It is a matter of debate if there is a direct (short-term) effect of elevated atmospheric CO2 concentration (Ca) on plant respiration in the dark. When Ca doubles, some authors found no (or only minor) changes in dark respiration, whereas most studies suggest a respiratory inhibition of 15–20%. The present study shows that the measurement artefacts – particularly leaks between leaf chamber gaskets and leaf surface, CO2 memory and leakage effects of gas exchange systems as well as the water vapour (‘water dilution’) effect on DCO2 measurement caused by transpiration – may result in larger errors than generally discussed. A gas exchange system that was used in three different ways – as a closed system in which Ca increased continuously from 200 to 4200 mmol (CO2) mol-1 (air) due to respiration of the enclosed leaf; as an intermittently closed system that was repeatedly closed and opened during Ca periods of either 350 or 2000 mmol mol-1, and as an open system in which Ca varied between 350 and 2000 mmol mol-1– is described. In control experiments (with an empty leaf chamber), the respective system characteristics were evaluated carefully. When all relevant system parameters were taken into account, no effects of short-term changes in CO2 on dark CO2 efflux of bean and poplar leaves were found, even when Ca increased to 4200 mmol mol-1. It is concluded that the leaf respiration of bean and poplar is not directly inhibited by elevated atmospheric CO2.
It has been reported that the global increase of atmospheric CO2 concentration (Ca) may affect dark respiration in plants. Doubling of Ca is said to reduce respiration rates by 15–20% (Drake et al. 1999; Gonzàlez-Meler & Siedow 1999), and the direct (short-term) effects of elevated CO2 on respiration are assumed to be greater than the indirect (long-term) ones (Amthor 1997). However, in various publications, a doubling of CO2 has been reported to cause a reduction of respiration rate, an absence of any effect, and even an enhancement of repiration (see Table 2·1 of Amthor 1997). Possible explanations for the inhibitory effect of CO2 are the inhibition of enzymes involved in respiration (Gonzàlez-Meler & Siedow 1999) and (re)fixation of CO2 (Amthor 1997; Wullschleger, Ziska & Bunce 1994). However, no report confirms that any of these mechanisms explain quantitatively the CO2 inhibition of respiration observed in gas exchange measurements. Thus, the underlying mechanisms of inhibition (when it occurs) are not understood (Drake et al. 1999; Gonzàlez-Meler & Siedow 1999).
The starting point of the present study was to investigate the possibility (based on previous work, e.g. van Oosten et al. 1995; Wullschleger & Norby 1992) that if the current atmospheric CO2 level of 370 mmol mol-1 were to more than double, there may be a total inhibition of leaf dark respiration in the short term, and perhaps a change to net CO2 uptake under these conditions. Preliminary experiments were undertaken on bean leaves. However, the results differed greatly, depending on the gas exchange method used. The inhibition of dark CO2 efflux by elevated CO2 was found using an open gas exchange system; when the same leaf was investigated in a closed system, no such inhibition was detected. This raised the suspicion that artefacts in CO2 measurement were involved; this hypothesis is supported by recent research showing that respiratory O2 consumption of Glycine leaves was not affected by elevated CO2 (Hunt et al. 2000). On the other hand, the technical problems of CO2 detection are assumed to explain only partially the observed direct CO2 effect (Amthor 1997; Drake et al. 1999; Gonzàlez-Meler & Siedow 1999).
The contrasting findings described earlier were the reason to look systematically for possible artefacts involved in the measurements. Gas exchange systems are affected by specific problems such as leaks within the gas tubing system, leakage at the interface between leaf chamber gaskets and leaf surface, adsorption/desorption of CO2 at the system surfaces, decreasing CO2 sensitivity of infra-red gas analysers (IRGAs) with increasing CO2 concentration, non-linearity of IRGA output signals, CO2/H2O cross-sensitivity of IRGAs, and the water vapour effect (WVE) on DCO2 measurements (Amthor 1997, 2000; Bunce & Ward 1985; Drake et al. 1999; Gonzàlez-Meler & Siedow 1999; Long & Hällgren 1993). Since respiratory CO2 exchange rates are usually small, technical errors may become more critical than is generally assumed.
The purpose of this study was to test the working hypothesis that the respiration of Phaseolus vulgaris and Populus tremula x P. alba leaves in the dark is not directly affected by elevated atmospheric CO2 concentration. Some details of the custom-made gas exchange system are described to illustrate how measurements were performed and to demonstrate the general technical problems involved in different techniques of gas exchange measurement.
