Annals of Botany 118: 1163–1173, 2016 doi:10.1093/aob/mcw165, available online at www.aob.oxfordjournals.org
Reduced plant water status under sub-ambient pCO2 limits plant productivity in the wild progenitors of C3 and C4 cereals Jennifer Cunniff1,*, Michael Charles2, Glynis Jones3 and Colin P. Osborne1 1
Department of Animal and Plant Sciences, Alfred Denny Building, University of Sheffield, Western Bank, Sheffield S10 2TN, UK, 2Institute of Archaeology, 36 Beaumont St, Oxford OX1 2PG, UK and 3Department of Archaeology, Northgate House, University of Sheffield, West Street, Sheffield S1 4ET, UK *For correspondence. Current address: CABI, Nosworthy Way, Wallingford, Oxfordshire OX10 8DE, UK. E-mail [email protected] Received: 28 February 2016 Returned for revision: 25 April 2016 Accepted: 13 June 2016 Published electronically: 29 August 2016
Background and Aims The reduction of plant productivity by low atmospheric CO2 partial pressure (pCO2) during the last glacial period is proposed as a limiting factor for the establishment of agriculture. Supporting this hypothesis, previous work has shown that glacial pCO2 limits biomass in the wild progenitors of C3 and C4 founder crops, in part due to the direct effects of glacial pCO2 on photosynthesis. Here, we investigate the indirect role of pCO2 mediated via water status, hypothesizing that faster soil water depletion at glacial (18 Pa) compared to postglacial (27 Pa) pCO2, due to greater stomatal conductance, feeds back to limit photosynthesis during drying cycles. Methods We grew four wild progenitors of C3 and C4 crops at glacial and post-glacial pCO2 and investigated physiological changes in gas exchange, canopy transpiration, soil water content and water potential between regular watering events. Growth parameters including leaf area were measured. Key Results Initial transpiration rates were higher at glacial pCO2 due to greater stomatal conductance. However, stomatal conductance declined more rapidly over the soil drying cycle in glacial pCO2 and was associated with decreased intercellular pCO2 and lower photosynthesis. Soil water content was similar between pCO2 levels as larger leaf areas at post-glacial pCO2 offset the slower depletion of water. Instead the feedback could be linked to reduced plant water status. Particularly in the C4 plants, soil–leaf water potential gradients were greater at 18 Pa compared with 27 Pa pCO2, suggesting an increased ratio of leaf evaporative demand to supply. Conclusions Reduced plant water status appeared to cause a negative feedback on stomatal aperture in plants at glacial pCO2, thereby reducing photosynthesis. The effects were stronger in C4 species, providing a mechanism for reduced biomass at 18 Pa. These results have added significance when set against the drier climate of the glacial period. Key words: Setaria viridis, Panicum miliaceum var. ruderale, Hordeum spontaneum, Triticum boeoticum, sub-ambient pCO2, origin of agriculture, C4 photosynthesis, C3 photosynthesis, water relations, crop progenitors.
INTRODUCTION Deglaciation at the end of the Pleistocene period was coupled to a rapid rise in atmospheric pCO2 from below 18 Pa to 27 Pa between 15000 and 12000 years ago (Jouzel et al., 1993; Petit et al., 1999). Humans began to cultivate plants in the Fertile Crescent of western Asia soon afterwards (Willcox et al., 2008) and, within five millennia, cultivation and subsequent domestication of plants had occurred in at least five primary centres across the globe (Purugganan and Fuller, 2009). This sequence of agricultural origins in widely separated regions of the world suggests the involvement of a global factor, and Sage (1995) proposed that pCO2 during the last glacial period may have been too low to support the level of productivity required for the successful establishment of agriculture. An experimental test of this hypothesis showed that the wild progenitors of C3 and C4 founder cereals from independent centres of origin displayed significant increases in vegetative biomass with a 50 % increase in pCO2 from 18 to 27 Pa (Cunniff et al., 2008, 2010). Biomass of the one C3 species in this experiment near-doubled and measurements of leaf gas exchange
revealed that photosynthesis (A) was strongly limited by glacial pCO2 in this species, highlighting a direct mechanism for biomass limitation. Biomass in the five C4 species showed a smaller, but still substantial, 40 % increase (Cunniff et al., 2008). However, measurements of leaf gas exchange immediately after watering revealed that A was only significantly limited by glacial pCO2 in two of these C4 crop progenitors, a finding consistent with the known CO2-concentrating mechanism of C4 photosynthesis (Pearcy and Ehleringer, 1984). Therefore, an alternative, indirect explanation for the biomass increase is required. The elevation of pCO2 from a glacial to postglacial level in this experiment was associated with large reductions in stomatal conductance (gs) and transpiration (E), resulting in a decrease in the use of water at leaf and canopy scales in all of the C4 species, and greatly improved water-use efficiency (WUE). These observations raise the possibility of an indirect feedback on biomass accumulation at sub-ambient pCO2 mediated via water relations because, although plants were not exposed to drought per se, they experienced a soil drying cycle between watering events on alternate days (Cunniff et al., 2008). This
C The Author 2016. Published by Oxford University Press on behalf of the Annals of Botany Company. V
All rights reserved. For Permissions, please email: [email protected]
1164
Cunniff et al. — Sub-ambient pCO2 and plant water status
hypothesized mechanism is consistent with studies investigating the effects of elevated atmospheric pCO2 (55–70 Pa) on the water relations of C4 plants, where reduced gs and E improve plant and soil water status, and extend the active period for photosynthesis and growth (Ghannoum et al., 2000). This indirect effect of elevated pCO2 is particularly important when C4 plants experience some kind of water deficit (Conley et al., 2001; Wall et al., 2001; Leakey et al., 2004, 2006; Leakey, 2009). It seems likely that these feedbacks of water status on stomatal conductance, and hence photosynthesis, should be greater at sub-ambient than elevated pCO2. This is because the response of A to intercellular pCO2 (Pi) in C4 plants is typically saturated at ambient pCO2 levels, but shows a steep, near-linear response to pCO2 at lower Pi (Pearcy and Ehleringer, 1984). Therefore, decreases in gs under water deficits could lead to substantial reductions in A. This effect has been recognized in studies using maize grown at ambient and elevated pCO2 (Samarakoon and Gifford, 1995, 1996). It has been demonstrated previously that WUE improves linearly across sub-ambient pCO2 gradients in both C3 and C4 species (Polley et al., 1993, 1996, 2002; Anderson et al., 2001). However, few studies have investigated the interacting effects of water limitation and sub-ambient pCO2. Polley et al. (2002) showed that mid-day xylem potentials were depressed by subambient pCO2 in C3 and