Increased ratio of electron transport to net assimilation rate supports elevated isoprenoid emission rate in eucalypts under drought.

Plants undergoing heat and low-CO2 stresses emit large amounts of volatile isoprenoids compared with those in stress-free conditions. One hypothesis posits that the balance between reducing power availability and its use in carbon assimilation determines constitutive isoprenoid emission rates in plants and potentially even their maximum emission capacity under brief periods of stress. To test this, we used abiotic stresses to manipulate the availability of reducing power. Specifically, we examined the effects of mild to severe drought on photosynthetic electron transport rate (ETR) and net carbon assimilation rate (NAR) and the relationship between estimated energy pools and constitutive volatile isoprenoid emission rates in two species of eucalypts: Eucalyptus occidentalis (drought tolerant) and Eucalyptus camaldulensis (drought sensitive). Isoprenoid emission rates were insensitive to mild drought, and the rates increased when the decline in NAR reached a certain species-specific threshold. ETR was sustained under drought and the ETR-NAR ratio increased, driving constitutive isoprenoid emission until severe drought caused carbon limitation of the methylerythritol phosphate pathway. The estimated residual reducing power unused for carbon assimilation, based on the energetic status model, significantly correlated with constitutive isoprenoid emission rates across gradients of drought (r(2) > 0.8) and photorespiratory stress (r(2) > 0.9). Carbon availability could critically limit emission rates under severe drought and photorespiratory stresses. Under most instances of moderate abiotic stress levels, increased isoprenoid emission rates compete with photorespiration for the residual reducing power not invested in carbon assimilation. A similar mechanism also explains the individual positive effects of low-CO2, heat, and drought stresses on isoprenoid emission.

The emission of volatile isoprenoids by plants (globally amounting to approximately 1,000 TgC year 21 , of which isoprene constitutes approximately 500 TgC year 21 ) plays a significant role in tropospheric oxidation chemistry (Amann et al., 2008). Plant isoprenoid emission at the ecosystem scale is determined not only by intrinsic biochemical and physiological controls but also by the relative abundances of species, each with characteristic baseline emission capacities and each subject to modification by environmental conditions (Harrison et al., 2013). While the effects of most environmental factors on isoprenoid emission have been documented (Loreto and Schnitzler, 2010), their interactions are likely to be complex and hold the key to accurate projections of global emissions (Arneth et al., 2007;Squire et al., 2014). The effect of soil water availability on plant volatile isoprenoid emission is crucial to the projections, especially as rainfall patterns are themselves subject to the impacts of climate change.
Volatile isoprenoid emission is notably insensitive to moderate drought (when the fraction of available soil water ranges from 40% to 70%; Fortunati et al., 2008;Centritto et al., 2011). Given the various other consequences of drought for plant function, such as stomatal closure leading to reduced photosynthesis (Lawlor and Cornic, 2002), increased leaf temperature (Jones, 2004), leaf shedding (Tyree et al., 1993), reduced growth and potential hydraulic failure (Maherali et al., 2004), reduced shoot-to-root ratio (Poorter et al., 2012), increased oxidative stress due to the activation of reactive oxygen species (Mittler and Zilinskas, 1994), and an increased Suc-to-starch ratio, affecting osmotic adjustment (Chaves, 1991), it is not surprising that there are large variations in experimental and field measurements of isoprenoid emission in response to drought (for review, see Laothawornkitkul et al., 2009;Niinemets, 2010).
Isoprene emission involves an energy-intensive biosynthesis through the methylerythritol phosphate (MEP) pathway in chloroplasts (for review, see Sharkey and Yeh, 2001). Photosynthesis contributes the required carbon skeletons and reducing power for isoprenoid biosynthesis, at least under stress-free conditions (for review,  Haupt-Herting and Fock (2002) and Skillman (2008). The share is determined by not only the absolute demand per pathway but also the actual instantaneous rates of each process (e.g. nitrate reduction needs a lot of reducing power, but its actual processing rate is very slow compared with core reactions of photosynthesis  (Heber, 2002).
see Loreto and Schnitzler, 2010). The photosynthetic energy and reducing power output of a chloroplast are shared in unequal proportions among colocalized pathways (Table I). Under favorable conditions, photosynthetic carbon reduction is the largest energy sink (approximately 50%). Abiotic stresses enhance the supply of energy and reducing power to nonphotosynthetic carbon reduction sinks in the chloroplast (Haupt-Herting and Fock, 2002). Examples include (1) increased photorespiration under drought, at least until Rubisco is directly affected by stress (Lawlor, 1976;Noctor et al., 2002), (2) increased photorespiration under low-CO 2 , low-light, or high-light stress (Kozaki and Takeba, 1996), and (3) increased photoreduction of oxygen (O 2 ) under photooxidative stress (Makino et al., 2002). It has been posited that isoprenoid emission is a mechanism to consume surplus reducing power in stressful high-light and/or low-CO 2 environments (Niinemets et al., 1999;Way et al., 2011). Increased accumulation of secondary metabolites such as phenols, alkaloids, and isoprenoids has been documented in plants under abiotic stresses (for review, see Wilhelm and Selmar, 2011). Decreased carboxylation and increased oxidative stress due to the oversupply of reducing equivalents is seen as the main driver of increased secondary metabolism under drought (Selmar and Kleinwächter, 2013). The postillumination behavior of isoprene emission (primary and secondary bursts) under O 2 -free, pure nitrogen (N 2 ) atmospheres have been attributed to the availability of reducing equivalents (Rasulov et al., 2011;Li and Sharkey, 2013). It has further been proposed that the MEP pathway competes with other sinks for reducing power, so that the flow of reducing power to isoprenoid biosynthesis is proportional to the energy unused for primary metabolism (Morfopoulos et al., 2013(Morfopoulos et al., , 2014Dani et al., 2014). However, the MEP pathway's requirement for reducing power is very small relative to that of photorespiration , and given the diversity in the relative sink strengths of intraplastidic processes (Table I), it is still unclear how the demands of these different processes influence one another. Eucalypts have been used as model systems to study plant isoprenoid emission since the early years of isoprenoid research (Guenther et al., 1991;Brilli et al., 2013). All eucalypts store monoterpenes as well as constitutively emit isoprene and some monoterpenes (He et al., 2000). Eucalyptus camaldulensis ssp. camaldulensis (river red gum) is a drought-avoiding mesic species that is tolerant to waterlogging and distributed in riparian, temperate southeastern Australia (Farrell et al., 1996). Eucalyptus occidentalis (swamp yate) is a drought-tolerant species found in saline environments in Mediterranean southwestern Australia (Benyon et al., 1999;Searson et al., 2004). E. camaldulensis ssp. obtusa is the most widespread eucalypt in subtropical Australia (Butcher et al., 2009). Exploiting this ecological contrast, we empirically tested the hypothesis that constitutive isoprenoid emission is driven by ATP and NADPH availability (Loreto and Sharkey, 1993;Niinemets et al., 1999) and could potentially compete for the same with carbon assimilation (Harrison et al., 2013;Morfopoulos et al., 2014). We manipulated the energy source-sink dynamics by imposing various abiotic stresses, including drought, heat, low CO 2 , and high O 2 . We investigated the relationship between three plastidic biochemical processes (carbon assimilation, photorespiration, and volatile isoprenoid emission) in eucalypts acclimated to drought for 4 to 6 months. Interactive effects of short-term exposure to five CO 2 concentrations, three O 2 levels, and heat stress on isoprenoid emission rates were also analyzed. It was hypothesized that the relative sink strength of various processes requiring reducing power in the chloroplasts could determine the variations of isoprenoid emission in plants experiencing abiotic stress. We started with the premise that the light-dependent and light-independent reaction components of photosynthesis have different susceptibilities to abiotic stress (particularly to drought) and that these susceptibilities vary across species.

