Journal of Experimental Botany, Vol. 55, No. 406, pp. 2313–2321, October 2004 DOI: 10.1093/jxb/erh239 Advance Access publication 13 August, 2004
RESEARCH PAPER
The photosynthetic limitation posed by internal conductance to CO2 movement is increased by nutrient supply Charles R. Warren* School of Forest and Ecosystem Science, The University of Melbourne, Water Street, Creswick, VIC 3363, Australia Received 9 March 2004; Accepted 30 June 2004
Abstract The internal conductance to CO2 supply from substomatal cavities to sites of carboxylation may pose a large limitation to photosynthesis, but little is known of how it is affected by nutrient supply. Knowing how internal conductance responds to nutrient supply is critical for interpreting the biochemical responses from A2Ci curves. The aim of this paper was to examine the response of gi and photosynthetic parameters to nutrient supply in glasshouse-grown seedlings of the evergreen perennial Eucalyptus globulus Labill. Seedlings were grown with five different nutrient treatments and gi was estimated from concurrent measurements of gas exchange and fluorescence. Internal conductance varied between 0.12 and 0.19 mol m22 s21 and the relative limitation of photosynthesis due to internal conductance was greater than the stomatal limitation. In most species these two limitations are rather similar, but in E. globulus stomatal limitations were abnormally low due to high stomatal conductance (0.31 to 0.39 mol m22 s21). The large positive response of photosynthesis to nutrient supply was not matched by changes in internal conductance, and thus the relative limitation of photosynthesis due to internal conductance increased with increasing nutrient supply. Failure to account for finite internal conductance led to estimates of Vcmax that were 60% of the true value, which, in turn, led to an underestimation of in vivo Rubisco specific activity (as Vcmax/Rubisco content). The specific activity of Rubisco in E. globulus (21 mol mol21 s21) was close to the maximum published estimates, and thus, despite these leaves containing a large fraction of N as Rubisco (38–44%) there was no evidence that Rubisco activity was down-regulated or that the enzyme was in excess.
Key words: Internal resistance, Jmax, mesophyll conductance, nitrogen, nutrient, photosynthesis, transfer conductance, Vcmax.
Introduction For photosynthesis (A) to occur, CO2 must diffuse from the atmosphere to the sites of carboxylation. The concentration of CO2 at the sites of carboxylation (Cc) is less than atmospheric (Ca) owing to a series of gas-phase (air) and liquid-phase (mesophyll cell) resistances. In the gaseous phase, CO2 must diffuse across a boundary layer in the air above the foliage surface, through a stomatal opening, and across intercellular air spaces in the substomatal cavity. In the liquid phase there are resistances as CO2 enters the liquid phase at the surface of mesophyll cells, as CO2 diffuses within the cell to the chloroplast membrane, and from there to the sites of carboxylation (Aalto and Juurola, 2002). It has been assumed that the concentration of CO2 in the substomatal cavities (Ci, for example, as measured by gas exchange) is uniform throughout the leaf and thus the same as Cc. There is mounting evidence that this assumption is invalid and Cc may be significantly less than Ci (Evans et al., 1986; Parkhurst and Mott, 1990; Lloyd et al., 1992; Epron et al., 1995; Ethier and Livingston, 2004). This drawdown in CO2 concentration between Ci and Cc is a function of rates of photosynthesis and the internal conductance ðgi = A=ðCi Cc Þ: Internal conductance affects the interpretation of the photosynthetic response to Ci and the biochemical photosynthesis model of Farquhar et al. (1980). The Farquhar model states that the rate of photosynthesis is limited by one of two (or in special cases three) processes, and that limitation shifts from one to another as CO2 concentration changes. At low CO2 concentrations, photosynthesis is
* Fax: +61 3 5321 4277. E-mail:
[email protected] Journal of Experimental Botany, Vol. 55, No. 406, ª Society for Experimental Biology 2004; all rights reserved
2314 Warren