C4 species during naturally occurring seasonal droughts. Furthermore, Ward et al. (1999) found that A was limited in the C4 annual Amaranthus retroflexus by subambient pCO2 and, when a drought treatment was applied experimentally, the CO2 limitation was stronger. Additionally, these plants showed the least recovery of leaf area and biomass upon re-watering at the end of the drought period. These responses at sub-ambient pCO2 were similar to those of the C3 annual Abutilon theophrasti and showed that C3 and C4 species were equally affected by drought at sub-ambient pCO2, which is predicted from other studies (Polley et al., 1993, 1995). Here we report a controlled environment experiment investigating the interacting effects of sub-ambient pCO2, plant and soil water status on the modern-day representatives of C3 and C4 crop progenitors. We focus on two C3 cereals from the Fertile Crescent and two C4 millets from the Loess Plateau in China, all of which were among the earliest wild plants to be brought into cultivation. Our experiment tested three hypotheses: (1) greater gs at glacial pCO2 increases the rate of soil water depletion and decreases leaf and plant water status compared with plants at post-glacial pCO2; (2) faster soil water depletion and reduced plant water status increase the rate at which gs and A decline in plants grown at glacial compared with post-glacial pCO2; and (3) water deficits amplify the limiting effect of glacial pCO2 on A to an equal extent in C4 and C3 crop progenitors. The aim was not to subject the crop progenitors to a drought treatment; rather, it was an investigation of how physiology changes between watering events in a controlled environment experiment. MATERIALS AND METHODS Growth conditions and plant material
The pCO2 treatments were applied in controlled environment chambers (Conviron BDR16, Conviron, Winnipeg, Manitoba,
Canada) at two levels throughout the full 24 d of plant growth: glacial (18 Pa) and post-glacial (27 Pa). pCO2 levels were maintained continuously throughout the day and night. The chambers were operated in a closed configuration, by connecting the outlet vent to the air inlet via a filter packed with a layer of activated charcoal and a layer of sodalime (Sofnolime 10–25-mm granules, Molecular Products Ltd, Mill End, UK); activated charcoal was employed to filter the air and remove any trace gases which could be emitted by plants or soil, and have the potential to affect plant development. The pCO2 level in each chamber was measured using a CO2 sensor (CARBOCAP Carbon Dioxide Probe GMP343, Vaisala, Finland) calibrated against a secondary standard. CO2 control was achieved by linking the sensor to a feedback system regulating the circulation of chamber air through the soda-lime scrubber. The pCO2 of air in the chambers was recorded every minute, giving overall mean values over the full growth period of 182 Pa (6 s.d. 048) and 271 Pa (6 s.d. 049). To maintain this tight control, the soda lime was changed as soon as pCO2 started to drift above the target level, which was approximately every 4 weeks. Treatment and plants were exchanged between the two controlled environment chambers every week from germination to avoid confounding the effects of chamber and growth environment. Seeds were obtained from germplasm holdings or commercial sources. They included two C4 species, Setaria viridis (L.) P. Beauv [(green foxtail millet) (Herbiseed, Twyford, UK; cat. no. 9602)] and Panicum miliaceum var. ruderale (Kitag.) [(wild broomcorn millet) (Herbiseed, cat. no. 9507)] from North China, and two C3 species, Hordeum spontaneum K. Koch [(wild barley) (Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany, accession no.: HOR 13791)] and Triticum boeoticum Boiss. [(einkorn wheat) (IPK, accession no.: TRI 17111)] from western Asia. Batches of 20 seeds of each species were sown into trays containing a 1:1 sand:John Innes no. 2 compost mix (seven parts loam, three parts peat, two parts sand N/P/K 20:10:10). This mix was chosen in an attempt to replicate an unimproved soil (Ivandic et al., 2000). Seeds were germinated under pCO2 treatments in the controlled environment chambers at 25/15 C (day/night), with an 8-h photoperiod and photosynthetic photon flux density (PPFD) of 300 lmol photons m2 s1. Once established, eight even-sized seedlings of each species were selected and planted into 32 individual 2-litre pots [132 cm (height) 17 cm (diameter)], containing the same growth media. The pots had eight holes in the base and were free to drain. Seedlings were returned to the same controlled environment chambers; and in both chambers temperature was maintained at 25/15 C (day/night), photoperiod was increased to 14 h with a maximum PPFD of 600 lmol m2 s1, and vapour pressure deficit (VPD) had a minimum value of 049 kPa during the night and a maximum value of 096 kPa during the day.
Physiology
Physiological measurements of gas exchange, canopy transpiration and gravimetric water content, and leaf water potential were taken over three separate drying cycles of 3 d duration: gas exchange [14–16 d after planting (DAP)], canopy
Cunniff et al. — Sub-ambient pCO2 and plant water status transpiration and gravimetric water content (17–19 DAP) and water potential (20–22 DAP). In total, physiological measurements occurred over 9 d (14–22 DAP). For each drying cycle, soil in the pots was watered to pot capacity before ‘dawn’ (0700ˆh) on day 1 (D1), and left to dry down over a 3-d cycle (D1–D3) before re-watering at the end of D3. We used a 3-d watering cycle as opposed to the 2-d cycle applied in our previous experiment (Cunniff et al., 2008) to exaggerate the hypothesized CO2 water interaction.
Gas exchange measurements
CO2 and H2O exchange were measured using a portable open gas exchange system (LI-6400P, LI-COR Biosciences, Lincoln, NB, USA) with the 6400-02B LED Light Source chamber. Chamber conditions were set with the aim of approximating the growth environment: PPFD was 600 lmol m2 s1, leaf temperature was 25 C, incoming air was maintained at a constant humidity to keep the leaf-air VPD at less than 1 kPa throughout all measurements, and pCO2 was set at 18 or 27 Pa as appropriate. Plants were watered to pot capacity before ‘dawn’ on D1 (14 DAP), allowed to drain and returned to the growth cabinets. Gas exchange was measured at nine time points: the beginning (0800ˆh), midpoint (1400ˆh) and end of the day (2000ˆh) from D1 to D3 (16 DAP). At each time point, measurements were commenced in a different cabinet, and were alternated between the two cabinets within each time point. This avoided confounding time with CO2 treatment, within the approximately 90 min that it took to complete all measurements. On each occasion, the same, most recently expanded leaf was clamped in the leaf chamber and allowed to equilibrate with chamber conditions for 90 s, before measuring leaf gas exchange. The aim of these measurements was to obtain ‘snapshots’ of photosynthesis (A), stomatal conductance (gs) and intercellular pCO2 (Pi) under growth conditions, and these were calculated using the equations of von Caemmerer and Farquhar (1981).
