope of the gs versus Ci response curve within a small finite interval of changing Ci within the range observed at the chosen I, Term IV is taken as the measured linear slope of the Ci versus I response curve within the range observed at the chosen I, and Term II is calculated as an unknown variable.

One key to using this approach is to choose a ΔI that is large enough to provide an accurately measured response in gs, yet small enough to minimize errors inherent in the assumptions of a linear slope. We chose an interval of ΔI = 100 µmol m−2 s−1, taken between 550 and 450 µmol  m−2 s−1, which is within the typical range of midday growth PPFDs determined by periodic measurement of incident PPFD within the greenhouse. Thus, the measured slope (ΔgsCi)I that was used in determining Term III is representative of the normal operating Ci for each species. An empirical evaluation of this approach is provided in the Results section.

Data analysis

Responses of individual leaves from a single run were used to calculate the components of a stomatal response to light. Thus, individual runs represent the statistical unit of interest and ensemble means and variances are determined from multiple runs on plants from the same species. Differences between species and photosynthetic pathways were determined by the use of a manova with the following response variables in the model: ΔgsI; (ΔgsI)Ci; (ΔgsCi)I; ΔCiI. Testing for differences with respect to individual parameters was accomplished by analysis of variance. Differences in the response curves of gs to Ci for different species and photosynthetic pathways were determined by the methods of Potvin, Lechowicz & Tardif (1990). Where the trends are consistent across photosynthetic pathways, we used combined data, but where species differences exist, we highlight species variation.

RESULTS

The decrease in gs with decrease in I for C4 species appeared to be less than that for C3 species measured in this study (Fig. 1). The C3 plants exhibited greater variation among species in conductance over the range of I measured. The C3-C4Flaveria showed a response in gs to changing I that was intermediate to the C3 and C4 congeners. Variation in assimilation rate with changing light intensity and internal CO2 concentration was more consistent in a relative response across species and photosynthetic pathways (Figs 1 & 2). The operating Ci at a light intensity of 500 µmol m−2 s−1 produced assimilation rates that were not CO2 saturated, as indicated by increases in assimilation with increasing Ci.

Figure 1.

The relationship between stomatal conductance (upper panel) or photosynthetic rate (lower panel) and light intensity for nine species of the genus Flaveria with different photosynthetic pathways. Stomatal conductance and photosymthetic rate plotted as relative values, where the maximum value at 800 µmol m−2 s−1 light is used to standardize the remaining numbers as light intensity is decreased. The left panel contains C3 species, the middle panel contains the C3-C4 intermediate species and the right panel is the C4 species. Data are plotted as means ± 1 SD.

Figure 2.

The relationship between assimilation rate and internal CO2 concentration for nine species of the genus Flaveria with different photosynthetic pathways. Net assimilation rate is plotted as a relative value, where the maximum value the highest internal CO2 concentration is used to standardize the remaining numbers. The left panels contain C3 species, the middle panels contain the C3-C4 intermediate species and the right panels are the C4 species. All individual data points are plotted from replicate curves. The arrow in each panel identifies the operating internal CO2 concentration.

The relationship between gs and Ci varied across the different species of Flaveria (Fig. 3). Using a best-fit regression approach, it was determined that in both C3 and C4 species, the response of gs to Ci was best described by a linear function. The slope of the line relating the relative change in gs (as compared to a maximum value at low Ci) to Ci was steeper in C4 species, compared to C3 species. The overall relationship for C3 and C4 species was significantly different (F = 40·4; d.f. = 2,76; P < 0·01). The relationship between gs and Ci for the C3-C4 species was not significantly different from that of the C3 species (F = 1·10; d.f. = 2,76; P = 0·31), but was significantly different than the C4 species response (F = 54·4; d.f. = 2,76; P < 0·01). Whereas each of the relationships for the C3 and C4 species was best characterized by a linear regression model, the C3-C4 response appears non-linear across the entire range. At relatively low Ci, individual data points tend to be within the 95% confidence intervals of the generalized C3 relationship, but at higher Ci, the individual data points tend toward a generalized C4 slope and some data extend to within the 95% confidence intervals of the C4 relationship. This pattern is most obvious in the stomatal responses of the C3-C4 species, F. floridana.

Figure 3.

