Characterization of Volatile Organic Compound Emissions and CO₂ Uptake from Eco-roof Plants

Abstract

Vegetation plays an important role in biosphere-atmosphere exchange, including emission of biogenic volatile organic compounds (BVOCs) that influence the formation of secondary pollutants. Gaps exist in our knowledge of BVOC emissions from succulent plants, which are often selected for urban greening on building roofs and walls. In this study, we characterize the CO₂ uptake and BVOC emission of eight succulents and one moss using proton transfer reaction – time of flight – mass spectrometry in controlled laboratory experiments. CO₂ uptake ranged 0 to 0.16 μmol [g DW (leaf dry weight)]⁻¹ s⁻¹ and net BVOC emission ranges −0.10 to 3.11 μg [g DW]⁻¹ h⁻¹. Specific BVOCs emitted or removed varied across plants studied; methanol was the dominant BVOC emitted, and acetaldehyde had the largest removal. Isoprene and monoterpene emissions of studied plants were generally low compared to other urban trees and shrubs, ranging 0 to 0.092 μg [g DW]⁻¹ h⁻¹ and 0 to 0.44 μg [g DW]⁻¹ h⁻¹, respectively. Calculated ozone formation potentials (OFP) of the succulents and moss range 4×10⁻⁷ − 4×10⁻⁴ g O₃ [g DW]⁻¹ d⁻¹. Results of this study can inform selection of plants used in urban greening. For example, on a per leaf mass basis, Phedimus takesimensis and Crassula ovata have OFP lower than many plants presently classified as low OFP and may be promising candidates for greening in urban areas with ozone exceedances.

2. Introduction

Many cities are embracing green infrastructure (e.g., green streets and alleys, green parking lots, green roofs, and green walls), and as the landscape changes from grey to green, more research is necessary to understand the effects of this change [1]. One often-claimed benefit of urban greening is the amelioration of air pollution [2–8]. However, the impact of vegetation on urban air pollution is complex, with bi-directional exchange of pollutants varying with environmental conditions, concentrations of air pollutants, plant growth, and plant phenology that require further study [9–13].

One approach used to introduce vegetation in urban settings is the installation of eco-roofs (also referred to as green roofs) and green walls. Prior studies explore the potential for these approaches to increase deposition of ambient air pollution compared to the “grey” urban hardscape. For example, a study of green roofs in Chicago estimated removal by deposition of 85 kg ha⁻¹ y⁻¹ of O₃, NO₂, PM₁₀, and SO₂ [14]. Studies have also estimated the life cycle impact of green roofs on air pollution, including the impact of manufacture of additional rooftop materials, finding that air pollution due to polymer production in the roof is offset by deposition to the roof over a period of 13–32 years [15]. Prior studies have also modeled the impact of green roofs on local ozone and CO₂ concentrations, estimating reductions in near-roof concentrations on the order of 1–2% [16,17]. At the city-scale, reductions in urban air pollution concentrations on the order of 1% or less are expected with tree canopy coverage of ~40% [11]. Due to higher air residence times and increased deposition, low-profile green surfaces may be more effective when placed in street canyons; one modeling study estimated that NO₂ and PM₁₀ may be reduced by 7–30% over the period of a year [2].

While ambient air pollutants may deposit to vegetated surfaces, plants also emit a variety of biogenic volatile organic compounds (BVOCs), including reactive species such as isoprene and monoterpenes [18,19]. Those BVOCs are involved in biosphere-atmosphere interactions such as ozone and secondary organic aerosol (SOA) formation [20–23]. A review of BVOCs emitted by urban trees shows the choice of tree species impacts urban air quality [24]. Prior studies characterize BVOC emissions from trees and plants in urban and other environments [25,26], often with a focus on isoprene. Broader characterization of the BVOC emission profile for plants used in urban greening will improve evaluation the air quality impacts of specific plant selections [27].

Succulent species are low-maintenance plants that are often selected for green roof, green wall, and other urban greening projects. Plant selection for green infrastructure projects, and especially green-roofs, generally focuses less on air quality impacts and more on other design goals such as enhancing biodiversity [28], stormwater management [29], and increasingly multiple benefits at once [30]. To our knowledge only one other study has considered emissions of BVOCs by plants commonly selected for green roofs [31]. This study characterized emissions from primarily herbaceous plants and shrubs presumed to be commonly found on green roofs; fifteen different species were selected from nine different plant families. Overall emissions from these species were found to be low according to the classifications assigned by Benjamin et al. [32], with low ozone forming potential. Despite this valuable reference, more information is needed about succulent species which tolerate extreme conditions and are therefore frequently selected as low-maintenance plants for green roofs and green walls [33].