Transpiration and gravimetric soil water content
Transpiration and gravimetric soil water content (hg) were determined via lysimetry. Plants were watered to pot capacity before ‘dawn’ on D1 17 DAP, allowed to drain and then weighed (g). Pots were then returned to the controlled environment chambers and reweighed (g) at each of the nine time points detailed for gas exchange measurements. Alongside the plants, empty pots containing only the growth media were placed in the controlled environment chambers to estimate soil evaporation. These were treated in the same way as the plants, being weighed at saturation and the nine measurement occasions. At the end of D3 19 DAP, leaf area per plant was measured non-destructively using a ruler. Length maximum width was measured for each leaf and used to estimate the total canopy area using an allometric relationship established before the experiment (Supplementary Data Table S1). The daily canopy transpiration of each plant (Eplant) was determined as:
Eplant ¼
ðpot þ plant weightÞ empty pot weight 18
1165 (1)
where 18 is the molar mass of water. Similarly, the instantaneous rate of leaf transpiration (Eleaf) was calculated as: Eleaf
Eplant 50400 1000 ¼ Leaf area
(2)
where 50ˆ400 is the number of seconds during the light period of 14 h, and 1000 is the conversion from moles to mmoles. To calculate hg, a core of known size was removed from the pots at the end of D3 and this was oven dried at 100 C over 72 h to a constant weight and the total dry weight of soil per pot was determined. The hg for each of the nine points was then calculated as: hg ¼
Mwet Mdry Mdry
(3)
where Mwet is the soil fresh weight (g) at each time point, and Mdry is the soil dry weight, oven-dried at 100 C over 72 h to a constant weight.
Leaf water potential
Leaf water potential (Wleaf) was measured on detached leaves using a Scholander pressure chamber (Model 1000 Pressure Chamber, PMS Instruments, Corvallis, OR, USA), immediately following lysimetry measurements. Plants were watered to pot capacity before ‘dawn’ on D1, 20 DAP, allowed to drain and returned to the growth cabinets. Wleaf was then measured at the end of D2 (2000 h) and before ‘dawn’ (pre-dawn) on D3, 23 DAP. In the controlled environment chambers, conditions remained constant throughout the day, and we therefore expected the maximum plant water deficit to occur towards the end of the day. Pre-dawn Wleaf measurements were used to provide an estimate of soil water potential as the plants returned to equilibrium with soil water during the dark (Tardieu and Simonneau, 1998). Only two water potential measurements were taken, due to the destructive nature of the technique, and small size of the plants (20 – 40 leaves).
Biomass
Total biomass was determined by a destructive harvest immediately following the completion of water potential measurements (24 DAP). Plants were divided into roots and shoots, washed clean of the growth medium, dried at 70 C for 7 d and weighed.
Experimental design and statistical analysis
Four species were used for the experiment, with four replicates at each pCO2 treatment. Statistical analysis was carried out using the computing package R (version 3.0.1, The R Foundation for Statistical Computing), with P ¼ 005 as the
Cunniff et al. — Sub-ambient pCO2 and plant water status 0·4
A
20
0·3
10
0·1
5
0
n.s
∗∗∗
∗∗∗
n.s
C
∗∗∗
∗∗
D
0 20
0·3
15
0·2
A (μmol m–2 s–1)
15
0·2
10 0·1
5 ∗∗∗
∗∗∗
0 0·9
n.s
∗∗∗
n.s
n.s
E
RESULTS
F
15 10
0·3
5 ∗
0
∗
∗∗∗
∗∗∗
n.s
G
∗∗∗
H
0 20
0·6
15
A (μmol m–2 s–1)
gs (mol H2O m–2 s–1)
0 20
0·6
Stomatal conductance and photosynthesis
10
0·3
5 3
0
D t-
D t-
t-
in
in id
po
po id M
∗
2
1 D
3 in po id M
M
id
po
in
t-
t-
D
D
2
1 D id
po
in
tin po id
∗∗∗
∗
n.s
M
∗
∗∗
0
M
At the midpoint of D1 in P. miliaceum, gs was 31 % lower under post-glacial pCO2 than glacial pCO2 (Fig. 1A). This difference in gs between pCO2 levels was maintained for D2, but then declined in glacial pCO2 and, by the midpoint of D3, was not significantly different from the value in post-glacial pCO2 (Fig. 1A). This strong decline in gs under 18 Pa pCO2 is supported by a significant interaction between pCO2 and day (F2,62 ¼ 88, P < 0001). Furthermore, gs showed a steeper decline at the end of each day, where, on all days, at the ‘p.m.’ measurements, there was no significant difference in gs between pCO2 levels (Supplementary Data Fig. S1). Photosynthesis was impacted directly by the growth pCO2 in P. miliaceum (F1,64 ¼ 598, P < 0001) but also indirectly via the faster decline in gs at 18 Pa, which fed back to limit A (Fig. 1B). At the midpoint of D1, A was equal between the pCO2 levels. However, by D2 A differed, and by the midpoint of D3 A was 46 % greater at post-glacial than at glacial pCO2 (Fig. 1B). This change led to a significant interaction between pCO2 and day (F1,64 ¼ 57, P < 0001). Pi appeared to track gs, decreasing from D1 to D3 at 18 Pa (Fig. 2A, F2,66 ¼ 39, P < 005), with a minimum Pi always being reached at ‘p.m.’ on each day (Supplementary Data Fig. S2). This decline in Pi is not significantly greater in the glacial than post-glacial treatment, but Pi is lower overall at glacial pCO2 (F1,66 ¼ 897, P < 0001), meaning that the same decline in Pi is more limiting for A in this environment than in plants grown at post-glacial pCO2. Values of gs in the second C4 species, S. viridis, showed a very similar pattern to those of P. miliaceum (Fig. 1C). At the midpoint of D1, gs was lower at 27 Pa in comparison to at 18 Pa pCO2. However, gs declined by a greater extent at glacial pCO2 (F2,64 ¼ 66, P < 001), and by D3 gs was not significantly different between the two levels of pCO2 (Fig. 1C). Higher pCO2 led to greater A overall (F1,62 ¼ 269, P < 0001) but the effect was not seen on D1, and only became significant on D2 during the ‘a.m.’ and ‘p.m.’ measurements (Fig. 1D, Fig. S1). Mirroring the response of gs, the decrease in A was more pronounced at 18 Pa than at 27 Pa pCO2 (F1,62 ¼ 38, P < 005). Pi was lower at 18 Pa than at 27 Pa pCO2 (Fig. 2B, F1,66 ¼ 998, P < 0001). Although Pi looked to decline at the midpoint of D3 at glacial pCO2, there was no significant interaction between pCO2 and day.
