Signiﬁcant emissions of dimethyl sulﬁde and monoterpenes by big-leaf mahogany trees: discovery of a missing dimethyl sulﬁde source to the atmospheric environment

Abstract. Biogenic volatile organic compounds exert a strong
influence on regional air quality and climate through their roles in the
chemical formation of ozone and fine-mode aerosol. Dimethyl sulfide (DMS),
in particular, can also impact cloud formation and the radiative budget as
it produces sulfate aerosols upon atmospheric oxidation. Recent studies have
reported DMS emissions from terrestrial sources; however, their magnitudes
have been too low to account for the observed ecosystem-scale DMS emission
fluxes. Big-leaf mahogany (Swietenia macrophylla King) is an agroforestry and natural forest tree known
for its high-quality timber and listed under the Convention on International
Trade in Endangered Species (CITES). It is widely grown in the American and
Asian environments (>2.4 million km2 collectively). Here,
we investigated emissions of monoterpenes, isoprene and DMS as well as
seasonal carbon assimilation from four big-leaf mahogany trees in their
natural outdoor environment using a dynamic branch cuvette system, high-sensitivity proton transfer reaction mass spectrometer and cavity ring-down
spectrometer. The emissions were characterized in terms of environmental
response functions such as temperature, radiation and physiological growth
phases including leaf area over the course of four seasons (summer, monsoon,
post-monsoon, winter) in 2018–2019. We discovered remarkably high emissions of
DMS (average in post-monsoon: ∼19 ng g−1 leaf dry weight h−1) relative to previous known tree DMS emissions, high monoterpenes
(average in monsoon: ∼15 µg g−1 leaf dry weight h−1, which is comparable to oak trees) and low emissions of
isoprene. Distinct linear relationships existed in the emissions of all
three BVOCs with higher emissions during the reproductive phase (monsoon and
post-monsoon seasons) and lower emissions in the vegetative phase (summer
and winter seasons) for the same amount of cumulative assimilated carbon.
Temperature and PAR dependency of the BVOC emissions enabled formulation of
a new parameterization for use in global BVOC emission models. Using the
measured seasonal emission fluxes, we provide the first estimates for the
global emissions from mahogany trees which amount to circa 210–320 Gg yr−1 for monoterpenes, 370–550 Mg yr−1 for DMS and 1700–2600 Mg yr−1 for isoprene. Finally, through the results obtained in this study,
we have been able to discover and identify mahogany as one of the missing
natural sources of ambient DMS over the Amazon rainforest as well. These new
emission findings, indication of seasonal patterns and estimates will be
useful for initiating new studies to further improve the global BVOC
terrestrial budget.


January 2019 at different humidities (∼ 40 % RH, 60 % RH and 70% RH) using a VOC standard (Apel-Riemer Environmental, Inc., Colorado, USA) by dynamic dilution with zero air at four different mixing ratios (in the range of 2-20 ppbv) for each VOC. The measured m/z ion signals in counts per second (cps) ( + ) for each VOC was converted to normalized cps (ncps) with respect to the sum of reagent H3O + ion signal ( 3 + ) and the first water cluster H3O + (H2O) signal ( 3 + ( 2 ) ) using the following normalization equation: = + × 10 6 3 + + 3 + ( 2 ) × 2 × 298.15 The normalized counts per second (ncps) thus calculated was corrected for dilution using zero air using the equation (2): These ncps corrected for dilution (ncpsg) were converted to sensitivity (ncps/ppb) by plotting it in y-axis with the introduced concentration of gas standard of each VOC in x-axis. The slope of the graph yielded the sensitivity factor for each VOC which was then used to calculate the mixing ratio (in ppb) from the measured counts per second of each VOC. The standard deviation 10 in ncpsg along with the error in the flows during calibration gives the uncertainty of each VOC measurement. The percentage instrumental uncertainty was then calculated using the root mean square propagation of individual uncertainties like the 5% accuracy error inherent in the VOC gas standard concentration, the 2σ instrumental precision error while sampling 10 ppbv of the VOC and error in the flow reproducibility (2%) of the two mass flow controllers. 15 The overall uncertainty in fluxes was calculated by propagating the error in each term in the flux calculation formula and the drift in sensitivity: where, EF is the emission flux, m out − m in is the difference in mixing ratios of the BVOC between input and output air, Q is the flow rate of air passing through the cuvette system in m 3 s -1 , V m is the molar volume of gas calculated using the cuvette 20 temperature and ideal gas law.
Following are the major steps in calculating the overall uncertainty of fluxes: Step 1: Let the error in measurement of m out and m in be s out and s in respectively. Since the percentage uncertainties associated with measurement of m out and m in are equal, we can say that Step 2. Uncertainty in measurement of BVOC of difference of input and output air from cuvette. 25 Let, y = m out − m in , s y = √s out 2 + s in 2 (2) Since we have percentage uncertainties instead of individual absolute uncertainties, s y can be written as: Further simplifying equation (3) we obtain that the maximum relative uncertainty (if m out = m in ) as: Therefore the maximum uncertainty ( if m out = m in ) is given as: 30 In the case of plant chamber experiments, m out >> m in , therefore the maximum uncertainty in difference (y)  Step 3: Now since the equation (1) contains only products and quotients to calculate the propagation of error, 5 We substitute for Eq. (5)  , s EF EF (%) = √ 8.9 2 + 4 + 1 + 1.67 2 + 3.8^2 = 13 % 20 , s EF EF (%) = √ 8.9 2 + 4 + 1 + 1.67 2 + 4.1 2 = 13 % , s EF EF (%) = √ 9.9 2 + 4 + 1 + 1.67 2 + 6.1 2 = 12 %

25
The total uncertainty in emissions flux measurements, while not being able to correct between 4 May and 4 October (which spans over 5 months including monsoon season) with new sensitivity, is less than equal to 13% for all the reported VOCs.
Thus the calculated total measurement uncertainty can be considered as the upper limit for monsoon season as well.

Isoprene measurements by Thermal Desorption-Gas Chromatography-Flame Ionization Detector (TD-GC-FID):
Isoprene was detected in output air from the branch cuvette using a gas chromatograph equipped with a flame ionization detector (GC-FID 7890B, Agilent Technologies, Santa Clara, United States) which is coupled to a thermal desorption unit 5 (CIA Advantage-HL and Unity 2, Markes International, UK) for sampling and pre-concentration. Water in the sample air was removed using a Nafion dryer which also removed the oxygenated VOCs such as alcohols, aldehydes and ketones (Badol et al., 2004;Gros et al., 2011). 1000 ml of dry sample air was then pre-concentrated at -30℃ at 20 ml min -1 on an ozone precursor trap (U-T17O3P-2S, Markes Internatioal, UK) which was then thermally desorbed by rapid heating to 325℃. The desorbed analytes were then transferred onto the GC instrument via a heated inlet (130℃) line. The GC instrument consisted of a 10 capillary column (Alumina PLOT, Al2O3/Na2SO4, 50 m x 0.32 mm, 8 µm film thickness, Agilent Technologies, Santa Clara, United States). The oven temperature was ramped from 30°C (hold for 12 min) to 200℃ at the rates of 5℃ min -1 (upto 170℃) and 15℃ min -1 (upto 200℃) for resolving the peaks.
Isoprene was resolved on Alumina PLOT column at a retention time of 37.5 min and identified based on the retention time of isoprene vapours injected into the TD-GC-FID system under identical instrument operational conditions as the sample. The 15 eluted isoprene was then detected using the FID. Unfortunately, due to the suspected transfer losses within the GC system, which could not be corrected, the data is only semi-quantitative and hence reported in arbitrary units.