Sorbent tubes (Tenax TA, 250 mg, 60–80 mesh, O.D. x L. 1/4 in. x 3 1/2 in.) are desorbed using an auto-sampler thermal desorber (TurboMatrix 650 ATD, PerkinElmer, Waltham, MA) and analyzed using a gas chromatography and mass spectrometry instrument (Clarus 600, PerkinElmer, Waltham, MA). The ATD system performed desorption of tubes at 40 mL/min for 15 min at 270°C, followed by a thermal flash from −30°C to 270°C, maintaining the temperature at 270°C for 5 min without split. Two reference solutions are used for monoterpene measurements from the cannabis terpene kit standard (100 μg/mL in methanol provided by SPEX CertiPrep, Rickmansworth, United Kingdom). Forty-one compounds (monoterpenes and sesquiterpenes mixed) are present in the kit. Furthermore, a gas standard composed of isoprene, toluene, benzene, and xylenes is analyzed from the National Physical Laboratory (NPL) standard bottle. The standard concentration limit of detection (LOD) values can be visualized in Supplementary Table S1 and is calculated, as indicated in the supplementary. The analysis method includes the following steps: the column temperature was maintained at 35°C for 5°min, increased at 5°C/min to 160°C, and then increased at 45°C/min to 270°C, where it was maintained for 5 min. Approximately 30 compounds were identified for these measurements and are listed in Supplementary Table S1. Uncertainties of concentrations (in pptv) for a species i were calculated using the following equation: u 2 ( concentration i   /   concentration i 2 = u 2 volume i   /   volume i 2 + u 2 standard i   /   standard i 2 + u 2 A i   sampled   /   A i   sampled 2 , where u² (volume_i)/volume_i ²: total absolute uncertainty on the sampled volume in the Tenqx tube; u² (standard_i)/standard_i ²: total absolute uncertainty on the standards; u² (A_{i sampled})/A_{i sampled} ²: total absolute uncertainty on the area measurement. Detailed calculation is provided in the Supplementary Material.

The PTR-MS technique has been used and optimized for VOC measurements for over two decades and described elsewhere (Lindinger and Jordan, 1998; Sekimoto et al., 2017; Yuan et al., 2017). PTR-ToF-MS (IONICON X6000) was operated in the H₃O⁺ mode at 1 Hz with an electric field strength (E/N) of 101.36 ± 0.11 Td (1 Townsend = 10^–17 V cm⁻² mol⁻¹). To obtain these standard drift tube conditions, the drift tube pressure was maintained at 2.590 ± 0.001 mbar and at a temperature of 119.96°C ± 0.24°C with a drift voltage E of 450.86 V. The extraction time rate of the ions inside the ToF part was fixed at 5.0 µs and the maximum time of flight at 32.0 µs, which allows the measurement of the mass spectrum between m/z 2 and m/z 475. During the experiment, we focused on compounds known to be emitted by plants and, in particular, the ones identified as pathways and catabolite markers (Fitzky et al., 2021). The list of 21 compounds and their respective sensitivities (ncps/ppbv) are listed below in Table 1. The background signal of the instrument was determined before and after each experiment with zero air provided by a HPZAG catalyst device and subtracted to the ambient signal for quantification.

As mentioned above, a gas standard from the National Physical Laboratory (NPL; 4 ppb ±5%, Supplementary Table S1) was used to determine the experimental sensitivities of the instrument. The gas standard was diluted using a liquid standard unit (LCU) with the HPZAG catalyst device mentioned earlier to perform a multiple point calibration with concentrations between 0.5 and 12 ppb; R² > 0.99 were determined. However, all the sensitivities cannot be obtained for all the target compounds. Indeed, due to the large number of VOCs that are identified and measured, there are no gaseous standards covering all the VOC spectra. Thus, alternative methods were used for the other VOCs. The one used in this study is based on the work of Sekimoto et al. (2017). A linear relationship between the experimental sensitivities from the gas standard and the ion molecule reaction rate constant (k-rate) of the compounds is obtained (slope of 7.85 with a confidence interval of 95%). For a large number of compounds, the k-rate constant is reported in the literature (Zhao and Zhang, 2004; Cappellin et al., 2012). For other compounds detected with the unknown k-rate, a value of 2 was chosen as a default value. The LOD was also determined for the standard gaseous compounds, following de Gouw and Warneke (2007), based on the ion per second of the background signal and the unnormalized sensitivity. LOD lies from 24 ppt (m/z 133.026) to 670 ppt (m/z 371.102). The range of uncertainties is 11%–16% except for m/z 47.049 (26%). The QA/QC of PTR-MS follows the Aerosol Cloud and Trace Gases Research Infrastructure (ACTRIS) recommendations (WP3, NA3, deliverable 3.20).

