the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Jun 2025
Variability in BVOC emissions and air quality impacts among urban trees in Montreal and Helsinki
Abstract. Many cities attempt to mitigate poor air quality by increasing tree canopy cover. Trees can indeed capture pollutants and reduce their dispersion, but they can also negatively impact urban air quality. For example, trees emit biogenic volatile organic compounds (BVOCs) that participate in both ozone (O3) and secondary organic aerosol (SOA) formation, yet these emissions have been little studied in urban contexts.
We sampled BVOCs from the leaves of mature urban trees using lightweight enclosures and adsorbent tubes in two cities: Montreal, Canada and Helsinki, Finland. In both cities, we targeted five common broadleaved species, comparing their standardised BVOC emission potentials 1) between parks and streets and 2) to nonurban BVOC emission potential estimates from emission databases. Finally, we calculated the potential O3 and SOA formation by urban trees at the leaf scale and upscaled to the neighbourhood.
We found that the BVOC emission potentials differed slightly between park and street trees. Compared to park trees, street tree emissions were higher in Montreal (specifically isoprene and sesquiterpenoids) and lower in Helsinki (specifically green leaf volatiles). However, the measured BVOC emission potentials generally deviated little from the emission database estimates, supporting the use of database estimates for urban trees. In addition, we found that O3 formation from urban tree BVOC emissions was dominated by isoprene, while SOA formation was also affected by lower monoterpenoid and sesquiterpenoid emissions. These findings highlight the importance of species selection and management strategies that protect trees from BVOC-inducing stresses.
- Preprint
(1260 KB) - Metadata XML
-
Supplement
(1837 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-2500', Anonymous Referee #1, 11 Jul 2025
General Comments
The research addresses the important and timely topic of biogenic volatile organic compound (BVOC) emissions from urban trees and their potential impact on air quality. The selection of two representative high-latitude cities, Montreal and Helsinki, and the novel comparison between park and street habitats, provides a valuable contribution to the field. The methodology, based on direct, in-situ measurements, is commendable and yields useful data. However, despite these strengths, the manuscript in its current form suffers from several critical flaws in experimental design, data analysis, and interpretation that undermine the reliability and generality of its conclusions. These deficiencies must be thoroughly addressed before the manuscript can be considered for publication.
Specific Comments
- The most significant flaw is the disparity in sampling strategy between the two cities. BVOC emissions were sampled twice in Montreal (June and August 2022) but only once in Helsinki (July 2022). Given the strong seasonal dynamics of BVOC emissions, this temporal mismatch makes any direct comparison between the cities scientifically unsound. For example, the higher emissions observed in Montreal in August could be due to late-season phenology or heat stress, a dynamic that was not captured in the single mid-summer measurement in Helsinki. This makes the discussion of inter-city differences highly speculative.
- While small sample sizes may have been common in early BVOC research, a sample of n=3 is statistically insufficient for a variable as notoriously heterogeneous as biogenic emissions. This limitation results in low statistical power for analyses like ANOVA and prevents generalizable conclusions. The extremely large error bars in Figures 1, 2, and 4 are a direct reflection of this high variability. The authors' own caution to interpret the results "with care" does not absolve this fundamental design weakness, which undermines the validity of the study's findings.
- The manuscript attributes differences in emissions to the binary park vs. street classification. However, these habitats represent a complex amalgam of confounding environmental factors (e.g., temperature, light, soil moisture and compaction, pollutant concentrations, human disturbance). The study itself found higher temperatures and O3 concentrations on Montreal streets but not in Helsinki, suggesting the "street effect" is not uniform and is city-dependent. The failure to systematically disentangle these factors makes the causal link between the "street environment" and emissions weak.
- The authors repeatedly invoke "stress" as a key driver of BVOC emissions, particularly in street environments. Yet, the only stress metric measured was leaf water potential, which showed that most trees were not experiencing significant drought stress. There is a critical lack of direct physiological indicators for heat stress, oxidative stress, or mechanical damage. This leaves the entire discussion of stress-induced emissions in the realm of speculation.
