Light dominates the diurnal emissions of herbivore-induced volatiles in wild tobacco

Background Timing is everything when it comes to the fitness outcome of a plant’s ecological interactions, and accurate timing is particularly relevant for interactions with herbivores or mutualists that are based on ephemeral emissions of volatile organic compounds. Previous studies of the wild tobacco N. attenuata have found associations between the diurnal timing of volatile emissions, and daytime predation of herbivores by their natural enemies. Results Here, we investigated the role of light in regulating two biosynthetic groups of volatiles, terpenoids and green leaf volatiles (GLVs), which dominate the herbivore-induced bouquet of N. attenuata. Light deprivation strongly suppressed terpenoid emissions while enhancing GLV emissions, albeit with a time lag. Silencing the expression of photoreceptor genes did not alter terpenoid emission rhythms, but silencing expression of the phytochrome gene, NaPhyB1, disordered the emission of the GLV (Z)-3-hexenyl acetate. External abscisic acid (ABA) treatments increased stomatal resistance, but did not truncate the emission of terpenoid volatiles (recovered in the headspace). However, ABA treatment enhanced GLV emissions and leaf internal pools (recovered from tissue), and reduced internal linalool pools. In contrast to the pattern of diurnal terpenoid emissions and nocturnal GLV emissions, transcripts of herbivore-induced plant volatile (HIPV) biosynthetic genes peaked during the day. The promotor regions of these genes were populated with various cis-acting regulatory elements involved in light-, stress-, phytohormone- and circadian regulation. Conclusions This research provides insights into the complexity of the mechanisms involved in the regulation of HIPV bouquets, a mechanistic complexity which rivals the functional complexity of HIPVs, which includes repelling herbivores, calling for body guards, and attracting pollinators. Supplementary Information The online version contains supplementary material available at 10.1186/s12870-021-03179-z.

HIPVs released from foliage are known to be diurnally regulated, particularly the terpenoid constituents. For example, the abundance of volatiles emitted from Picea abies foliage after methyl jasmonate elicitation, which mimic herbivore-induced emissions, is greater during day than night [17]. Similar patterns are reported for Phaseolus vulgaris HIPVs elicited by attack from L. huidobrensis larvae [18]. Furthermore, in A. annua the gene QH6, which encodes (−)-α-PINENE/(−)-β-PINENE SYNTHASE, showed diurnal patterns of transcript abundance [19]. GLV emissions were induced by herbivory both during the day and during the night from Phaseolus lunatus [20], during night from Nicotiana tabacum [21], while in P. vulgaris (kidney bean), some constituents are released during light periods while others are released during dark periods [18].
In previous research with the native diploid tobacco N. attenuata, we found that foliar HIPV constituents, including (E)-α-bergamotene, (E)-β-ocimene and linalool, were diurnally released [22][23][24][25]. These volatiles play a central role in N. attenuata's herbivory, defense and pollination interactions with insects and the same compounds play different roles depending on the tissue from which they are released and their timing. For example, (E)-α-bergamotene, produced by NaTPS38, is emitted from flowers at night and increases the probing time and thus pollination success of M. sexta moths visiting flowers. However, when emitted from attacked leaves during the day, it attracts natural enemies of M. sexta larvae [24] to mediate a potent indirect defense [26]. The emission of (E)-β-ocimene, produced by NaTPS25, was correlated with (E)-α-bergamotene emission from diverse N. attenuata accessions, and the two genes are co-localized on the N. attenuata genome [27]. In contrast, a QTL for the emission of linalool, a monoterpene produced by NaLIS, was mapped to a locus different from that of (E)-α-bergamotene and (E)-β-ocimene. Linalool also attracts natural enemies, and can deter oviposition of M. sexta moths in a context-dependent manner [23]. GLVs have also been found to play important roles in the defense of N. attenuata against herbivory in nature [26]. GLVs released in the morning elicit greater predatory activity by day-active Geocoris spp. predators than does the evening-released blend of GLVs even when both blends are experimentally applied to plants during the window of Geocoris spp. activity [28]. Furthermore, transcripts of genes involved in GLV biosynthesis, including NaHPL and NaLOX2, were regulated by the plant's circadian clock [25,29].
Here we confirmed that terpenoid volatile emission is greater during light periods while emission of GLVs is greater during dark periods, by manipulating the light environment of leaves independently of day-night cycles. The emission patterns of terpenoids were not altered by silencing the expression of photoreceptor genes, the supplementation of far-red light, or by increasing stomatal resistance by external abscisic acid (ABA) application. The emission pattern of GLVs, however, was changed by silencing NaPhyB1 transcripts, and increased by exogenous ABA treatments. The genes involved in terpenoid and GLV biosynthesis all showed diurnal patterns of transcript accumulation, and the promoter regions of these genes were populated by various cis-acting regulatory elements involved in light-, stress-, phytohormoneor circadian-regulated processes.