25
B
M
critical level of significance. Throughout the analysis the species were treated separately. Analysis of variance (ANOVA) with repeated measure factors was used to analyse the data. For the parameters A, gs, hg and Pi, a three-factor mixed design (pCO2, time and day) with repeated measures on time and day was used, and for the parameters Eleaf and Wleaf a two-factor design (pCO2 and day) with repeated measures on day was employed. The estimable function [(library(gmodels))] was used to apply a contrast matrix to the data, by computing a significance value between the levels of pCO2 at each time point. The data collected at the final harvest were evaluated using Student’s t-test to investigate the effect of pCO2 on biomass and partitioning.
gs (mol H2O m–2 s–1)
1166
FIG. 1. Changes in stomatal conductance [(gs) A, C, E and G] and photosynthesis [(A) B, D, F and H] at the midpoint of each day over a 3-d drying cycle (D1– D3) in two C4 crop progenitors: (A, B) P. miliaceum and (C, D) S. viridis; and two C3 crop progenitors: (E, F) H. spontaneum and (G, H) T. boeoticum grown at pCO2 levels of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s. ¼ not significant. Stomatal conductance is set to a different scale for the C3 and C4 species. See Fig. S1 – for measurements taken at nine time points (a.m., midpoint and p.m.) over the 3 d.
Measurements of gs were significantly greater at 18 Pa pCO2 for the C3 species, H. spontaneum (Fig. 1E, F1,64 ¼ 117, P < 001). At the midpoint of D1, gs was 23 % lower at post-glacial pCO2, and gs then declined from D1 to D3; with the decline being greater under glacial pCO2 (Fig. 1E, F1,64 ¼ 58, P < 001). There was a strong direct effect of pCO2 on A (Fig. 1F, F1,62 ¼ 2601, P < 0001), but the indirect effect mediated via the reduction in gs was less clear. Photosynthesis declined marginally from D1 to D3 under both pCO2 levels, and the decline was not significantly greater at glacial compared to post-glacial pCO2 (Fig. 1F). Similarly, Pi was strongly affected by pCO2 (Fig. 2C, F1,64 ¼ 3614, P < 0001) and showed a small decline on D3, particularly at the ‘p.m.’ measurement (Fig. S2); however, the reduction in Pi was no more rapid at 18 Pa pCO2. In the second C3 species, T. boeoticum, gs was 38 % greater in glacial pCO2 at the midpoint of D1, but by the midpoint of
Cunniff et al. — Sub-ambient pCO2 and plant water status 12
2·0
A
1167
A
B
1·6
10
1·2
6 4
Pi (Pa)
2 ∗∗∗
0
∗
∗
B
10
Eleaf (mmol H2O m–2 s–1)
8
0·8
n.s
0·4 ∗
0 4
∗∗∗
2
6
2 ∗∗
0
∗∗
∗∗∗
∗∗∗
3
0
4
∗∗∗ D
C
n.s
1
8
n.s
∗
∗∗∗
D1
D2
D3
∗∗
n.s
n.s
D1
D2
D3
FIG. 3. Changes in daily transpiration at the leaf scale (Eleaf) in the C4 species (A) P. miliaceum and (B) S. viridis and the C3 species (C) H. spontaneum and (D) T. boeoticum, over a 3-d drying cycle (D1–D3). Plants were grown at pCO2 of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s ¼ not significant. Values of Eleaf are set to a different scale for the C3 and C4 species.
C 20
15 10
Pi (Pa)
5 ∗∗∗
∗∗∗
∗∗
0
D 20
D3 the difference between pCO2 levels had declined to 18 % and was not significant (Fig. 1G). However, the decline of gs at 18 Pa was not strong and the interaction between pCO2 and day was not significant. The 18 Pa pCO2 level did, however, show an end-of-day decline at the ‘p.m.’ measurement, which became stronger from D1 to D3 (Fig. S1). Photosynthesis was significantly lower at glacial pCO2, and showed little change at each pCO2 between D1 and D3 (Fig. 1H, F1,68 ¼ 703, P < 0001). At glacial pCO2, A fluctuated marginally until the end of D3, when it dropped to 87 lmol m2 s1, and this corresponded with the greatest decrease in gs (Fig. S1). Consistent with these results, values of Pi showed no overall decline in either pCO2 (Fig. 2D).
15
Transpiration 10 5 ∗∗∗
∗∗∗
∗∗ 3 D tM
id po
in
tin id po M
M
id
po i
nt -
D
D
1
2
0
FIG. 2. Changes in intercellular pCO2 (Pi) at the midpoint of each day over a 3-d drying cycle (D1–D3), in (A) P. miliaceum, (B) S. viridis, (C) H. spontaneum and (D) T. boeoticum grown at pCO2 levels of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s. ¼ not significant. Values of Pi are set to a different scale for the C3 and C4 species. See Fig. S2 for measurements taken at nine time points (a.m., midpoint and p.m.) over the 3 d.
