Introduction

Isoprene (2-methyl-1,3-butadiene) is a volatile organic compound emitted by a number of plant species and with important effects on atmospheric chemistry (Fuentes et al. 2000; Monson and Holland 2001; Carlton et al. 2009). Isoprene biosynthesis has been proposed as an important adaptation for plants, potentially contributing to tolerance of high leaf temperatures, high photon flux densities and/or metabolic homeostasis (Loreto and Schnitzler 2010). Isoprene emission rates increase with rising temperatures (Monson and Fall 1989; Monson et al. 1994; Sharkey et al. 1999; Singsaas et al. 1999; Petron et al. 2001), but decline with increasing CO2 concentrations (Sanadze 1964; Rosenstiel et al. 2003; Wilkinson et al. 2009). Thus, the potential for isoprene to function in an adaptive role may have changed during past geologic epochs and may differ in the future as the atmosphere and climate continue to change.

One proposed biological function for isoprene (as well as mono- and sesquiterpenes) is abiotic stress protection. Isoprene acts as a thermoprotective molecule, potentially stabilizing chloroplast membranes during high temperature events (e.g., Singsaas et al. 1997; Singsaas and Sharkey 2000; Behnke et al. 2007). Because isoprene production is enhanced by both high temperatures and light, it has been proposed to maintain photosynthetic capacity during rapid leaf temperature fluctuations caused by sunflecks in the canopy (Sharkey and Singsaas 1995; Behnke et al. 2010). Indeed, an in-silico experiment has demonstrated that isoprene can interact with the phospholipid bilayer of membranes to maintain membrane stability during high temperature events (Siwko et al. 2007). Isoprene also appears to act as an antioxidant compound, reducing damage caused by ozone (Loreto and Velikova 2001; Vickers et al. 2009) and reactive oxygen species (Affek and Yakir 2002; Velikova et al. 2004). All these functions should be more important during simultaneous light and heat stress, when thermotolerance and reactive oxygen quenching mechanisms are needed. Isoprene production has also been proposed as a means of maintaining metabolic homeostasis in the face of past changes in atmospheric CO2 concentration (Rosenstiel et al. 2004). In this role, isoprene production may facilitate the metabolic breakdown of excess pyruvate in chloroplasts of certain plants. It was proposed that, as atmospheric CO2 concentration declined in past geologic epochs, the import of phosphoenolpyruvate (PEP) into chloroplasts may have increased due to reduced cytosolic PEP carboxylase activity, thus creating a build-up of chloroplastic pyruvate that could not be accommodated through the highly regulated biosynthetic processes that normally utilize pyruvate in the chloroplast.

While predicting how rising CO2 concentrations will impact isoprene production is an important scientific goal, the biological importance of isoprene may be best evaluated under low CO2 conditions. This suggestion is because: (1) emission rates are highest at low CO2 (thereby maximizing both potential benefits and costs); (2) atmospheric CO2 concentrations have been low for much of the recent evolutionary history of plants (Petit et al. 1999; Zachos et al. 2001; Siegenthaler et al. 2005); and (3) the need for photosynthetic tolerance of heat and light stress should be of greatest importance at low CO2 (Cowling and Sage 1998; Sage and Kubien 2007).

The goal of our study was to determine how growth at both low and elevated atmospheric CO2 affected photosynthetic tolerance to simulated sunflecks and photosynthesis above the thermal optimum in isoprene-emitting and non-emitting plants. We examined stress tolerance in wild-type (and empty-vector control) poplar trees in which isoprene was produced at high rates, as well as in mutants in which isoprene biosynthesis was suppressed by silencing the formation of the isoprene synthase (ISPS) protein. We hypothesized that isoprene would provide a greater level of stress tolerance to photosynthesis in trees grown and measured at low CO2 concentrations than in high CO2-grown trees. In light of our hypothesis, we then discuss the potential implications of our results for the evolution of isoprene biosynthesis and the future utility of isoprene in a high CO2 atmosphere.