The software ioniTOF of IONICON was used for data acquisition and for controlling the different instrumental parameters during the experiment. Different steps are used for data processing. First, we controlled the quality of the peak integration and any drift in the “timebin-masses” relationship. Then, the PTR-MS viewer (IONICON v. 3.4) was used. Finally, the data were quantified using Igor Pro v 6.37 and RStudio.

The Puy de Dôme station (PUY, 45.77°N, 2.96°E, 1,465 m a.s.l.) is a high-altitude station of the instrumented site CO-PDD (Cézeaux-Aulnat Opme Puy de Dôme, Baray et al., 2020). Regular measurements of atmospheric composition are performed there in the long term for the study and monitoring of the climate and atmospheric composition in the framework of the Global Atmosphere Watch (GAW) network and of the European Research Infrastructure ACTRIS, respectively.

Samples were collected in the forest downstream of the Puy de Dôme station (PUY) during summer (July 2020 and 2021) into the forest (Figure 2). The first sampling was performed at point 1 near the Col de Ceyssat, and the second sampling (2 on the map) was performed near the PUY train station.

Measurements in the vegetation enclosure system were performed on three species of trees: European beech tree (F. sylvatica L., n = 1), common hazel tree (C. avellana L., n = 1), and Norway spruce (P. abies L., n = 1) branches. These species are important components of Puy de Dôme inventory, 2010 vegetation with a frequency of 58%, 75%, and 27%, respectively (Inventaire-forestier-national, 2010); their emissions have been previously investigated in other research endeavors (see Section 3). These species are, therefore, good candidates to evaluate the capacities of VELVET in qualifying and quantifying BVOC emissions and to compare to the literature. The temperature inside the laboratory room was fixed at ∼20°C ± 1°C.

Reunion Island is constituted of almost 40% of tropical forest (managed by the Office National des Forêts, ONF) and has an original flora comprising 835 native vascular plant species, of which 28.0% are strictly endemic to the island and 18.6% are regional endemics (Ah-Peng et al., 2010; Boullet and Picot, 2012). Reunion Island is often considered an open-air laboratory for the atmospheric studies (FARCE, BIO-MAIDO campaign, for example) and favorable for multi-disciplinary projects (ocean–vegetation–atmosphere interactions–geosphere) on the critical zone. During the OCTAVE in 2018 (BR/175/A2/OCTAVE) and BIO-MAIDO in 2019 (ANR-18-CE01-0013) campaigns, first studies of atmospheric composition have been performed in different places on the island: Mare-Longue and Bélouve forests, at the Observatoire de Physique de l’Atmosphère or OPAR-Maïdo (Duflot et al., 2019; Rocco et al., 2020; Rocco et al., 2022; Verreyken et al., 2021). First ambient measures of BVOC were performed during the FARCE campaign in 2015 (Duflot et al., 2019). Measurements and sampling were carried out at the Maïdo observatory, in the native Acacia heterophylla (Lam.) Willd. forest, Cryptomeria japonica (L.f.) D. Don planted forest and primary cloud forest (Bélouve forest), and lowland Mare-Longue Forest on a 10-m mast by sampling with adsorbent cartridges. This study showed high ambient concentrations of emitted isoprene in the Mare-Longue, Bélouve, and A. heterophylla forests (>50 pptv). In 2019, the BIO-MAÏDO campaign enabled transects to be carried out in order to understand the distribution of VOCs on the slopes of OPAR-Maïdo, which presents a diversified subalpine vegetation (Strasberg et al., 2005). These measurements made it possible to acquire an ambient VOC profile for the whole island. However, these studies did not take into account the VOC directly emitted by the plants in each site. In other words, this does not allow the assessment of the specific chemical signature (i.e., speciation of VOCs at emission) for each plant species present on site.