- The method to upscale leaf-level measurements to the neighborhood scale is overly simplistic and introduces massive uncertainty. It relies on multiple unvalidated assumptions, including a fixed LAI, approximate canopy areas derived from Voronoi polygons, and a simplified shading correction factor. Most importantly, the calculation only includes the study species, which represent a small fraction of the total canopy area in the test sites (23% in Montreal, 36% in Helsinki). The resulting neighborhood-scale emission maps (Fig. 5) are therefore of low accuracy and potentially misleading.
- A central conclusion is that the measured emission potentials show little deviation from database values, thus supporting the use of these databases for urban trees. This conclusion paradoxically diminishes the novelty and necessity of the present study. Furthermore, the data show significant deviations for some species. Instead of a blanket statement that "database estimates are generally usable," the authors should provide a deeper analysis of why these deviations occur.
- The authors state that for a portion of the samples, the incoming replacement air was not scrubbed for ozone. Although a post-hoc correction was applied based on literature values, this introduces an unquantified source of uncertainty. The accuracy of this correction, without rigorous validation for this specific experimental setup, is questionable and could have systematically biased the measurements of highly reactive terpenes.
- The manuscript uses fixed MIR and FAC values to calculate ozone and SOA formation potentials. However, these coefficients are highly sensitive to the ambient chemical regime, especially NOx concentrations. The high-NOx environment typical of urban atmospheres can significantly alter BVOC oxidation pathways and product yields.
- The study is located in two cities with similar cool, humid continental climates. The observed patterns, such as higher street emissions in Montreal versus higher park emissions in Helsinki, are likely not transferable to cities in other climatic zones (e.g., Mediterranean, arid, or tropical) where the dominant environmental stressors are entirely different. The authors must be more explicit about these geographical and climatic limitations in their discussion and conclusions.
- The manuscript's narrative vacillates between two somewhat contradictory messages: 1) that urban environments have complex, city-specific effects on BVOC emissions, and 2) that existing non-urban databases are generally adequate for urban trees. The authors must clarify what their single most important and robust scientific finding is and rebuild the manuscript's narrative to unequivocally support it.
Technical Corrections
- In the graphical abstract, the '2.5' in PM2.5 should be a subscript.
- The text relies heavily on species abbreviations (QM, PC, AP, etc.). A list or table of abbreviations should be provided at the beginning of the manuscript to aid readability.
Citation: https://doi.org/10.5194/egusphere-2025-2500-RC1 -
RC2: 'Comment on egusphere-2025-2500', Anonymous Referee #2, 19 Jul 2025
General
The study by Rissanen et al. investigated the BVOC emission potentials of urban trees (growing in parks vs. streets) and compared them with estimates of BVOC potentials of non-urban trees from emission databases. The overall topic fits well within the scope of ACP and is highly relevant, as there is still a lack of knowledge regarding the impact of urban environmental conditions on BVOC fluxes. In general, the manuscript is well structured and written, and the study's aims and hypotheses are clearly defined. The literature cited is timely and reflects the current state of knowledge. However, I have some major concerns regarding the methodology used to evaluate the stress status of trees and the conclusions drawn from the study.
Major comments
- The authors argue in the introduction, that urban trees are expected to release higher amounts of BVOCs compared to non-urban trees (L. 103 ff.), due to environmental stressors such as drought, heat, high ozone levels, mechanical damage (L. 80 ff.). However, with the exception of leaf water potentials, which did not indicate any water stress (L. 195), no physiological parameters were analyzed, which, in my opinion, would have been necessary, to evaluate the hypothesis.