limited by the maximum rate of Rubisco carboxylation (Vcmax), whereas at high CO2 concentrations photosynthesis is limited by the electron transport limited rate of RuBP regeneration (Jmax). Analysis of the A Ci response has become enormously popular and there are now data for a great many species growing under various conditions (Wullschleger, 1993; Wohlfart et al., 1999). The fashion in which the A Ci response is fitted assumes implicitly that there is no draw-down in CO2 concentration from Ci to Cc. It is now known that this assumption is incorrect (see above) and failure to account for finite gi results in erroneously low estimates of Vcmax (Epron et al., 1995; Bernacchi et al., 2002; Centritto et al., 2003; Warren et al., 2003a; Ethier and Livingston, 2004). The author is aware of only one paper that has examined gi in plants grown with varying nutrient supply (Triticum; von Caemmerer and Evans, 1991), and it reported that gi was positively related to foliage N content per unit area. A subsequent paper suggested that the increase in gi was smaller than the increase in Rubisco carboxylation and thus the limitation posed by gi was greater in leaves with a high N content (Poorter and Evans, 1998). This type of experiment has not been replicated with other species and, in general, little is known of how gi responds to N supply. While not explicitly considering the effect of nutrient supply, several previous studies have reported gi at a range of N or Rubisco contents in leaves grown under different light intensities (Lloyd et al., 1992; Hanba et al., 1999; Piel et al., 2002; Warren et al., 2003a), in a developmental series (Hanba et al., 2001), in leaves of wild-type and antiSSU Nicotiana (Evans et al., 1994), and among species (Lloyd et al., 1992; Hanba et al., 1999, 2001; Piel et al., 2002; Singsaas et al., 2004). It is not known if the relationship of gi with N or Rubisco varies among species or depends on the underlying cause of variation in N or Rubisco (e.g. light versus nutrient supply), and thus one aim of this paper was to determine if there are ‘universal’ Ngi and Rubiscogi relationships based on a compilation of literature data. The response of photosynthetic parameters to nutrient supply is widely reported (Walcroft et al., 1997; Warren et al., 2003b), but in none of these previous reports was gi determined. The underlying biochemical response (of photosynthesis) to nutrient supply may be confounded by the response of gi. If, for example, elevated nutrient supply increases Vcmax but gi is unaffected, the relative limitation due to gi will increase. A common observation in plants grown with a high N supply is a reduction in Vcmax/Rubisco content (i.e. specific activity), which is seen as evidence of over-investment in Rubisco (Warren and Adams, 2004). However, these studies were based on Vcmax determined from Ci, and thus the reduction in apparent Vcmax/Rubisco could simply indicate that the limitation due to gi is greater in high N plants, as suggested by Poorter and Evans (1998). The aim of this paper was to examine the response of gi and
photosynthetic parameters to nutrient supply in glasshousegrown seedlings of Eucalyptus globulus Labill. Seedlings were grown with five different nutrient treatments and gi was estimated from concurrent measurements of gas exchange and fluorescence (Harley et al., 1992). Materials and methods Plant material and treatments Seed of Eucalyptus globulus ssp. globulus (CSIRO ATSC seedlot 18725) was obtained from the Australian Tree Seed Centre (Kingston, ACT, Australia). On 25 August (late winter) seed was germinated in moist vermiculite in a glasshouse at the Forest Science Centre (Creswick, Victoria, Australia). Approximately one month later (26 September) germinants had one or two pairs of true leaves. At this stage germinants were carefully transferred to 6.4 l plastic pots (16 cm square325 cm high) filled with coarse sand. For the first month after transfer to pots, seedlings were irrigated to field capacity three times per week with balanced nutrient solution at one-quarter its normal concentration, while for the second month they received the same nutrient solution at half its normal concentration. The full-strength nutrient solution contained 8 mM N (as 4 mM NH4NO3), 0.67 mM P, 4 mM K, 2.9 mM Ca, 1.5 mM Mg, 1.5 mM S, 52 lM Mn, 49 lM Fe, 20 lM Cu, 15 lM Zn, 8 lM B, and 0.03 lM Mo. Previous studies have shown this nutrient solution produces optimal growth of E. globulus and other evergreen perennials (C Warren, unpublished data). On 20 November, differential nutrient treatments were imposed. 