In the continuity of these measurements, with the VELVET-RUN project, we performed the measurement of BVOC emissions from dominant species (endemic and exotic) in the forests of La Réunion in order to characterize their chemical signature and utilize them in emission registries, thereby enabling a better modeling of VOC emissions in the atmosphere. For this purpose, we performed measurements with the VELVET chamber on native and exotic species (Table 2).

At the first stage, we performed the characterization of the signature of plant species for three different habitats (Figure 3) in the lowland forest (Mare-Longue Forest, MLF), cloud forest (Plaine des Fougères, PLF), and subalpine vegetation at the OPAR-Maïdo site in May 2022 (Table 2). Then, we characterized the endemic species of Reunion Island within a single plot in the lowland forest through intra-species comparisons (n = 3 replicates) to validate emission measurements for Labourdonnaisia calophylloides Bojer and Syzygium cymosum Lam. DC.

During chamber measurements, the temperature and humidity (MadgeTech sensors, Warner, United States) varied from 24.3°C to 26.4°C and 66%–86% in the chamber in MLF, 22.1°C–23.4°C and 56%–89% in the chamber in PLF, and 20.8°C to 22.9°C and 63%–83% in the chamber at OPAR-Maïdo.

Figure 4 shows the evolution of the concentration measured with the PTR MS and cartridge sampling for all the species. The average concentration during blank and TENAX tube sampling is presented in Supplementary Table S3. There are three kinds of evolution in the concentration: stabilized, growing, and decreasing concentrations. During measurement, we show that isoprene and monoterpene concentrations are decreasing, indicating that there are no high emissions from plants of these species. During the measurement in the chamber, the isoprene concentrations are stable after 10 min for P. abies and F. sylvatica. The monoterpene concentration decreases with time except for P. abies where a peak is observed at 10 min. This species is not considered monoterpene emitters.

High isoprene concentrations have been recorded for all species with an average 301 ± 36 pptv (Figure 5). Furthermore, high concentrations of MACR + MCK (up to 422 ± 37 pptv) have been measured, showing that oxidation processes occur from isoprene in the chamber. These concentrations could result from the high oxidation of isoprene in the chamber by the OH radical. Higher concentrations of sesquiterpenes are shown for C. avellana (211 ± 24 pptv). High total BVOC concentrations have been measured for the species C. avellana with concentrations up to 400 pptv, 42 pptv, and 100 pptv for isoprene, MACK + MVK, and monoterpenes, respectively. The high proportion of acetaldehyde and methanol (52% and 33%) has been measured for C. avellana (Supplementary Figure S1). The species with a large panel of VOC is F. sylvatica, for which high proportions of methanol and butyl have been recorded (20% and 50%, respectively). Similar to F. sylvatica, P. abies shows great proportions of butyl (74%). Butyl is a GLV and a signature of drought and herbivory stress emissions (Copolovici et al., 2014). Additional GLVs are detected in our measurements, including hexanal and methyl salicylate, but they are present in low concentrations (<2%, see Supplementary Figure S1). This implies that the emission of butyl compounds is likely originated from a different pathway.

Similar proportions are visualized between P. abies and F. sylvatica, except for concentrations that are less important for P. abies than F. sylvatica and acetone proportion higher than F. sylvatica. This result is in concordance with the VOC emission rates (%) of Fitzky et al. (2021) for F. sylvatica with a methanol emission of of 0.061 ± 0.111 nmol m⁻² s⁻¹ (or 2.02 ± 0.33 ng m⁻² s⁻¹). Fluxes of monoterpenes in Fitzky et al. (2021) have been estimated to be 0.390 ± 0.359 nmol m⁻² s⁻¹ (or 53.04 ± 48.82 ng m⁻² s⁻¹). Fitzky et al. (2023) showed emissions from F. sylvatica within control and stress conditions. They find fluxes of 0.42 ± 0.27 ng m⁻² s⁻¹, 0.07 ± 0.07 ng m⁻² s⁻¹, 62.12 ± 50.61 ng m⁻² s⁻¹, and 2.46 ± 2.26 ng m⁻² s⁻¹ for isoprene, MACR + MVK, monoterpenes, and sesquiterpenes, respectively. Fluxes from PTR-ToF-MS were calculated as Eq. 2 and shown in Table 3. Our fluxes of F. sylvatica are in line with this study for isoprene and sesquiterpene concentrations (0.62 ± 0.07 ng m⁻² s⁻¹ and 0.28 ± 0.06 ng m⁻² s⁻¹). Our estimation of monoterpene fluxes is less significant with a flux of 1.65 ± 0.31 ng m⁻² s⁻¹. Furthermore, higher fluxes have been estimated for MACR + MVK compounds in our study (0.91 ± 0.042 ng m⁻² s⁻¹).