RESULTS
The results are from three independent experiments (see Table II).

Photosynthesis
Acclimation to Drought in Paired Species (at 20% O 2 ) E. occidentalis had a significantly higher stomatal conductance (g s ; P = 0.043) when watered to field capacity (FC) than E. camaldulensis ssp. camaldulensis, but the difference in transpiration rate (T r ) was not significant (P = 0.193). Net assimilation rate (NAR) of both E. occidentalis (16.8 6 2.28 mmol m 22 s 21 ) and E. camaldulensis ssp. camaldulensis (18.1 6 1.86 mmol m 22 s 21 ) was comparable (test of equal means, P = 0.001). During acclimation to severe drought stress (FC # 50%), E. camaldulensis ssp. camaldulensis showed a significant decline in all photosynthetic parameters (P , 0.001), whereas net assimilation in E. occidentalis (15.3 6 1.81 mmol m 22 s 21 ), although decreased, remained comparable to control values despite a significant decrease in g s (Fig. 1, A and D). Under well-watered conditions, estimates of photosynthetic linear electron transport rate (ETR) based on chlorophyll fluorescence for E. occidentalis and E. camaldulensis ssp. camaldulensis showed a consistent proportionality with their respective NARs. For E. occidentalis, the ETR-NAR ratio remained unchanged even at 50% FC, while the ratio increased significantly (doubled) for E. camaldulensis ( Fig. 2A). The carbon cost of isoprenoid emission as a proportion of net assimilation increased more than 10fold as drought intensified and was highest in both species at 25% FC (Fig. 2B). ETR was measured independently using fluorescence and was not coupled with LiCor measurements (experiment 1; Table II). Hence, the observations should be treated only as indicative and not as absolute. ETR-NAR ratios (approximately 7:1 under 20% O 2 ) were more realistic when obtained after fitting A-C i curves (experiment 3; Supplemental  Fig. S1). ETR-NAR ratios across drought treatments followed a near-significant quadratic regression with total emission rates (r 2 = 0.75, P = 0.12; Fig. 2C).

Response to Heat and CO 2 in E. camaldulensis ssp. obtusa
Heat caused a significant decline in photosynthesis under elevated CO 2 (1,000 mmol mol 21 or greater; P , 0.001). Leaves at 38°C transpired at a significantly higher rate than those at 28°C (P , 0.0001), except under drought. Heat did not cause a significant change in either g s or T r under normal O 2 in plants under drought (50% FC). This was true across most of the CO 2 range (400-1,800 mmol mol 21 ), except at 180 mmol mol 21 CO 2 and at the photorespiratory compensation point (60 mmol mol 21 ; normal O 2 ). Leaves of well-watered plants did not show a decline in NAR when subjected to 38°C under present-day ambient CO 2 (400 mmol mol 21 ) and normal O 2 (Fig. 3A). Positive effect of heat on T r was pronounced at CO 2 # 400 mmol mol 21 (Supplemental Fig. S2).