Quantification of BVOC emissions from succulent species also contributes to our theoretical understanding of crassulacean acid metabolism (CAM) [34] via which plants minimize stomatal conductance and water loss under drought stress by utilizing carbon fixed under more favorable conditions and stored as malate. Though it is suspected that, for example, isoprene emissions from CAM species would be modest compared to other plants, it is acknowledged that work on this topic is limited [35,36]. Even less is known about other secondary metabolites produced by CAM species, though drought stress in C3 plants, i.e., plants that fix carbon during the day, is recognized to enhance production of other reduced compounds, such as phenols and alkaloids, in addition to isoprenoids [37–39]. Many succulent species planted on green roofs, especially those in the family Crassulaceae, utilize CAM metabolism to varying degrees [40]; facultative CAM species can switch from CAM to C3 photosynthesis in response to environmental conditions [41,42]. For example, Bacheraeau et al. [43] found that S. album stopped CAM cycling under UV-stress and increased production of phenols.

Emissions from plants are important drivers of urban air quality and quantification of BVOC emissions from a variety of urban plants are needed to guide both the design of green surfaces and inform modeling of air quality in cities [44]. Here we aim to characterize BVOC emissions and ozone formation potentials (OFPs) for a range of succulent species and one moss commonly selected or present as vegetation on green roofs and green walls. These data contribute to inventories of BVOC emissions and inform the selection of vegetation in exterior building design.

3. Materials and Methods

BVOC emissions from plants were measured under controlled conditions (~25 °C, ~1000 PAR, and ~30–50 % RH) occurring within a consistent time range of each day (between 10:00 AM and 1:00 PM, local time) allowing for comparison with other similar measurements from the literature.

3.1. Plant material and preparation

Experiments were performed on four replicates of eight succulents and one moss species on December 14–17, 2020. Seven of the eight succulent species were selected from five different clades in the Crassulaceae family [45–47]: Sedum album (ALB, Sempervivum clade); Sedum cauticolum (CAU, Telephium clade); Crassula ovata (CRA, Crassula clade), Sedum dasyphyllum (DAS, Leucosedum clade); Sedum oreganum (ORE, Leucosedum clade); Sedum sexangulare (SEX, Acre clade); and Phedimus takesimensis (TAK, Telephium clade). A review of CAM studies showed these species are facultative CAM plants [48]; we assume this is the case for our studied species. Delosperma cooperi (DEL) was selected from the Azoiaceae family and other plants in this genus are listed obligate CAM plants [48]. The moss species Polytrichum juniperinum (POL) was also studied in four replicates and was used as our model C3 photosynthesis. Plants were purchased or donated from local nurseries in August 2020; roots were washed free of all organic matter content, and then plants were transplanted into either 200 mL (ALB, ORE, SEX, POL and TAK) or 350mL (CAU, CRA, DAS) of green roof substrate (Phillips Soil Products, Canby Oregon) in food grade plastic containers based on their growth habit and size. Photos of the plants in the chambers are provided in Fig. 1. Note that all replicates of S. cauticolum were in late stage of flowering while the other plants were in the growth stage.

Figure 1.

Photos of the plants in the chambers during snapshot experiments.

Prior to and after chamber testing, the plants were maintained in a greenhouse with environmental conditions over four days following the experiment (avg. ± std. error): maximum temperature = 24.0 ± 1.0 °C, minimum temperature = 15.1 ± 0.1°C, maximum relative humidity (RH) = 80.7 ± 1.7%, minimum RH = 47.6 ± 5.0%, and maximum photosynthetic active radiation = 78 ± 37 lux (PPFD = 1.4 ± 0.7 μmol/m²/s). Outdoor weather was similar every day of the week of experiments. Plants were watered to full capacity approximately every four days and last irrigated on December 10, 2020. At the end of the experiment, average substrate volumetric water content (VWC) ranged from 37–43% for ALB, ORE, SEX, and TAK; the range was higher for CAU, CRA, and DAS (44–62%). Here, VWC is reported as sample water volume divided by the volume of substrate the plants were initially planted in. The water volume in each sample was determined as the difference between substrate wet weight and dry weight, not including the weight of above ground plant material. A small number of individual aphids were observed on some plants while in the greenhouse; they were systematically removed by hand from the plants (plant vitality is shown in a representative image of each plant prior to experiments in Figure 1).

3.2. Experimental setup

Eight custom chambers (material: poly(methyl methacrylate), dimensions: 19 cm height by 14.6 cm diameter) were used as enclosures for uptake and emission testing. Each chamber has a removable flanged top, sealable with screws and nuts, with the flange subsequently wrapped (Parafilm^® M) to reduce leakage. A pump (Welch, Model WOB-L 2511B-01), operating at 10 L/min, pushed laboratory air through an activated carbon filter, then the flow was directed to the inlet of each chamber. Needle valves were installed upstream each chamber allowing for flow adjustment to 1.20 ± 0.06 L/min in each chamber. Chambers were slightly positively pressurized, and leakage was determined from inlet and outlet measurements with a primary flow calibrator (Gilian, Gilibrator 2) and were less than 10% of the inlet flow across all chambers. Fittings and needle valves were stainless steel (Swagelok^®) and perfluoroalkyl (PFA) 0.635 cm outer diameter tubing was used for the sampling lines.