Light deprivation decreased terpenoid, but increased GLV emissions
Predation assays performed with a biparential RIL population of N. attenuata plants grown in a field plot in the summer of 2017 revealed that M. sexta eggs and larvae were predated by big-eyed bugs (Geocoris spp.) during the day time (Fig. S1), consistent with previous findings [27]. Because plant volatiles play essential roles in mediating this tri-trophic interaction [23,26] and to get a better understanding of their diurnal/nocturnal regulation, we collected headspace day and night time volatiles from N. attenuata leaves on glasshouse-grown plants elicited by wounding plus M. sexta larval regurgitant (W + R) using PDMS tubing pieces (Kallenbach et al. [30], Fig. 1A, B). The assayed leaf was enclosed in a plastic cup which could be wrapped using aluminum foil to locally deprive the treated leaf of light (Fig. 1A, B). We found that the emission of most terpenoid volatiles corresponded strongly to light periods, especially linalool and (E)-α-bergamotene (Fig. 1C). The emission of these terpenoid volatiles was very low during the night period (22:00-6:00), increased in the morning (6:00-10:00), and peaked during the middle of the day (10:00-14:00), then decreased in the afternoon (14:00-18:00) and returned to low levels in the evening (18:00-22:00). When the leaf was enclosed in a wrapped cup during the middle hours of the day (10:00-14:00), the emission of most terpenoids immediately decreased to low levels, comparable to those measured at night. After the foil wrapping was removed in the afternoon and evening, the emission of these terpenoid volatiles returned to control levels (Fig.  1C). In contrast to the day-emitted terpenoids, many GLVs were abundantly emitted at night and steadily decreased during the day, although some, such as (Z)-3hexenyl acetate and (Z)-3-hexenyl propanoate, continued to be released at high levels in the morning. Local light deprivation during the middle hours of the day did not alter GLV emissions but up-regulated them in the afternoon, which likely continued through the evening hours (Fig. 1C). The decreased total terpenoid volatile emissions and increased total GLV emissions associated with light deprivation were further confirmed in single 24 h collections after W + R treatments (Fig. 1D). Thus terpenoids and GLVs showed opposing responses to local light deprivation.
Because the volatile emission might also be affected by temperature and humidity caused by wrapping the cups, we monitored these two factors outside of cups, in clean cups and in wrapped cups (Fig. S2A). We found that there was only 0.8-2.4°C drop in temperature in wrapped cups compared to clean cups during the wrapping period 10:00-14:00 (Fig. S2B). The average humidity in wrapped cups were about 9% higher than in clean cups, but it was more variable depending on the enclosed leaf (Fig. S2B). These difference in temperature and humidity were not supposed to significantly affect the volatile emission in our experimental setups because they were in the comfortable zone of the plants.
Silencing NaPhyB1 expression and ABA treatments altered GLV but not terpenoid emissions We assessed the volatile emissions elicited by W + R treatment in a series of plants individually silenced (by stable transformation with inverted repeat-ir-RNAi constructs) in the following photoreceptor genes: NaPhyB1, NaPhyB2, NaPhyB1 x NaPhyB2, NaCrypt1 and NaCrypt2 (previously characterized by Oh et al. [31]). The emission pattern of volatiles in all the plants were largely unchanged ( Fig. 2A, Fig. S3) with the exception of those of irPhyB1 plants. Particularly, in plants generated by crossing irPhyB1 with irPhyB2 to silence both phytochrome genes, (Z)-3-hexenyl acetate emissions increased in the afternoon. Supplementation of far red light did not alter total terpenoid or GLV emissions as assessed in 24 h collections ( Fig. 2B) or collections taken every 4 h (Fig. S4).