Response to Varying O 2 Concentrations
Net assimilation rate in both species increased significantly (greater than 30%) during low O 2 . The gain was proportional to their respective basal rates under normal O 2 except for E. camaldulensis ssp. camaldulensis at FC # 50% ( Fig. 1B; Supplemental Fig. S3). During low O 2 , Figure 1. Photosynthesis and isoprenoid emission rates over a drought gradient in two eucalypts, E. occidentalis (solid line, diamonds) and E. camaldulensis ssp. camaldulensis (dotted line, triangles), subjected to 20% O 2 (left) and 2% O 2 (right). A, g s ; B, leaf C i ; C, T r ; D, NAR.; E, I e ; F, constitutive M e . Each point represents n = 4 plants; values are means 6 SE. *P # 0.05, **P , 0.01, ***P , 0.001, comparison within species relative to 100% FC control. Note that the pronounced decline in NAR with drought in E. camaldulensis ssp. camaldulensis reflected the fact that its stomata were sensitive to soil water status, a mechanism that presumably achieves a minimum necessary transpiration rate per unit of leaf area during drought stress (White et al., 2000).
T r did not change significantly at 100% FC (equal means, P = 0.002) despite decreases in g s and (as a result) a decrease in leaf internal CO 2 (C i ). High O 2 (50% O 2 ) caused a significant increase in C i and a severe decline in net assimilation rate in E. camaldulensis ssp. camaldulensis, and the effect was persistent and amplified under drought (Supplemental Table S1).

Response to Drought
Branch-level basal isoprene emission rate (I e ) at 25°C, 1,200 mmol m 22 s 21 photosynthetically active radiation, and present-day CO 2 and normal O 2 was 40% higher (P = 0.004) in E. camaldulensis ssp. camaldulensis (5.9 6 1.48 nmol m 22 s 21 ) than in E. occidentalis (3.4 6 1.99 nmol m 22 s 21 ), and the rates remained unchanged in both species despite acclimation to moderate drought (70% FC). The trends were conserved in leaf-level measurements. There was a tiny decrease (relative to control) in NAR in E. occidentalis at 50% FC (21.5 mmol m 22 s 21 ), and it was accompanied by a marginally significant increase in I e (+1.4 nmol m 22 s 21 ; P = 0.05; Fig. 1E [note that there is a 3 orders of magnitude difference between amounts of carbon fixed via photosynthesis and carbon lost via emission]). The biggest increase in I e for these two species occurred at two different drought intensities. I e peaked significantly at 50% FC for E. camaldulensis (P , 0.005; Fig. 1E) and at 25% FC for E. occidentalis (P , 0.001). These emission peaks coincided with the first noticeable increase in their respective ETR-NAR ratios (Fig. 2, A and C), but the emission declined for E. camaldulensis ssp. camaldulensis even though ETR-NAR increased further at 25% FC. Constitutive monoterpene emission rate (M e ; pinenes and D-limonene) in E. camaldulensis ssp. camaldulensis behaved in a manner similar to isoprene, while M e in E. occidentalis did not respond to drought even at 25% FC. I e and M e in both E. occidentalis and E. camaldulensis ssp. camaldulensis were comparable in magnitude across the drought gradient ( Fig. 1, E and F): the I e :M e molar ratio was roughly 10:1 in both species. E. camaldulensis ssp. obtusa showed a basal I e :M e molar ratio of approximately 2:1.

Response to Heat and CO 2 in E. camaldulensis ssp. obtusa
Heat was the most significant factor causing an increase in I e (P , 0.0001) and M e (P , 0.001) across all treatments. At 28°C, 100% FC (n = 6), and 20% O 2 , I e showed a significant peak at 180 mmol mol 21 CO 2 (P , 0.001) and almost full inhibition at the saturating CO 2 level of 1,800 mmol mol 21 (Figs. 3B and 4). This response completely disappeared at 38°C (Fig. 3B). Under normal O 2 , the I e response at 50% FC without heat stress (28°C) was equivalent in magnitude to I e observed in 100% FC plants subjected to heat stress (38°C; P , 0.001) irrespective of CO 2 acclimation (Supplemental Fig. S4). Plants acclimated to drought (50% FC) and exposed to heat stress (38°C) showed the highest isoprenoid emission. High-CO 2 -induced inhibition of isoprene emission at 28°C disappeared at 38°C and was accompanied Figure 2. A, ETR-NAR ratio. B, Carbon cost. C, ETR-NAR ratio versus total isoprenoid emission rates at 20% O 2 . The relative response of linear ETR and NAR in two eucalypts over a drought gradient is shown. Note the greater decline in NAR in E. camaldulensis subsp. camaldulensis at 50% FC, which results in a large ETR-NAR ratio. Emission is almost twice as expensive under all conditions on a carbon basis in E. camaldulensis (the drought-sensitive species). The regression fits in C encompass data points from both species (n = 4; values are means 6 SE for emission rates; quadratic P = 0.12, linear P = 0.18). If E. occidentalis were subjected to harsher droughts (FC # 10%), it is also likely to have followed a quadratic regression, with peak emission at 25% FC and declining emission thereafter due to carbon limitation despite a favorable ETR-NAR ratio. by a significantly low g s and low leaf internal CO 2 (Supplemental Fig. S3).