Materials and methods

We used four lines of the poplar hybrid Populus × canescens (syn. Populus tremula × P. alba): wild-type plants (WT); two well-characterized mutants where ISPS expression was silenced by RNA interference (RNAi) (R2 and R22); and a line transformed with the empty vector (C) to act as a control for the transgenic manipulation; for more details on the plant lines, see Behnke et al. (2007). The four lines were grown from cuttings placed in small pots filled with sterile sand. Cuttings were grown at 24°C and 200 μmol photons m−2 s−1 photosynthetic photon flux density (PPFD) with a 12-h photoperiod and ambient CO2 in misting rooms with 70% relative humidity. Once roots formed, each plant was transferred to a 10 × 10 × 36 cm pot filled with 1:1:1 (v:v:v) sand:perlite:peat and transferred to one of two growth chambers (Model M-13; Environmental Growth Chambers, Chagrin Falls, OH, USA). Plants were grown at 27:23°C day:night temperatures and 16:8 h day:night photoperiods with 700 μmol photons m−2 s−1 PPFD at canopy level from parallel sets of metal halide lamps and incandescent bulbs (Phillips MH400 and A19 100 W bulbs). At least five trees from each line were grown for 3 months at either low (190 ppm) or high (590 ppm) CO2 concentrations. CO2 concentrations were measured with an infra-red gas analyzer (LI-COR 6252, Lincoln, NE, USA) every 2–5 min. Elevated CO2 was achieved by injecting pure CO2 into the ambient airstream as needed, while low CO2 concentrations were maintained by using soda lime to scrub CO2 from the incoming air. Treatments were rotated between chambers every 3 weeks to minimize chamber effects.

Sunfleck simulations

Gas exchange was measured using an open photosynthesis system (Walz GFS-3000; Effeltrich, Germany) with a dual LED/PAM (pulse amplitude modulation) fluorometer module, programmed to provide light and leaf temperature conditions as outlined below. Simultaneous isoprene emission rates were measured by diverting a fraction of the outgoing air from the leaf cuvette to a chemi-luminescence based fast isoprene sensor (Hills Scientific, Boulder, CO, USA; described in Hills et al. 1991). Measurements on the eighth to tenth leaf from the top of each sapling were made on five trees from each line (WT, C, R2 and R22) from both CO2 treatments at growth CO2 (190 or 590 ppm) and a constant water concentration of 15,000 ppmv (50% relative humidity at 30°C). Leaves were placed in a dark cuvette programmed to provide a leaf temperature of 30°C for 30 min to measure dark respiration (measurement R d). At the end of the 30-min dark period, leaves were exposed to a 0.8-s saturating light pulse (4,500 μmol photons m−2 s−1) to measure F v/F m (the maximum efficiency of photosystem II). Light was then increased to representative growth levels (700 μmol photons m−2 s−1) with a constant leaf temperature of 30°C for another 30 min to assess unstressed net CO2 assimilation rates (measurement AU). Gas exchange was then measured while exposing leaves to two sequential, 10-min light- and heatflecks designed to simulate a sunfleck: leaf temperatures were rapidly increased to 38–39°C while PPFD was simultaneously raised to 1,600 μmol photons m−2 s−1 (measurement AS1). We maintained leaf temperature below 40°C to study reversible heat stress effects and avoid thermal damage. Samples were then given a 15-min recovery period at 30°C and 700 μmol photons m−2 s−1 PPFD (measurement AR1). After the recovery period, leaves were exposed to a second sunfleck stress (measurement AS2) and then left for a second recovery period of 25 min (measurement AR2). Electron transport rates [J = (F m′ − F/F m′)  × PPFD × 0.5 × 0.8] (from Genty et al. 1990, where 0.5 accounts for the fraction of light to photosystem II and 0.8 accounts for leaf absorptivity) were assessed before the first sunfleck and at the end of both stress events and both recovery periods. Non-photochemical quenching (NPQ = (F m − F m′)/F m′) was calculated according to Bilger and Björkman (1990).

Photosynthetic temperature response curves

The temperature response of net CO2 assimilation rates (A net) was assessed with an open photosynthesis system [either a LI-COR 6400 (Lincoln, NE, USA) or a Walz GFS-3000] for six isoprene-emitting trees and six non-emitting plants from each growth CO2 treatment (three trees from each line from each CO2 treatment). The cuvette was maintained at growth CO2 concentrations (190 or 590 ppm), saturating light (1,000 μmol photons m−2 s−1) and a constant water concentration (15,000 ppmv) to examine conditions similar to the sunfleck experiment. A net was measured on the eighth to tenth leaf from the top of the plants at 3°C intervals from 30 to 42°C with a 30-min acclimation time to achieve stable readings. Simultaneous isoprene emission rates were measured for three isoprene emitters and non-emitters from both CO2 treatments as described above; isoprene emissions were not measured from the LI-COR system due to equipment constraints.