Our estimation fluxes for P. abies are higher for isoprene flux than the estimation from bark fluxes (Jaakkola et al., 2023). However, they estimate fluxes eight times higher for monoterpene fluxes and ∼3 times for sesquiterpene fluxes (8.06 ± 14.17 ng m⁻² s⁻¹ and 0.58 ± 0.89 ng m⁻² s⁻¹, respectively). The flux of monoterpenes under stress conditions during infested periods increases by 6,500%–22,400%. Spruce fluxes are important due to their contribution to the SOA formation in the atmosphere (Furnell et al., 2023).

Processes of emissions at the leave or at the bark are different and imply that the emissions are not similar. Furthermore, the measurement conditions are not similar and are individual tree-dependent.

Although our measurements are in line and complement those previously conducted in other studies, they require additional replicates. Moreover, additional measurements without inducing additional stress (such as cut branches) are necessary.

In this section, we will look at the speciation of the VOC according to the tree species studied and their ER. First, Figure 6 shows the distribution of VOCs according to the three species measured in the chamber in 2020. Two rather identical profiles can be seen: P. abies and C. avellana. These two species have a higher proportion of limonene, α-pinene, and camphene. Beech tree (F. sylvatica) has a very different profile. A majority of sabinene and isoprene are measured for this species (920 ± 377 pptv and 115 ± 47 pptv, respectively). These profiles are different from what was measured in the forest environment where the dominant compound was isoprene (Supplementary Figure S2). Some external parameters, such as solar radiation, are not well-reproduced in the chamber as we use normal light inside the laboratory during the experiments (PAR = ∼50 μmol m⁻² s⁻¹). This could explain the difference in the proportion between the measurements in the chamber and real conditions. In these cases, the isoprene and monoterpene concentrations are underestimated compared to its real emissions by a ratio of ∼20 (Zeng et al., 2023).

Discrepancies in the concentrations of isoprene, total of monoterpenes, and the sum of sesquiterpenes between the 2020 and 2021 measurements are illustrated in Supplementary Figure S3. We can see a range of 74% to 15,000% for isoprene, sum of monoterpenes, and sesquiterpenes. These variations are associated with distinct environmental conditions during the measurements.

Second, Table 4 shows the ER for isoprene and the sum of monoterpenes. The ER for the second series of measurements was calculated only for isoprene, sum of monoterpenes, and sum of sesquiterpenes as we did not perform a sorbent tube blank for individual monoterpenes (Supplementary Appendix SA). All ERs are summarized in Supplementary Table S4. According to our measurements, a higher ER is observed for the sum of terpenes. Isoprene is not the most emitted compound in the chamber. Many terpenes were measured in the speciation principle of BVOCs emitted by the selected species.

We would like to remind that the ER calculation results from concentrations measured in the chamber during blank and species measurements. Then, ERs are dependent on the species and can vary significantly and may be attributed, as suggested by previous studies (Niinemets et al., 2010; van Meeningen et al., 2017, and references therein), to differences in the species’ environment, acclimation, or both local climate and genetic diversity that can impact observed emission patterns. This can explain the differences encountered between our measurements and the literature values.