Response to Varying O 2 Concentrations
Exposure to 2% O 2 (10 min) resulted in a marginal increase in I e and M e across the drought gradient (Fig. 1, E and F), and these trends were conserved in both E. occidentalis and E. camaldulensis ssp. camaldulensis, except that the latter showed no significant change in I e at 100% and 70% FC. Low O 2 on its own did not significantly affect M e in E. occidentalis (Fig. 1F) or in E. camaldulensis ssp. obtusa (P . 0.6; Fig. 3C). Low O 2 significantly increased I e at CO 2 = 1,800 mmol mol 21 (no heat stress) compared with lower CO 2 levels, which was opposite to the effect under normal O 2 (P , 0.0001). Low O 2 also had a positive effect on the I e of well-watered plants at 38°C. In E. camaldulensis ssp. camaldulensis, I e increased when well-watered plants were exposed to 50% O 2 (relative to plants at 20% O 2 ). Emission rate decreased significantly when the plants simultaneously experienced drought (50% FC) and high O 2 (50% O 2 ). The effect of low O 2 on I e was more pronounced than on M e in all three taxa (P , 0.001), and increased I e under low O 2 resulted in a decreased M e in well-watered plants (Fig. 5, A and C). , I e (B), and constitutive M e (C) in response to shortterm heat stress under 20% O 2 (left) and 2% O 2 (right) over a CO 2 concentration span (60-1,800 mmol mol 21 ) in E. camaldulensis ssp. obtusa (experiment 2) acclimated to well-watered conditions (100% FC) and drought (50% FC) at two independent temperature treatments (28˚C and 38˚C). n = 6; values are means 6 SE. Figure 4. The energetic status model. A, Relationship between residual ETR J r (not used for light-independent reactions of photosynthesis) and isoprenoid emission rate at different drought stress levels in E. occidentalis (black rhomboids) and E. camaldulensis ssp. camaldulensis (black triangles). B, J/V cmax versus isoprenoid emission rate in both species. n = 3 for J r and J/V cmax , values are means 6 SD; n = 4 for isoprenoid emission rate, values are means 6 SE (P , 0.05).

The Relationship between Leaf Energetic Status and Isoprenoid Emission Rate
The estimated ETR not used for light-independent reactions (J r ) correlated significantly and positively with isoprenoid emission rate in both species (r 2 = 0.81, P = 0.014) and was consistent across the drought gradient (Fig. 4A). The positive relationship between electron transport rate ( J )/maximum carboxylation rate (V cmax ; derived from A-C i curves) and isoprenoid emission rate (Fig. 4B) was consistent with the ETR-NAR ratio, peaking with increased emission in both species. The relationship between J r and I e estimated at three different O 2 concentrations was significantly positive (r 2 . 0.91, P = 0.02) for E. camaldulensis ssp. camaldulensis (Fig. 5A). Plants experiencing drought (50% FC) exposed to 50% O 2 were the only exception to this trend and showed a significant decline in isoprenoid emission despite large J r (Fig. 5A). E. camaldulensis ssp. camaldulensis exposed to extreme photorespiratory stress without drought (100% FC, 50% O 2 ) and drought without extreme photorespiration (50% FC, 20% O 2 ) resulted in large increases in J/V cmax (Fig. 5B).

Drought, ETR-NAR Ratio, and Constitutive Isoprenoid Emission
The insensitivity of ETR to drought in both species in this study confirmed that the photosystems and the electron transport chain are not susceptible to moderate drought stress (Ben et al., 1987) and that the relative decrease in ETR under drought is proportionately less when compared with decrease in CO 2 assimilation (Cornic and Briantais, 1991;Bota et al., 2004). The absolute rates of isoprene and constitutive monoterpene emission did not change provided that the simultaneous assimilation rate of CO 2 remained unchanged.
The Mediterranean species E. occidentalis maintained almost an unchanged NAR and I e even at 50% FC (Fig. 1,  D and E). E. camaldulensis ssp. camaldulensis showed a gradual and more marked decline in NAR with increasing water deficit. ETR-NAR ratio increased significantly at 25% FC for the former and at 50% FC for the latter, the point where the species showed its highest isoprene and monoterpene emission rates (Figs. 1 and 2, A and C). We attribute the increased I e under drought to the increased ETR-NAR ratio and the increased availability of reducing power to the MEP pathway among other nonphotosynthetic carbon reduction sinks (Fig. 4). Under severe drought (25% FC), despite a favorable ETR status of its leaves, isoprenoid emissions of E. camaldulensis ssp. camaldulensis declined, suggesting carbon limitation despite 2% O 2 (see decline in NAR at FC # 50%; Supplemental Fig. S3). It was later confirmed that low-O 2 exposure did not significantly affect the ETR-NAR ratio under both well-watered and drought conditions (Supplemental Fig. S1). Similarly, imposing severe photorespiratory stress (50% O 2 ) on well-watered plants resulted in increased I e , and the rates plummeted under drought despite a large pool of residual reducing power (Fig. 5). These results suggested that reactions other than photosynthetic carbon reduction (especially photorespiration and the MEP pathway) compete for reducing power not allocated to carbon assimilation reactions under situations of suboptimal carbon assimilation due to abiotic stress. Although eucalypts may have only a Figure 5. Photorespiration, drought, and isoprenoid emission. A, J r versus I e (r 2 = 0.91, P = 0.02) and B, J/V cmax versus I e (r 2 = 0.85, P = 0.06), in E. camaldulensis ssp. camaldulensis (experiment 3) acclimated to wellwatered (100% FC) and droughted (50% FC) conditions and exposed to three different levels of oxygenated atmospheres. Two percent and 50% O 2 exposure were maintained for 10 to 15 min before volatile sampling (n = 5; values are means 6 SE). The correlation remains significant (P , 0.05) even when both isoprene and monoterpenes are considered together. C, Response of constitutive M e to drought and varying O 2 levels. Drought (50% FC) had a significant positive impact on M e at both 2% and 20% O 2 , which was consistent with the isoprene-like response of monoterpenes in E. camaldulensis ssp. camaldulensis (Fig. 1F). n = 5; values are means 6 SE (**P , 0.05).
modest capacity for nonphotochemical quenching (NPQ) due to their typically high photosynthetic capacities and acclimation to high-light habitats, the proportion of energy dissipated through NPQ, which is one of the primary mechanisms to mitigate oxidative stress, increased in E. camaldulensis ssp. camaldulensis as drought intensified (Supplemental Fig. S5). NPQ is likely to be a significant sink for ETR and may also account for reduced emissions under severe stress (Fig. 5A).

