Elevated CO2 effects on mesophyll conductance and its consequences for interpreting photosynthetic physiology



    Corresponding author
    1. Department of Plant Biology, University of Illinois Urbana-Champaign, 265 Morrill Hall, 505 S. Goodwin Avenue, Urbana, IL 61801, USA and
      Eric L. Singsaas, (present address) University of Wisconsin – Stevens Point, Department of Biology, Room 167 C.N.R. Building, Stevens Point, WI 54481, USA. Fax: +1 715 346 3624; e-mail: Eric.Singsaas@uwsp.edu
    Search for more papers by this author
  • D. R. ORT,

    1. Department of Plant Biology, University of Illinois Urbana-Champaign, 265 Morrill Hall, 505 S. Goodwin Avenue, Urbana, IL 61801, USA and
    2. USDA/ARS Photosynthesis Research Unit, 190 ERML, 1201 W. Gregory Drive, Urbana, IL 61801, USA
    Search for more papers by this author

    1. Department of Plant Biology, University of Illinois Urbana-Champaign, 265 Morrill Hall, 505 S. Goodwin Avenue, Urbana, IL 61801, USA and
    Search for more papers by this author

Eric L. Singsaas, (present address) University of Wisconsin – Stevens Point, Department of Biology, Room 167 C.N.R. Building, Stevens Point, WI 54481, USA. Fax: +1 715 346 3624; e-mail: Eric.Singsaas@uwsp.edu


Mesophyll conductance (gm) generally correlates with photosynthetic capacity, although the causal relationship between the two is unclear. The response of gm to various CO2 regimes was measured to determine its relationship to environmental changes that affect photosynthesis. The overall effect of CO2 growth environment on gm was species and experiment dependent. The data did not statistically differ from the previously shown Agm relationship and was unaffected by CO2 treatment. The consequences of the CO2 effect on gm for interpreting photosynthesis in individual cases were investigated. Substantial effects of assumed versus calculated gm on leaf properties estimated from gas-exchange measurements were found. This differential error resulted in an underestimation in ratio of maximum carboxylation to electron transport, especially in plants with high photosynthetic capacity. Including gm in the calculations also improved the agreement between maximum carboxylation rates and in vitro Rubisco measurements. It is concluded that gm is finite and varies with photosynthetic capacity. Including gm when calculating photosynthesis parameters from gas-exchange data will avoid systematic errors.


net photosynthetic CO2 assimilation


[CO2] at the site of carboxylation inside the chloroplast


[CO2] inside the leaf airspaces


Free Air CO2 Enrichment


mesophyll conductance


maximum potential rate of RuBP regeneration


light harvesting complexes



Vc max

maximum potential rate of RuBP carboxylation.


A better understanding of the mechanism by which elevated atmospheric CO2 affects photosynthesis is necessary to predict plant responses to future environments (Long 1998). Growth at elevated CO2 frequently leads to a reduction in photosynthetic capacity (e.g. Curtis 1996; Drake, Gonzàlez-Meler & Long 1997). This reduction is often quantified by monitoring changes in the maximum rates of RuBP carboxylation (Vc max) and regeneration (Jmax) through leaf gas-exchange measurements (e.g. Sage 1994) and is considered an acclimation response (Gunderson, Norby & Wullschleger 2000). These changes are part of a cascade of acclimation responses to growth at elevated CO2 affecting the biochemistry, morphology, and phenology of the plant (Curtis & Wang 1998). Measurements of the effect of elevated CO2 on mesophyll conductance (gm) are scarce. Mesophyll conductance (Harley et al. 1992; Loreto et al. 1992) is a measure of the transfer capacity of CO2 between the leaf internal airspaces and the site of carboxylation in the chloroplast and is a fundamental property of leaves that may influence photosynthetic capacity (Epron et al. 1995; Evans & Loreto 2000). Changes in gm may contribute to the acclimation of photosynthesis to elevated CO2. The primary goal of this study was to investigate the effects of growth at different ambient CO2 levels on gm and its relationship with photosynthetic capacity.

The secondary goal of this study was to investigate the relationship between in vitro biochemical and in vivo gas-exchange measurements of photosynthetic CO2 acclimation, and to determine any possible differential effects of a change in gm on the estimations of photosynthetic acclimation. This potential for error results from the common use of Ci as a basis for calculations of photosynthetic parameters such as Vc max and Jmax even though these parameters are defined based on the CO2 at the site of RuBP carboxylation (Cc; Farquhar, von Caemmerer & Berry 1980). This interchangeable use of Ci and Cc carries with it the implicit assumption that gm is effectively infinite; any effect of growth CO2 on gm may alter the interpretation of CO2 acclimation studies. For example, if two leaves differed only in mesophyll conductance, the difference between Ci and Cc will be greater for the leaf with the lower gm. Thus the initial slope of the photosynthesis–Ci relationship would be lower in this plant even when the photosynthesis–Cc relationship is identical between the two. This would make the calculated apparent Vc max erroneously lower for the plant with lower gm. Conversely, if a leaf has a lower Vc max than its counterpart, a compensating decrease in gm would erroneously increase its apparent Vc max (measured based on Ci), and mask changes in the underlying biochemistry.