- The conclusions drawn from the study are somewhat inconsistent: On the one hand, significant differences in BVOC emission potentials between trees growing in parks and streets were found for some species and compounds; on the other hand, it was concluded, that there was no deviation in the BVOC emission potentials between urban and non-urban trees. Clearly, the low number of replicates (n = 2–3), combined with the high intraspecific variability in BVOC emissions and the variations in BVOC potentials across the two measurement periods in Montreal, makes the results difficult to interpret. However, given the differences in street vs. park trees for some species and compounds, I wonder if the authors have considered to analyze the difference between emission potentials of BVOCs measured and the reference values obtained from databases separately for each site type, instead of taking means across street and park trees (Fig. 3).
- The authors make extensive use of the supplement (21 pages!), with the results of the statistical analysis mainly presented there instead of in the main manuscript. I would suggest including the results of the statistical analysis in the main document wherever possible (e.g. Figures 1, 2 and 3). In my opinion, this would improve the article's overall readability and strengthen the results section.
Minor comments
- L. 115 ff: Please provide a brief explanation of why these two cities were chosen. Additionally, the introduction/conclusion should state that 'common urban tree species of the northern temperate zone' were analyzed, since the chosen tree species are not representative of urban trees worldwide.
- L. 153: Was the flow rate of 0.08 L/min enough to transport transpired water out of the plant enclosures? In figure S2 it seems, that water condensed inside the bags, which may have impacted concentrations of oxygenated VOCs.
- L. 211: How stable was the incoming air? I would expect BVOC concentrations in urban environments to be highly variable, so only sampling the background once a day carries the risk of subtracting the wrong background.
- Methods S2, please revise the sentence: “The mean (SD) concentration for α-pinene without a scrubber…” There a several values given for “α-pinene without scrubber”.
Citation: https://doi.org/10.5194/egusphere-2025-2500-RC2 - The authors argue in the introduction, that urban trees are expected to release higher amounts of BVOCs compared to non-urban trees (L. 103 ff.), due to environmental stressors such as drought, heat, high ozone levels, mechanical damage (L. 80 ff.). However, with the exception of leaf water potentials, which did not indicate any water stress (L. 195), no physiological parameters were analyzed, which, in my opinion, would have been necessary, to evaluate the hypothesis.
-
AC1: 'Comment on egusphere-2025-2500', Kaisa A. Rissanen, 08 Aug 2025
Author response comment to Referee #1 (citation https://doi.org/10.5194/egusphere-2025-2500-RC1)
Thank you for your time and effort to go through the manuscript and write the constructive feedback. The comments presented points that were indeed challenges in building the statistical testing and the storyline of the manuscript, and they are very helpful to further develop the manuscript for better usability by the readers and community. We answer each specific and technical comment below, the numbered answers being the same as those in the referee comments.
Specific comments:
1. We agree that the sampling schedule between the two cities, which was caused by technical limitations, is not suitable for comparisons. Direct comparisons between the two example cities were not within our initial aims, and during the data processing, we also decided against doing any direct comparisons between the two cities. Seeing the slightly differing results between the cities, we were, however, curious to explore whether some of the differences could be connected to the variations in city size and climate. As evoked by the referee, it is possible that the differing results were also influenced by seasonality on top of the effects of the city and species selection (which also differed between the two cities), and we agree that this is a valuable point to add to the discussion on the results by city.
2. The large tree-to-tree variation was an important lesson learnt through the research process and will be taken into account in any further projects. Yet, as we sampled 6 trees per species and in total 30 trees per city – in the field – we believe that the data is still interesting to the community. The statistical tests, while low in power, give indications of the potential trends within urban areas and directions for future research. To alleviate the small sample size per species and site type, we did pool all species in the ANOVA tests, and both site types for testing against BVOC database values, so in no case did we do tests with only three trees.