The eight chambers were housed within a larger insulated, stainless-steel interior enclosure which maintained environmental conditions for the duration of testing. Two LED lamps were installed in the enclosure (Advanced Platinum Series P300 and P450, respectively at 24 and 30 cm from the top of the chambers). The photosynthetic photon flux density (PPFD) was measured (Onset, S-LIA-M003) inside each chamber, where the plants were installed, and the value was 1046 ± 161 μmol/m²/s. Temperature and relative humidity were measured continuously (Onset, S-THB-M002) in the enclosure, corresponding to the inlet air condition of each plant chamber.

During experiments, inlet air temperature was maintained at 25.5 ± 0.3 °C using an air conditioner (McLean thermal, 29-4121-10) installed in the enclosure. Inlet relative humidity was not directly controlled but was stable due to the surrounding laboratory and measured 37 ± 2 % across experiments. Outlet relative humidity was calculated from the measurements from the CO2/H2O analyzer (LI-840, Licor Inc., USA) and was 53 ± 4 % across experiments and plants. Outlet airflow from the chambers was connected to an eight-port flow-through actuator and selector (model EUTB-2VLSF8MWE2, Valco Instruments, USA) that rotated the actively sampled outlet line through all chambers. The eight-port selector was controlled via a MATLAB script. Note that when not being actively sampled, each chamber’s airflow rate was maintained due to the flow-through sampling valve that allowed continuous, independent flow through each chamber. Each chamber was analyzed for three minutes during testing. BVOC and CO₂ concentrations were measured in-line via the common port from the selector. (Fig. 2.A). Each set of experiments started and ended with the analysis of one empty chamber allowing for background measurements. Two chambers contained substrate alone and the remaining five chambers had one replicate of each species. The sampling sequence across the eight chamber is shown in Fig. 2.B. Data collected during the first and last 30 seconds from each 3-minute measurements were removed from analysis to account for potential cross-contamination in the short distance of perfluoroalkyl (PFA) and polyetheretherketone (PEEK) tubing (<0.5 m) that connected the selector to the PTR-ToF-MS and CO₂ analyzer.

Figure 2.

A. Schematic of the experimental setup. B. Example of the timeline of one set of experiments.

3.3. BVOCs, CO₂, and H₂O measurements.

BVOCs were measured via proton transfer reaction – time of flight – mass spectrometry (PTR-ToF1000, Ionicon Analytik GmbH, Innsbruck, Austria) with H₃O⁺ as reagent ion. The operating principle of the PTR-ToF-MS has been well-described previously [49–51]. A description of operating conditions is available in the Appendix. A CO₂/H₂O analyzer (LI-840, Licor Inc., USA) was placed in-line with the PTR-ToF-MS measuring the common outlet from the selector. A span check for CO₂ and H₂O was performed prior to the experiments. Organic matter was processed to determine the dry weight (DW) and leaf area of above ground biomass, further described in the Appendix.

3.4. BVOCs identification and quantification.

A peak auto-search was performed using PTR-MS Viewer 3.2.8 and thirty-seven ions were detected above a threshold of 0.5 cps in the range of m/z 21 to 490. Chemical formulas were attributed to masses by the software using high-mass-resolution data. Twenty-seven ions were kept for further calculations after removing ions from operating conditions (H₃O⁺, NO⁺, O2⁺, (H₂O)H₃O⁺, C₆H₄I₂ fragment and their isotopologues). The software was also used to pre-process the raw data (in counts per second) into corrected data (corrected counts per second, corrected for transmission efficiency of the mass filter). Peak areas were then converted into concentrations (ppb) using the primary ion H₃O⁺ measured by its ¹⁸O isotopologue. A rate coefficient of 2×10⁻⁹ cm³/s in the formula determined by Lindinger and Jordan [50] is used for compounds for which rate coefficients are not available in the literature [52,53]. The PTR-MS was calibrated using a multicomponent mixture containing a mixture of VOCs (16 components) at 2 ppmv in N₂ (Airgas, Plumsteadville, PA, USA) certified to ± 5% accuracy. This mixture allows calculation of a transmission efficiency curve and a calibration factor for isoprene (m/z = 69.070, (C₅H₈)H⁺). It has to be noted that m/z = 69.070 could also be attributed to a pentenol fragment or a monoterpenes fragment. A correlation test was performed on the BVOC dataset including all the species and the replicates. The results showed a high correlation (r² = 0.99498) between m/z = 69.070 and m/z = 41.040 (C₃H₄)H⁺ a well-known fragment of isoprene [54] as well as a correlation (r² = 0.97287) between m/z = 69.070 and m/z = 87.081 (C₅H₁₀O)H⁺ which could be putatively attributed to pentenol. No correlation was found between m/z = 69.070 and m/z = 137.133 (monoterpenes) (r² = 0.068), implying that m/z = 69.070 unlikely corresponds to a monoterpenes fragment. This correlation implies the signal m/z = 69.070 is associated with isoprene, though we acknowledge that there may be a contribution from a pentenol fragment that we are not able to quantify. Monoterpenes are known to be fragmented across m/z = 137.133 and m/z = 81.070 under similar ionization conditions to that used here with minimal interferences [55,56], leading us to sum the concentration from those two ions and report as monoterpenes. This fragmentation pattern was confirmed by injection of limonene in the same conditions as the experiments. Total quantified BVOCs concentration is defined here as the sum of the concentration of the twenty-seven ions detected and quantified above the threshold of 0.5 cps. Total quantified BVOCs net emissions is defined here as the sum of all the emission (positive) and uptake (negative) of the twenty-seven compounds.