The relationship between stomatal conductance and internal CO2 concentration for nine species of the genus Flaveria with different photosynthetic pathways. Stomatal conductance is plotted as a relative value (relative to the maximum value for each species at low Ci). The top panel is a comparison of C3 species (open symbols) versus C4 species (closed symbols). The bottom panel illustrates the C3-C4 species response relative to the general trend of either C3 or C4 species. The predicted values for the regression lines for both C3 and C4 species are presented (y = −0·0010x + 1·06, r2 = 0·45 and y = −0·0014x + 0·99, r2 = 0·56, respectively), along with their respective 95% confidence intervals for comparative purposes. Data plotted are the C3-C4 intermediate species.

From the responses to I and Ci, and using gain analysis, we evaluated the magnitudes of the indirect and direct components of the gs versus I response. It is important to note that the full range of values in Figs 1, 2 and 3 were constrained to points providing a linear relationship around the operating conditions of each species at a PPFD of 500 µmol m−2 s−1 and a CO2 concentration of 340 µmol  m−2 s−1 to determine the slope for Eqn 2. There were significant differences among species with different photosynthetic pathways (Wilk's Lambda = 0·349; d.f. = 1,8; P < 0·05). The C4 and C3-C4 species exhibited lower sensitivity to changes in I when evaluated across the entire range of I. For the C3-C4 and C4 species, within the range 450–550 µmol m−2 s−1, the less sensitive response, compared to C3 species, is due to slight changes in both the calculated direct response (ΔgsI)Ci, and the two measured components of the indirect response (ΔgsCi)I and ΔCiI (Tables 1 & 3).

Table 1.  Analysis of variance for different stomatal characteristics determined for C3, C4 and C3-C4 intermediate Flaveria species by gain analysis. Included are the overall change in stomatal conductance (ΔgsI), the direct response of stomata to a change in light ((ΔgsI)Ci), and the two components of the indirect response (ΔCiI and (ΔgsCi)I)
Parameterd.f.FP
ΔgsI2,292·790·07
gsI)Ci2,299·2<0·01
gsCi)I2,292·770·08
ΔCiI2,294·60·01
Table 3.  Mean responses (± 1 SD) for the different species and photosynthetic pathways as determined from gain analysis of stomatal behaviour in Flaveria species. Included are the overall change in stomatal conductance (ΔgsI), the direct response of stomata to a change in light ((ΔgsI)Ci), and the two components of the indirect response (ΔCiI and (ΔgsCi)I). It is important to note that the indirect response is a combination of the previous two parameters
SpeciesStomatal parameters
ΔgsIgsI)CigsCi)IΔCiI
  1. Significant differences between photosynthetic pathway means are indicated by different upper case letters before a value in a row.

C4A.0·087 ± 0·03A.0·005 ± 0·01A.−0·39 ± 0·17A.−0·23 ± 0·05
 F. vaginata  0·096 ± 0·028  0·007 ± 0·01 −0·33 ± 0·08−0·26 ± 0·01
 F. bidentis  0·044 ± 0·025  0·003 ± 0·01 −0·15 ± 0·05−0·24 ± 0·08
 F. trinervia  0·129 ± 0·030  0·005 ± 0·01 −0·88 ± 0·40−0·14 ± 0·01
C3-C4A.0·081 ± 0·03 B.0·054 ± 0·11A.−0·18 ± 0·07A.−0·64 ± 0·04
 F. floridana  0·050 ± 0·020  0·013 ± 0·01 −0·17 ± 0·04−0·19 ± 0·03
 F. sonorensis  0·126 ± 0·011  0·103 ± 0·01 −0·23 ± 0·08−0·11 ± 0·02
 F. chlorafolia  0·059 ± 0·023  0·035 ± 0·01 −0·12 ± 0·05−0·19 ± 0·03
C3B.0·151 ± 0·06C.0·108 ± 0·05A.−0·26 ± 0·10B.−0·15 ± 0·04
 F. cronquitii  0·207 ± 0·010  0·154 ± 0·07 −0·28 ± 0·13−0·16 ± 0·45
 F. pringlei  0·184 ± 0·055  0·114 ± 0·04 −0·38 ± 0·11−0·20 ± 0·02
 F. robusta  0·133 ± 0·010  0·111 ± 0·02 −0·21 ± 0·05−0·10 ± 0·02