Statistics

Because there were no differences in measured variables between WT and C lines or between R2 and R22 lines (Student’s t test, all p > 0.05), data were pooled into two groups: isoprene emitters and non-emitters. Temperature response curves of photosynthesis were analyzed using a repeated-measures ANOVA. Gas exchange from the sunfleck experiment was analyzed with ANOVAs, and differences within a CO2 treatment were compared with Student’s t tests. The specific gas-exchange measurements analyzed were dark respiration rates after 30 min of dark acclimation (R d), as well as net CO2 assimilation rates (A net): (1) after 30 min of unstressed light acclimation (AU); (2) at the end of the first light- and heatfleck stress (AS1); (3) at the end of the first recovery period (AR1); (4) at the end of the second fleck stress (AS2); and (5) at the end of the second recovery period (AR2) (see Fig. 1). We also ran a repeated-measures ANOVA on points from the simulated sunfleck experiment that were taken at the same measurement conditions (unstressed, 30°C, 700 μmol photons m−2 s−1 PPFD: AU, AR1 and AR2; and stressed, 38–40°C, 1,600 μmol photons m−2 s−1 PPFD: AS1 and AS2). The goal of the repeated-measures ANOVA was to determine if repeated temperature and light stress affected A net, J and NPQ over time. All statistics were performed in JMP 7.0.1 (SAS, Cary, NC, USA).

Fig. 1
figure 1

a Leaf temperature and light levels during the simulated sunfleck experiment, and isoprene emission rates in isoprene-emitting (filled symbols) and non-isoprene-emitting (empty symbols) leaves, measured at b low (190 ppm) and c high (590 ppm) growth CO2 concentrations. Measurements began at 30°C in the dark and light was increased to 700 μmol photons m−2 s−1 after 30 min. These conditions were maintained except for the two flecks (gray bars) where leaf temperature was increased to <40°C and light to 1,600 μmol photons m−2 s−1. Letters and arrows indicate times where representative gas exchange measurements were made for dark respiration (Rd), unstressed net CO2 assimilation rates (AU), the first light- and heatfleck stress (AS1), recovery from the first fleck (AR1), the second light- and heatfleck stress (AS2) and recovery from the second fleck (AR2). Mean ± SE, n = 20 for (a), 10 for (b and c)

Full size image

Results

Sunfleck simulations

Light levels increased rapidly during light- and heatflecks (sunflecks) while leaf temperature continued to increase over the 10-min stress (Fig. 1a). Suppressed lines showed no detectable isoprene emissions at any temperature; for isoprene-emitting trees, plants from low CO2 generally exhibited twice the isoprene emission rates of plants from high CO2 (p < 0.001; Fig. 1b, c). Isoprene emissions were suppressed in the dark, began when light was provided, and showed rapid increases during the transient high light and heat stresses of the simulated sunflecks. In both CO2 treatments, the maximum isoprene emission rate increased from the first to the second sunfleck.

Elevated CO2 increased A net at all temperatures (Fig. 2a, b; Table 1). Isoprene emitters had higher pre-stress A net (measurement AU) than non-emitting leaves as a group (Table 1), but especially at low CO2 (Fig. 2a, b). At low CO2, A net declined during the first sunfleck, increased during the recovery period, decreased more sharply during the second fleck, and rose again during the last recovery phase (Fig. 2a). At high CO2, A net increased during both flecks, declining slightly during the first recovery period compared to the initial measurement (AU), but remaining stable after the second fleck, such that measurements AR1 and AR2 were similar (Fig. 2b). Compared to A net measured at the end of the first sunfleck (AS1), A net after the second sunfleck (AS2) was reduced by an extra 0.5 μmol CO2 m−2 s−1 in isoprene-emitting leaves, but by twice this amount (1.05 μmol m−2 s−1) in non-emitting leaves, regardless of the CO2 concentration (Fig. 2a, b).