3. Our aim was, most importantly, to sample trees in both common urban environments in order not to bias the results by either street or by park and to gain more representative mean values per species. We will reformulate the relevant parts of the introduction, results and discussions to better reflect this aim rather than the street-park dichotomy as needed. However, we were of course interested in exploring whether the park and street differed in BVOC emission rates, i.e. if there was any fundamental site-type effect that would override the heterogeneity within each site type (as seen, for example, for urban tree water use and water potential in Montreal, https://doi.org/10.1016/j.ufug.2025.128690). We agree that to further understand the potential street-park along with smaller spatial scale differences, more environmental data (especially soil, temperature, light environment and pollutants) would be needed in each site to disentangle these effects. For this, a study with a single urban tree species would potentially be more fruitful.
As to the potential “street effect” not being uniform across cities, we found this interesting to report as it goes partly against the common intuitive opinion that street trees must be more stressed and thus, potentially larger sources of BVOCs. The non-uniformity is also reflective of the city-to-city variation reported elsewhere (see e.g. https://doi.org/10.1016/j.envpol.2013.04.003). We think that reporting this result can encourage further research in other cities to potentially understand which city characteristics (climate, infrastructure..) may explain the BVOC emission patterns in different urban environments.
To extend our data analysis beyond the street-park dichotomy and explore more precisely the potential characteristics within each site type that could impact the tree-to-tree variations in BVOC emissions, we performed a cross-site correlation analysis with factors of local environment that can be regarded as sources of stress. We discuss the results more in the answer to referee #1 comment 4.
4. We agree that more measurements of physiological stress indicators, such as Fv/Fm or photosynthetic capacity, would have been interesting additions to try an explain a larger portion of the tree-to-tree variation, including the observed differences between park and street environments. Of potential sources of stress, we measured WP, which, as noted also by the referee, showed no substantial drought and ambient O3 concentrations that did vary between trees and in Montreal between street and park environments. We did not have reliable quantification of invisible mechanical stressors (such as trucks hitting the tree canopy), but we did note any visual signs, including stem wounding and dry branches. In two cases, the higher emission potentials of a tree individual did not match with visual signs of stress or damage: one street PC with high emission potentials of most measured compounds both in Periods I and II did have dry shoot tips in the canopy close to the measured branch (see in supplement Figure 3 alternative version and answer to referee #2 comment 2), and one street TC with relatively high monoterpenoid emissions in Period II had had a branch pruned in earlier summer. We will add these details in the results and discussion to deepen the analysis of the potential stress factors.
Additionally, as mentioned in the answer to referee #1 comment 3, we performed a cross-site correlation analysis with factors of local environment that can be regarded as sources of stress (see in supplement Figure R1). With this analysis, we intend to bring into the analysis details on the specific growth conditions and potential sources of stress that were called upon in the referee #1 comments 3 and 4, and referee #2 comment 1.
The analysis shows that the correlations between the potential stress factors or indicators with the BVOC emission potentials seem species-specific. However, as a commonality, high sun exposure (measured as PAR) or ambient temperature often correlated with higher emission potentials of terpenoids, even though the emission potentials are standardised to a certain temperature and PAR using G97 algorithm.
Among species-specific correlations was the correlation of higher O3 concentrations with higher sesquiterpenoid emission potentials among AP in Montreal, both in Periods I and II. This is even though O3 concentrations measured in the study were generally low. Surface impermeability degree in 10-m radius around the tree (in Helsinki) or as a mean of the block (in Montreal) correlated positively with PC isoprene emission potentials that were higher on the street than park, but interestingly negatively with sesquiterpenoid and GLV emission potentials of BP in Helsinki.
BVOC emission potentials often did not correlate with WP, but higher monoterpene emission potentials among QM and AP in Montreal in Period I and higher sesquiterpene emission potentials among TE in Helsinki were related to higher (less negative) water potential. This suggests that we did not see drought-induced emissions (as there was no drought), but good water status potentially supported higher emission potentials in certain species.
Even with the addition of this analysis, we are willing to reformat the discussion part to emphasise less the potential effects of stress factors on the measured BVOCs. However, as certain VOCs such as GLVs or sesquiterpenes are generally stress-induced, we find it reasonable to assume that their high peaks may relate to a stress event even without a detailed knowledge of the type and severity of the potential stress.