Materials and methods
Plant material and growth conditions
Plants of Phaseolus vulgaris L. cv. Saxa were grown from seeds, and cuttings were made from Populus tremula L. x P. alba L. seedlings (seedlings supplied by Prof. Rennenberg, Universität Freiburg, Germany). The bean and poplar plants grew up in soil (70% peat, 30% clay; Einheitserde Typ 0, Balster Einheitserdewerk, Fröndenberg, Germany) mixed with perlite (4 : 1, v/v) in 1·0 and 1·5 L pots, respectively. The plants were watered periodically with a nutrient solution adjusted to pH 5·8: 4 mm KNO3, 8 mm Mg(NO3)2·4H2O, 3 mm KH2PO4, 2 mm MgSO4·H2O, 1 mm NaCl, 4·7 mm CaSO4·0,5H2O, 22 mm M Fe-EDTA (Fetrilon, BASF), 60 mm H3BO4, 14 mm MnSO4·H2O, 0·6 mm CuSO4·5H2O, 1 mm ZnSO4·7H2O, 0·3 mm H2MoO4 and 0·02 mm CoCl2·6H2O. Controlled environmental cabinets provided day/night regimes of 14 : 10 h, temperature 23/20 C and relative humidity (RH) 60/70%[i.e. water vapour pressure deficit (VPD) 1·12/0·70 kPa], respectively. A photosynthetically active photon flux density (PPFD) of 400–550 mmol (photons) m-2 s-1 at upper leaf level was provided by various lamps (Krypton bulbs and HQI-400 W/D, Osram, München, Germany; SON-T AGRO, Philips, Köln, Germany). In the cabinets, Ca was measured by IRGAs (WMA-2; PPSystems, Hutchin, UK) and was regulated day and night at 355 10 mmol mol-1.
Gas exchange system
Figure 1a illustrates the air flow in the gas exchange system. Different CO2 concentrations ([CO2]) were produced by scrubbing CO2 in the incoming air (Soda Lime Carbosorb; Promochem, Wesel, Germany) and mixing it with CO2 from a pressure cylinder by using mass flow controllers (Tylan FC-260; Millipore, Eschborn, Germany). Moisture was adjusted with two custom-made devices, a humidifier (Hm1) and a dewpoint condensor (DP1), in which pressure was kept only a little higher than atmosphere (+0·1 kPa). Part of the incoming gas was vented to the reference cell of differential IRGA2 (CIRAS 1, PPSystems).
Loop 1 consisted of the leaf chamber with a gas circulation bypass (not shown), a site to apply radioactive 11CO2 (CS2) and humidity sensors (H1, H2; HMP 133 Y, Vaisala, Hamburg, Germany). Gas flow was measured by mass flow meter MFM1 (Tylan FM-360) and kept constant by controlling the pressure pump GP1 (WISA 300, ASF Thomas Industries, Puchheim, Germany). To keep pressure differences between the leaf chamber and atmosphere small, a differential pressure transducer PD1 (143PC05D, Honeywell, Offenbach, Germany) controlled the suction pump GP3. Process controllers (Sipart DR20, Siemens, Essen, Germany) were used for the control circuits.
In the closed system, loop 2 was connected to loop 1 through the valve MV2. Loop 2 mainly consisted of absolute IRGA1 (Binos 4·1, Rosemount, Hanau, Germany) and a dewpoint trap (DP2) to keep moisture in the closed system constant. At connection site CS1, a CO2 compensation system (CComp), which was constructed for a null-balance mode, was inserted. It was used to calculate the volume of the closed system by injecting a known volume of bicarbonate solution into perchloric acid, thus liberating a known amount of CO2. Alternatively, a test unit (LinT) was connected to CS1 to check output linearity of IRGA1.
The gas exit consisted mainly of a suction pump (GP3) and a dewpoint trap (DP3). The outlet was normally open to the atmosphere at the ‘pressure reference point’ (PRP). In experiments with radioactive CO2, gas tubing was prolonged beyond PRP to scrub CO2 on absorber columns. The resulting back-pressure to the analyser cells of IRGA2 was avoided by keeping pressure at PRP constant at -0·2 kPa compared with atmosphere with the aid of a suction pump (not shown).
The gas exchange system was used in three modes: as a continuously closed system, an intermittently closed system and an open system. Simple schemes of the different system modes and experimental protocols are illustrated in Figs 1b, 1c and 2, respectively.