3.5. Calculations

Emission rates (or removal rates, if negative) are calculated from a mass balance shown in equation (1):

$$\frac{dC}{dt}=\lambda C_0-\lambda C+\frac{E}{V}$$ (1)

where C is the concentration of the compound in the chamber that contains the plant (converted from mixing ratios measured as ppb or ppm for BVOCs and CO₂, respectively, to μg/m³ or mg/m³), t is time (h), $\lambda$ is the chamber air change rate (h⁻¹), ${\mathrm{C}}_0$ is the concentration of the compound in the chamber that contains the substrate alone (taken to be the background, and accounting for impact of the substrate and exposed chamber surfaces) (μg/m³ or mg/m³ respectively for BVOCs or CO₂), E is the emission rate of the compound (μg/h or mg/h respectively for BVOCs or CO₂), and V is the volume of the chamber (m³).

At steady state, equation (1) becomes:

$$E=\lambda \times \left(C-C_0\right)\times V$$ (2)

where all terms are as defined previously.

Emissions factors for BVOCs were calculated by normalizing the emission rate, E, from equation (2) by the leaf dry weight, shown in equation (3):

$$EF=\frac{E}{DW}$$ (3)

where $\mathrm{E}\mathrm{F}$ is the emission factor (μg [g DW]⁻¹ h⁻¹) and $\mathrm{D}\mathrm{W}$ is the dry weight of the leaves (or total dry weight in the case of the moss, POL) (g).

Our measurements were performed at T = 25 °C and PAR ~1000 μmol/m²/s (1046 μmol/m²/s as noted previously). The temperature differs from the standard conditions for isoprene basal emission factors; isoprene emission factors were corrected for both temperature and light following the process described in Guenther et al [57]. Ambient air temperatures and light conditions are provided in the appendix (Table A1); note that leaf temperatures were not measured and so this correction was conducted using measured air temperatures.

${\mathrm{C}\mathrm{O}}_2$ uptake normalized by leaf dry weight is calculated using equation (4):

$${CO}_{2}uptake=\frac{\left(\frac{-E}{MW}\times \frac{1000}{3600}\right)}{DW}$$ (4)

where ${\mathrm{C}\mathrm{O}}_2$ uptake is reported in μmol [g DW]⁻¹ s⁻¹, $\mathrm{E}$ is the ${\mathrm{C}\mathrm{O}}_2$ emission calculated using equation (2) (mg/h), $\mathrm{M}\mathrm{W}$ is the molecular weight of ${\mathrm{C}\mathrm{O}}_2$ (g/mol) and $\mathrm{D}\mathrm{W}$ is as defined previously. Note that the ${\mathrm{C}\mathrm{O}}_2$ emission rate, $\mathrm{E}$, has been multiplied by −1 to reflect the parameter defined as “uptake”. A positive ${\mathrm{C}\mathrm{O}}_2$ uptake means ${\mathrm{C}\mathrm{O}}_2$ is removed from air.

Ozone formation potentials (OFPs) are calculated (equation 5) following the procedure outlined by Benjamin and Winer [58], with modification to result in OFPs normalized per leaf mass to facilitate comparison between the small plants studied here and urban trees and shrubs reported previously in the literature:

$$OF{P}_{DW}=\left[E_{\text{iso}}{R}_{iso}+E_{mono}{R}_{mono}\right]\times \frac{24h}{d}\times \frac{1g}{{10}^6\mu g}$$ (5)

where $\mathrm{O}\mathrm{F}{\mathrm{P}}_{\mathrm{D}\mathrm{W}}$ is the ozone formation potential normalized by leaf dry weight (g O₃ [g DW]⁻¹ d⁻¹), ${\mathrm{E}}_{\text{iso}}$ and ${\mathrm{E}}_{\text{mono}}$ are measured emission factors of isoprene and monoterpenes, respectively (μg VOC [g DW]⁻¹ h⁻¹) and ${\mathrm{R}}_{\text{iso}}$ and ${\mathrm{R}}_{\text{mono}}$ are reactivity factors for isoprene and monoterpenes (g O₃ formed [g VOC]⁻¹), respectively.