Fig. 2
figure 2

The effect of two sequential simulated sunflecks on: a, b net CO2 assimilation rates (A net); c, d net CO2 assimilation rates relative to rates measured at point AU (A rel); e, f electron transport rates (J); g, h electron transport rates relative to rates measured at point AU (J rel); i, j non-photochemical quenching (NPQ); k, l non-photochemical quenching relative to NPQ measured at point AU (NPQ rel) in isoprene-emitting (filled symbols) and non-isoprene-emitting (empty symbols) leaves, measured at their growth CO2 concentration (low, 190 ppm; high, 590 ppm). Measurements taken at points as shown in Fig. 1. AU unstressed conditions, AS 1 sunfleck 1, AR 1 recovery 1, AS 2 sunfleck 2, AR 2 recovery 2. Mean ± SE, n = 10. ns non-significant, *p < 0.10, **p < 0.05, ***p < 0.01

Full size image
Table 1 ANOVA results for the response of net CO2 assimilation (A net) during simulated sunflecks for isoprene-emitting and non-emitting poplars grown at either low (190 ppm) or high (590 ppm) CO2 concentrations
Full size table

Photosynthesis was reduced proportionally more at low CO2 than high CO2 when isoprene synthesis was suppressed. Because of differences between A net at different CO2 levels, A net values were normalized relative to pre-stress A net (measurement AU) to generate relative net CO2 assimilation rates (A rel) (Fig. 2c, d). By the end of the first 10-min sunfleck (AS1), A rel was reduced by 20% in both emitters and non-emitters at low CO2 (Fig. 2c). However, A rel continued to decline after the first sunfleck was finished, so that while leaf temperatures recovered towards 30°C, it eventually fell by 29 and 38% in isoprene-emitting and non-isoprene-emitting leaves, respectively (data not shown). By the end of the second sunfleck (AS2), A rel was reduced by 27% in emitters and 47% in non-emitters from low CO2 (Fig. 2c). At high CO2, the first fleck (AS1) increased A rel by 9 and 7% in isoprene-emitting and non-isoprene-emitting leaves, respectively, but by the second fleck (AS2), A rel was increased by 4% in emitting trees and reduced by 5% in non-emitters (Fig. 2d). After recovering from both flecks (AR2), A rel in low CO2-grown trees was only reduced by 4% in isoprene-emitting leaves, but was 14% lower in non-isoprene-emitting leaves (Fig. 2d); at high CO2, A rel was 4–7% lower in both groups at AR2 (Fig. 2d).

Electron transport rates (J) and non-photochemical quenching (NPQ) recovered better from sunflecks in isoprene-emitting leaves (Fig. 2e, f, i, j). Electron transport rates were lower in non-isoprene-emitting trees than in emitters (Table 2), and while J was reduced during sunflecks at low CO2, J increased in non-emitting leaves during each sunfleck at high CO2 (AS1 and AS2; Fig. 2e, f). Non-emitting poplars lost a greater proportion of their initial J capacity (relative J, J rel) after recovering from sunflecks, but this effect was exacerbated at low CO2; while J rel in isoprene-emitters from both CO2 concentrations was reduced by less than 4% at point AR2, J rel in non-isoprene-emitters was 8% lower at high CO2 and 14% lower at low CO2 (Fig. 2g, h). At both CO2 concentrations, isoprene-emitters showed lower NPQ values than non-emitters at 30°C (points AU, AR1 and AR2) (Fig. 2i, j; Table 2). When normalized relative to pre-stress NPQ (NPQrel), non-isoprene emitters had higher NPQrel values after recovering from sunflecks (points AR1 and AR2) and less of an increase in NPQ during heat stress (points AS1 and AS2) than isoprene-emitting plants (Fig. 2k, l; Table 2). There was no difference in F v/F m between CO2 treatments or isoprene lines (data not shown).

Table 2 Results from repeated measures ANOVAs from simulated sunflecks for A net, A rel, J, J rel, NPQ and NPQrel
Full size table

Dark respiration rates (R d) were reduced in isoprene-emitting leaves from low CO2 compared to non-isoprene-emitting leaves and to both groups at high CO2 (Fig. 3a, c; Table 1). When normalized to account for differences in A net, relative dark respiration rates were greater (p < 0.01) in non-isoprene emitting leaves from low CO2 than low CO2 emitters or either group from high CO2 (Fig. 3b, d).