These experiments were designed to determine whether growth CO2 substantially affects gm, and whether using Cc rather than Ci influences the calculation of photosynthetic parameters and the interpretation of their response to elevated CO2. We made gas-exchange measurements on trees growing in two different elevated CO2 experiments. These included a complete set of gas-exchange and chlorophyll fluorescence measurements to quantify gm using the constant J method (Loreto et al. 1994). Further measurements were made on herbaceous plants grown in environmental chambers at elevated CO2. We analysed the CO2 effects on gm and on the relationship between photosynthetic capacity (assessed as both net photosynthesis rate at a standard Ci as well as by Rubisco content) and gm. We used gm measurements to re-analyse gas-exchange derived measurements of photosynthesis, Vc max and Jmax, to compare the estimates based on Ci and Cc. We also investigated the relationship between acclimation to CO2 treatments as calculated from gas-exchange measurements compared with those measured by biochemical methods.


Plant growth conditions

Mesophyll conductance was estimated on sweetgum (Liquidambar styraciflua L.) and aspen (Populus tremuloides L.) trees growing under field conditions but exposed to elevated and ambient levels of CO2. In the FACTS-1 experiment near Chapel Hill, NC, sweetgum trees had naturally sprouted in the understorey of a 17-year-old Loblolly pine (Pinus tadaea L.) plantation. Three 30-m-diameter plots in this plantation are continuously fumigated with CO2 to raise the ambient CO2 levels to 200 µmol mol−1 above atmospheric levels (to about 560 µmol mol−1). An additional three rings were fully instrumented to serve as controls. The treatment had been applied at the site for 3 years at the time of measurements. Additional details on the site are provided elsewhere (DeLucia et al. 1999; Singsaas, Ort & DeLucia 2000). Field measurements on aspen were made at the FACTS-II field site in Rhinelander, WI, which had an array of treatment and control fumigation rings similar to those at FACTS-I. Aspen trees were planted as 6-month-old rooted cuttings propagated from greenhouse stock. All trees were between 2 and 3 m tall and approximately 3-year-old at the time of measurement. Additional details on the FACTS-II experiment can be found in Dickson et al. (2000).

To study the CO2gm relationship under a greater range of CO2 conditions, we performed additional experiments on potted plants grown indoors. Linden bean (Phaseolus vulgaris L. var. Linden), cucumber (Cucumis sativus L.), and spinach (Spinacia oleracea L.) were grown in controlled-environment chambers (Model PGW36; Conviron, Winnipeg, Manitoba, Canada). Light was provided by high-intensity fluorescent lamps and averaged 530 µmol m−2 s−1 at 1 m above the floor throughout the experiment. Ambient temperature averaged 26.4 °C throughout the experiment. Temperature and illumination in both chambers were monitored weekly with a thermocouple (Type T; Omega Inc., Stamford, CT, USA) and quantum photometer (Model LI-189; Li-Cor Inc., Lincoln, NE, USA).

Ambient CO2 and dewpoint were monitored in the growth chambers by an automated measurement system. Air was pumped from the chamber to a valve system that allowed airflow from each chamber to be alternately sent through an infrared gas analyser (Model 6262; Li-Cor Inc.). Ambient CO2 was elevated in the one chamber by injection of 100% CO2 through a valve controlled by a feedback loop based on the sampled CO2. A data logger (Model CR10x; Campbell Scientific, Logan, UT, USA) was used for system control. The elevated CO2 chamber was maintained at 745 µmol mol−1 during the first and second experimental blocks and 737 µmol mol−1 during the third. Because both cabinets were located in a small room, the lower CO2 treatment was 478 µmol mol−1 in the first and second blocks and 501 µmol mol−1 in the third. For simplicity in data reporting, we have labelled the treatments 750 and 500 µmol mol−1, respectively. To further minimize any chamber effects other than CO2, plants and CO2 control systems were switched weekly between the two chambers.

Plants were germinated from seed in 3 L pots, watered daily to maintain adequate soil moisture, and fertilized weekly with approximately 500 mL full strength Hoagland's solution. Plants were grown for 4 to 6 weeks before measurements began and then all measurements were made within 1 week. The youngest fully expanded leaf was measured in all cases.