25 seedlings were randomly assigned to one of five nutrient treatments. The five nutrient treatments were obtained by dilution of full-strength nutrient solution, and thus all nutrients (not just N) varied. However, for the sake of convenience nutrient treatments are referred to based on their elemental N concentrations: 0.5, 1.0, 2.0, 4.0, and 8.0 mM N. Seedlings were irrigated to field capacity with nutrient solution four times per week, and with water on the remaining days. At no time were soil water deficits allowed to develop, and thus larger seedlings were irrigated with water or nutrient solution more frequently than smaller seedlings. From October until the end of the experiment, temperatures inside the glasshouse varied between 17 8C and 23 8C, while leaf-to-air vapour pressure deficit was typically less than 1 kPa. Seedlings were exposed to full (glasshouse) sunlight at all times, which was approximately 70% of external sunlight, with weekly average PPFD varying between 33 and 41 mol m2 d1. Photosynthesis measurements were made on five seedlings per treatment between 29 December and 2 January (6 weeks after nutrient treatments commenced). All measurements were made on the youngest fully expanded leaves. Specific leaf area and leaf nitrogen concentration The specific leaf area (SLA, m2 leaf area kg1 dry mass) was determined for a leaf opposite the one used for photosynthesis measurements. Leaves were excised, placed in damp plastic bags and transported to the laboratory where projected area was determined with a planimeter (LI-3000A/LI-3050A, Li-Cor). Dry mass was determined after 72 h at 70 8C. Leaves were subsequently ground to a fine powder in a mixer mill (MM301, Retsch, Haan, Germany) and analysed for total N by Dumas combustion. Leaf Rubisco content Four leaf discs (0.56 cm2 each) were punched from the leaves used for photosynthesis measurements, placed in a 2 ml Eppendorf microfuge tube (Safe-Lock tube 2.0 ml, Eppendorf AG, Hamburg, Germany), frozen in liquid N and stored at 80 8C until analysis. Rubisco was quantified by capillary electrophoresis using a method
Effect of nutrient supply on internal conductance
Simultaneous fluorescence and gas exchange measurements Simultaneous gas exchange and chlorophyll fluorescence measurements were made with an open gas exchange system (LI-6400, Li-Cor, Lincoln, NE, USA) and integrated fluorescence chamber head (LI-6400–40). The photochemical efficiency of photosystem II (/PSII) was determined by measuring steady-state fluorescence (F9) and maximal fluorescence (F9m ) during a light-saturating pulse (Genty et al., 1989):
/PSII = ðF9m F9Þ=F9m
ð1Þ
The rate of linear electron transport (Jf) is related to /PSII:
Jf = /PSIIa0:5 PPFD
ð2Þ
where a is the total leaf absorptance, the factor 0.5 describes the distribution of light between the two photosystems and is assumed to be 0.5. Jf is not strictly related to linear electron transport because fluorescence primarily measures upper cell layers and is thus not representative of the whole leaf. Furthermore, the distribution of light between photosystems is hard to determine. Owing to these uncertainties, no a priori assumptions were made regarding relationships between fluorescence and linear electron transport, and instead chose to determine an empirical relationship. This was done by measuring a light-response curve under non-photorespiratory conditions and rests
on the assumption that under non-photorespiratory conditions linear electron transport should be wholly associated with gross photosynthesis. Such a ‘calibration’ procedure obviates the need for measuring leaf absorptance and assumptions regarding the distribution of light between photosystems and representativeness of fluorescence to whole-leaf processes. In E. globulus, this relationship was highly linear (R2=0.96) with a non-significant intercept for all leaves (Fig. 1), suggesting there were no significant alternative electron sinks. To construct a ‘calibration’ curve, reference gas was supplied from a cylinder containing a mixture of 450 lmol mol1 CO2 in 1% O2 (Linde Gas, Australia). Leaf temperature was controlled at 25 8C, leaf-to-air vapour pressure deficit was maintained