Gas tubing and leaf chamber design
The gas exchange system was constructed mostly from inert materials and was kept as clean as possible. Tubing was from stainless steel (6·0 ¥ 0·5; 3·0 ¥ 0·2 mm) and was connected with stainless steel fittings (Swagelok; B.E.S.T., Dortmund, Germany). Flexible tubing (Polyamide 12; Deutsche Tecalemit, Bielefeld, Germany) was used only upstream of the DP1 (Fig. 1a) and downstream of the PRP. In the gas pumps, the diaphragms were from Viton and parts originally made from plastic material were replaced by stainless steel replicas. The leaf chamber bottom was from anodized aluminium and the chamber lid from glass (Fig. 3a). Natural rubber profiles (NR 15, 8 ¥ 8 mm; Meteor Gummiwerke K.H. Bädje, Bockenem, Germany) were used as gaskets and fixed by a flexible jointing compound (Hylomar; Marston -Domsel, Zülpich, Germany). To avoid leakiness between chamber gaskets and the leaf surface, the gaskets were covered with thin layers of a silicon-based putty (Optosil P Plus; Heraeus Kulzer, Dormagen, Germany). Sealing quality was tested by applying 11CO2 to the closed system while a leaf was enclosed in the chamber. Without the putty, 11CO2 leaked out and was detected at low levels by scintillation counters (Jahnke et al. 1998). No leakage was measured when the putty was used.
Specifications of the experimental conditions
The inner diameter of the leaf chamber was 70 mm, the inner height 6 mm, and the gaskets were 8 mm in width. Only the apical part of intact leaves was enclosed (with areas of 20–35 cm2). Gas pressure in the leaf chamber was 0·03 0·01 kPa (mean SD) compared with atmosphere. The volume of the closed system was 198 cm3 and gas flow was 25 cm3 s-1. Gas flow in the open system was 6·7 cm3 s-1. Boundary layer conductance () was then 210 mmol m-2 s-1; with gas circulation in the chamber bypass, it was 1750 mmol m-2 s-1. Temperature in the leaf chamber was 23·0 0·5 C. At the leaf, it was 0–0·5 C lower, depending on the transpiration rate. Dewpoint temperature of the incoming air was 16·0 0·1 C. The experiments were performed in a cabinet providing 23·0 0·5 C, 60 5% RH (VPD = 1·12 kPa) and [CO2] of 355 10 mmol mol-1.
Calibrations and control measurements in the closed gas exchange system
The absolute CO2 gas analyser (IRGA1) was calibrated by venting through dry and wet CO2-free air, and dry and wet CO2 reference gas (1900 mmol mol-1; Messer-Griesheim, Oberhausen, Germany). During calibration, the humidity of the respective gases was measured. From the IRGA readings (Fig. 4a), linear coefficients for the dry and wet reference gases were drawn:
where b(0) and b(1) are linear coefficients and [CO2] is the CO2 concentration (mmol mol-1). Subscripts: d, dry gas; meas, measured value; ref, reference gas; w, wet gas.
Since the differences between dry and wet gases were not large, the coefficients b(0) and b(1) were assumed to be affected linearly by water vapour. A correction factor for the coefficients (Fcoeff) was calculated by considering water vapour pressures both in the experiments (VPmeas) and during calibration (VPref,d of the dry and VPref,w of the wet reference gases):
The ‘true’ CO2 concentration under normalized dry conditions in the closed system was then calculated:
Output linearity of absolute IRGA1 was checked with a simple physical test (applied for patent). In loop 2 (Fig. 1a), a test chamber (LinT) was inserted, in which the stainless steel tubing of the gas circuit was bridged by silicon tubing (2–3 mm in length). At the outer side, the silicon tubing was ventilated with pure CO2 gas (Fig. 3b). Loop 2 was flushed with CO2-free air before it was closed through valve MV2 (Fig. 1a). Because of the high CO2 differencies between the inner (about 4000 mmol mol-1 at maximum; cf. Fig. 4b) and outer side (1 mol mol-1) of the silicon tubing, diffusion of CO2 into the closed system was taken as linear (temperature and pressure were constant). Through this approach, the output signal of the IRGA1 was tested continuously from low to high [CO2]. From the CO2 slopes (cf. Fig. 4b), linearity factors were obtained (Fig. 4c) by calculating the slope at reference [CO2] (1900 mmol mol-1) to unity. They were used to correct the measured CO2 data (e.g. in Figs 7 & 8). Later in this study, IRGA1 was modified to acquire directly the unlinearized CO2 signal. The linearity factors were then much more stable and were used, for example, in the experiments of Fig. 6.