Fig. 3
figure 3

Dark respiration rates in isoprene-emitting (emit, filled bars) and non-isoprene-emitting leaves (non-emit, empty bars), measured at their growth CO2 concentration (low, 190 ppm; high, 590 ppm). a, c Dark respiration rates; b, d dark respiration rates relative to net CO2 assimilation rates measured at point AU. Mean ± SE, n = 10. ns non-significant, *p < 0.10, **p < 0.05, ***p < 0.01

Full size image

Photosynthetic temperature response curves

Isoprene-emitting leaves from low CO2-grown plants produced two to three times more isoprene across all leaf temperatures than leaves grown and measured at high CO2 (Fig. 4a; Table 3). Isoprene emission, however, peaked at 39°C at low CO2, whereas it increased linearly in plants grown and measured at high CO2. As in the sunfleck experiment, no detectable isoprene was produced by the suppressed lines. Leaves from elevated CO2 displayed higher A net than low CO2 leaves, with A net declining above a leaf temperature of 30–33°C (Fig. 4b; Table 3). Both isoprene-emitting and non-emitting lines had similar A net across the temperature range measured (Fig. 4b; Table 3). At 42°C, A net was 50% lower in non-isoprene-emitters than in emitting leaves at low CO2, while at high CO2, A net was only 22% lower in non-isoprene-emitters than in isoprene-emitters (Fig. 4b). Measurements of A net were normalized relative to A net measured at 30°C to generate A rel. A rel decreased with increasing leaf temperature more at low CO2 than high CO2, and more in non-emitting leaves than in isoprene-emitting lines (Fig. 4c; Table 3). While maintaining a constant water vapor content resulted in an increase in vapor pressure deficit at the higher temperatures, there were no significant differences in stomatal conductance (g s) between emitters and non-emitters in general or between groups over the temperature response curve measurements (data not shown).

Fig. 4
figure 4

Response of a isoprene emission rates, b net CO2 assimilation rates, and c net CO2 assimilation rates relative to 30°C to leaf temperature in isoprene-emitting (filled symbols) and non-emitting (empty symbols) lines measured at their growth CO2 concentration (low, 190 ppm, circles; high, 590 ppm, squares). Mean ± SE; a n = 3 trees, b, c n = 6 trees

Full size image
Table 3 Results from a repeated-measures ANOVA for the temperature response (30–42°C) of A net in isoprene-emitting and non-emitting poplar leaves grown at either low (190 ppm) or high (590 ppm) CO2 concentrations
Full size table

Discussion

Our analysis shows that isoprene provides greater heat and light stress tolerance to photosynthesis at low CO2 than at high CO2. Leaves with suppressed isoprene formation lost a greater fraction of their photosynthetic capacity than isoprene-emitting leaves at high temperatures (>40°C), especially at low CO2 concentrations. Even at more moderate leaf temperatures, isoprene-emitting leaves maintained higher A net than non-emitting leaves at low CO2. When heat stress was applied concurrently with light stress to mimic sunflecks, isoprene-emitting leaves from low CO2 lost less and recovered more of their pre-stress photosynthetic capacity than non-emitting leaves. This photosynthetic response was mirrored by the ability of isoprene-emitting plants to recover pre-stress levels of J and NPQ after two sunfleck events.

Low CO2-grown leaves emitted twice as much isoprene as those from high CO2-grown plants. Since the effect of isoprene on thermotolerance depends on the dose of isoprene at ambient CO2 (Singsaas et al. 1997), the greater effect of isoprene at low than at high CO2 likely relates to differences in emission rates. If isoprene stabilizes protein–protein interactions in membranes at high temperatures (Sharkey and Singsaas 1995; Sharkey and Yeh 2001), increased isoprene emission rates may yield greater protection from stress-induced electron transport declines. At low CO2, declines in A net with rising leaf temperature are generally caused by increasing photorespiration and decreasing Rubisco carboxylation capacity, rather than by impairment of photosynthetic electron transport, but photosynthesis can be electron transport-limited at low CO2 as leaf temperatures rise (Sage and Kubien 2007). Thus, isoprene-induced maintenance of J would allow isoprene-emitters to quickly recover CO2 fixation after sunfleck stresses. Because isoprene emission rates are largely independent of g s (low g s causes intercellular isoprene concentrations to build up, increasing the diffusive gradient, while high g s allows for rapid diffusion, but a low build-up of isoprene concentrations in the intercellular airspace; Niinemets et al. 2004; Loreto and Schnitzler 2010), any changes in g s from changing light levels, temperatures, or vapor pressure deficits will have little effect on differences in isoprene emission rates and, therefore, photosynthetic sunfleck tolerance. The higher maximum isoprene emission rates achieved in the second sunfleck, compared to the first, are consistent with increases seen in successive heat stresses or simulated sunflecks in other studies (Behnke et al. 2007, 2010), and may reflect incomplete recovery of pre-stress isoprene emission rates prior to a rapid reapplication of the stress.