ABA regulated HIPVs but not via stomatal conductance
We tested if the day/night patterns of volatile emission were controlled by changes in stomatal resistance, which might influence the release of volatiles from plant tissues (Fig. 3A, [32,33]). We used ABA treatments to increase the stomatal resistance of leaves of empty vector control (EV) and irMPK4 plants, which, as previously described [34], have stomata that are largely in the open state and are not responsive to ABA treatments (Fig. 3A, B). ABA treatment of EV plants increased stomatal resistance during the day, but the diurnal emissions of the terpenoids, linalool and (E)-α-bergamotene, were not altered in either EV or irMPK4 plants (Fig. 3C). In contrast, emissions of (Z)-3-hexenyl acetate were upregulated at the sampling period that followed the ABA treatment and returned to control levels at the end of the day in both EV and irMPK4 plants (Fig. 3C). Interestingly, measurements of the internal leaf volatile pools revealed that ABA treatments decreased the amounts of linalool accumulated in leaves during the middle part of the day in both EV and irMPK4 plants, although it did not decrease amounts released to the headspace (Fig. 3C, D). Internal (E)-α-bergamotene pools tended to increase but were not significantly changed by ABA treatment, consistent with the headspace emissions. Internal (Z)-3-hexenyl acetate pools tracked headspace emissions, and both were enhanced by ABA treatments in EV plants, with smaller enhancements in irMPK4 plants (Fig. 3D).
Thus ABA treatment enhanced the emission of a GLV, which is more abundantly emitted in the dark, despite reducing stomatal conductance. Similarly, impaired red/far red light perception also increased GLV emission. However, neither ABA treatment nor impairment of plant light perception had a significant effect on emission of terpenoids.

Genes responsible for terpenoid and GLV biosynthesis are diurnally transcribed
A previously published microarray data set revealed dynamic transcriptome changes in N. attenuata after W + R treatment [35]. From this data set, NaLIS transcripts, encoding the linalool synthase 23, were abundant in the daytime and barely detectable at night (Fig. 4A). Transcripts of the (E)-α-bergamotene synthase gene, NaTPS38 [24], were strongly increased in both day and night, however, the rate of up-regulation was greater during the day. Intriguingly, transcripts of NaHPL and NaLOX2, both involved in GLV biosynthesis, were also more abundant during the day, even though the emission of most GLVs is nocturnal. Using qRT-PCR we further confirmed that light deprivation strongly downregulated NaLIS transcript levels, as well as those of NaGPPS1 and NaGPPS2, genes responsible for the biosynthesis of monoterpenoid substrates upstream in the pathway (Fig. 4B).
Cis-acting regulatory elements involved in light-, stress-, phytohormone-and circadian-regulation were identified in the biosynthetic genes We scanned the 2 kb promoter regions of NaLIS, NaTPS38, NaHPL and NaLOX2 for cis-acting regulatory elements (Fig. 5, Table 1, Table S2). NaLIS has six and NaTPS38 five elements identified as potentially involved in responsiveness to light in their promoters, and eight and fourteen were found in the promotors of NaHPL and NaLOX2, respectively. Box 4 [36] was present in all the promoters with different copy numbers. Some of the elements were found in two or three of the promoters such as the AE box [37], G-box [38,39], GT1-motif [40], and TCT-motif [41]. The others were only found in a specific promoter, e.g. the ACE [42] element was only in the promoter of NaLOX, chs-CMA1a [43] in the promoter of NaLIS, GA-motif [44] and I-box [45] in the promoter of NaHPL. The promoter regions of these genes also contained numerous elements responsive tostress and/or phytohormones including JA, MeJA, SA and ABA, specifically: seven for NaLIS, eight for NaTPS38, 11 for NaHPL and 13 for NaLOX. Notably, the ABRE element (ACGTG/CACGTG/AACCCGG) involved in ABA regulation [46,47] was specifically found in the NaLIS, NaHPL and NaLOX2 promotor regions, but not for NaTPS38. This is consistent with ABA treatment enhancing (Z)-3-hexenyl acetate emissions and internal pools, and decreasing linalool internal pools, but not affecting (E)-α-bergamotene (Fig. 3D). A single circadian element (CAAAGATATC, [48]) was identified for NaLIS, NaTPS38, and NaLOX2, but not for NaHPL.