Drought, Photorespiration, and Constitutive Isoprenoid Emission
The direction of change in the rates of isoprene emission and photorespiration in response to many environmental factors is the same, despite the absence of a biochemical (carbon-based) link between the two pathways (Monson and Fall, 1989;Hewitt et al., 1990;Loreto and Sharkey, 1990). In this study, net assimilation and I e increased by a small yet significant extent in wellwatered E. occidentalis exposed to short-term low O 2 (Fig. 1, D and E; in agreement with Hewitt et al., 1990), and such an increase persisted under drought. The increased de novo carbon pool and decreased competition for ATP and NADPH may explain the small increase in isoprene emission under low O 2 in well-watered plants, given that ATP could be limiting emissions when carbon is plentiful (Loreto and Sharkey, 1993). Although increased isoprene emission in low O 2 under most conditions may be physiologically important, it is not comparable to the large difference in emission between well-watered plants (low emission) and plants experiencing drought (high emission) under 20% O 2 (Fig. 1E). Increased emission under drought is sustained so long as the intensity of drought is within a species-specific tolerance threshold. When such a threshold was exceeded, I e decreased significantly (at 25% FC for E. camaldulensis ssp. camaldulensis; Fig. 1E). Since we did not directly estimate photorespiration rates under drought (which are known to increase significantly under abiotic stress), we increased photorespiratory stress in well-watered plants to mimic alternative scenarios. When photorespiratory stress is extreme (50% O 2 ) and is coupled with reduced carbon assimilation capacity due to drought (50% FC), the MEP pathway suffers a double jeopardy and is likely deprived of both carbon and reducing power (Fig. 5A).
The cost of carbon due to isoprenoid emission increased nearly 15-fold, from 0.46% of freshly fixed carbon in well-watered plants to 7.2% in severely stressed E. camaldulensis ssp. camaldulensis (Fig. 2B). Although alternative carbon imported from the cytosol avoids the carbon limitation of the MEP pathway under moderate abiotic stress (Funk et al., 2004;Brilli et al., 2007;Trowbridge et al., 2012), under extreme stress carbon could be limiting due to irreparable biochemical impairment of both the photosynthetic carbon reduction cycle and rates of carbon import into plastids. Whether drought just inhibits carbon assimilation rate through stomatal diffusional limitation or there is a clear biochemical down-regulation is highly debatable . Under severe drought stress (especially for E. camaldulensis ssp. camaldulensis), even those limited numbers of leaves that were retained by the plants could not have recovered fully if the plants were rewatered. In such cases, diffusional limitation was likely compounded by an impairment of photosynthetic biochemical machinery. Limited catalytic activity of the MEP pathway, particularly isoprene synthase (Brilli et al., 2007), could have contributed to decreased emissions under extreme stress.

Drought, Low CO 2 , Heat, and Constitutive Isoprenoid Emission
In E. camaldulensis ssp. obtusa, short-term acclimation to heat stress (38°C) caused significantly higher emissions with no significant change in net assimilation despite drought (50% FC; Fig. 3). CO 2 inhibition of isoprene emission also disappeared at high temperatures (for review, see Sharkey and Monson, 2014). It is known that moderate heat stress can suppress ETR yet increase both V cmax and I e (Dreyer et al., 2001;Darbah et al., 2008). Cold and heat treatments have been shown to selectively suppress PSII (linear electron transport) and thus reduce NADPH availability and up-regulate PSI (cyclic electron transport) to increase ATP production (Huner et al., 1993;Zhang and Sharkey, 2009). All of these observations appear to contradict the view that Figure 6. Relationship between percentage FC and leaf water potential in E. occidentalis and E. camaldulensis ssp. camaldulensis (experiment 1). reducing power availability (however small the requirement may be) influences variation in volatile isoprenoid emission. For the moment, if we ignore cold stress, which is not relevant to isoprene emission, at least to the extent that we understand the phenomenon today, the energetic status model may be inadequate to explain emission behavior at high temperatures for the following reasons: (1) prolonged heat stress reduces net assimilation rate (despite an increase in V cmax ), primarily due to decreased CO 2 solubility and decreased Rubisco-CO 2 affinity (Sage and Kubien, 2007); (2) prolonged heat and drought stress (when imposed together) reduce emission, possibly due to heat sensitivity of the cytosolic carbon pool (Fortunati et al., 2008;Centritto et al., 2011); and (3) extreme stress not only reduces net assimilation rates but also increases the photorespiratory drain on carbon and reducing power (Fig. 5).
Unlike isoprene emission, the response of constitutive monoterpene emission did not follow a consistent pattern across any of the treatments (Figs. 1F and 3C). In E. camaldulensis ssp. camaldulensis, constitutively emitted monoterpenes behaved like isoprene, while in E. occidentalis, monoterpene emission was not sensitive even to severe drought. This could be partly due to sustained monoterpene synthase activity during drought, as reported in evergreen oaks (Quercus spp.; Lavoir et al., 2009). Oddly, isoprene emission increased in leaves simultaneously exposed to 28°C, 1,800 mmol mol 21 CO 2 (not saturating for eucalypts), and 2% O 2 (Fig. 3B). For reasons unknown, the same leaves also showed a significant (16%) decrease in net assimilation rate (Fig. 3A). We speculate that the possible (temporary) inhibition/ down-regulation of other sinks of reducing power, such as photoassimilation of N 2 under very high CO 2 (Bloom et al., 2002), also could have contributed to increased isoprene emission. However, it is acknowledged that both the photosynthetic carbon reduction cycle and nitrate assimilation do not compete directly for reducing power (Robinson, 1988). A tradeoff between isoprene and monoterpene emissions in response to CO 2 (28°C; Supplemental Fig. S6) and low O 2 (Fig. 5C) indicates that the emission of isoprene and monoterpenes is inversely related (Harrison et al., 2013).
Increased secondary metabolism and specifically increased isoprene emission under drought could be involved in protecting photosystems against transient periods of oxidative and heat stress, as seen in some transgenic studies, although the mechanisms are unclear (Behnke et al., 2007;Velikova et al., 2011;Selmar and Kleinwächter, 2013;Ryan et al., 2014). However, the tiny increase in isoprene emission when photorespiration (the largest photoprotective sink) is suppressed, despite the down-regulation of photosynthetic carbon reduction under drought, suggests that the MEP pathway has limited capacity to oxidize the pool of excess reductants available under abiotic stress. The increased isoprenoid emission rates among plants experiencing drought (without heat stress), heat (without drought stress; Supplemental Fig. S4), and artificially increased photorespiratory stress (without drought and without heat at 50% O 2 ) were quantitatively equivalent, and more experiments are needed to differentiate the underlying mechanisms between these responses.