A similar perspective of the stomatal response to I emerges when interspecific comparisons are made. When combined, the variation in the parameters resulted in significant differences for the nine Flaveria species (Wilk's Lambda = 0·013; d.f. = 1,36; P < 0·05). For the overall stomatal response, ΔgsI there were no differences at the species level, but significant variation occurred between photosynthetic groups. For the change in Ci associated with the change in I, ΔCiI, significant variation in the data were identified by analysis of variance, but pairwise differences were not present between any group of species (Tables 2 & 3). The significant effect of species on (ΔgsCi)I only resulted in a single pairwise difference. The C4 species, F. trinervia, has a significantly more sensitive and negative response, in comparison with all other species. The overall difference among the species in stomatal parameters was driven by changes in the direct component of the response (ΔgsI)Ci. The C3-C4 and other C4 species, F. vaginata, F. bidentis, F. trinervia and F. floridana, had significantly lower direct responses to I, compared to the C3 species F. cronquistii, F. pringlei and F. robusta.

Table 2.  Analysis of variance of the different stomatal characteristics determined for nine species of Flaveria with different photosynthetic pathways by gain analysis. Parameters are the same as those presented in Table 1
Parameterd.f.FP
ΔgsI9,220·920·52
gsI)Ci9,222·570·03
gsCi)I9,227·5<0·01
ΔCiI9,222·640·03

The overall response of gs to Ci in the C4 species was steeper than that for the C3 species (Fig. 3), reflecting the potential for a higher relative contribution of the indirect component of the PPFD response. The C3-C4 species, F. floridana and F. chloraefolia, exhibited gs versus Ci responses that tended to be less sensitive than the C3 species at low Ci, but more sensitive that the C3 species at high Ci, with the switch point being near the range of normal operating Ci. The C3-C4 species, F. sonorensis, exhibited a gs versus Ci response similar to the C3 species. When assessed solely within the range of the normal operating Ci, the portion of the stomatal response to I that is due to the indirect component was significantly greater in the C4 species, F. vaginata, F. bidentis, and F. trinervia, in comparison with the C3 species, F. cronquistii, F. pringlei and F. robusta (Fig. 4). The C3-C4 species, F. floridana, exhibited a C4-like indirect response, whereas the C3-C4 species, F. sonorensis and F. chloraefolia, exhibited C3-like indirect responses. The pattern of variation in the relative indirect response did not correlate with the variation among species in the normal operating Ci or gs, consistent with the hypothesis that variation in the indirect response is a function of photosynthetic pathway.

Figure 4.

The operational internal CO2 concentration, stomatal conductance and percentage indirect stomatal response to a hypothetical change in light intensity for nine species of the genus Flaveria with different photosynthetic pathways. Dark bars are C4 species, grey bars are C3 species and open bars are C3-C4 intermediates. Data plotted are means + 1 SE.

We observed a significant correlation (y = 29·3x − 64; r2 = 0·57) between the relative contribution of the indirect component of the gs versus I response and water use efficiency (Fig. 5). The correlation was founded primarily on differences between the C3-C4 and C4 species. The C3-C4 intermediate species exhibited lower water use efficiencies than the C3 species, but comparable contributions of the indirect component of the gs versus I response.

Figure 5.

The relationship between the percentage indirect response to a hypothetical change in light intensity and water use efficiency at the operational Ci (calculated as in Ehleringer & Osmond 1989) and a light intensity of 500 µmol m−2 s−1 for nine species of the genus Flaveria. Symbols for species are listed on the figure, with closed symbols being C4 species, open symbols being C3 species, and shaded symbols being C3-C4 intermediates.