To enable comparison to previous estimates of OFP, we assign a reactivity factor of 9.1 for isoprene and 3.8 for monoterpenes, following the rationale provided by Benjamin and Winer [58]. Note that biomass factors are provided in Table A1 of the appendix to facilitate OFP calculations on a per plant or per leaf area basis.

3.6. Statistical tests

Normality tests were performed using the Shapiro-Wilk (S-W) test on CO₂ uptake, total quantified BVOCs net emissions, isoprene emissions, and monoterpenes emissions for each plant. The CO₂ data was determined to be normal (p = 0.06) while BVOCs (total quantified BVOCs, isoprene, and monoterpenes emissions) were non-normal (p <0.0001, p = 0.04, and p = 0.0001 respectively). For CO₂ data, a one-way analysis of variance (ANOVA) followed by a pair-wise comparison with Tukey HSD post-hoc analysis was performed while two-sided permutation tests [59] with a confidence level of set at 0.05 were used for the BVOCs, isoprene, and monoterpenes data. We make statistical comparisons with four replicate measurements of BVOC emissions for each plant, in-line with prior chamber studies [60,31,61]. However, as will be discussed, observed variability in BVOC emissions indicates future experiments should use findings presented here to design studies of BVOC emissions from succulents with increased replication. The small sample sizes input to the permutation tests allowed us to run all tests as exact, where all possible combinations of measured emissions in each comparison are considered; 70 permutations were performed, which give a minimum p-value of 0.0141. We note that our study involves multiple comparisons of BVOC emissions or select BVOCs across studied plants; researchers often adjust thresholds of significance to reduce the chance of Type I error, often using the Bonferroni correction. However, such a correction may be overly conservative, and its appropriateness is a subject of debate [62]. We opt not to apply such a correction as we contend it creates an unacceptably high probability of Type II error. However, we acknowledge that the Bonferroni-corrected level for significance (0.0007) indicates the need for larger sample sizes if greater confidence in statistical comparisons across succulents is necessary. We present calculated p-values throughout the manuscript when statistical testing is discussed.

4. Results and Discussion

4.1. CO2 uptake and BVOC emission

Mean values of CO₂ uptake and total quantified BVOC net emissions for each plant are shown in Fig. 3.A and 3.B respectively. CO₂ uptake ranged from 0 to 0.16 μmol/g DW/s. C. ovata had the lowest CO₂ uptake (~ 0 μmol/g DW/s) while S. sexangulare showed the highest CO₂ uptake (0.16 ± 0.05 μmol/g DW/s). Knowing that the genus Crassula (C. ovata, CRA) is a facultative CAM plant, the zero CO₂ uptake indicates the plant is using CAM photosynthesis during our experiments, assuming the stomates are closed in the light. Baraldi et al. [31] report carbon assimilation of one S. spectabile of 6.00 ± 0.28 μmol/m²/s under CO₂ concentration, PAR, and RH levels similar to this study while temperature was 30 °C compared to 25 °C here. Using the measured leaf area (see Table A1) we calculate CO₂ uptake flux ranging 0.03 μmol/m²/s (CRA) − 31 μmol/m²/s (SEX) and an average across all species of 7.5 μmol/m²/s. Our estimate for S. cauticolum, the closest relation to S. spectabile, is 4.0 ± 0.64 μmol/m²/s, reasonably consistent with the estimate by Baraldi et al. [31]. It is worth noting that while S. spectabile is the most closely related to S. cauticolum and from the same clade, the plants are two different species, which may explain the slight differences. Other plants such as S. sexangulare belong to different clades than S. spectabile, which may explain why results are not in closer agreement.

Figure 3.

CO₂ uptake (panel A) and total quantified BVOC net emission rates (panel B) for each species averaged over 4 replicates and normalized by leaves dry weight (DW). ALB = S. album, CAU = S. cauticolum, CRA = C. ovata, DAS = S. dasyphyllum, DEL = D. cooperi, ORE = S. oreganum, SEX = S. sexangulare, TAK = P. takesimensis, POL = P. juniperinum. Data were normalized by total dry weight. Error bars correspond to the standard errors of the means of the 4 replicates. Measurements were taken in the morning (10 AM – 1 PM) at T = 25°C under PPFD ≈ 1100 μmol/m²/s. Shared letters across bars denote no statistical difference.

Differences in CO₂ uptake between species were significant (1-way ANOVA, F = [16.40], p < 0.001). Results of Tukey’s HSD Test for multiple comparisons are represented by letters on Fig. 3A. CO₂ uptake was highest for SEX and significantly different from the other species except ALB and DEL. A summary of the adjusted p-values of comparisons across all measured CO₂ uptakes reported in Table A3 in the appendix. Knowing that the genus Delosperma (D. cooperi, DEL) is an obligate CAM plant, we expected to see low CO₂ uptake and low/no BVOC emissions during the day assuming the stomates are closed in the light (daytime condition). We observe low/no BVOC emissions but some CO₂ uptake that could be explained by the fact that Delosperma is known to perform C3 photosynthesis during the juvenile period [63]. Also, obligate CAM species still use CO₂ in the daylight during the transitional phases.