CONCLUSION
The energy (ATP) and reducing power (NADPH) budgets of the chloroplast are used in a hierarchical fashion. The photosynthetic carbon reduction cycle dominates, while, possibly, all other reducing sequences colocalized in the chloroplasts must compete for the remaining pool (Table I). The equilibrium between the source (light reactions) and major sinks (carbon reduction and photorespiration) of energy, as well as the sources (de novo and stored) and sinks (all anabolic processes) of carbon, becomes distorted under drought stress. Drought-induced reduction in the photosynthetic carbon reduction cycle is accompanied by an increase in ETR-NAR ratio and a significant increase in volatile isoprenoid emission. The qualitative response of isoprenoid emission under drought may be similar among species, but the degree of drought-induced shift in J/V cmax is species specific. While energy availability is clearly the common factor that underpins the individual effects of low CO 2 (Morfopoulos et al., 2014), heat, drought, and photorespiratory stress (this study) on isoprenoid emission, the complex interactive effects of heat, CO 2 , and drought seem to defy simple assumptions and remain largely uncertain (Fig. 3). Variation in atmospheric O 2 concentration between 10% and 35% during the last 200 million years (Falkowski et al., 2005) and its influence on carboxylation efficiency also could have played an important role in regulating global isoprene emissions on a macroevolutionary time scale. All of these indicate a need for a wider experimental analysis across different functional plant types and ecosystems if we are to reliably scale up volatile isoprenoid emissions from plants to large regions.

MATERIALS AND METHODS
This study included three independent and yet mutually supporting experiments (Table II).

Rationale for Selecting Species
A group of 15 species of eucalypts belonging to distinct biomes within Australia were screened for photosynthetic performance and isoprenoid emission potential in March 2012. The first experiment involved a paired comparison of isoprenoid emission rates and photosynthesis in Eucalyptus occidentalis and Eucalyptus camaldulensis ssp. camaldulensis in response to drought acclimation. E. camaldulensis ssp. camaldulensis and E. occidentalis were studied as a pair in experiment 1, as both had comparable photosynthetic capacities and both predominantly emitted isoprene and some monoterpenes at significant levels. A desirable contrast in the physiology of their water relations is already highlighted (see introduction).
The second experiment tested whether known CO 2 and heat responses of isoprenoid emission are consistent under drought acclimation. The idea was to test any potential pathway discrimination toward either isoprene or monoterpene emission under abiotic stresses. E. camaldulensis ssp. obtusa was selected for experiment 2 because it emitted comparable quantities of isoprene and constitutive monoterpenes.
The third experiment was a supplementary exercise that was inspired by the results of the first experiment. The third experiment explicitly tested the relationship between photorespiration and isoprenoid emission under drought in E. camaldulensis ssp. camaldulensis.
Eucalypts store monoterpenes, and it is difficult to estimate instantaneous carbon and energy invested in monoterpenoid biosynthesis. However, we took necessary precautions to rule out monoterpene emissions from stored pools (from leaf glands). Even if one considers both constitutive and stored monoterpenes together, the quantities are smaller (in the paired species of experiment 1) than that of isoprene emission by 1 order of magnitude. Besides, it was recently shown that stored monoterpenes are quantitatively insensitive to drought stress in eucalypts, although there are clear qualitative variations and inconsistent trends in secondary metabolite accumulation (phenolics and terpenoids) in various plants under drought (Brilli et al., 2013;Selmar and Kleinwächter, 2013).