Photosynthesis measurements

Gas-exchange measurements were made at the FACTS-I site on shaded sweetgum leaves < 3 m from the ground and on fully sunlight leaves in the upper canopy accessed from canopy towers and hydraulic lifts. At the FACTS-II site, leaves were selected for measurement from the top 1 m of the canopy. On chamber-grown plants, measurements were made on the youngest fully expanded leaf during the measurement period that began 6 weeks after germination.

Measurements were made using an open gas-exchange system (Model LI-6400; Li-Cor Inc.) where the partial pressure of CO2 in the cuvette (Ca) was controlled using a CO2 injection system controlled by the instrument. Chlorophyll fluorescence (ΔF/Fm′) was measured simultaneously using a portable pulse-modulated fluorometer (Model OS-500; OptiScience Corporation, Tyngsboro, MA, USA). Light was provided using a 100 W metal halide lamp attenuated to the desired PPFD with neutral-density filters. The PPFD levels were selected to provide saturating light without causing photo-inhibition, by comparison with A versus PPFD measurements made separately before each experiment cycle (data not shown). Saturating PPFD was 2000 µmol m−2 s−1 for leaves grown in full sunlight, 800 µmol m−2 s−1 for leaves grown in shade, and 1500 µmol m−2 s−1 for leaves grown in controlled-environment chambers. Because of the need for an external lamp when making combined gas-exchange and chlorophyll fluorescence measurements, leaves at each of the FACTS sites were excised with the petiole submerged in distilled water and brought to the measurement apparatus inside a portable laboratory trailer. These measurements were used to calculate gm.

Two input parameters were required to calculate gm; the rate of mitochondrial respiration (Rd) and the CO2 compensation point in the absence of mitochondrial respiration (Γ*). These were calculated from the common intersection points of three A versus Ci response curves (Laisk 1977; Brooks & Farquhar 1985; Villar, Held & Merino 1994, 1995). Briefly, the CO2 response of photosynthesis was measured at five points below a Ci of 200 µmol mol−1. Three such curves were measured at different PPFD levels (150, 100, and 75 µmol m−2 s−1). The parameters are determined from the co-ordinates of the intersection point of the three lines on a graph of A versus Ci. Measurements for these parameters were made in situ using attached leaves at both of the FACTS sites and in the environmental chambers. Average Γ* values (µmol mol−1) determined for each species and used in subsequent calculations were: 42.49 ± 2.9, 46.83 ± 3.0, 42.52 ± 2.1, 41.38 ± 1.2, 43.02 ± 1.7, and 36.8 ± 0.8 for sweetgum sun leaves, sweetgum shade, aspen, bean, cucumber, and spinach, respectively. From the same calculations, daytime respiration (Rd) values (µmol m−2 s−1) were −0.92 ± 0.2, −0.32 ± 0.1, −3.73 ± 0.7, −1.74 ± 0.2, −1.16 ± 0.2 and −2.07 ± 0.1 for each species, respectively.

Leaf gas-exchange parameters were calculated using the equations of von Caemmerer & Farquhar (1981). The maximum rate of RuBP carboxylation (Vc max) and RuBP regeneration (Jmax) were calculated from CO2 response curves with 11 points measured at ambient CO2 levels between 1000 and 20 µmol mol−1. These data were fit to the Farquhar et al. (1980) model by non-linear least squares regression as described in Harley & Tenhunen (1991). We used in vitro model constants from Harley & Baldocchi (1995). We made a further comparison of approaches to determine the correct value of Vc max by re-fitting the ACi data using in vivo Rubisco parameters (Bernacchi et al. 2001). To compare photosynthesis rates among blocks on an equal basis, we standardized the photosynthesis measurements by choosing light-saturated photosynthesis measured at a Ci of approximately 400 µmol mol−1.

Mesophyll conductance was calculated using the constant J method (Loreto et al. 1992). Data were selected from CO2-response measurements in the region where ΔF/Fm′ was constant with increasing CO2. The gas-exchange data (A, Ci) and constants (Γ* and Rd; determined separately for each experimental block and species) were used to calculate the rate of electron transport needed to support CO2 assimilation and photorespiration, Jp, for each point (Loreto et al. 1992). Electron transport through PSII was monitored independently from Jp using chlorophyll fluorescence measurements ΔF/Fm′ and is referred to as Jf. The variance in Jp across all the points of known constant Jf was calculated as described in Harley et al. (1992), using the Γ* and Rd values calculated previously for that experimental block. Conductance values were determined by least-squares regressions, minimizing the variance across the selected data points by substituting values of gm into the Jp calculations. We report gm in units of mol m−2 s−1 bar−1 to remain consistent with units in most publications (Harley et al. 1992; Loreto et al. 1992, Evans & Loreto 2000).