at 0.8–1.3 kPa using a dew-point generator (LI-610, Li-Cor). Leaves were acclimated to a PPFD of 2000 lmol m2 s1 for at least 30 min, or until stomatal conductance, net photosynthesis, and fluorescence were steady. Thereafter, PPFD was decreased, in ten steps, from 2000 lmol m2 s1 to 0 lmol m2 s1. At each PPFD, leaves were acclimated for at least 5 min until stomatal conductance, net photosynthesis, and fluorescence reached a steady-state, and then gas exchange and fluorescence parameters were recorded. After the last point was measured (i.e. at 0 lmol m2 s1 PPFD), the gas supply was changed to 450 lmol mol1 CO2 in 21% O2, and PPFD was increased to 2000 lmol m2 s1. These measurements were subsequently used to estimate gi. Leaves were allowed to acclimate to these non-photorespiratory conditions for at least 30 min, or until gas exchange and fluorescence were steady. Thereafter, a light response curve was generated in the same fashion as for nonphotorespiratory conditions. The CO2 and H2O sensitivity of the LI-6400 is affected by O2 concentration of the analysis gas. The effect of O2 on CO2 sensitivity is small enough to be ignored, whereas the change in H2O sensitivity of approximately 3% (between 1% and 21% O2) has a measurable effect on the estimation of stomatal conductance and Ci. Therefore, the measured H2O concentration was corrected using an empirical correction reported by Ghannoum et al. (1998). gi was estimated with the ‘variable J method’ (Harley et al., 1992):
gi = A=fCi fC ½Ja + 8ðA + Rd Þg=fJa 4ðA + Rd Þgg ð3Þ gi was determined at a PPFD of 1000 lmol m2 s1. This PPFD was chosen for two reasons: (i) fluorescence yield is greater at 1000 lmol
140 J a (µmol electrons m-2 s-1)
modified from Warren et al. (2000). Samples were removed from the freezer and kept frozen while adding two 3 mm diameter stainless steel ball bearings and approximately 0.2 ml of polyvinylpolypyrrolidone. Frozen samples were rapidly ground in a mixer mill (MM301, Retsch, Haan, Germany). 1.0 ml of cold (0–4 8C) extraction buffer (50 mmol l1 TRIS-HCl, 0.1 mol l1 2-mercaptoethanol, 1% w/v SDS, and 15% v/v glycerol) was added to each tube and samples were extracted by shaking for 90 s with the mixer mill. The extract was centrifuged for 2 min in a microfuge, the supernatant removed, and the pellet was re-extracted with an additional 1.0 ml of extraction buffer. The supernatants were pooled, mixed and then recentrifuged at room temperature. A 150 ll aliquot was precipitated with the methanol/chloroform/water procedure (Wessel and Flu¨gge, 1984), taken up in 400 ll of extraction buffer and denatured by heating at 95–100 8C for 10 min in a water bath. Samples were stored at 0–4 8C for no greater than 6 h prior to analysis. Capillary electrophoresis was performed with a Beckman P/ACE MDQ system (Beckman-Coulter, Fullerton, CA) fitted with a photo-diode array detector and controlled by a computer running System Gold software (Beckman-Coulter). The separation of proteins was performed in SDS 14–200 gel buffer (Beckman-Coulter) in a SDS-coated fused-silica capillary (100 lm i.d.331.2 cm long, eCap SDS 14–200 capillary, Beckman-Coulter). To reduce analysis time, samples were injected from the outlet side, resulting in an effective length from injection to detection window of 11.2 cm. Electrophoresis was conducted at 20 8C, a constant voltage of 9 kV, with 0.5 psi of pressure applied at both ends of the capillary. Samples were injected at 0.5 psi for 15 s, and protein–SDS complexes were detected at 220 nm. The capillary was rinsed sequentially between successive electrophoretic runs with 1 M HCl for 60 s, followed by SDS 14-200 gel buffer for 120 s. Standard curves for purified Rubisco were highly linear (r2=0.99) over the range of 0.05 to 1 mg ml1. Standard curves constructed by serial dilution of a leaf extract were also highly linear (r2=0.99), and this, combined with 95% recovery of Rubisco in a spike and recovery test, suggests quantification was unaffected by the complex matrix of E. globulus leaves (Warren et al., 2000). Rubisco concentrations were initially calculated on a leaf area basis, but were subsequently converted to a dry mass basis using measured SLA. The fraction of total foliar nitrogen present in Rubisco was calculated using the assumption that Rubisco contains 16.7% N.