In order to evaluate our empirical approach for partitioning the direct and indirect components of the gs versus I response, we compared the results of our revised protocol to a data set collected previously, which was conducted according to the traditional protocol that requires two days of measurements (e.g. Wong et al. 1978; Ramos & Hall 1982). The traditional protocol was conducted for sample of the same species that were used in the current study [(F. linearis (C3-C4), P. vulgaris (C3), F. floridana (C3-C4), F. trinervia (C4), and Z. mays (C4)]; however, the traditional protocol was conducted several years prior to implementation of the newer protocol, and the plants were grown under slightly different conditions (in a glasshouse during the winter with constant supplemental light from metal halide lamps). Stomatal parameters, estimated from both methods, were similar in form and relative rank for the different photosynthetic pathways. For example, the stomatal response of the C3 species was dominated by the direct effect of light regardless of technique. The contribution of the indirect component to the overall response of gs to I was between 1 and 10% for the traditional technique and between 10 and 40% for the current protocol. The C3-C4 species had slightly greater indirect responses (from 20 to 80% of the total response of gs to I for the new method and 10 to 15% for the two C3-C4Flaveria species in the older protocol). The two C4 species had the highest indirect response (80 to 90% of the total response of gs to I for the new method and 30 to 35% for the old protocol). Both methods identified the trend of greater (ΔgsI)Ci in C3 as compared to C3-C4 and C4 species, along with a similar trend in the overall response (ΔgsI). However, the older method consistently estimated values for the direct component that were up to two times greater than those in the new method. The differences in the calculated percentages of the direct and indirect responses between the two methods occurred as a result of proportionately greater values of ΔgsI, estimated from the old protocol for C4 and C3-C4 species. These differences may have been due to plant-growth differences between the two experiments, differences in gas-exchange measurement protocol, and even differences in the equations used to calculate gas-exchange parameters. Although the purpose of the current study was not to stringently compare the methods for empirical evaluation of stomatal gain theory, we were interested in learning whether the general conclusions concerning C3 and C4 species and their stomatal responses to light and CO2, which were derived from the new methods would hold if evaluated using the older method. In fact, the conclusions appear to hold in form, but not with quantitative precision.

DISCUSSION

In this study we evaluated the stomatal responses of different species of the genus Flaveria to changes in light intensity, focusing on the relationship between a direct response that is presumably mediated by photosensitive transducers in the guard cells, and an indirect response that is mediated through changes in intercellular CO2 concentration. With regard to the direct component, clear differences emerged among the species, with most of the differences segregating according to photosynthetic pathway type (Table 3). The C3 species exhibited the strongest contribution to the overall gs versus I response from the direct component followed by the C3-C4 species and finally, the C4 species. In fact, in all three C4 species, less than 20% of the total stomatal response to I could be explained by the direct component.

With regard to the indirect component, there was more variation among representatives of the different photosynthetic pathway types, and clear patterns were more difficult to distinguish. For example, the C4 species, F. bidentis, exhibited a weaker response to changes in Ci than the C3 species, F. pringlei and F. cronquistii (Table 3); even though the weaker response in the C4 species explained most of the overall stomatal response to I in this species. The C3-C4 species, F. floridana, exhibited a stronger response to changes in Ci than all three of the C3 species, whereas the other two C3-C4 species exhibited a weaker response.

If we assume that the C3-C4Flaveria species represent evolutionary intermediates that have evolved from fully expressed C3 ancestors (Monson 1989a, 1999; Monson & Moore 1989), then the results of the current study demonstrate a clear evolutionary reduction in the contribution of the direct, photosensitive component of the gs versus I response, as taxa evolve toward the C4 extreme. Of the C3-C4 species we examined, F. floridana has been found to be the most C4-like in past studies (Monson et al. 1986; Ku et al. 1991). It is also the species with the greatest reduction in the contribution of the direct component and the greatest increase in the contribution of the indirect component, compared to C3 species. Despite these C4-like features in the stomatal response to I and Ci, F. floridana is strongly C3-like in many features, including its normal operating Ci value and 13C/12C stable isotope ratios, and it has poorly compartmentalized C4 metabolism (Monson et al. 1986, 1988; Monson 1989b). The results of this study provide clear evidence that evolutionary changes in the response of stomata to I and Ci, which normally distinguish C3 and C4 dicot species, occur well before the evolution of fully expressed C4 metabolism. This conclusion, and past studies that have demonstrated the lack of full biochemical development of the C3-C4 species (Monson et al. 1984; Moore, Ku & Edwards 1987; Moore et al. 1988), are taken as compelling evidence that the evolution of C4 photosynthesis is not a simple single step in the development of novel plant function, but a syndrome of traits that tends to form in sequential phases over evolutionary time (Edwards et al. 2001).