Discussion
HIPVs emitted from N. attenuata plants varied in composition and abundance across the day/night cycle. Terpenoid constituents of the volatile bouquet showed emissions which peaked in the middle of the day and responded quickly to light deprivation, while GLVs tended to be emitted more at night and were enhanced by light deprivation. Rhythms of the terpenoid volatiles were not altered by the constitutive silencing of photoreceptor genes, supplementation by far-red light, or stomatal closure elicited by ABA treatments. GLV emissions, including (Z)-3-hexenyl acetate, were increased by PhyB silencing and ABA treatments. Transcript abundances of the genes involved in the biosynthesis of terpenoids and GLVs varied over a day/ night cycle, and various light-or JA-responsive cis-acting regulatory elements, as well as some circadian clock and ABA regulation elements were identified in their promoter regions.

Correlations of HIPV emissions with herbivory and tritrophic interactions
In nature, plants must cope with environments characterized by periodic changes that accompany day/night cycles that include fluctuations in many factors in addition to light: temperature, humidity, and innumerable biotic interactions. For instance, plants need to adopt to rhythms of attacks by pathogens and herbivores, or activity patterns of beneficial species, such as pollinators and natural enemies of herbivores [4,6,57,58]. To cope with this environmental rhythmicity, plants have evolved both endogenous rhythms and the ability to coordinate responses to changes in specific environmental factors. Much of this can be seen in changes in the abundance and composition of their metabolites [59]. Different HIPV constituents of N. attenuata plants are associated with activities of different species that interact with the plants in nature. The larvae of the specialist Manduca spp. are known to feed diurnally. Experimental measures of predation rates of eggs and larvae were correlated with the diurnal emissions of terpenoid volatiles, which is consistent with previous reports by Joo et al. [28]. If we understand the mechanisms responsible for these emission patterns, these mechanisms could provide means of experimentally manipulating the patterns and hence provide a way to test such correlations observed in field studies.

Rhythms of the emission of terpenoid volatiles and GLVs
Herbivore-induced mono-or sesquiterpene terpenoids are diurnally regulated in many species such as G. hirsutum [60], A. annua [19], A. majus [61] and Phaseolus vulgaris [18] upon herbivory. Similarly, in N. attenuata at least (E)-β-ocimene, (E)-α-bergamotene and linalool showed diurnal emission following herbivore elicitation [22,23]. Interestingly, (E)-α-bergamotene is diurnally emitted from foliage of N. attenuata plants and functions in attracting natural enemies that protect plants from herbivory by larvae of M. sexta. However, the same compound is nocturnally released from flowers and attracts adults of the same insect, which provides pollination services [24]. Further study showed that (E)-βocimene and (E)-α-bergamotene are correlated with each other across different N. attenuata genotypes, but linalool is independently regulated [23,27]. Furthermore, unlike the emission of linalool, the endogenous leaf pools of conjugates derived from linalool, such as linalool glycosides, lack the diurnal pattern of emissions [23]. In this study we confirmed the diurnal emission pattern of terpenoid volatiles, including (E)-α-bergamotene and linalool [28], and further found that this diurnal pattern was not altered by the silencing of photoreceptor genes or increased stomatal resistance resulting from ABA treatments. Interestingly, ABA treatment suppressed internal linalool, but had little influence on internal (E)-α-bergamotene. The contradictory influences of ABA treatments on emission and internal linalool might be attributed to opposing effects of ABA on biosynthesis or conjugation of linalool, and the mechanisms of the emission process.
In contrast to conserved diurnal emission of terpenoids across different species, GLVs showed both diurnal and nocturnal release. For example, mechanical damage to lima bean (Phaseolus lunatus) plants during the light phase induced high emissions of both βocimene and (Z)-3-hexenyl acetate, while damage during the night only upregulated the emission of (Z)-3-hexenyl acetate [20]. In elicited kidney bean, GLV (Z)-3-hexenol was abundantly emitted during the night while the opposite was true for (Z)-3-hexenyl acetate [18]. In this study, we revealed a high nocturnal emission of GLVs, which is in agreement with previous studies [28,29]. Additional work is required to understand the potential ecological significance of these opposing patterns for different constituents, and for these inferences, basic natural history information about the N. attenuata plants and activities of their surrounding organisms in the field plays a central role. For example, GLVs emitted during night might be used as feeding cues by nocturnally active herbivores in the field, as has been shown for M. sexta larvae in both laboratory and field studies [62].