Rubisco and chlorophyll

We measured Rubisco activity using a NADH-linked enzyme assay modified from Sharkey, Savitch & Butz (1991). Leaf punches (1.7 cm2) were excised with a cork borer and immediately ground in extraction buffer in a ground-glass tissue homogenizer at 0 °C. The extraction buffer contained 100 mm bicine–NaOH (pH 7.8), 100 mm Na2B4O7, 20 mm MgCl2, 1 mm ethylenediaminetetraacetic acid (EDTA), 4 mm amino-N-caprioic acid, 0.8 mm benzamidine, 0.1% (w/w) Triton-X-100, 0.02% (w/v) bovine serum albumin, 150 mm NaHCO3, 5 mm dithiothreitol (DTT), and 30 mg poly(vinylpolypyrrolidone) (insoluble). The crude extract was transferred to a 1.5 mL microcentrifuge tube and spun for 30 s. Initial activity was measured using 10 µL of supernatant assayed immediately in 1 mL of assay buffer (50 mm Bicine–NaOH (pH 8.0), 15 mm MgCl2, 1 mm EDTA, 19 mm NaCl, 9.3 mm NaHCO3, 9.3 mm DTT, 0.2 mm RuBP, 0.1 mm NADH, 4.7 mm photsphocreatine, 4.7 mm ATP, 1.4 U mL−1 creatine-P-kinase, 1.4 U mL−1 glyceraldehyde-3-P-dehydrogenase, and 2.9 U mL−1 phosphoglycerokinase); the reaction was monitored at ΔA340 for at least 3 min. A 1 mL aliquot of crude extract was incubated for 10 min with 80 mm MgCl2 and 150 mm HCO3 to fully activate Rubisco, and then assayed as described for the crude extract. Five aliquots of activated Rubisco extract were then incubated with CABP (concentrations of 0, 0.58, 1.1 and 1.8 µm) for 10 min to inhibit Rubisco activity, and assayed as above. Rubisco activity was calculated based on the slope of ΔA340 versus time. Rubisco content was calculated from the y-intercept of a plot of activity versus [CABP]−1.

Chlorophyll was measured spectrophotometrically. Leaf punches (1.7 cm2) were taken immediately after gas exchange measurements and ground in 96% EtOH using a chilled mortar and pestle. After centrifuging, the optical density of the supernatant was measured at 665, 649 and 654 nm. Chlorophyll concentration was calculated using the specific absorption coefficients in Wintermans & DeMots (1965).

Experimental design and statistical analyses

The FACTS-I experiment consisted of six rings enclosing plants in paired control (ambient CO2) and treatment (ambient + 200 µmol mol−1 CO2) conditions. The controls were fully instrumented. The FACTS-II experiment consisted of 12 rings in a crossed CO2 × O3 experiment. Measurements were only made on plants in the six rings not receiving ozone treatment, which consisted of three ambient CO2 and three ambient + 200 µmol mol−1 CO2 rings. Both experiments were designed with three blocks of paired (ambient + elevated) rings, and three replicate measurements were averaged within blocks. Blocked means were calculated across each species and treatment. The CO2 effects on gm were analysed for the FACE results using mixed anova (JMP; SAS, Inc., Cary, NC, USA) with gm as the main effect, treatment as a fixed factor and species and block as random factors. Post-hoc comparisons of treatment effects were performed within each species using the Tukey adjustment.

Growth chamber experiments were conducted in a pair of chambers, and blocks were replicated through time. Each block consisted of three to five individual plants of each species and ran for 6 to 8 weeks, after which the chambers were emptied, and new seedlings were started for the next experimental block. Results were analysed using anova with treatment as a fixed factor and species as random. Unequal numbers made the block effects untestable. Tukey-adjusted post-hoc comparisons were made within species to investigate treatment effects.

We made several comparisons of relationships common to all experiments. Because Evans & Loreto (2000) found a consistent relationship between net photosynthesis rate and gm we made the same comparison across our different experiments. We also compared the Vc max : Jmax ratio across our experiments. The slopes and intercepts of regression lines were compared using analysis of covariance (ancova) as described by Underwood (1997).


Photosynthesis increased with increasing gm consistently across all species and observations (Fig. 1). We used a standardized measurement (the light saturated rate of net CO2 assimilation at Ci ≈ 400 µmol mol−1) as a consistent metric of the leaf capacity for photosynthesis. The linear regressions between gm and photosynthetic capacity on the elevated and ambient CO2 data differed neither in slopes (ancova; F = 0.76, P = 0.47) nor intercepts (ancova; F = 0.13, P = 0.88), so we show only a single regression for the combined data set. There were similar, apparently linear, relationships between both leaf Rubisco and chlorophyll content and gm in the growth chamber experiment (Fig. 2). The difference in CO2 from the intercellular airspaces to the site of carboxylation did not vary in a systematic fashion with net CO2 assimilation rate (Fig. 3). This relationship also showed no discernable dependence on growth CO2 or plant species.

Figure 1.