Total quantified BVOC net emissions ranged −0.10 to 3.13 μg/g DW/h, shown in Figure 3B. C. ovata and P. takesimensis were the lowest emitters (−0.10 ± 0.43 μg/g DW/h and 0.01 ± 0.91 μg/g DW/h respectively). A two-sided permutation test showed statistically significant differences only between P. juniperinum and C. ovata (p = 0.0141), D. cooperi (p = 0.0423) and P. takesimensis (p=0.0141) in terms of total quantified BVOC net emissions, although the high variability of the data implies that a larger sample size would have been more suitable to make such a comparison. High variations were observed between replicates, which can be expected with biological samples, even though all replicates were held under the same conditions before and during experiments. Considering water dependent rates of stomatal conductance, variations in soil moisture in the substrate and RH conditions in each individual chamber could account for variability in emission magnitude as the water use strategy of the sample will depend on these environmental factors. Interestingly, there was no relationship between total quantified BVOC net emissions and CO₂ uptake (see Pearson’s correlation in Table A4 in the appendix).

Speciated BVOC emissions are shown in Figure 4. For the highest net BVOCs emitter (Fig. 3.B), S. album, emissions were dominated (73%) by C₄H₈O (see Fig. 4), a compound which we speculate is 2-butanone (methyl ethyl ketone - MEK), as prior studies have observed 2-butanone emissions by plants experiencing heat stress[64]. Notably, S. album showed some signs of heat stress during the experiments. We observe C₄H₈O present with lower emissions for S. dasyphyllum, S. oreganum and S. sexangulare. C. ovata had the lowest emissions, which may be of interest in selecting urban vegetation with low emissions of reactive BVOCs. The dominant positive emissions from C. ovata are attributed to acids, putatively identified as ~60% from acetic acid and associated fragment, and ~20% formic acid.

Figure 4.

Details of BVOCs emissions and uptakes for each species normalized by leaves dry weight and averaged over the 4 replicates. P. juniperinum data were normalized by total dry weight. ALB = S. album, CAU = S. cauticolum, CRA = C. ovata, DAS = S. dasyphyllum, DEL = D. cooperi, ORE = S. oreganum, SEX = S. sexangulare, TAK = P. takesimensis, POL = P. juniperinum.

Methanol was emitted by all plants studied here. Methanol is known to be emitted from the leaves of plants during growth [65,66]. This emission might also be in response to cellulose breakdown as a result of plant wounding [67]. Some plants may have experienced some minor damage during transport to the experimental chambers or as a result of the aforementioned aphids observed in the greenhouse that were, as noted previously, removed by hand prior to experiments.

Acetaldehyde was removed at the highest rate, ranging 0.07 – 1.23 μg/g DW/h across plants. Some studies of vascular plants show a source of acetaldehyde [68,69] while others report bidirectional exchange, e.g., there is a nighttime sink and a daytime source in a red oak forest [70]. Jardine et al. [71] show net uptake of acetaldehyde to forest canopies in Michigan and North Carolina during the daytime. Uptake of acetaldehyde by plants is thought to be controlled by stomata, rather than deposition to plant surfaces [72–74]. Acetaldehyde uptake implies our experiments occurred above the compensation point as described by Jardine et al. [71] and defined by Kesselmeier [75]. Acetaldehyde levels in inlet air were in the range 3.1 – 4.3 ppb, a result of low capacity of activated carbon for acetaldehyde [76]. Compensation points in forest environments have been determined as 0–3 ppb [70] and 3.4 ppb [71]. Note that compensation points of VOCs specific to succulent species studied here are not present in the literature.

4.2. Ozone formation potential of studied sedums and moss

Reactive organic compounds, including isoprene and monoterpenes, are of particular interest in this work given the motivation of understanding the potential for urban green surfaces to emit compounds that participate in urban photochemistry [77,78,23]. For instance, Li et al. [78] showed that an increase of biogenic isoprene emissions resulted in increases in ozone concentration in the urban Houston area. Isoprene and monoterpene emission factors of the studied succulents and moss are compared to vascular plant species present in the literature in Table 1 [60,31,61]. In general, the studied succulent species have low isoprene emission factors compared to vascular plants. Isoprene emission factors for studied succulents ranged from 0 to 0.09 μg/g DW/h while for some tropical plants isoprene emissions have been measured more than three orders of magnitude higher [60]. Isoprene emissions factors measured here were lower than values reported for isoprene for S.spectabile in Baraldi et al. [31]. While this study did not include S. spectabile, we measured isoprene emission from S. cauticolum, a related species, of approximately one order of magnitude lower than for S. spectabile in Baraldi et al. [31]. Monoterpene emissions from the studied succulents ranged from 0 to 0.44 μg/g DW/h. Monoterpene emissions from S. cauticolum are 0.011 μg/g DW/h, compared to 0.06 μg/g DW/h for S.spectabile by Baraldi et al. [31]. Note that the Baraldi et al. [31] study conducted measurements at the leaf scale, while we studied intact plants here. Further, the temperature as well as the time of day of when the plants were analyzed were different across the two studies; plant BVOC emissions are known to vary across a range of timescales, including diurnally [79,70,80].