Isoprene emissions are strongly associated with temperature and light (Zeng et al., 2023). According to Figure 8, predominantly isoprene emitter species are found around Maïdo. This explains the observed concentrations and diurnal variations reported in previous studies at the Maïdo observatory (Rocco et al., 2020; Verreyken et al., 2021; Rocco et al., 2022). The concentrations presented in this figure represent values measured in the chamber after subtracting the chamber blank and are resumed in Supplementary Table S5. Therefore, only two species show isoprene emission (E. reunionensis or Syzygium jambos). The distribution of concentrations at Plaine des Fougères Forest (PLF) for the species A. heterophylla reveals higher sesquiterpene emissions compared to OPAR-Maïdo (242 ± 121 pptv and 62 ± 22 pptv, respectively). The differences observed in emissions from the same species could be explained by the local environment of this species. Indeed, emissions from trees are linked to their environment and could play a great role in the emissions (Staudt et al., 2003; Staudt and Visnadi, 2023). Furthermore, temperature over the measurements ranged between 20°C and 26°C (Section 2.3.2) with high relative humidity inside the chamber (up to 89%). These recorded differences in temperature and relative humidity could influence the measured emissions.

Species E. reunionensis is a full isoprene emitter with concentration emitted by this species being 76 ± 14 pptv at OPAR and 87 ± 9 pptv at PLF. No isoprene emission is recorded for B. penduliflora at Mare-Longue forest (MLF) compared to PLF (43 ± 6 pptv). ERs have been estimated and are under the LOD.

The exotic species Ulex europaeus shows a concentration of 173 ± 24 pptv. This species was also measured during the sampling at the Puy de Dôme and showed high isoprene emission (Supplementary Figure S7) but less diversity in emission of monoterpenes at the Puy de Dôme. The emissions for this exotic species are well-documented for its properties of bioherbicide (Pardo-Muras et al., 2018; Pardo-Muras et al., 2019). We calculated ER for tropical species in Supplementary Table S6. ERs are very low for all species. Our result is in line with Boissard et al. (2001) and Owen et al. (2001). They showed high ER of isoprene (range of 0.004–20 µg g_DW ⁻¹ h⁻¹ compared to 0.004 µg g_DW ⁻¹ h⁻¹ in our study).

Psidium cattleyanum is a bioindicator species of atmospheric pollutants, notably NO_x and SO₂ (Kateivas et al., 2018; Kateivas et al., 2022; Agrawal et al., 2023). In plants exposed to NO_x and SO₂, inhibited growth was observed in terms of both height and size, along with a reduction in the quantity of leaves. Additionally, symptoms of chlorosis and necrosis were evident.

In the literature, no isoprene ERs were found for P. cattleyanum but monoterpenes and sesquiterpenes (Llusià et al., 2010) that are in line with our measurements in MLF. ERs of monoterpenes and sesquiterpenes were, respectively, 0.057 and 0.022 µg g_DW ⁻¹ h⁻¹ in MLF.

Then, exotic species S. jambos shows only the concentration of sesquiterpenes (6 ± 8 pptv). Despite this measurement, the literature shows the emission of isoprene (0.11–10.81 µg g_DW ⁻¹ h⁻¹ in Zeng et al., 2023; 199 µg g_DW ⁻¹ h⁻¹ in Klinger et al., 2002) and monoterpenes (0.29 ± 0.05 µg g_DW ⁻¹ h⁻¹ in Leung et al., 2010).

Average concentrations of monoterpenes for L. calophylloides (n = 3) and S. cymosum (n = 3) are initially depicted in Supplementary Figure S8, prior to the exclusion of blank chamber concentrations. Subsequent to subtracting the measured concentrations of both species by the blank (as illustrated in Figure 9), it was observed that the concentrations of the sum of monoterpenes in the blank exceeded those in the species measurements. Consequently, we have opted not to report these concentrations any further. Higher emission can be shown of sesquiterpenes for both species (19 ± 21 and 14 ± 12 pptv for L. calophylloides and S. cymosum, respectively).

Considering only the positive monoterpene concentrations, we estimated monoterpene ERs of 0.023 ± 0.015 µg g_DW ⁻¹ h⁻¹ for L. calophylloides and 0.003 ± 0.001 µg g_DW ⁻¹ h⁻¹ for S. cymosum. Estimated sesquiterpene ERs are 0.003 ± 0.001 µg g_DW ⁻¹ h⁻¹ for L. calophylloides and 0.004 ± 0.004 µg g_DW ⁻¹ h⁻¹ for S. cymosum. Finally, isoprene ERs are very low with ER < 0.001 µg g_DW ⁻¹ h⁻¹ and 0.001 ± 0.001 µg g_DW ⁻¹ h⁻¹ for L. calophylloides and S. cymosum, respectively.