The relationship between net CO2 assimilation (A) and mesophyll conductance (gm). Mesophyll conductance was estimated using the constant J method. Photosynthesis was measured at Ci between 390 and 450 in all cases. Symbol shapes represent species; sweetgum (▪,□) aspen (•, ○) cucumber (▴,▵) bean (▾,▿) and spinach (◆,◊). Open symbols represent plants grown at elevated CO2, and closed symbols represent plants grown at ambient CO2. The solid line represents a linear regression of the combined (all species and treatments) data (y = A + Bx; A = 6.4, B = 60, r2 = 0.51).

Figure 2.

The relationship between mesophyll conductance and leaf Rubisco content (top panel) and chlorophyll (bottom panel). Mesophyll conductance was estimated using the constant J method, Rubisco was measured using CABP binding, and chlorophyll was measured spectrophotometrically after extraction in 96% ethanol. Symbols represent species and treatments as in Fig. 1. Linear regressions were calculated with data from all species and treatments (y = A + Bx, top panel: A = 3.2, B = 11, r2 = 0.67; bottom panel A = 963, B = 279, r2 = 0.67

Figure 3.

The difference in pCO2 from the leaf internal airspaces to the sites of carboxylation, CiCc versus net CO2 assimilation rate, A. Measurements and symbols are as in Fig. 1.

The overall effect of the CO2 treatment on photosynthesis generally co-varied with the change in gm except in spinach and linden bean (Table 1). The largest CO2 effects were seen in cucumber and sweetgum (sun leaves) followed by spinach. An analysis of the growth chamber experiment data (top three rows of Table 1) showed significant CO2[F = 5.59, P = 0.0248, degrees of freedom (d.f.) = 1] and species (F = 39.63, P < 0.001, d.f. = 3) effects. In the FACE experiments, gm was significantly affected by species (F = 31.22, P < 0.001, d.f. = 3) and block (F = 3.87, P = 0.03, d.f. = 3). Mean CO2 effects were significant only at the 10% level (F = 1.28, P = 0.09, d.f. = 1). Pair wise comparisons of CO2 effects within each species were significant at the 5% level in cucumber and at the 10% level in sunlit sweetgum leaves.

Table 1.  The effect of growth CO2 on mesophyll conductance and photosynthesis
SpeciesCO2 treatment
(µmol mol−1)
(µmol m−2 s−1)
(µmol mol−1)
(mol m−2 s−1 bar−1)
  1. To compare measurements on an equal basis, all photosynthesis measurements were compared at a common Ci. Data are means of all observations (SE). Probability (P) values represent the Tukey-adjusted pairwise comparisons between the CO2 treatments for each species, and n designates the sample size. aSun leaves; bshade leaves.

Cucumber50014.6 (3.2)400 (11)0.18 (0.08)0.066
750 9.9 (0.3)407 (8)0.08 (0.02) 6
75033.34130.39 5
Linden bean50015.2 (2.3)421 (11)0.21 (0.02)0.476
75016.7 (1.1)411 (5)0.19 (0.03) 8
Aspen36026.6 (1.6)415 (5)0.18 (0.02)0.959
56026.4 (2.0)412 (3)0.18 (0.01) 9
Sweetguma36014.8 (2.1)417 (5)0.17 (0.01)0.039
56016.7 (1.4)415 (6)0.32 (0.12) 9
Sweetgumb3608.3 (1.3)414 (4)0.08 (0.02)0.498
5608.8 (0.5)415 (5)0.10 (0.01) 9

To examine the potential error in estimating photosynthetic parameters we calculated Vc max and Jmax based on Ci and (using our values of gm) based on Cc. Calculated values of Vc max were affected more than Jmax by the gm correction, increasing between 20 and 70% when calculated based on Cc rather than Ci (Table 2). The gm effects often were numerically uneven, affecting the ambient and elevated CO2 treatments differently, thus revealing that the CO2 effect on Vc max could be over- or under-estimated depending on the direction of change in gm. In sweetgum, aspen, and cucumber this meant that the change in Vc max was always less than predicted based on Ci. In the case of sweetgum shade leaves, a 5% increase at elevated CO2 became a 9% decrease when recalculated based on Cc. Spinach and bean behaved differently, increasing the CO2 effect slightly. Sensitivity of Jmax to gm was generally smaller than the sensitivity of Vc max (Table 3); values of Jmax varied by approximately 10% with the inclusion of gm in the calculations, with the exception of shade-grown sweetgum. Using Ci caused a slight underestimate in Jmax for aspen

Table 2.  The effect of growth CO2 on Vc max from gas-exchange measurements
 Vc max (µmol m−2 s−1)
AnalysisAmbient CO2Elevated CO2CO2 effect (%)
  1. Calculations based on Ci were done with standard gas-exchange equations. Cc was calculated from mesophyll conductance and Ci. All parameters were calculated separately for each leaf. aSun leaves; bshade leaves.