Independent measurements of in situ water loss per unit leaf area, calculated using lysimetry, were consistent with the observed relationship between pCO2 and gs (Figs 1 and 3). In P. miliaceum and S. viridis, Eleaf was lower at post-glacial pCO2 on D1 by 25 and 43 %, respectively, but a steeper decline in Eleaf at glacial pCO2 meant that, by D3, there was no difference between the pCO2 levels (Fig. 3A, B). As a result, there was a significant interaction between pCO2 and day for both P. miliaceum (F2,18 ¼ 61, P < 001) and S. viridis (F2,18 ¼ 184, P < 0001). Similarly, in the C3 species H. spontaneum, Eleaf was 25 % lower at post-glacial pCO2 on D1 yet, by D3, there was no significant difference between the pCO2 levels and this was due to Eleaf declining more rapidly under glacial than post-glacial pCO2 (Fig. 3C, F2,18 ¼ 128, P
Reduced plant water status under sub-ambient pCO2 limits plant productivity in the wild progenitors of C3 and C4 cereals Jennifer Cunniff1,*, Michael Charles2, Glynis Jones3 and Colin P. Osborne1 1
Department of Animal and Plant Sciences, Alfred Denny Building, University of Sheffield, Western Bank, Sheffield S10 2TN, UK, 2Institute of Archaeology, 36 Beaumont St, Oxford OX1 2PG, UK and 3Department of Archaeology, Northgate House, University of Sheffield, West Street, Sheffield S1 4ET, UK *For correspondence. Current address: CABI, Nosworthy Way, Wallingford, Oxfordshire OX10 8DE, UK. E-mail [email protected] Received: 28 February 2016 Returned for revision: 25 April 2016 Accepted: 13 June 2016 Published electronically: 29 August 2016
Background and Aims The reduction of plant productivity by low atmospheric CO2 partial pressure (pCO2) during the last glacial period is proposed as a limiting factor for the establishment of agriculture. Supporting this hypothesis, previous work has shown that glacial pCO2 limits biomass in the wild progenitors of C3 and C4 founder crops, in part due to the direct effects of glacial pCO2 on photosynthesis. Here, we investigate the indirect role of pCO2 mediated via water status, hypothesizing that faster soil water depletion at glacial (18 Pa) compared to postglacial (27 Pa) pCO2, due to greater stomatal conductance, feeds back to limit photosynthesis during drying cycles. Methods We grew four wild progenitors of C3 and C4 crops at glacial and post-glacial pCO2 and investigated physiological changes in gas exchange, canopy transpiration, soil water content and water potential between regular watering events. Growth parameters including leaf area were measured. Key Results Initial transpiration rates were higher at glacial pCO2 due to greater stomatal conductance. However, stomatal conductance declined more rapidly over the soil drying cycle in glacial pCO2 and was associated with decreased intercellular pCO2 and lower photosynthesis. Soil water content was similar between pCO2 levels as larger leaf areas at post-glacial pCO2 offset the slower depletion of water. Instead the feedback could be linked to reduced plant water status. Particularly in the C4 plants, soil–leaf water potential gradients were greater at 18 Pa compared with 27 Pa pCO2, suggesting an increased ratio of leaf evaporative demand to supply. Conclusions Reduced plant water status appeared to cause a negative feedback on stomatal aperture in plants at glacial pCO2, thereby reducing photosynthesis. The effects were stronger in C4 species, providing a mechanism for reduced biomass at 18 Pa. These results have added significance when set against the drier climate of the glacial period. Key words: Setaria viridis, Panicum miliaceum var. ruderale, Hordeum spontaneum, Triticum boeoticum, sub-ambient pCO2, origin of agriculture, C4 photosynthesis, C3 photosynthesis, water relations, crop progenitors.
INTRODUCTION Deglaciation at the end of the Pleistocene period was coupled to a rapid rise in atmospheric pCO2 from below 18 Pa to 27 Pa between 15000 and 12000 years ago (Jouzel et al., 1993; Petit et al., 1999). Humans began to cultivate plants in the Fertile Crescent of western Asia soon afterwards (Willcox et al., 2008) and, within five millennia, cultivation and subsequent domestication of plants had occurred in at least five primary centres across the globe (Purugganan and Fuller, 2009). This sequence of agricultural origins in widely separated regions of the world suggests the involvement of a global factor, and Sage (1995) proposed that pCO2 during the last glacial period may have been too low to support the level of productivity required for the successful establishment of agriculture. An experimental test of this hypothesis showed that the wild progenitors of C3 and C4 founder cereals from independent centres of origin displayed significant increases in vegetative biomass with a 50 % increase in pCO2 from 18 to 27 Pa (Cunniff et al., 2008, 2010). Biomass of the one C3 species in this experiment near-doubled and measurements of leaf gas exchange
revealed that photosynthesis (A) was strongly limited by glacial pCO2 in this species, highlighting a direct mechanism for biomass limitation. Biomass in the five C4 species showed a smaller, but still substantial, 40 % increase (Cunniff et al., 2008). However, measurements of leaf gas exchange immediately after watering revealed that A was only significantly limited by glacial pCO2 in two of these C4 crop progenitors, a finding consistent with the known CO2-concentrating mechanism of C4 photosynthesis (Pearcy and Ehleringer, 1984). Therefore, an alternative, indirect explanation for the biomass increase is required. The elevation of pCO2 from a glacial to postglacial level in this experiment was associated with large reductions in stomatal conductance (gs) and transpiration (E), resulting in a decrease in the use of water at leaf and canopy scales in all of the C4 species, and greatly improved water-use efficiency (WUE). These observations raise the possibility of an indirect feedback on biomass accumulation at sub-ambient pCO2 mediated via water relations because, although plants were not exposed to drought per se, they experienced a soil drying cycle between watering events on alternate days (Cunniff et al., 2008). This
C The Author 2016. Published by Oxford University Press on behalf of the Annals of Botany Company. V
All rights reserved. For Permissions, please email: [email protected]
1164
Cunniff et al. — Sub-ambient pCO2 and plant water status