Light is essential for the regulation of HIPVs
A major conclusion of this study is that light strongly and locally regulates HIPV emissions, for both terpenoids and GLVs, although this could not be attributed directly to specific photoreceptors. This finding is consistent with earlier reports demonstrating that HIPV emissions from maize plants were light-dependent [63]. Light could influence the biosynthesis of terpenoids in at least two ways. First, pyruvate and glyceraldehyde 3phosphate, the precursors for terpene biosynthesis, are derived from the Calvin cycle [64,65], which requires the immediate products of photosynthesis. For example, biosynthesis of β-ocimene in lima bean plants dependeds on the CO 2 fixation by photosynthesis [20]. Perhaps this explains the immediate reduction of terpenoid emission by light deprivation in our study, because it shuts down the primary source of substrates for the terpenoid biosynthesis pathway. However, in the sink parts of a plant, such as floral tissues, the biosynthesis of terpenoids is not limited by photosynthesis, and N. attenuata flowers emit (E)-α-bergamotene during the night [24]. Second, light as signal, can regulate the expression of the genes involved in the biosynthesis of terpenoids. Light is necessary for the expression of genes in the MEP/DOXPpathway, which synthesizes the terpenoid precursors [66]. In seedlings of A. thaliana almost all the genes in the pathway are repressed in darkness [67]. In the current study, we also found that synthase genes of (E)α-bergamotene, linalool, and the precursor GPP had lower transcript levels or were less induced by herbivory during night, were also downregulated by light deprivation in the middle of the day, and were associated with light-responsive regulatory elements.
Surprisingly, we found that GLVs were also strongly and temporarily affected by light and ABA treatment in N. attenuata. In response to herbivory, the increase in GLV emissions is more rapid than the response in terpenoid volatiles [28,68]. Here, study, light deprivation in the middle of the day enhanced GLV emission with a delay during the subsequent hours, in contrast with the immediate reduction of terpenoid emission. This might be because the precursors for GLV biosynthesis, polyunsaturated fatty acids, are not immediate de novo products of photosynthesis. Therefore, light, probably together with the circadian clock, may affect biosynthesis of GLVs through regulating the expression of biosynthesis genes, with some delay in the accumulation of active proteins. This enhancement could be viewed as consistent with the effect of NaPhyB1 silencing in the down-regulation of GLV (Z)-3-hexenyl acetate emissions, and suggests that far-red light could function as a signal regulating GLV emissions. However, volatile collections from leaves exposed to supplemental far-red light did not support this inference, implying that the regulation might be very transient or occur only at a specific interval of the day/night cycle. Furthermore, we found that transcript levels of NaHPL and NaLOX2 sharply decreased at the start of the dark phase, in conflict with the higher nocturnal emission of most of the GLVs. These often contradictory results from physiological manipulations underscore the inference that emissions are under complex regulation. A detailed examination of the promotors of HIPV biosynthetic genes can be a valuable exercise so as to better understand the environmental factors that contribute to the patterns of transcript abundances.
Complex regulation of HIPVs in N. attenuata HIPV emissions are known to be influenced by herbivory elicitation [7], light conditions, the endogenous circadian clock [61], temperature [69], water availability [70] and the still mysterious emission process [71]. Many volatiles are only emitted or increased emission upon elicitation such as by herbivory and mechanical damage [67,72,73]. These inductions are usually mediated by the JA signaling pathway [11,29,[74][75][76][77][78]. The circadian clock is another major regulator of HIPVs. For instance, genes involved in the biosynthesis of terpenoid volatiles, including TPSs and the genes upstream in the pathways, are known to be regulated by circadian clock components [61,64,65]. The transcription of GLV biosynthesis genes, NaHPL and NaLOX2, is also regulated by the circadian clock [25,29]. The cis-acting regulatory elements identified in the promoter regions of the synthase genes of terpenoids and GLVs provide additional evidence for the concerted regulation of these genes by herbivory, light and the circadian clock. Multiple stress-and phytohormonesignaling elements were identified in the promoter regions of the genes involved in terpenoid and GLV biosynthesis and could provide a means of coordinating HIPVs with other JA-, MeJA, SA, and ABA-mediated chemical defenses. A number of light responsive elements and circadian related elements were found for both terpenoid and GLV genes. Interestingly, an ABAresponsive element present in the promoter of NaLIS but not NaTPS38 might account for the specific suppression of NaLIS transcription and linalool biosynthesis by ABA treatments. These cis-acting regulatory elements suggests the regulation by different transcription factors that could coordinate HIPV biosynthesis with various environmental factors such as herbivory, light and the internal clock.