CucumberCi 36.3 26.0−28
Cc 48.7 40.1−18
SpinachCi 64.8 85.0 31
Cc 83.0112 35
Linden beanCi 44.0 39.9 −9
Cc 57.8 54.2 −6
AspenCi 79.9 78.1 −2
Cc102 92.9 −9
SweetgumaCi 38.8 46.1 19
Cc 55.3 62.1 12
SweetgumbCi 24.8 26.1  5
Cc 43.0 39.0 −9
Table 3.  The effect of growth CO2 on Jmax from gas-exchange measurements
 Jmax (µmol m−2 s−1)
AnalysisAmbient CO2Elevated CO2CO2 effect (%)
  1. Calculations based on Ci were done with standard gas-exchange equations. Cc was calculated from mesophyll conductance and Ci. All parameters were calculated separately for each leaf. aSun leaves; bshade leaves.

CucumberCi 91.8 65.1−29
Cc 98.4 76.5−22
SpinachCi163192 18
Cc173235 36
AspenCi230216 −6
Cc217200 −8
Linden beanCi109 88.8−19
Cc120 99.2−13
SweetgumaCi 94.0106 13
Cc109119  9
SweetgumbCi 57.5 64.2 12
Cc 75.5 86.2 14

Because the choice of using Ci versus Cc in photosynthesis calculations affected Vc max more than Jmax, the relationship between the two parameters was sensitive to gm as seen by the change in the slope of the Jmax : Vc max relationship (Fig. 4). The regression lines for the values calculated from Ci and Cc had significantly different slopes (ancova; F = 27.9, P < 0.001, d.f. = 52). There was no systematic effect of CO2 treatment on the Jmax : Vc max ratio using either calculation, so data from both treatments were grouped for the regressions. We took an alternative approach to ‘correcting’Ci-based calculations of Vc max and Jmax by calculating Vc max and Jmax using Rubisco constants determined from in vivo measurements (Bernacchi et al. 2001). This approach gave similar results to the Cc-based calculations at low photosynthesis capacity, but deviated from those data at higher values (Fig. 4).

Figure 4.

The relationship between maximal carboxylation and electron transport rates as calculated from gas-exchange measurements (ACi curves). Vc max and Jmax were fitted using non-linear regression using in vitro Rubisco constants. Open symbols represent values calculated based on Ci, and closed symbols represent values calculated based on Cc. Symbol shapes represent species as in Fig. 1. The same raw data, re-fitted using in vivo Rubisco parameters, are included for comparison (*). The lines were fitted to each data set by linear regression (y = A + Bx; Ci: A = −14.2, B = 2.87, r2 = 0.96; Cc: A = 5.06, B = 2.17, r2 = 0.97).

In the growth chamber experiments, we used measurements of Rubisco activity and chlorophyll content as secondary indicators of photosynthetic capacity. Using Cc in calculations of Vc max somewhat changed the relationship between in vivo and in vitro carboxylation capacity (Fig. 5; a 1 : 1 line is shown for comparison). RuBP regeneration capacity increased with total leaf chlorophyll content in a seemingly linear fashion (Fig. 6). As the Jmax calculations were not strongly affected by the gm, the differences between the two data sets are relatively small.

Figure 5.

Maximal Rubisco activity (Vmax) determined by in vitro assay versus measured by gas-exchange based on Ci and Cc. Open symbols represent values calculated based on Ci, and closed symbols represent values calculated based on Cc. Symbol shapes represent species as in Fig. 1. The solid line marks a 1 : 1 relationship.

Figure 6.

The relationship between leaf chlorophyll content and maximum electron transport activity (Jmax) estimated based on Cc and Ci. Open symbols represent values calculated based on Ci, and closed symbols represent values calculated based on Cc. Symbol shapes represent species as in Fig. 1.


There was a consistent relationship between gm and photosynthetic capacity across species, growth location, and CO2. This apparently linear relationship (Fig. 1) is similar to that reported by Evans & Loreto (2000), where they summarized the results of several earlier gm studies. We calculated a regression line to determine the slope of the Agm relationship from those data (not shown) and tested it against the regression of our data. The slopes were not significantly different at the 5% level (ancova: F = 3.5, P = 0.06, d.f. = 80) although they were different at the 10% level. The aspen data were the main outliers in our data that affected the slope, and without these data the two data-sets were indistinguishable. Excluding these data had a minimal effect on the slope while reducing the y-intercept to 4.3 and increasing the r2 to 0.84.