Plant Material
Seeds of E. camaldulensis ssp. camaldulensis and E. occidentalis were obtained from the Australian Tree Seed Centre at the Commonwealth Scientific and Industrial Research Organization and germinated in May 2012. Twoto 3-month-old seedlings (eight per species) were transplanted to large pots comprising approximately 80 kg of red clay loam (from the Robertson area in New South Wales) and the required quantities of Osmocote slow-release fertilizer. An independent group of E. camaldulensis ssp. obtusa (n = 6) was established in similar large pots. The plants were grown under the open sun with regular watering. Six-month-old saplings (December 2012) were transferred to and kept until the end of the experiment in a glasshouse maintained at a 25°C/18°C diurnal temperature cycle and a natural photoperiodic regime. A third independent group of E. camaldulensis ssp. camaldulensis (n = 5) were germinated in March 2013 and grown in a similar manner for 1 year (used for experiment 3). These plants were maintained at 100% FC, and isoprenoid emission rates were determined at three different O 2 levels (2%, 20%, and 50%) in April and May 2014. After the measurements were complete, the plants were droughted to achieve 50% FC and maintained (over 10 d). Gas-exchange measurements and volatile sampling were repeated.

Water Relations
The soil water-holding capacity was determined by water saturation and weighing. Five-month-old saplings were watered and weighed (pot + plant) to obtain the 100% FC reference point. Plants were grouped into two sets of four biological replicates per species. One set was maintained at 100% FC throughout the experiment, while the other received reduced water to achieve 70% FC (within 2 weeks) and was maintained thereafter for 3 months. After volatiles were sampled from plants acclimated to 70% FC, watering was further reduced to achieve 50% FC and acclimated for 2 weeks followed by volatile sampling (repeated for 25% FC). One batch of E. camaldulensis ssp. obtusa was acclimated to 50% FC for 1 month before volatile sampling. The difference in acclimation period between experiments 2 and 3 is due to the species involved. In experiment 2, we had E. camaldulensis ssp. obtusa, which comes from lower latitudes of Australia and grows in some of the driest places on the continent. It could endure longer periods of drought and also took longer to acclimate (checked by measuring g s on randomly selected leaves), while E. camaldulensis ssp. camaldulensis stabilized more quickly as well as reflected the effects of drought sooner. In experiment 3, we had only E. camaldulensis, and we could shorten the acclimation period. The duration of severe stress was shorter than the duration at 70% FC to avoid severe defoliation (especially in E. camaldulensis ssp. camaldulensis), which would have otherwise hampered emission measurements. Leaf water potential was determined (experiment 1) using a 12-channel thermocouple psychrometer (JRD Merril Specialty Equipment) calibrated at 25°C using standard sodium chloride solutions (Lang, 1967). Leaf discs (diameter = 0.5 cm) were bored out at predawn from half of a fully expanded leaf and transferred and sealed in the psychrometer (10 leaves per treatment group). The chambers were equilibrated in a water bath at 25°C for 3 h before signal acquisition. The procedure was repeated on the following midday using a leaf disc punched out from the other half of the same leaf. Measurements were repeated at three time points during drought treatment (for the relationship between leaf water potential and percentage field capacity for a physiological assessment of drought intensity, see Fig. 6).

Monitoring Photosynthesis under Drought
Photosynthetic gas-exchange measurements were made using a LI-6400XT (Li-Cor Biosciences) infrared gas analyzer. During branch-level sampling, photosynthesis was measured between 12 noon and 3 PM in parallel to isoprenoid sampling on independent branches of the same plant. Leaf temperature was 25°C, light intensity was 1,200 mmol m 22 s 21 , and relative humidity ranged from 39% to 47%.
A-C i curves were obtained from one or two of the best performing healthiest leaves from each biological replicate (n = 4) per treatment group using a Li-Cor6400 at 25°C and normal O 2 (also at 2% and 50% O 2 for experiment 3). V cmax and J were estimated using the curve-fitting tool described by Sharkey et al. (2007;minimized errors), and the mean values were used as model input parameters. J r (given C i and V cmax ) was calculated following Harrison et al. (2013) and Morfopoulos et al. (2014): Given V cmax , we calculated J v and then substituted Equation 1 in Equation 2, where J v = proportion of electron transport used for dark reactions, V cmax = maximum carboxylation rate by Rubisco, K m = effective Michaelis-Menten coefficient for carboxylation by Rubisco (700 mmol mol 21 at 25°C), G* = photorespiratory compensation point (60 mmol mol 21 ), and C i = species-specific leaf internal CO 2 concentration (mmol mol 21 ) at ambient CO 2 = 400 mmol mol 21 . Chlorophyll fluorescence was monitored using a Pocket PAM chlorophyll fluorescence meter (Gademann Instruments). Kautsky dark-light fluorescence induction curves were obtained from dark-adapted leaves (predawn) between 3 and 5 AM and replicated on 15 fully expanded leaves per treatment group (n = 4; biological replicates). The measurements were made in a dark room (photosynthetic photon flux density , 10 mmol m 22 s 21 ). Predawn leaf temperature was 15.5°C 6 0.3°C. During the day (between 2 and 4 PM), leaves experiencing moderate light levels (350 , photosynthetic photon flux density , 450 mmol m 22 s 21 ) were used to estimate linear ETRs using steady-state light induction curves by gradually increasing the pulse intensity from 0 to 1,500 mmol m 22 s 21 . Predusk leaf temperature was 24.5°C 6 0.8°C. The data for ETR-NAR ratio were measured only after the dry group had been acclimated to 50% FC, and the control group (100% FC plants) was retained throughout the experiment.