hypothesized mechanism is consistent with studies investigating the effects of elevated atmospheric pCO2 (55–70 Pa) on the water relations of C4 plants, where reduced gs and E improve plant and soil water status, and extend the active period for photosynthesis and growth (Ghannoum et al., 2000). This indirect effect of elevated pCO2 is particularly important when C4 plants experience some kind of water deficit (Conley et al., 2001; Wall et al., 2001; Leakey et al., 2004, 2006; Leakey, 2009). It seems likely that these feedbacks of water status on stomatal conductance, and hence photosynthesis, should be greater at sub-ambient than elevated pCO2. This is because the response of A to intercellular pCO2 (Pi) in C4 plants is typically saturated at ambient pCO2 levels, but shows a steep, near-linear response to pCO2 at lower Pi (Pearcy and Ehleringer, 1984). Therefore, decreases in gs under water deficits could lead to substantial reductions in A. This effect has been recognized in studies using maize grown at ambient and elevated pCO2 (Samarakoon and Gifford, 1995, 1996). It has been demonstrated previously that WUE improves linearly across sub-ambient pCO2 gradients in both C3 and C4 species (Polley et al., 1993, 1996, 2002; Anderson et al., 2001). However, few studies have investigated the interacting effects of water limitation and sub-ambient pCO2. Polley et al. (2002) showed that mid-day xylem potentials were depressed by subambient pCO2 in C3 and C4 species during naturally occurring seasonal droughts. Furthermore, Ward et al. (1999) found that A was limited in the C4 annual Amaranthus retroflexus by subambient pCO2 and, when a drought treatment was applied experimentally, the CO2 limitation was stronger. Additionally, these plants showed the least recovery of leaf area and biomass upon re-watering at the end of the drought period. These responses at sub-ambient pCO2 were similar to those of the C3 annual Abutilon theophrasti and showed that C3 and C4 species were equally affected by drought at sub-ambient pCO2, which is predicted from other studies (Polley et al., 1993, 1995). Here we report a controlled environment experiment investigating the interacting effects of sub-ambient pCO2, plant and soil water status on the modern-day representatives of C3 and C4 crop progenitors. We focus on two C3 cereals from the Fertile Crescent and two C4 millets from the Loess Plateau in China, all of which were among the earliest wild plants to be brought into cultivation. Our experiment tested three hypotheses: (1) greater gs at glacial pCO2 increases the rate of soil water depletion and decreases leaf and plant water status compared with plants at post-glacial pCO2; (2) faster soil water depletion and reduced plant water status increase the rate at which gs and A decline in plants grown at glacial compared with post-glacial pCO2; and (3) water deficits amplify the limiting effect of glacial pCO2 on A to an equal extent in C4 and C3 crop progenitors. The aim was not to subject the crop progenitors to a drought treatment; rather, it was an investigation of how physiology changes between watering events in a controlled environment experiment. MATERIALS AND METHODS Growth conditions and plant material
The pCO2 treatments were applied in controlled environment chambers (Conviron BDR16, Conviron, Winnipeg, Manitoba,
Canada) at two levels throughout the full 24 d of plant growth: glacial (18 Pa) and post-glacial (27 Pa). pCO2 levels were maintained continuously throughout the day and night. The chambers were operated in a closed configuration, by connecting the outlet vent to the air inlet via a filter packed with a layer of activated charcoal and a layer of sodalime (Sofnolime 10–25-mm granules, Molecular Products Ltd, Mill End, UK); activated charcoal was employed to filter the air and remove any trace gases which could be emitted by plants or soil, and have the potential to affect plant development. The pCO2 level in each chamber was measured using a CO2 sensor (CARBOCAP Carbon Dioxide Probe GMP343, Vaisala, Finland) calibrated against a secondary standard. CO2 control was achieved by linking the sensor to a feedback system regulating the circulation of chamber air through the soda-lime scrubber. The pCO2 of air in the chambers was recorded every minute, giving overall mean values over the full growth period of 182 Pa (6 s.d. 048) and 271 Pa (6 s.d. 049). To maintain this tight control, the soda lime was changed as soon as pCO2 started to drift above the target level, which was approximately every 4 weeks. Treatment and plants were exchanged between the two controlled environment chambers every week from germination to avoid confounding the effects of chamber and growth environment. Seeds were obtained from germplasm holdings or commercial sources. They included two C4 species, Setaria viridis (L.) P. Beauv [(green foxtail millet) (Herbiseed, Twyford, UK; cat. no. 9602)] and Panicum miliaceum var. ruderale (Kitag.) [(wild broomcorn millet) (Herbiseed, cat. no. 9507)] from North China, and two C3 species, Hordeum spontaneum K. Koch [(wild barley) (Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany, accession no.: HOR 13791)] and Triticum boeoticum Boiss. [(einkorn wheat) (IPK, accession no.: TRI 17111)] from western Asia. Batches of 20 seeds of each species were sown into trays containing a 1:1 sand:John Innes no. 2 compost mix (seven parts loam, three parts peat, two parts sand N/P/K 20:10:10). This mix was chosen in an attempt to replicate an unimproved soil (Ivandic et al., 2000). Seeds were germinated under pCO2 treatments in the controlled environment chambers at 25/15 C (day/night), with an 8-h photoperiod and photosynthetic photon flux density (PPFD) of 300 lmol photons m2 s1. Once established, eight even-sized seedlings of each species were selected and planted into 32 individual 2-litre pots [132 cm (height) 17 cm (diameter)], containing the same growth media. The pots had eight holes in the base and were free to drain. Seedlings were returned to the same controlled environment chambers; and in both chambers temperature was maintained at 25/15 C (day/night), photoperiod was increased to 14 h with a maximum PPFD of 600 lmol m2 s1, and vapour pressure deficit (VPD) had a minimum value of 049 kPa during the night and a maximum value of 096 kPa during the day.
Physiology
Physiological measurements of gas exchange, canopy transpiration and gravimetric water content, and leaf water potential were taken over three separate drying cycles of 3 d duration: gas exchange [14–16 d after planting (DAP)], canopy
Cunniff et al. — Sub-ambient pCO2 and plant water status transpiration and gravimetric water content (17–19 DAP) and water potential (20–22 DAP). In total, physiological measurements occurred over 9 d (14–22 DAP). For each drying cycle, soil in the pots was watered to pot capacity before ‘dawn’ (0700ˆh) on day 1 (D1), and left to dry down over a 3-d cycle (D1–D3) before re-watering at the end of D3. We used a 3-d watering cycle as opposed to the 2-d cycle applied in our previous experiment (Cunniff et al., 2008) to exaggerate the hypothesized CO2 water interaction.
Gas exchange measurements
CO2 and H2O exchange were measured using a portable open gas exchange system (LI-6400P, LI-COR Biosciences, Lincoln, NB, USA) with the 6400-02B LED Light Source chamber. Chamber conditions were set with the aim of approximating the growth environment: PPFD was 600 lmol m2 s1, leaf temperature was 25 C, incoming air was maintained at a constant humidity to keep the leaf-air VPD at less than 1 kPa throughout all measurements, and pCO2 was set at 18 or 27 Pa as appropriate. Plants were watered to pot capacity before ‘dawn’ on D1 (14 DAP), allowed to drain and returned to the growth cabinets. Gas exchange was measured at nine time points: the beginning (0800ˆh), midpoint (1400ˆh) and end of the day (2000ˆh) from D1 to D3 (16 DAP). At each time point, measurements were commenced in a different cabinet, and were alternated between the two cabinets within each time point. This avoided confounding time with CO2 treatment, within the approximately 90 min that it took to complete all measurements. On each occasion, the same, most recently expanded leaf was clamped in the leaf chamber and allowed to equilibrate with chamber conditions for 90 s, before measuring leaf gas exchange. The aim of these measurements was to obtain ‘snapshots’ of photosynthesis (A), stomatal conductance (gs) and intercellular pCO2 (Pi) under growth conditions, and these were calculated using the equations of von Caemmerer and Farquhar (1981).