The non-zero intercept of our regression differs from Evans & Loreto (2000), which probably results different methods of determining photosynthesis for the y-axis. We defined this as the light-saturated photosynthesis rate at Ci ≈ 400 µmol mol−1 to make consistent comparisons among all treatments. Although the conditions are not specified in Evans & Loreto (2000) most authors use light-saturated photosynthesis rates measured at ambient CO2 for photosynthetic capacity. Thus all our measurements were made at CO2 levels between 600 and 1000 µmol CO2 mol−1, resulting in higher net A. To demonstrate this we re-fit our data using photosynthesis measured at the growth CO2 in each case, and calculated an intercept of 2.83 µmol m−2 s−1(data not shown). This approach also improved the agreement between aspen and all other species. This strongly suggests that the intercept of this relationship depends solely on the conditions under which photosynthetic capacity is defined, and has no physiological interpretation.

By similar comparison with the data in Evans & Loreto (2000), the calculated difference between Ci and Cc determined from gm and A values averaged 106 ± 6 µmol mol−1, but by excluding the aspen data were indistinguishable from the 78 µmol mol−1 value of the data summarized by Evans & Loreto (2000). The relationship between A and gm, and the consistency of the CiCc draw-down, appear to be general features of leaves that are empirically predictable based on photosynthetic capacity measurements. This is especially notable considering the values from different studies used different methods to measure gm, seemingly confirming the consistency of gm measurement regardless of measurement technique (Loreto et al. 1992). Thus, this relationship might in principle be used to improve the estimated CO2 effect on Vc max and Jmax even when gm was not explicitly measured.

Growth at different CO2 concentrations caused a small effect on gm in almost all species studied which was statistically significant at either the 5 or 10% levels (Table 1). Since the magnitude and direction of photosynthetic acclimation to CO2 varied considerably across treatments, and gm was consistently related to photosynthesis, the use of block means is misleading in cases where the direction of acclimation differed between blocks. This was the case in sweetgum (one block showed a decrease in photosynthesis rates whereas the others showed an increase), linden bean (one decrease and one increase) and aspen (two decrease and one increase). Thus the general effect of CO2 enrichment on gm is to cause it to move up and down the A versus gm regression line without substantially changing the relationship between the two.

There remains uncertainty about the mechanism relating gm to any particular anatomical, biochemical or physical feature of a leaf. Substantial work has gone into relating leaf thickness and cell wall structure to gm in various species, but the changes in gm over the growing season and during leaf senescence (Loreto et al. 1994; Evans & Vellen 1996), when cell wall structure is relatively fixed, suggest wall structure may only be minimally involved. Gas diffusion within the leaf may contribute to gm in some instances (Parkhurst & Mott 1990; von Caemmerer & Evans 1991; Syvertsen et al. 1995) although this effect has only been found in thick, hypostomatus leaves. A more consistent predictor of gm is the surface area of chloroplasts appressed to the cell surface exposed to the leaf airspaces (von Caemmerer & Evans 1991; Evans et al. 1994). The relationship between chlorophyll content, Rubisco content and gm (Fig. 2) may be indicative of this relationship, although it is impossible to determine a mechanistic relationship using these data.

Although no single anatomical measurement accurately predicts gm (Evans et al. 1994; Syvertsen et al. 1995), empirical models relating several anatomical measurements to gm have shown promise (Syvertsen et al. 1995). Using stable isotope discrimination analysis to partition leaf conductance into its components, Gillon & Yakir (2000) found cell wall conductance was most limiting to overall leaf conductance in oak, whereas resistances in the chloroplasts were more limiting in tobacco and soybeans. Liquid-phase dynamics are likely to be more limiting to gm, and thus enzymatic processes such as leaf carbonic anhydrase activity and aquaporins (Coleman 2000; Gillon & Yakir 2000, Ono & Terashima 2002) may also play important roles in determining gm. These conclusions are consistent with the observed temperature response of gmin vivo (Bernacchi et al. 2002). Given the substantially different conclusions of the various studies, it seems that growth CO2 may well affect the various determinants of gm differently across species. This added complexity might mean that a mechanistic prediction of mesophyll conductance is not possible in the general case, and must be considered on a species-by-species basis.

Given the complex interacting factors contributing to gm, the consistency of the Agm relationship is puzzling. It is likely that this consistency results from covariance of several attributes. For example, the relationship between Rubisco and gm (Fig. 2) may result from a correlation between Rubisco and carbonic anhydrase (CA; Coleman 2000) rather than through any direct mechanism, given the relationship between CA activity and gm. Since the relationship between Rubisco and gm is broken in antisense-Rubisco-transformed plants (Evans et al. 1994), we conclude that it is coincidental and not causal.