In the continuously closed system, memory and leakage of the system were tested as described in Fig. 4d. In the intermittently closed system, the same protocol as in the plant experiments (cf. Figs 7 & 8) was performed in controls (empty leaf chamber) to test CO2 memory effects and leakage after short-term changes in CO2. Knowing the volume of the closed system, net carbon exchange rates (NCERs) were calculated from the [CO2] slopes according to the following equation:
where s is the projected surface area of the enclosed leaf part (m-2), t is the time (s) and Vm is the molar volume of the closed system (mol).
Calculations and control measurements in the open gas exchange system
The CO2 measurement of differential IRGA2 was calibrated with dry reference gas. H2O detection was checked by dewpoint settings of the incoming air. The water vapour effect on DCO2 detection was tested as described in the legend for Fig. 5a. The ‘true’ CO2 difference, not being dependent on any WVE, was calculated according to the following equations:
where D[CO2] is the CO2 concentration difference between outgoing and incoming air [mmol (CO2) mol-1 (air)]; pH2O is the partial pressure of water vapour (kPa); D(pH2O) is the water vapour pressure difference between outgoing and incoming air (kPa); P is the atmospheric pressure (kPa); pAir is the partial pressure of dry air (kPa); WVE is the water vapour effect [mmol (CO2) mol-1 (air)] on D[CO2] measurement, and Error(%) is the percentage error if WVE is not considered. Subscripts: in, incoming air; out, outgoing air; meas, measured value; calc, calculated value; true, true value that might be measured when there is no WVE.
The WVE depends exclusively on the true water vapour difference between reference and analyser cells of the IRGA. It makes no difference what finally causes DH2O. In controls, DH2O was generated by altering the temperature of DP2 compared with DP1 (Figs 1d & 5a). In Eqns 6–12, the measured DH2O was considered directly (not the transpiration rate of the investigated leaf; see also von Caemmerer & Farquhar 1981; Long & Hällgren 1993). In the open system, NCER was then calculated as:
where ue is the molar flow of air entering the leaf chamber (mol s-1).
The NCERs are presented on the negative scale for net release of CO2(Figs 6–12). To avoid any problems caused by WVE experimentally, the dewpoint traps in the incoming (DP1) and outgoing gas streams (DP3) were kept at the same temperature. DH2O was then zero at IRGA2, and transpiration was measured by humidity sensors H1 and H2.
Data from the different instruments were acquired by a personal computer via a PCL-714 Super-Lab card (Advantech Europe, Freiburg, Germany). The software for data acquisition (DAQ), calibration routines and data scaling was written in Turbo Pascal 6·0 (Borland Software Corp., Langen, Germany). In the course of this study, the DAQ system was replaced by a new signal-conditioning unit based on SCXI modules (National Instruments, München, Germany). With the graphical programming language LabVIEW (National Instruments), a program was developed to operate the gas exchange system either manually or automatically, control system components (valves, gas pumps, etc.), generate analogue setpoints (CO2 concentrations, gas flows, etc.), acquire analogue and digital data, calculate data online, and visualize the results on screen. More details of the new system are described elsewhere (Jahnke & Proff 2001). For presentation and further calculations, the graphing program SigmaPlot 2000 was used, together with TableCurve 2D v. 4 and SigmaStat v. 2·0 (all SPSS Science Software, Erkrath, Germany). Some data presented here were obtained with the old DAQ system (Figs 5a,b, 7, 8, 9 & 11), whereas others were gained from the new one (Figs 5c, 6 & 10).
The leaves of 10 different bean and nine poplar plants were investigated with the three modes of the gas exchange system. Some were measured for several days in the dark with repeated periods of low and high [CO2]. Using continuously closed system measurements, NCERs were drawn from the corrected data at 350 and 4000 mmol mol-1 (cf. Fig. 6b), and the response ratios (NCER4000/NCER350) were calculated (Table 2). In the intermittently closed and open system experiments, NCERs at 2000 mmol mol-1 were calculated from linear regressions; at 350 mmol mol-1, they were calculated (together with the 95% prediction intervals) from linear regressions of data points before and after the respective high [CO2] periods (cf. Fig. 7). Corresponding NCERs (at 350 and 2000 mmol mol-1) of individual studies were calculated at the end of the high CO2 periods and are shown in Fig. 12. The calculated response ratios were pooled and are shown in Table 2.