Conclusions
In summary, this study revealed that light is necessary for the full induction of terpenoid volatiles and the suppression of GLVs, and regulates the transcription of HIPV genes in concert with many other factors.

Plant materials and growth
Plant seeds used in this study were from a seedbank of the Department of Molecular Ecology, MPICE (http:// www.ice.mpg.de/). The numbers of transgenic lines used in a glasshouse experiment are listed in Table S1. Plants grown in the field plot for the predation assay were the F12 generation of an advanced intercross-recombinant inbred line (AI-RIL) population [24], which was constructed from a cross between two inbred lines separately originating from collections in Arizona (AZ) and Utah (UT) by Baldwin and colleagues [79,80]. The field experiments were performed at the Walnut Creek Center for Education and Research (WCCER)(34°55′17.8″N 112°50′42.2″W) located in Arizona, southwestern United States. The plants grown in the field are native to the area, not endangered species, non-transgenic material and non-regulated. No samples were collected from the wild. The field experiments were permitted and supported by WCCER. Seeds were germinated on Gamborg's B5 medium in Petri dishes under conditions previously described by Krügel and colleagues [81]. Two weeks later, seedlings were transferred into peat plugs (Jiffy 703, www.jiffypot.com). During the next 2 weeks seedlings were increasingly exposed to outdoor conditions and subsequently transplanted into the field plot. The plants were watered regularly by a drip irrigation system [82].
Plants used for the different light treatments and dynamic volatile collections were grown in a glasshouse. The generation and screening of the transgenic lines (listed in Table S1) silenced in the expression of photoreceptor genes and NaMPK4 were previously described [31,34]. Seedlings were germinated in the described way and transferred into potting soil (www.klasmanndeilmann.com) in small pots (TEKU JJP 3050 104 pots, Poeppelmann GmbH & Co. KG, Lohne, Germany) for 10 days and then transplanted into 1 L pots and kept in a glasshouse (19°C-35°C, 55% humidity, light period: 6: 00-22:00, supplemental lighting by Philips Sun-T Agro 400 W and 600 W sodium lights) [81,83]. Plants were watered and fertilized as previously described by Schuman et al. [84]. Experimental research and field studies on plants complies with relevant institutional, national, and international guidelines and legislation.
M. sexta rearing and regurgitant collection M. sexta used in this study were from an in-house colony maintained on artificial diet (Bell RA & Joachim FG 1976) in a climate chamber (24°C, light period: 0:00 to 13:00, relative humidity: 70%). For collection of regurgitant, caterpillars were reared on N. attenuata UT WT plants. Collection and storage of regurgitant were previously described [78,85].