In previous work, pre-stress A net at ambient CO2 was similar in these emitting and non-emitting mutants (Behnke et al. 2007, 2009). This result is consistent with our high CO2 data, but at low CO2, isoprene-emitting leaves had higher A net and J than non-emitters in our study. The slightly higher A net in isoprene-emitters at low CO2 implies either an alleviation of Rubisco carboxylation limitations on photosynthesis, which limit gross CO2 assimilation (A gross) rates at low CO2 concentrations, or lower day respiration rates (R day), since A net = A gross − R day. Emitting leaves operated at lower intercellular CO2 concentrations (C i) than non-emitters (data not shown), so they did not increase photosynthesis by operating at a higher C i. The possibility of a difference in R day is plausible, given that emitters have 50% lower dark respiration rates at low CO2 and that day and dark respiration rates are usually correlated (Loreto et al. 2007; Way and Sage 2008). If R day was 10–30% lower than R dark (as it was in Populus alba grown at ambient and high CO2; Loreto et al. 2007), this would reduce, but not fully account for, the difference in A net; isoprene-emitters at low CO2 had 31% higher A net than non-emitters, but would still have 8–12% higher A gross if differences in R day were taken into account. Regardless of the cause for lower carbon fixation rates in non-emitters, their ability to recover from sunflecks was not inherently hindered. Individual leaves with the lowest pre-stress photosynthetic rates (2.6 μmol m−2 s−1 for both emitters and non-emitters at low CO2) were fully capable of recovering pre-stress A net values, suggesting that non-emitting poplar leaves, with mean A net rates of 4.1 μmol m−2 s−1, had more than sufficient capacity to fully recover from sunflecks.

In isoprene-emitting lines, high CO2 increased respiration rates but suppressed isoprene emissions compared to low CO2 (Griffin et al. 2001; Wang et al. 2001; Wilkinson et al. 2009). However, isoprene-emitters had lower dark respiration rates at low CO2 than non-emitters, while there was no such effect at high CO2. Following current metabolic theory (Rosenstiel et al. 2004), competition for phosphoenolpyruvate (PEP) between the cytosol and chloroplast is mediated by PEP carboxylase. Reduced isoprene emission rates at high CO2 might result from higher consumption rates of cytosolic PEP through PEP carboxylase, thus restricting the chloroplast import of PEP (and thus pyruvate).

Monson and co-workers (Monson et al. 2009; Wilkinson et al. 2009) proposed a metabolic control scheme for C3 plants with competition between cytosolic PEP carboxylase and Rubisco controlling carbon flow to the MEP (methylerythritol 4-phosphate) pathway in response to changes in CO2 concentration. In line with this scheme, increased mitochondrial respiration, as shown here, can constitute a growing sink for cytosolic PEP under rising CO2 (as proposed by Loreto et al. 2007); higher respiration demand would then compete with the chloroplast import of PEP for isoprene biosynthesis and lower isoprene emission rates (Rosenstiel et al. 2003). This possibility is supported by our low CO2 data, where non-emitting mutants exhibited increased absolute and relative respiration rates, implying that the lack of isoprene had diminished chloroplast requirements for PEP, freeing up more PEP substrate to be channelled to mitochondrial respiration. Thus, in both cases, lower respiration rates correlated with higher isoprene emission rates. In contrast to our data, Loreto et al. (2007) found that in mature Populus alba leaves, rates of dark respiration and isoprene emission were positively correlated; however, the two rates were negatively correlated in developing leaves.

Competition between cytosolic and plastidic demands for substrate likely contributed to differences in isoprene emissions at high temperatures between treatments. At low CO2 concentrations, isoprene emission rates of emitting lines leveled off at temperatures above 39°C, but continued to increase linearly at high CO2. Under these conditions—low A net and high metabolic demand of the MEP pathway—the metabolic flux to dimethylallyl diphosphate (DMAPDP) is likely limited, resulting in a substrate limitation of the ISPS enzyme. In contrast to other terpene synthases, ISPS enzymes display Michaelis constants for their substrate in the millimolar range (Schnitzler et al. 2005), making this catalytic reaction in vivo highly sensitive to a depletion of the plastidic DMADP pool (shown for poplar in Magel et al. 2006).