Transpiration and gravimetric soil water content
Transpiration and gravimetric soil water content (hg) were determined via lysimetry. Plants were watered to pot capacity before ‘dawn’ on D1 17 DAP, allowed to drain and then weighed (g). Pots were then returned to the controlled environment chambers and reweighed (g) at each of the nine time points detailed for gas exchange measurements. Alongside the plants, empty pots containing only the growth media were placed in the controlled environment chambers to estimate soil evaporation. These were treated in the same way as the plants, being weighed at saturation and the nine measurement occasions. At the end of D3 19 DAP, leaf area per plant was measured non-destructively using a ruler. Length maximum width was measured for each leaf and used to estimate the total canopy area using an allometric relationship established before the experiment (Supplementary Data Table S1). The daily canopy transpiration of each plant (Eplant) was determined as:
Eplant ¼
ðpot þ plant weightÞ empty pot weight 18
1165 (1)
where 18 is the molar mass of water. Similarly, the instantaneous rate of leaf transpiration (Eleaf) was calculated as: Eleaf
Eplant 50400 1000 ¼ Leaf area
(2)
where 50ˆ400 is the number of seconds during the light period of 14 h, and 1000 is the conversion from moles to mmoles. To calculate hg, a core of known size was removed from the pots at the end of D3 and this was oven dried at 100 C over 72 h to a constant weight and the total dry weight of soil per pot was determined. The hg for each of the nine points was then calculated as: hg ¼
Mwet Mdry Mdry
(3)
where Mwet is the soil fresh weight (g) at each time point, and Mdry is the soil dry weight, oven-dried at 100 C over 72 h to a constant weight.
Leaf water potential
Leaf water potential (Wleaf) was measured on detached leaves using a Scholander pressure chamber (Model 1000 Pressure Chamber, PMS Instruments, Corvallis, OR, USA), immediately following lysimetry measurements. Plants were watered to pot capacity before ‘dawn’ on D1, 20 DAP, allowed to drain and returned to the growth cabinets. Wleaf was then measured at the end of D2 (2000 h) and before ‘dawn’ (pre-dawn) on D3, 23 DAP. In the controlled environment chambers, conditions remained constant throughout the day, and we therefore expected the maximum plant water deficit to occur towards the end of the day. Pre-dawn Wleaf measurements were used to provide an estimate of soil water potential as the plants returned to equilibrium with soil water during the dark (Tardieu and Simonneau, 1998). Only two water potential measurements were taken, due to the destructive nature of the technique, and small size of the plants (20 – 40 leaves).
Biomass
Total biomass was determined by a destructive harvest immediately following the completion of water potential measurements (24 DAP). Plants were divided into roots and shoots, washed clean of the growth medium, dried at 70 C for 7 d and weighed.
Experimental design and statistical analysis
Four species were used for the experiment, with four replicates at each pCO2 treatment. Statistical analysis was carried out using the computing package R (version 3.0.1, The R Foundation for Statistical Computing), with P ¼ 005 as the
Cunniff et al. — Sub-ambient pCO2 and plant water status 0·4
A
20
0·3
10
0·1
5
0
n.s
∗∗∗
∗∗∗
n.s
C
∗∗∗
∗∗
D
0 20
0·3
15
0·2
A (μmol m–2 s–1)
15
0·2
10 0·1
5 ∗∗∗
∗∗∗
0 0·9
n.s
∗∗∗
n.s
n.s
E
RESULTS
F
15 10
0·3
5 ∗
0
∗
∗∗∗
∗∗∗
n.s
G
∗∗∗
H
0 20
0·6
15
A (μmol m–2 s–1)
gs (mol H2O m–2 s–1)
0 20
0·6
Stomatal conductance and photosynthesis
10
0·3
5 3
0
D t-
D t-
t-
in
in id
po
po id M
∗
2
1 D
3 in po id M
M
id
po
in
t-
t-
D
D
2
1 D id
po
in
tin po id
∗∗∗
∗
n.s
M
∗
∗∗
0
M
At the midpoint of D1 in P. miliaceum, gs was 31 % lower under post-glacial pCO2 than glacial pCO2 (Fig. 1A). This difference in gs between pCO2 levels was maintained for D2, but then declined in glacial pCO2 and, by the midpoint of D3, was not significantly different from the value in post-glacial pCO2 (Fig. 1A). This strong decline in gs under 18 Pa pCO2 is supported by a significant interaction between pCO2 and day (F2,62 ¼ 88, P < 0001). Furthermore, gs showed a steeper decline at the end of each day, where, on all days, at the ‘p.m.’ measurements, there was no significant difference in gs between pCO2 levels (Supplementary Data Fig. S1). Photosynthesis was impacted directly by the growth pCO2 in P. miliaceum (F1,64 ¼ 598, P < 0001) but also indirectly via the faster decline in gs at 18 Pa, which fed back to limit A (Fig. 1B). At the midpoint of D1, A was equal between the pCO2 levels. However, by D2 A differed, and by the midpoint of D3 A was 46 % greater at post-glacial than at glacial pCO2 (Fig. 1B). This change led to a significant interaction between pCO2 and day (F1,64 ¼ 57, P < 0001). Pi appeared to track gs, decreasing from D1 to D3 at 18 Pa (Fig. 2A, F2,66 ¼ 39, P < 005), with a minimum Pi always being reached at ‘p.m.’ on each day (Supplementary Data Fig. S2). This decline in Pi is not significantly greater in the glacial than post-glacial treatment, but Pi is lower overall at glacial pCO2 (F1,66 ¼ 897, P < 0001), meaning that the same decline in Pi is more limiting for A in this environment than in plants grown at post-glacial pCO2. Values of gs in the second C4 species, S. viridis, showed a very similar pattern to those of P. miliaceum (Fig. 1C). At the midpoint of D1, gs was lower at 27 Pa in comparison to at 18 Pa pCO2. However, gs declined by a greater extent at glacial pCO2 (F2,64 ¼ 66, P < 001), and by D3 gs was not significantly different between the two levels of pCO2 (Fig. 1C). Higher pCO2 led to greater A overall (F1,62 ¼ 269, P < 0001) but the effect was not seen on D1, and only became significant on D2 during the ‘a.m.’ and ‘p.m.’ measurements (Fig. 1D, Fig. S1). Mirroring the response of gs, the decrease in A was more pronounced at 18 Pa than at 27 Pa pCO2 (F1,62 ¼ 38, P < 005). Pi was lower at 18 Pa than at 27 Pa pCO2 (Fig. 2B, F1,66 ¼ 998, P < 0001). Although Pi looked to decline at the midpoint of D3 at glacial pCO2, there was no significant interaction between pCO2 and day.