The results of the current study provide clear evidence that in evolving the C4 syndrome, a change occurs in the way that stomata sense light and Ci. There are possible advantages of both a weaker direct component and a stronger indirect component in the response of gs to I in C4 leaves. The CO2-concentrating mechanism of C4 leaves allows them to achieve higher quantum efficiencies than C3 leaves (Long 1999), and the higher quantum efficiencies are largely independent of Ci (Ehleringer & Monson 1993). Thus, as I increases, and assuming similar investments in the production of ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco), concomitant increases in CO2 assimilation rate (A) will be larger for C4 leaves than C3 leaves, but the increases in A will be less constrained by the response of gs to I. The strong dependence of quantum yield on Ci in C3 leaves, requires that gs increase proportionately with A as I increases, or accommodate the requisite compromise of lower A. This means that in evolving from the ancestral C3 state, the direct component of the stomatal response to I could be relaxed in C4 leaves, allowing for the maintenance of lower Ci and higher water use efficiencies as I increases. The CO2-concentrating mechanism of C4 leaves, also allows them to achieve higher carboxylation efficiencies. Thus, smaller changes in Ci will have a greater effect on A in C4 leaves than C3 leaves. A stronger stomatal sensitivity to Ci may be advantageous for C4 leaves, allowing them to mitigate the effects of changes in Ci on A. In evolving a stronger sensitivity to Ci, C4 plants would most likely exhibit increases in the indirect component of the response of gs to I. To a large extent, this complex interaction of photosynthetic mechanisms can be summarized in an analysis of trade-offs between photosynthetic nitrogen use efficiencies and water use efficiencies. The direct response of gs to I has the primary effect of maximizing the photosynthetic nitrogen use efficiency of leaves by maximizing the amount of CO2 substrate available for use by Rubisco. The indirect response of gs to I has the primary effect of maximizing photosynthetic water use efficiency by minimizing the Ci required to achieve a certain A. The evolution of the CO2-concentrating mechanism in C4 leaves provides for a way of maximizing photosynthetic nitrogen use efficiency without increases in the response of gs to I, allowing C4 taxa to evolve stomatal mechanisms that more strongly respond to Ci, and less to I, thus maximizing the water use efficiency.

Recently, it was shown that elevated levels of leaf abscisic acid (ABA) can cause the stomata of C3 leaves to exhibit increased sensitivity to Ci, similar to the natural response of C4 leaves (Franks & Farquhar 2001). It is possible that the causal evolutionary event that produces C4-like stomatal sensitivity to Ci involves the constitutive up-regulation of leaf ABA levels, changes in the intraleaf compartmentation of ABA, or changes in the sensitivity of guard cells to ABA. Considering the role that ABA plays in the signal transduction of environmental stimuli and stomatal response (Franks & Farquhar 2001; Schroeder, Kwak & Allen 2001), it is not inconceivable for such a simple change in ABA sensitivity to occur and be maintained in conjunction with C4 photosynthetic biochemistry.

The results of our studies are similar to several past studies that have shown higher sensitivity of gs to Ci, and a concomitantly greater contribution of the CO2 feedback loop to the overall response of gs to I, in C4 species (Dubbe et al. 1978; Sharkey & Raschke 1981; Ramos & Hall 1982). Our results as well as those of the past studies stand in contrast, however, to those reported by Morison & Gifford (1983). The latter study focused on C3 and C4 monocot species. It is possible that differences among the studies are related to differences between monocot and dicot C4 species. The phylogenetically constrained group of species used in the present study has provided us with the opportunity to interpret interspecific differences in stomatal function, and evolutionary trajectories, without complication of markedly divergent phylogenetic history. Future studies of similarly constrained groups of plants will be required to fully interpret the relationship between photosynthetic pathway type and stomatal response to the environment.

ACKNOWLEDGMENTS

The authors thank Jed Sparks and Kim Sparks for helping with establishment of the Flaveria cuttings, and Erik Hamerlynck, Todd Rosentsiel and Peter Harley for discussions and advice. Much of the experimental approach that was used in this study was designed during a trip in 1984 by R. Monson to the University of California, Riverside to work with Anthony Hall, who contributed greatly to the ideas and approach. Subsequent discussion with Graham Farquhar (Australian National University) helped focus some of the critical issues. We are grateful for the contributions of these two scientists. Two anonymous referees provided critiques that significantly improved the contents of this manuscript.

Received 12 June 2002;received inrevised form 12 August 2002;accepted for publication 15 August 2002

Ancillary