Table 2. . Overview of the CO2 exchange experiments on bean and poplar leaves in the dark. The CO2 concentration in the gas exchange system was altered between 350 (low) and 2000 or 4000 (high) mmol mol-1. CCS, continuously closed system; ICS; intermittently closed system; OS, open system; mean, arithmetic mean; SD, standard deviation; NCER, net CO2 exchange rate. Response ratios of NCERs at high versus low CO2 concentrations were obtained from individual treatments. A paired t-test was used for normally distributed data (p); a Wilcoxon signed rank test was performed for data that failed the normality test (f)
Control measurements and responses of the gas exchange system itself to short-term changes in CO2
The CO2 calibration protocol for absolute IRGA1 (cf. Figure 4a) included both the specific effects of water vapour on CO2 detection (like signal broadening) and the general WVE (‘dilution’ of CO2). IRGA outputs are generally linearized to compensate for lower sensitivity at higher CO2 concentrations. When the accuracy of linearization was tested, the calculated linearity factors varied by 30% within the CO2 detection range of the instrument (Fig. 4b,c). For closed system measurements, controls were performed to test both memory and leakage effects of the empty system (no leaf). At low [CO2], at which the ratios between the subsequent CO2 steps were high (cf. Fig. 4d), memory effects were larger than at high [CO2] when leakage became dominant. Considering that the volume of the closed system was about 200 cm-3 (including the leaf chamber and a dewpoint trap), the rate of CO2 leakage was about 60 mmol mol-1 h-1 at a [CO2] of 4000 mmol mol-1. This is much smaller than that reported for the LI-6200 system (cf. McDermitt et al. 1989; McDermitt 1991). When controls were performed in the intermittently closed system with the same protocol as in the experiments, the results looked similar to the uncorrected data of Fig. 7b. Shortly after [CO2] was altered, memory effects due to CO2 adsorption and desorption were largest, but decreased afterwards.
System parameters were also studied in the open system. The WVE on DCO2 measurements was tested (Fig. 5a) according to the sheme of Fig. 1d. After careful calibration of the CO2 and H2O channels, there was little difference between measured DCO2 and the DCO2 calculated from measured DH2O (Eqn 10), which, in general, is used to correct the WVE. However, at high CO2 (2000 mmol mol-1), conformity was not as good (Fig. 5a). To test the possible effects of CO2 on H2O detection, the water vapour pressure of the incoming air was kept constant while [CO2] was increased stepwise. The absolute H2O values obtained by humidity sensors were hardly affected, whereas those of IRGA2 changed by 0·04 kPa H2O when [CO2] increased from 0 to 2000 mmol mol-1 (Fig. 5b). Within the same CO2 range, measurement of DH2O was not affected by changes in CO2 (not shown). To test memory effects and leakage of the open system, [CO2] was repeatedly altered between 350 and 2000 mmol mol-1. Every time, it took about 30 min for the new equilibrium to be established (Fig. 5c).
Dark respiration of Phaseolus and Populus leaves under short-term changes in CO2
When the NCER rate was steady after a prolonged time in the dark (92 h), the experiment of Fig. 6a,b was performed on a Phaseolus leaf. The CO2 concentration in the closed system was raised from about 200 to 4200 mmol mol-1 within 14 h. During this time, the true CO2 efflux rate of 0·17 mmol m-2 s-1 was constant at a slope of 6·4E-07 mmol m-2 s-1 (mmol mol-1)-1 (Fig. 6b). When [CO2] was highest (at the end of the experiment), the maximum correction due to memory and leakage effects was 0·03 mmol m-2 s-1, which represented 23% of the uncorrected data. In a similar experiment, performed in the dark (27 h) on a Populus leaf much earlier, the NCER of 0·76 mmol m-2 s-1 showed a slope of 3·5E-06 mmol m-2 s-1 (mmol mol-1)-1, i.e. it was not affected by CO2 (Fig. 6c,d). When [CO2] had reached 4000 mmol mol-1, the correction for memory and leakage was 0·04 mmol m-2 s-1 which represented 6% of the uncorrected data (Fig. 6d).