2315
120 100 80 60 40 20 0 0
20
40
60
80
100
120
J f (µmol electrons m-2 s-1)
Fig. 1. The relationship between apparent rates of linear electron transport estimated from fluorescence (Jf) and actual rates of linear electron transport determined from gas exchange measurements under non-photorespiratory conditions (Ja). Data were determined from lightresponse curves and simultaneous measurements of gas exchange and chlorophyll fluorescence measured on 25 E. globulus seedlings grown with one of five nutrient treatments.
2316 Warren m2 s1 than at 2000 lmol m2 s1, yielding more precise estimates of Jf and thus gi; and (ii) 1000 lmol m2 s1 is above the lightsaturation point for these plants. A and Ci were measured directly by gas exchange and Ja was estimated from the empirical leaf-specific relationship between Jf and Ja. The rate of mitochondrial respiration in the light (Rd) was assumed to be the same as measured dark respiration, and that C*=38.7 (von Caemmerer et al., 1994; Bernacchi et al., 2002) Measurement of the CO2 response of photosynthesis The CO2 response of photosynthesis was determined 1 d after fluorescence measurements. Leaves were enclosed in a 233 cm broadleaf chamber with integrated light source (LI-6400–02B, Li-Cor). Using the broadleaf chamber, as opposed to the fluorescence chamber, increased the enclosed leaf area and the signal-to-noise ratio by a factor of 3, a critical factor when making photosynthesis measurements at low Ca and PPFD. Measurements were made at 2561 8C and a leaf-to-air vapour pressure deficit of 0.8–1.3 kPa. The molar flow rate through the chamber was varied between 300 and 750 lmol s1 depending on the rates of photosynthesis and transpiration. For each leaf an A Ci curve was generated at 2000 lmol m2 s1 PPFD. Leaves were exposed to 400 lmol mol1 CO2 in air and a PPFD of 2000 lmol m2 s1 until rates of photosynthesis and transpiration were steady. After this, Ca was increased to 1500 lmol mol1 and an A Ci curve was generated by decreasing, stepwise, Ca to 50 lmol mol1. Over the Ca range from 150 to 50 lmol mol1, measurements were made at 25 lmol mol1 intervals so as to ensure adequate characterization of the initial portion of the curve. At each Ca photosynthesis was allowed to stabilize for at least 3 min, and three successive measurements were made at 1 min intervals to ensure stability. Data were not corrected for diffusion of CO2 into and out of the leaf chamber because this was insignificant with this experimental set-up. Preliminary experiments established that with a molar flow rate of 300 lmol s1 (the lowest flow rate used in these experiments) there was never more than 1.5 lmol mol1 diffusion of CO2 into an empty chamber maintained at 50 lmol m1 CO2. Vcmax and Jmax were determined from CO2 response data fitted to the photosynthesis model of Farquhar et al. (1980), essentially as described previously (Warren et al., 2003b). For ½Kc ð1 + Oc =Ko Þ; a value of 736 lmol mol1 was used (von Caemmerer et al., 1994), while 0.24 was assumed for the apparent quantum yield of electron transport. Estimation of Cc and relative limitation on photosynthesis imposed by gi and gs Using estimated gi and measured A and Ci, Cc was calculated as:
Cc = Ci ðA=gi Þ
ð4Þ