25
B
M
critical level of significance. Throughout the analysis the species were treated separately. Analysis of variance (ANOVA) with repeated measure factors was used to analyse the data. For the parameters A, gs, hg and Pi, a three-factor mixed design (pCO2, time and day) with repeated measures on time and day was used, and for the parameters Eleaf and Wleaf a two-factor design (pCO2 and day) with repeated measures on day was employed. The estimable function [(library(gmodels))] was used to apply a contrast matrix to the data, by computing a significance value between the levels of pCO2 at each time point. The data collected at the final harvest were evaluated using Student’s t-test to investigate the effect of pCO2 on biomass and partitioning.
gs (mol H2O m–2 s–1)
1166
FIG. 1. Changes in stomatal conductance [(gs) A, C, E and G] and photosynthesis [(A) B, D, F and H] at the midpoint of each day over a 3-d drying cycle (D1– D3) in two C4 crop progenitors: (A, B) P. miliaceum and (C, D) S. viridis; and two C3 crop progenitors: (E, F) H. spontaneum and (G, H) T. boeoticum grown at pCO2 levels of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s. ¼ not significant. Stomatal conductance is set to a different scale for the C3 and C4 species. See Fig. S1 – for measurements taken at nine time points (a.m., midpoint and p.m.) over the 3 d.
Measurements of gs were significantly greater at 18 Pa pCO2 for the C3 species, H. spontaneum (Fig. 1E, F1,64 ¼ 117, P < 001). At the midpoint of D1, gs was 23 % lower at post-glacial pCO2, and gs then declined from D1 to D3; with the decline being greater under glacial pCO2 (Fig. 1E, F1,64 ¼ 58, P < 001). There was a strong direct effect of pCO2 on A (Fig. 1F, F1,62 ¼ 2601, P < 0001), but the indirect effect mediated via the reduction in gs was less clear. Photosynthesis declined marginally from D1 to D3 under both pCO2 levels, and the decline was not significantly greater at glacial compared to post-glacial pCO2 (Fig. 1F). Similarly, Pi was strongly affected by pCO2 (Fig. 2C, F1,64 ¼ 3614, P < 0001) and showed a small decline on D3, particularly at the ‘p.m.’ measurement (Fig. S2); however, the reduction in Pi was no more rapid at 18 Pa pCO2. In the second C3 species, T. boeoticum, gs was 38 % greater in glacial pCO2 at the midpoint of D1, but by the midpoint of
Cunniff et al. — Sub-ambient pCO2 and plant water status 12
2·0
A
1167
A
B
1·6
10
1·2
6 4
Pi (Pa)
2 ∗∗∗
0
∗
∗
B
10
Eleaf (mmol H2O m–2 s–1)
8
0·8
n.s
0·4 ∗
0 4
∗∗∗
2
6
2 ∗∗
0
∗∗
∗∗∗
∗∗∗
3
0
4
∗∗∗ D
C
n.s
1
8
n.s
∗
∗∗∗
D1
D2
D3
∗∗
n.s
n.s
D1
D2
D3
FIG. 3. Changes in daily transpiration at the leaf scale (Eleaf) in the C4 species (A) P. miliaceum and (B) S. viridis and the C3 species (C) H. spontaneum and (D) T. boeoticum, over a 3-d drying cycle (D1–D3). Plants were grown at pCO2 of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s ¼ not significant. Values of Eleaf are set to a different scale for the C3 and C4 species.
C 20
15 10
Pi (Pa)
5 ∗∗∗
∗∗∗
∗∗
0
D 20
D3 the difference between pCO2 levels had declined to 18 % and was not significant (Fig. 1G). However, the decline of gs at 18 Pa was not strong and the interaction between pCO2 and day was not significant. The 18 Pa pCO2 level did, however, show an end-of-day decline at the ‘p.m.’ measurement, which became stronger from D1 to D3 (Fig. S1). Photosynthesis was significantly lower at glacial pCO2, and showed little change at each pCO2 between D1 and D3 (Fig. 1H, F1,68 ¼ 703, P < 0001). At glacial pCO2, A fluctuated marginally until the end of D3, when it dropped to 87 lmol m2 s1, and this corresponded with the greatest decrease in gs (Fig. S1). Consistent with these results, values of Pi showed no overall decline in either pCO2 (Fig. 2D).
15
Transpiration 10 5 ∗∗∗
∗∗∗
∗∗ 3 D tM
id po
in
tin id po M
M
id
po i
nt -
D
D
1
2
0
FIG. 2. Changes in intercellular pCO2 (Pi) at the midpoint of each day over a 3-d drying cycle (D1–D3), in (A) P. miliaceum, (B) S. viridis, (C) H. spontaneum and (D) T. boeoticum grown at pCO2 levels of 18 Pa (closed symbols) and 27 Pa (open symbols). Data are means 6 s.e. of four replicates. Significance codes are ***P < 0001, **P < 001, and *P < 005, n.s. ¼ not significant. Values of Pi are set to a different scale for the C3 and C4 species. See Fig. S2 for measurements taken at nine time points (a.m., midpoint and p.m.) over the 3 d.
Independent measurements of in situ water loss per unit leaf area, calculated using lysimetry, were consistent with the observed relationship between pCO2 and gs (Figs 1 and 3). In P. miliaceum and S. viridis, Eleaf was lower at post-glacial pCO2 on D1 by 25 and 43 %, respectively, but a steeper decline in Eleaf at glacial pCO2 meant that, by D3, there was no difference between the pCO2 levels (Fig. 3A, B). As a result, there was a significant interaction between pCO2 and day for both P. miliaceum (F2,18 ¼ 61, P < 001) and S. viridis (F2,18 ¼ 184, P < 0001). Similarly, in the C3 species H. spontaneum, Eleaf was 25 % lower at post-glacial pCO2 on D1 yet, by D3, there was no significant difference between the pCO2 levels and this was due to Eleaf declining more rapidly under glacial than post-glacial pCO2 (Fig. 3C, F2,18 ¼ 128, P