The most common anatomical changes resulting from growth at elevated CO2 are associated with stomatal and epidermal cell density (Ferris et al. 1996; Masle 2000), neither of which is likely to substantially affect gm or photosynthetic capacity, at least when measured independently of stomatal conductance. Elevated CO2 can cause changes in leaf thickness (Kürschner et al. 1998), increasing airspace diffusion limitations that in hypostomatous leaves may reduce gm. Significant CO2 effects have also been noted on total mesophyll cell cross-sectional area (Ferris et al. 1996; Masle 2000), which may correspond to an increase in mesophyll cell surface area and cause an increase in gm (Evans et al. 1994). Studies of growth CO2 effects on carbonic anhydrase activity have shown either substantial changes in Rubisco and carbonic anhydrase (Majeau & Coleman 1996), or changes in Rubisco with no changes in carbonic anhydrase (Sicher, Kremer & Rodermel 1994). The uncertainty of which leaf properties substantially affect gm makes the use of these as a proxy for determining CO2 effects on gm difficult.

Although mesophyll conductance defies a complete physical or mathematical description, its consideration when calculating photosynthetic parameters is necessary especially when a treatment is anticipated to affect gm. Including gm in photosynthesis calculations changed our interpretation of the effect of growth CO2 on photosynthesis. In one case, an apparent increase in Vc max with CO2 treatment was revealed to actually be a decrease when analysed on a Cc basis (Table 1). In most cases the apparent CO2 effect was smaller when gm was considered. The effects of growth CO2 on Jmax were quite small in all cases (Table 2). As Vc max was affected more by the recalculation than Jmax, the relationship between the two parameters changed and the slope of Jmax : Vc max was reduced (Fig. 4). In both analyses, however, there was no systematic CO2 effect across species on the ratio of the parameters. A change in the relationship between these parameters was predicted by the mechanistic analysis of CO2 acclimation and photosynthesis (Medlyn 1996). In the model, lower gm at elevated CO2 increased the difference between Ci and Cc, thus reducing the need to reallocate of N from carboxylation (reflected in Vc max) to RuBP regeneration (reflected in Jmax). This could happen if CO2 affected gm independently of any acclimation in photosynthesis, but our data indicate that this is not the case. The CiCc relationship remains unchanged because gm and photosynthesis change proportionally (Figs 1 & 3). These observations support the work showing that the relationship between carboxylation and RuBP regeneration rates is not affected by growth CO2 made in other gas-exchange and biochemical analyses (Maxwell, Griffiths & Young 1994; Hymus et al. 1999).

Mechanistic photosynthesis models are based on [CO2] in the chloroplast, yet the majority of studies report values calculated based on Ci. The implicit assumption in such cases is that gm is infinite. When gm is not considered, treatment effects such as growth at elevated CO2 that affect diffusion within the leaf may be falsely attributed to changes in leaf biochemistry (Parkhurst & Mott 1990). This principle is illustrated by Delfine et al. (1998) who show apparent treatment differences in the ACi response of photosynthesis that are not seen in the ACc relationship. The inclusion of gm in our calculations improved the agreement between gas exchange and biochemical measurements of carboxylation capacity (Fig. 5). This effect was relatively small for the relationship between chlorophyll and RuBP regeneration capacity (Fig. 6). This is more difficult to evaluate, however, because there is no reason to expect a linear relationship between the two values.

Bernacchi et al. (2001) avoid the problems associated with the assumption of infinite gm by measuring Rubisco kinetic parameters from in vivo measurements. This approach simplifies gas-exchange analysis because gm is included in the Rubisco parameters since they were determined from Ci-based measurements. We re-analysed our gas-exchange data from both experiments using these parameters in place of the Baldocchi & Harley (1995) constants, and found the Vc max : Jmax relationship matched the ratios we calculated based on Cc at low Vc max : Jmax, but the two approaches systematically deviate from one another at higher Vc max (Fig. 4). The deviation between these approaches occurs because the in vivo calculations assume gm is constant (although not infinite), whereas our approach of determining Cc for each set of measurements accounts for the increasing gm as photosynthetic capacity increases (Fig. 1).

We conclude that there was a potentially important effect of growth CO2 on gm that corresponded with photosynthetic acclimation to CO2 through the consistent linear relationship between photosynthetic capacity and gm. Many gas-exchange studies calculate Vc max and Jmax based on A versus Ci measurements rather than A versus Cc, implicitly assuming that gm is infinite. We found gm is neither infinite nor constant in field or growth-chamber experiments. Including gm in the analysis of photosynthetic responses to the CO2 environment significantly changed the relationship between parameters estimated by gas-exchange measurements and improved the agreement of Rubisco activity measured with gas-exchange with biochemical assays.


This work was funded by a NSF traineeship in the Integrative Photosynthesis Research Training Program (DBI 96-02240), and by a grant to E.H.D. from the US Department of Energy, Office of Biological and Environmental Research (FG02-95 ER62127). CABP was generously provided by Dr Archie Portis. The authors thank Dr Peter Curtis for helpful comments on the manuscript.