W + R treatment and headspace sampling
The first stem leaf of elongated plants was chosen for elicitation with wounding by a pattern wheel and the immediate addition of 20 μL of 20% diluted regurgitant in water to the puncture wounds in the leaf lamina. The treated leaf was gently rubbed with fingers in clean gloves. This leaf was enclosed in a ventilated PET cup (ca. 650 mL). Headspace volatiles were collected as described by Kallenbach et al. [30] with two pieces of silicone laboratory tubing (ST, 1 mm i.d. × 1.8 mm o.d.; Carl Roth, catalog number: 9555.1; 5 mm for each piece) placed into the cup under the leaf for 4 h (time-resolved) or 24 h (long term collection). For the quantification of internal volatile pools, the sampled leaf was flash-frozen and ground in liquid N 2 . One hundred mg tissue was placed into a 1.5 mL glass GC vial (Sigma-Aldrich) and 0.8 mL saturated CaCl 2 solution was added. One piece of ST was placed into the vial and fully immersed into the solution to extract free volatiles [29]. STs with collected volatiles were stored in a clean glass vial in − 20°C freezer.
Light deprivation, far-red supplementation and ABA treatment Temporary, local light deprivation was applied to elicited leaves by wrapping the collection cup with aluminum foil. Temperature and humidity in the glasshouse, clean and wrapped cup were recorded by PRO-USB-2 data Logger with humidity, temperature sensor (RS Components GmbH). Additional far-red was supplied as previously described [31]: five-Watt far-red light emitting diodes (LED, 720 ± 10 nm) were directed to the adaxial side of the targeted leaf. ABA (Sigma-Aldrich) was diluted to 2 μM and sprayed on the leaf in the morning at around 6:00. Stomatal resistance was measured by an AP4 Porometer (AP4-UM-3, Delta-T Devices) following the manufacturer's instructions.

TD-GC-QMS analysis
STs with absorbed volatiles were analyzed on a quadrupole GC-MS-QP2010Ultra equipped with a TD-20 thermal desorption unit (Shimadzu). Desorption and analysis were performed as previously described [30]. Peaks were integrated using the target and reference ions as previously described [23] and tentative compound identifications were based on comparison with an in-house library of N. attenuata volatiles prepared in the same lab including the spectra, retention times and relative retention orders of these volatiles from standards and samples analyzed using the same method. Identification of volatiles in the reference library was based on pure standards, Kovats retention indices and comparison of spectra to NIST MS libraries [30].

Measuring transcript abundance of genes
The transcription levels of NaLIS, NaGPPS1 and NaGPPS2 in leaves under standard light or light deprivation conditions were quantified using qPCR as previously described [23]. Total RNA was extracted using the NucleoSpin® RNA Plant kit (MACHEREY-NAGEL). After digestion of genomic DNA, 1 μg total RNA was transcribed with the PrimeScript™ RT reagent Kit (TAKARA). Primer sequences were described previously [23] and the qPCR was run in a MX3005P PCR cycler (Stratagene).

Identification of cis-acting regulatory elements
The 2Kb sequences upstream of the start codons of NaLIS, NaTPS38, NaHPL and NaLOX2 were extracted from the genomic sequences of N. attenuata [86]. Then these sequences were searched online through the Plant Cis-acting Regulatory Element database (http:// bioinformatics.psb.ugent.be/webtools/plantcare/html/, Lescot et al. [2002]).  Fig. S2. Difference in temperature and humidity caused by wrapping the collecting cups. Fig. S3. Volatile emissions in plants with NaCrypt and NaCrypt2 silenced (related to Fig. 2). Fig. S4. Supplementation with extra far-red light did not alter the dynamics of volatile emissions. Table S1. Transgenic lines used in this study. Table  S2. Detailed information of cis-acting regulatory elements identified in the promoter sequences of NaLIS, NaTPS38, NaHPL and NaLOX.