The results ofFigs 7 and 8 were obtained with the intermittently closed system, which was repeatedly closed and opened at [CO2] levels of either 350 or 2000 mmol mol-1. Every time that [CO2] was altered, the measured (uncorrected) NCERs changed (small gray circles; Figs 7b & 8b), and the direction of changes was dependent on whether [CO2] was low or high before it was switched. Even some time after a change in CO2, the uncorrected data showed an apparent effect of CO2 on respiration; however, this was artefactual. When the Phaseolus plant of Fig. 7 was in the dark for 51 h (at the end of the second high CO2 period), corrections of 0·04 mmol m-2 s-1 had to be made, corresponding to 26% of the small NCER of -0·19 mmol m-2 s-1. Independently of the time in the dark, during which NCER decreased steadily, the true respiration rate was not affected by CO2. Furthermore, for Populus, no evidence was found that short-term changes in [CO2] between 350 and 2150 mmol mol-1 had any direct effect on true CO2 efflux (Fig. 8).
Open system measurements on Phaseolus leaves are shown in Figs 9 and 10. NCERs decreased with time in the dark, and every time that [CO2] was altered, it took about 30 min until a new equilibrium was reached. Respiration rates were calculated directly from measured DCO2 (large circles, Fig. 9b) because the WVE was avoided experimentally. However, in order to compare the results with what might have been measured when water vapour was not controlled, values of D[CO2]meas (Eqn 11) were recalculated as if there was a change in water vapour according to the measured transpiration rate (Fig. 9a). When water vapour is not controlled, the potential errors may be very large, as calculated from the data shown in Fig. 9 (see also Table 1). In the experiment of Fig. 10, a Phaseolus plant was measured for 72 h in the dark and [CO2] was changed between 350 and 2000 mmol mol-1 every 3 h. Respiration rate, starting at 0·7 mmol m-2 s-1, decreased steadily after the lights were shut off and it took almost 36 h in the dark until it was constant at about 0·2 mmol m-2 s-1. Apart from the scattering of NCERs (mainly because of the low precision of the IRGA at high [CO2]), there was no indication that CO2 had any effect on respiration at any time in the dark (Fig. 10b,c). In the Populus experiment of Fig. 11, transpiration rates varied largely, which might have caused large variations in NCERs due to the WVE (small circles, Fig. 11b). However, the WVE was avoided experimentally and the measured gas exchange data (large circles) did not show any reduction of CO2 efflux under elevated CO2 in the dark. The results were confirmed in a series of experiments, which are summarised in Fig. 12 and Table 2.
Table 1. Potential errors in open system measurements if the water vapour effect on DCO2 measurement is not considered. The data were obtained from the Phaseolus experiment shown in Fig. 9 for specified times at low and high CO2 concentrations. For abbreviations and calculations, see Eqns 6–12
Time in dark (h)
D[CO2]meas (mmol mol-1)
WVE (mmol mol-1)
D[CO2]true (mmol mol-1)
Potential error (%)
Technical aspects of gas exchange systems and their responses to short-term changes in CO2
The conflicting reports given in the literature about direct CO2 effects on respiration are occasionally thought to arise (at least partially) from artefacts in measurements. The reduced CO2 sensitivities of IRGAs at elevated CO2, as well as system leakage, were considered (Amthor 1997). Gonzàlez-Meler & Siedow (1999) mentioned possible diffusive leaks of CO2 from the surrounding air. Drake et al. (1999) focused on two potential systematic artefacts, namely ‘dilution’ of CO2 by transpired water and leaks between the leaf chamber and the surrounding air. However, possible errors are rarely tested in detail and, to my knowledge, potential artefacts of gas exchange measurements have not been investigated systematically in dark respiration studies to date. As stated by some authors (Drake et al. 1999; Amthor 1997), most studies give insufficient technical details to conclude whether – or to what extent – systematic measurement errors were involved in the experiments.
Gas exchange of plants can be measured with either closed or open systems, and each one has its own merits and demerits (reviewed by Long & Hällgren 1993). In general, open systems are used but, as shown here, closed system can also be useful tools, particularly when NCERs are small. Independently of the method used, controls through which the specific system properties were fully characterized turned out to be absolutely essential. Controls can be extremely time-consuming, but checking all components of a system is indispensible. Leakage between the leaf cuvette and the surrounding air was avoided by using a sealing putty; however, even in the gas lines of the IRGAs, leaks were identified. CO2 memory effects (cf. Bernacchi et al. 2001) can be based on different system properties, such as dead volumes (e.g. within solenoid valves), the materials used (see Long & Hällgren 1993), total area and quality (e.g. smooth or clean) of the inner surfaces of the system, and the size and mechanical construction of dewpoint traps or desiccants (if included).