Branch Enclosure Method
Tedlar bags (25-L capacity; Sigma) were modified to make a volatile collection chamber with a polytetrafluoroethylene base for air-tight sealing. The gas-exchange line was plumbed with Teflon tubing and stainless steel air-tight connectors (Swagelok). High-purity instrument-grade air (BOC; 78% N 2 , 21% O 2 , and 1% argon) was mixed with CO 2 (b-mix 5% 6 0.1% in N 2 ) to achieve ambient CO 2 (400 6 10 mmol mol 21 ) concentration in the headspace containing the branch. The unit was flushed at 20 L min 21 for 10 min before each sampling to remove memory effects . A branch was inserted into the chamber and sealed around at the base. Plants were provided with natural photosynthetically active radiation at 800 to 1,200 mmol m 22 s 21 during sampling. The chamber temperature was maintained at 30°C 6 2°C, and leaf temperature was 26°C 6 2°C measured using an infrared thermometer (Agri-Therm III; Everest Interscience). Relative humidity varied from 29% to 52%. Sterile fritted glass thermal desorption (TD) tubes comprising Carboxen 1016 and Carbopack X adsorbents (Sigma Supelco) were conditioned at 250°C through helium purging (100 mL min 21 , 60 min). Volatiles were collected into TD tubes in April and May 2013 using an oil-free pump connected to a mass flow controller (Brooks 5850E). Chamber blank control, TD tube secondary desorption control, and branch memory effect (preflush blank) screenings were also performed.

Cuvette-Based Leaf-Level Sampling
An LI-6400XT portable gas-exchange system was suitably modified to sample volatiles directly from the leaf cuvette onto the thermal adsorbent bed described above. The LiCor was supplied with volatile free humidified air mixed with CO 2 . Within-cuvette ambient CO 2 was 400 mmol mol 21 , leaf temperature was 25°C, humidity ranged from 40% to 62%, and the light intensity was 1,200 mmol m 22 s 21 . Isoprenoids were also sampled in a low-O 2 (2%), ambient-CO 2 (400 mmol mol 21 ) atmosphere generated by mixing N 2 , high-purity O 2 , and CO 2 at the required ratio and humidified to achieve 40% to 50% relative humidity in the cuvette. Leaves were exposed to low O 2 or high O 2 for 5 to 10 min (stable readings) prior to sampling. The effect of varying O 2 exposure on photosynthesis was also studied independently (although simultaneously) on many leaves.
Sampling from E. camaldulensis ssp. obtusa under Controlled CO 2 , O 2 , Temperature, and Water Availability One-year-old saplings (n = 6) were maintained at 100% FC (volatiles were sampled), then gradually dried down to 50% FC, and again acclimated for 15 d (volatiles were sampled again). Before sampling, individual leaves were exposed for 10 min to five possible atmospheric CO 2 concentrations (60, 180, 400, 1,000, and 1,800 mmol mol 21 ), two temperatures (28°C and 38°C), two O 2 concentrations (20% and 2%), and saturating light intensity (1,500 mmol m 22 s 21 ). The CO 2 compensation point remained close to 60 mmol mol 21 in all treatments except in well-watered plants at 28°C (low O 2 ), where it was 10 mmol mol 21 . CO 2 treatment was randomized so that on a given day, some leaves from different biological replicates received low-CO 2 treatment while others received high-CO 2 treatment to avoid either stimulation or limitation of photosynthesis.

Thermal Desorption Gas Chromatography-Mass Spectrometry Analysis
A Shimadzu GCMS-QP 2010 fitted with an auto thermal desorption system (TD-20) was used for offline volatile analysis. Ultrapure helium (BOC) was used as the carrier gas. Isoprene and D-limonene (analytical grade; Sigma) were injected into sterile 5-L Tedlar bags comprising N 2 to generate standard mixing ratios. Then, the standard mixture was adsorbed onto TD tubes (described above), which were used to calibrate the instrument at regular intervals. Isoprene sampled from the plant chamber was desorbed from the TD tube at 220°C (60 mL min 21 for 5 min). A 30-m, 25-mm i.d., 25-mm RTX-5 Sil MS (Restek) capillary column was used for gas chromatography. The temperature regime for the gas chromatography run was 28°C (3 min) to 110°C (3 min) at 5°C min 21 and finally to 180°C at 5°C min 21 . The chromatographic peaks were identified by comparing them with isoprene and monoterpene standards (a-pinene and D-limonene) and reference mass spectrographs in the National Institute of Standards and Technology Standard Reference Database 1A (National Institute of Standards and Technology, 2008). I e was calculated by utilizing sampling flow rate, total leaf area in the sampling chamber (for branches), and quantified isoprene standards.

Statistical Analysis
The statistical tests were performed using Minitab (version 16 statistical package). The equality of means in responses within species between two treatments was analyzed using paired Student's t tests. Differences in mean responses between two species to the same treatment were subjected to twosample Student's t tests. The CO 2 , O 2 , temperature, and drought intensity interactions were analyzed using a multilevel general full factorial model with ANOVA (Montgomery, 2004). Experiments had four (between species) to six (sequential) biological replicates and eight to 15 independent leaf-level measurements (technical replicates) per treatment group.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Electron transport rate to net assimilation rate ratio in response to varying levels of oxygen concentration in E. camaldulensis ssp. camaldulensis.
Supplemental Figure S2. Photosynthesis response to short-term heat stress.
Supplemental Figure S3. Short-term response of photosynthesis to low O 2 .
Supplemental Figure S4. Comparison of emissions under heat without drought and drought without heat.
Supplemental Figure S6. Isoprene and monoterpene emission rates peaking at two different CO 2 concentrations.
Supplemental Table S1. Photosynthesis parameters of E. camaldulensis ssp. camaldulensis in response to drought and photorespiratory stress.