Table 1.

Comparison of isoprene and monoterpenes emissions factors (per dry weight and per leaf area) between species in our study^# 363 and other species and plants from other studies.

	Isoprene (μg/g DW/h)	Monoterpenes (μg/g DW/h)	Isoprene (μg/m2/h)	Monoterpenes (μg/m2/h)
P. takesimensis (TAK)	0.00 ± 0.02	0.01 ± 0.01	0.21 ± 1.66	0.77 ± 0.58
Crassula ovata (CRA)	0.00 ± 0.00	0.00 ± 0.00	0.34 ± 0.39	−0.04 ± 0.09
S. sexangulare (SEX)	0.04 ± 0.04	0.14 ± 0.10	9.47 ± 8.60	26.76 ± 21.53
S. cauticolum (CAU)	0.03 ± 0.01	0.01 ± 0.00	1.79 ± 0.36	0.59 ± 0.12
S. dasyphyllum (DAS)	0.05 ± 0.02	0.47 ± 0.15	3.79 ± 1.24	48.50 ± 33.09
S. album (ALB)	0.06 ± 0.03	0.04 ± 0.02	2.36 ± 1.34	1.20 ± 0.34
S. oreganum (ORE)	0.09 ± 0.02	0.27 ± 0.15	6.24 ± 1.69	17.64 ± 8.97
D. cooperi (DEL)	0.09 ± 0.07	0.04 ± 0.01	0.13 ± 4.15	2.88 ± 0.73
P. juniperinum* (POL)	1.97 ± 1.17	0.00 ± 0.00	N.A.	N.A.
S. Spectabile (1)	0.17 ± 0.08	0.06 ± 0.03	N.A.	N.A.
Big-leaf mahogany (2)	0.02 – 1.93	2.20 to 14.70	7.27 – 540.40	225.06 – 1217.16
tropical & mediterranean plants (3)	0.10 to 63.20	0.13 to 52.50	14.3 – 1649.52	37.83 – 2252.25

^#Conditions in this study were 1046 ± 161 μmol/m²/s, 25.5 ± 0.3 °C and 37 ± 2 % relative humidity (RH). Isoprene emission factors were normalized to standard conditions (PAR = 1000 μmol/m²/s and leaf temperature = 30 °C) with R as a constant (8.314 J/K/mol) and the empirical coefficients α (0.0027), C_L1 (1.066), C_T1, (95,000 J/mol), C_T2 (230,000 J/mol), TS (= 303 K), and T_M (314 K) according to Guenther et al. (1993).

^*A moss analyzed with the same setup, in the same conditions, total dry weight instead of leaves dry weight.

⁽¹⁾Baraldi et al. [31], 1000 μmol/m²/s, 30 °C, 30–50 % RH,

⁽²⁾Vettikkat et al. [61]. Species: Swietenia macrophylla, PAR ranged 170 – 338 μmol/m²/s, temperature ranged 13.5 – 34.9 °C, isoprene emission factor were normalized to standard conditions, emission factor based on leaf area were calculated using Table S1

⁽³⁾Bracho-Nunez et al. [60] (values already normalized to standard conditions: PAR = 1000 μmol/m²/s and T = 30 °C), emission factors based on leaf area were calculated using data provided in Bracho-Nunez et al. [60].

Differences in isoprene and monoterpenes emissions across plants studied here were significant. P. juniperinum has the highest isoprene emission, significantly higher than isoprene emissions from all other plants tested. This result was expected as P. juniperinum was chosen here as a reference plant known to emit isoprene [79,80]. S. Dasyphyllum showed the highest emissions of monoterpenes, significantly higher than other plants with the exception of S. oreganum (p=0.2958) and S. sexangulare (p=0.0704). Interestingly, C. ovata monoterpenes emissions were the lowest (negative) and also significantly lower than those from the other plants except P. Juniperinum (p = 0.0704). Summaries of the p-values for BVOC, isoprene and monoterpene comparisons across plants are reported in Tables A5, A6 and A7 of the supporting information.

Calculated ozone formation potential for each plant studied in this work is shown in Figure 5, along with previously determined OFPs for select plants taken from Benjamin and Winer [77]. Plants selected for comparison from the literature (orange circles) are species that are i) relevant to the Northwestern United States, ii) present in local (Portland, OR, USA) street tree planting recommendations [81], or iii) are representative of high (Goldrain Tree) and low (Hollyleaf Cherry) ranges of observable OFP_DW [77].

Figure 5.