In non-isoprene-emitting poplar leaves, neither J nor NPQ showed full recovery to pre-stress levels within 25 min, implying a reduced ability to recover from the two sunflecks imposed. In a heat stress experiment performed at ambient CO2, J rel and NPQ of wild-type poplars fully recovered from these events, while the J rel of non-emitting trees decreased and NPQ increased with each successive heatfleck imposed (Behnke et al. 2007). Also, measurements at ambient CO2 indicated that while J in isoprene-emitting leaves recovered within 30 min, non-emitting grey poplar needed 90 min or longer to recover from a series of six sunflecks following the cessation of the stress (Behnke et al. 2010). While we did not examine longer recovery periods, since J and A net are usually correlated (for example, a 14% decrease in both A rel and J rel at point AR2 in non-emitting, low CO2 trees), photosynthesis is likely to remain inhibited in these trees for up to an hour and a half after sunfleck-type stress in the absence of isoprene production.

Potential evolutionary implications regarding atmospheric CO2

It appears that isoprene production evolved independently in various groups of higher plants (Harley et al. 1999; Sharkey et al. 2005), a degree of convergent evolution implying a common selective pressure. The recent evolutionary history of land plants took place in a low CO2 atmosphere: although atmospheric CO2 concentrations were generally above 1,000 ppm for most of the last 600 million years, they decreased to modern, low levels of ~300 ppm by the early Miocene (20–24 mya) (Zachos et al. 2001; Tipple and Pagani 2007). CO2 levels have stayed low since, ranging from 180 ppm during the last glacial maximum 21,000 years ago to 280 ppm before the industrial revolution (Jansen et al. 2007).

We propose that the evolutionary pressure for isoprene synthesis in higher plants may have been partly attributable to the relatively low CO2 concentrations of the past 25 million years. Since photosynthesis is more susceptible to heat and high light damage at low CO2 (Cowling and Sage 1998), plants that emit isoprene may have had a competitive advantage over non-emitting species at low CO2 in environments.

Our proposal for the importance of a low CO2 environment for the evolution of isoprene biosynthesis is also consistent with other researchers’ theories regarding why plants emit isoprene. Isoprene may act as a ROS-scavenging molecule, when excess light energy absorption at the thylakoid membranes cannot be channeled into photosynthesis (Affek and Yakir 2002; Peñuelas et al. 2005; Vickers et al. 2009); this role should be more important at low CO2 concentrations, where low substrate availability reduces the photosynthetic sink for light energy. In the metabolic homeostasis hypothesis, isoprene biosynthesis was already proposed to have evolved in low CO2 environments because of a decrease, relative to prior high CO2 regimes, in the rate at which cytosolic PEP carboxylase used PEP substrate (Rosenstiel et al. 2004). In the absence of control over PEP partitioning between the cytosol and chloroplast at the level of the PEP/Pi antiporter in the chloroplast envelope, a shift toward higher chloroplastic PEP influx at low CO2 may have led to pyruvate accumulation in chloroplasts (Rosenstiel et al. 2004). Enhanced isoprene biosynthesis at low CO2 may provide a means of metabolizing accumulated pyruvate, similar to the role proposed for the alternative mitochondrial oxidase (Plaxton and Podesta 2006).

Our results also provide insight into how rising atmospheric CO2 concentrations might alter any photosynthetic stress-tolerance advantage of isoprene biosynthesis in the future. We found few differences in photosynthesis between emitting and non-emitting poplar leaves at high CO2. There was no significant difference at high CO2 in pre-stress A net between emitters and non-emitters, although emitters had slightly higher A net at 30°C in general (Table 2), and both groups showed similar abilities to recover from sunflecks. While J and NPQ both recovered from the second sunfleck (AR2) more fully in isoprene-emitting than non-emitting leaves at high CO2, the relative difference between the two groups was much smaller than at low CO2. Since the benefit of producing isoprene is substantially reduced at high CO2, isoprene emission may be less adaptive in a future, high CO2 atmosphere.