The reduced CO2 sensitivity of IRGAs at high [CO2] (Amthor 1997; McDermitt, Welles & Eckles 1992) deteriorates the accuracy of CO2 detection (absolute as well as differential) when [CO2] increases. IRGA outputs are generally linearized. However, the quality of linearization may vary between individual instruments, and linearity tests are rarely mentioned in the literature (e.g. Baker et al. 2000). New analysers should provide better linearity than the one used here (the ‘fluctuations’ in Fig. 3c reflected exactly the number of linearization units of the IRGA1). In any case, it would be best to acquire the unlinearized signals of the CO2 detectors and to check linearity as described above. Unfortunately, such signals are generally not available from the IRGAs.
In open systems, serious systematic measurement errors may be caused by ‘dilution’ of CO2 because of transpired water. When the WVE is calculated, the error is defined by the accuracy of measurements of DH2O, DCO2 and absolute CO2. The error is definitively zero only when DH2O is zero (Eqns 6–12). Even in the dark, when transpiration is generally small, not considering water dilution may produce serious errors (cf. Table 1) much higher than those stated by Drake et al. (1999). When the WVE was tested in control experiments, measured and true DCO2 (Eqn 11) were identical only when CO2 and H2O calibrations were performed most carefully (cf. Fig. 5a). It is reported that properties of individual instruments may be different and theoretical corrections of the CO2 output for water dilution may be erroneous by a few mmol mol-1 (Bunce & Ward 1985). In addition, cross-sensitivity of IRGAs concerning CO2 and H2O causes errors that may be corrected only with an accuracy of about 1 mmol CO2 mol-1 (McDermitt, Welles & Eckles 1992). For low DCO2 readings (as in dark respiration experiments), the correction values due to WVE may be in the order of the measured DCO2 or even higher (cf. Table 1). Therefore, it is absolutely necessary to calibrate the DH2O measurement of IRGAs at least as precisely as that of DCO2. In the present study, any problems potentially caused by water vapour were avoided by keeping the incoming and outgoing air at the same dewpoint temperature. In particular, when DCO2 signals are small, it is better to avoid the WVE on CO2 detection experimentally rather than to calculate it. When CO2 and H2O channels of an IRGA are accessible separately, it is simplest to first measure water vapour of the incoming and outgoing gas in the H2O channels and then to scrub the water before CO2 is measured in the CO2 channels (cf. Farquhar & Raschke 1978). Unfortunately, miniaturization of modern differential IRGAs (as in the one used here) does not allow the use of such a simple method to avoid the WVE in DCO2 measurement.
Response of dark respiration of leaves to short-term changes in CO2
Most reports about the direct effects of elevated CO2 on respiration in the dark deal with CO2 concentrations in a range of doubled ambient (700 mmol mol-1) to about 1000 mmol mol-1 (Amthor 1997; Wullschleger, Ziska & Bunce 1994). However, even for that (small) elevation in CO2, the results are conflicting. On leaves of Populus euramericana, a 30% reduction in dark respiration was reported for plants that were grown at a Ca of 330 mmol mol-1 and were measured at 660 mmol mol-1 the next day (Gaudillère & Mousseau 1989). No direct CO2-dependent inhibition of respiration was found for the grass species Lolium perenne (Ryle, Powell & Tewson 1992), whereas inhibition was observed on whole rice canopies (Baker et al. 2000). In Glycine max, leaf respiration was always reduced after a short-term increase in [CO2] (Thomas & Griffin 1994), but no such effect was measured when soybeans grew up at 700 mmol mol-1 (Bunce 1990). The direct effect of elevated CO2 on the leaf respiration of nine tree species was found to be small when [CO2] was increased from 400 to 800 mmol mol-1 (Amthor 2000). Moreover, Bunce (1995) concluded from indirect evidence that measured CO2 efflux rates of soybean plants in the dark did not reflect accurately the average 24 h rates of growth and maintenance processes. Respiration measurements on roots have also shown conflicting and inconsistent results. Qi et al. (1994) reported on a clear respiratory reduction in roots of Pseudotsuga menziesii under high [CO2], but in Phaseolus vulgaris, CO2 efflux of roots was affected by neither CO2 pretreatment (up to 20 000 mmol mol-1) nor [CO2] during measurements (200–2000 mmol mol-1; Bouma et al. 1997).