The limitation of A imposed by finite gi and gs was based on estimates of the potential rate of photosynthesis assuming these conductances were either infinite or as measured (Farquhar and Sharkey, 1982; Warren et al., 2003a). Estimates of A were based on CO2-response curves and measured gi and gs. Rates of net photosynthesis were estimated assuming gi and gs were as measured (An, the light-saturated rate of photosynthesis at Ca=360 lmol mol1), assuming gi was infinite and gs as measured (Ail, the light-saturated rate of photosynthesis at Cc = Ci ), or assuming gi as measured and gs was infinite (Asl, the light-saturated rate of photosynthesis at Ci=360 lmol mol1). The relative limitations due to internal resistances (Li) and stomatal resistances (Ls) were estimated as:
Li = ðAil An Þ=Ail
ð5Þ
Ls = ðAsl An Þ=Asl
ð6Þ
Results Specific leaf area and leaf content of nitrogen and Rubisco Specific leaf area (SLA) was not significantly affected by differential nutrient treatments, despite a trend for SLA to increase with increasing N supply (Table 1). Nitrogen content per unit area varied significantly among nutrient treatments, increasing from 0.7 g m2 to 1.3 g m2 with an increase in N supply from 0.5 mM to 8.0 mM. This difference was proportionally greater on a mass basis (12– 27 mg g1; data not shown). The content per unit area of Rubisco increased from 1.8 g m2 to 3.0 g m2 with increasing N supply, but this was only marginally significant (P=0.08) owing to large variation within treatments. On a mass basis, Rubisco increased from 32 to 61 mg g1, and this difference was clearly significant (P=0.01; data not shown). There was no effect of nutrient supply on the proportion of total N allocated to Rubisco, which varied between 38% and 44% among treatments. Response of gi and photosynthetic parameters to nutrient supply
The rate of net photosynthesis was significantly affected by nutrient supply, increasing from 10 lmol m2 s1 to 16 lmol m2 s1 as nutrient supply increased from 0.5 mM N to 8.0 mM N (Table 2). Stomatal conductance was unaffected by nutrient supply, but there were significant differences in Ci among treatments with Ci generally lower in those plants receiving a greater nutrient supply. There were significant differences in internal conductance (gi) among treatments, with gi being noticeably lower in seedlings receiving 0.5 mM N compared with those receiving a more concentrated nutrient supply. However, gi did not vary in a monotonic fashion with nutrient supply, and thus there was a (minor) reduction in gi at the highest nutrient supply. Cc was, on average, 81 lmol mol1 lower than Ci, and there was a general trend for Cc to decrease with Table 1. Effect of nutrient supply on specific leaf area (SLA), leaf content per unit projected area of nitrogen (N) and Rubisco, and the percentage of total N present as Rubisco Seedlings of E. globulus were grown for 6 weeks with one of five different concentrations of a balanced nutrient solution. The nutrient treatments are referred to on the basis of their elemental N concentration (mM of N). Data are means (sd) of five replicate plants per treatment. The final row shows the significance (P) of differences among treatments as indicated by one-way ANOVA. Treatment (mM of N)
SLA (m2 kg1)
N (g m2)
Rubisco (g m2)
Rubisco N (% of total N)
0.5 1.0 2.0 4.0 8.0 P
17 (2) 18 (2) 18 (1) 19 (2) 20 (2) 0.139
0.7 (0.1) 0.8 (0.2) 0.9 (0.2) 1.1 (0.1) 1.3 (0.2)