Ozone formation potentials for succulents and moss studied in this work (blue circles) and for urban trees and shrubs determined previously (orange circles). ALB = S. album, CAU = S. cauticolum, CRA = C. ovata, DAS = S. dasyphyllum, DEL = D. cooperi, ORE = S. oreganum, SEX = S. sexangulare, TAK = P. takesimensis, POL = P. juniperinum.

Figure 5 shows a general trend of increasing ozone formation potential with increasing isoprene and monoterpene emission, expected given the formulation of equation 5. We report OFP normalized per leaf dry weight rather than per plant or per horizontal projected area to enable a comparison between the small plants (succulents and moss) that are the focus of this work and larger trees and shrubs that are more commonly studied in the literature as possible urban greening species. This normalization enables a more meaningful comparison of OFP across plants that better acknowledges the substantially different impact to greening from one succulent plant vs. a tree or shrub.

Several succulents studied here (DAS, ORE) and the moss (POL) have OFP in the range of “medium-OFP” identified in Benjamin and Winer [77] while four succulents (CAU, ALB, SEX, DEL) are in-line with “low-OFP” plants. Two studied succulents, TAK and CRA, appear particularly promising as candidates for urban greening with low impact to urban ozone; these plants have OFP substantially lower than Hollyleaf Cherry, a tree representative of the lowest non-zero range of “low OFP” plants listed by Benjamin and Winer [77]. Notably, realizing a similar extent of greenery (in this case, indicated by leaf dry weight) would require a planting campaign for succulents that differs from that of an urban tree like a Hollyleaf Cherry. Low OFP succulents could complement urban greening campaigns through integration into building walls, roofs, or other engineered green structures in urban environments.

The OFPs calculated in this work consider the emissions of monoterpenes and limonene only, consistent with prior evaluations of the impact of BVOC emissions on ozone formation [77,31]. Our aim is to compare the OFP of succulent species studied here, for which there exists no emission inventories or OFP estimates, to plants presently for which OFP is currently estimated from monoterpenes and isoprene emission factors. Compounds beyond monoterpenes and isoprene engage in ozone formation chemistry, with complex dependencies subject to local conditions (e.g. VOC levels, NOx levels, temperature, other meteorological factors, etc.). For example, a study of ozone formation potential of organic compounds in Beijing [82] showed a significant contribution to ozone formation from some carbonyls, such as acetaldehyde. Interestingly, our measurements show succulents studied in this work act as sinks for a compound we putatively identify as acetaldehyde (m/z 45.033). Future work should be conducted to 1) more broadly characterize BVOC emissions of succulents, including dynamic behavior and parallel GC-MS measurement for more confident compound identification and 2) evaluate emission inventories of succulents in chemically explicit models of urban and regional ozone formation to more fully capture the impact of the observed source and sink behavior of BVOCs emitted from these plants. The present work implies these efforts may further bolster succulents as promising candidates for urban greening, as we show specific species have very low isoprene and monoterpenes emission and, further, appear to remove some compounds (e.g., acetaldehyde) that can contribute to ozone formation.

5. Conclusions

BVOC emissions from green infrastructure can play a role in urban air quality; this paper fills a gap in the literature for succulents and provides data of practical use to those selecting vegetation for urban greening. The emission of BVOCs from green infrastructure should be considered given the potential negative consequences of secondary air pollution. Green roof plants are often selected due to their tolerance of low-water and high-heat conditions, but BVOC emissions from these heat tolerant succulents, many of which utilize CAM photosynthesis, are previously not available in the literature. The emission factors and ozone formation potentials determined here support selection of succulents with reduced impact to secondary formation of urban air pollution.

Emissions of total quantified BVOCs were generally low but varied by greater than an order of magnitude across species studied here. Under the conditions studied, we found, on a per leaf mass basis, succulents studied range from “low” to “medium” OFP. In suitable climates, our work indicates specific succulents are low-OFP and may be promising candidates for greening of urban environments. For example, P. takesimensis (TAK) and C. ovata (CRA) have ozone formation potential lower than many plants presently classified as low-OFP.

Diurnal experiments allowing for the observation of CO₂ uptake and BVOCs emissions variations as a function of time were attempted but not presented here. These longer duration experiments should be repeated in future research with larger sample sizes and with careful control of chamber environmental conditions and downstream analytical impacts, e.g., control of soil moisture and chamber RH in concert with sampling line RH. Future work should also further investigate how BVOC emission and CO₂ uptake from these plants vary under changing environmental conditions, including light, temperature, environmental stressors like drought, and developmental cues like flowering or other stressors that may be present in an urban setting. Such data would enable more careful consideration of the various factors that trigger BVOC emissions by different succulent species.

Highlights.

• Succulents are stress-tolerant plans often selected for urban greening

• VOC emission from succulents are quantified using real-time mass spectrometry

• Total VOC emissions from succulents are low, ranging −0.10 – 3.11 μg/g DW/h

• Ozone formation potentials of succulents range from very low to medium

• Succulents may support urban greening with low VOC impact