Because effects on respiration in the dark were hardly found when Ca was doubled in the preliminary experiments of the present study, [CO2] was elevated to much higher values to obtain clear effects. The experiments shown in Fig. 6 were performed on Phaseolus and Populus leaves with the continuously closed system. They clearly indicate that even high CO2 concentrations of up to 4200 mmol mol-1 had no effect on NCERs. An advantage of the closed system is that the release of respiratory CO2 can be measured continuously from low to high [CO2] and the WVE is not relevant. However, memory effects in the continuously closed system are not simple to characterize because they are dependent on the changing rates of [CO2], which again are determined by the NCER of the investigated leaf. In the second set of experiments, performed with the intermittently closed system, the system properties were defined unequivocally by identical protocols in the controls and experiments. As with the continuously closed system, no differences in NCERs between low (350 mmol mol-1) and high (2000 mmol mol-1) CO2 concentrations were measured, either for Phaseolus or for Populus (Figs 7 & 8, respectively).
The open system experiments performed here were similar to the ones reported by Amthor, Koch & Bloom (1992), in which leaves of Rumex crispus were investigated and Ca was altered repeatedly between 350 and 650 mmol mol-1. Doubling of ambient [CO2] was found to inhibit respiration by 25–30%, and the responses to short-term changes in CO2 were readily reversible (Amthor, Koch & Bloom 1992). In the open system measurements in this study, when Ca was altered between 350 and 2000 mmol mol-1, no such effect was observed, either for Phaseolus (Figs 9 & 10) or for Populus (Fig. 11). The discrepancies between the open and closed system measurements, as observed in the preliminary experiments of this study, disappeared completely when all relevant system properties were taken into account.
It has been shown that changes in CO2 have larger effects on maintenance than on growth respiration in various plants and plant organs (Gale 1982; Reuveni & Gale 1985; McDowell et al. 1999). For that reason, experiments were performed at different times in the dark for between 3 and 148 h (Table 2). The respiration rates declined for about 36–48 h in the dark until they were steady at quite low rates (cf. Fig. 10) and mainly reflected maintenance (cf. Reuveni & Gale 1985). Potential (relative) measurement artefacts might then have caused much larger effects than when the leaves were in the dark for a short time with much higher respiration rates. However, no CO2-dependent effect was observed, even when NCERs were very low. One might argue that after such a long time in the dark (e.g. 107 h in Fig. 6a,b), leaf metabolism might have changed already to start with senescence. On the other hand, there was no indication that the time at which CO2 was changed in the dark had any influence on the results because there was no significant effect at any time in the dark between 3 and 148 h (Table 2).
The working hypothesis that dark respiration might not be affected directly by elevated CO2 is supported by evidence coming from recent research using a high-resolution oxygen analyser that showed that O2 uptake of soybean leaves in the dark was not affected by CO2 elevation of up to 2000 mmol mol-1 (Hunt et al. 2000; S. Hunt, personal communication). In the same experiments, CO2 efflux was affected, and from the discrepancy between the O2 and CO2 results it was concluded that there might be some artefacts in the CO2 measurements. That leaf respiration rates of Glycine max and Rumex crispus were not affected by CO2 elevation in the dark was confirmed when the O2 uptake of leaves was quantified by two different methods whereas the CO2 efflux was slowed in some experiments (Amthor et al. 2001). Additionally, that CO2 efflux might be much less affected by elevated CO2 than reported earlier (e.g. by Amthor et al. 1992) has been shown recently for different species (Amthor 2000; Tjoelker et al. 2001); however, an effect is still reported. In contrast to this, no CO2-dependent reduction of leaf CO2 efflux was found here, even under highly elevated CO2 concentrations. The results suggest that there is no direct effect of elevated CO2 on respiration in the dark, at least for the leaves of the two Phaseolus and Populus species examined.
I am indebted to Reinhard Frese, Dirk Krauss-Stangenberg, Alfred Lenk, Karl-Heinz Menze and Michael Neugebauer for their help in developing and improving the gas exchange system and for technical assistance during the experiments. Many thanks to Bernd Proff for the help with LabVIEW programming. The friendly and excellent collaboration of the technical divisions of the Universität Essen (in particular, the workshops for precision mechanics, electronics, optics and glass-blowing) is greatfully acknowledged. Thanks also to Stephen Hunt for critical comments on the manuscript and Donald MacLean for reading the English text. This work has been supported by the Deutsche Forschungsgemeinschaft (Ja 447/3) and by the Alfried Krupp von Bohlen und Halbach-Stiftung.