5. We agree that the upscaling exercise is simple, and it is so because the introduction of a more complex models would not yield more accurate results due to the nature of the underlying data. The idea of the exercise was not to provide an accurate representation of the neighbourhood emissions, because as mentioned by the referee, it lacks the other tree species and importantly, any private trees, but to compare the effects of species emissions and the species density in the landscape. We thank the referee for noting that this may not have been worded clearly enough in the manuscript, and we will clarify the intent of the exercise in the methods, results and discussion parts to avoid misleading the reader to interpret the values as real neighbourhood scale BVOC emission potentials. We feel that even with the presented limitations, the exercise still provides an interesting insight into the importance of low BVOC emissions of trees that are present in large numbers. However, the exercise is not crucial to the conclusions of the manuscript, so if it is seen as not providing additional value to the manuscript, we are willing to submit the manuscript also without it if deemed necessary.
6. Adding more analysis on why certain trees deviate more from the mean than others is a good point, thank you. We can add to this analysis via the findings in Figure R1 (see supplement, presented in the answer to referee #1 comment 4), showing some of the specific conditions in which BVOC emission potentials can be higher than expected based on the database values. In addition, the alternative version of Figure 3 (see supplement and the answer to referee #2 comment 2) illustrates more clearly the single-tree deviations from the database estimates and can be used as a basis for deeper analysis, including the noticed potential effects of dry branches or pruning mentioned in the answer to referee #1 comment 3. We do also find that the mean emission potentials generally being close to database values is a useful result for further research, modelling efforts and urban planning, so we prefer also presenting this conclusion.
7. For all samples, the incoming replacement air going into the shoot chamber was not scrubbed for ozone (to allow for any effects of ozone on the tree leaves). However, we normally scrubbed ozone from air entering the adsorbent tubes (which was a side stream of the incoming replacement air or outgoing sample air), but in a portion of these samples, we could not use scrubber before the adsorbent tube taking air from incoming replacement air. As the theory-based and previously tested corrections (Helin et al. 2020) that we applied ended up being very small, and the terpene concentrations in the incoming air samples were in any case low in comparison to the outgoing sample air, we believe that any systematic bias left after the corrections would be a very minor effect on the final emission results.
7. We agree that taking into account the NOx regime (as well as O3 and NOx intake by leaves, and aerosol deposition on the leaves) would be crucial for modelling the realistic effects of BVOC emissions in urban atmosphere O3 or SOA concentrations. Our intent was to compare tree species and explore potential differences between site types in their potential to contribute to O3 and SOA formation, we think that it is reasonable to assume the same NOX regime for all species and sites and thus use the fixed MIR and FAC.
In further studies and as MIR and FAC (or other ways to represent a single BVOC potential to contribute to O3 or SOA formation) become available for more individual BVOCs, a more detailed representation of urban tree effects on air quality should become possible. We still had to rely on a mean MIR or FAC of the compound group for many monoterpenes and sesquiterpenes, although as seen in table S7, they can differ largely between two compounds of the same group.
9. Our study focused on the high-latitude cities as these remain somewhat underrepresented in studies on urban tree functioning and also BVOC emissions. We did address the generalisability of our results in the discussion (“the BVOC emission potential estimates do not appear to serve as a large source of error in creating urban BVOC emission budgets in cities like Montreal and Helsinki (small to mid-sized cities with cool, humid climates)”) and in the conclusions (“further direct measurements of urban tree BVOC emissions in various cities with differing sizes and climates, including during acute stress events such as high O3 concentrations, heat waves, or prolongued droughts, are needed to scope where and when the database values can be applied”), but if deemed necessary we can state this even more clearly in both cases. Especially if it would encourage further studies from the mentioned regions with higher temperatures and aridity, or larger population densities.
10. Our conclusion was indeed that 1) there seems to be large tree-to-tree variation in BVOC emissions by urban trees, a part of which seems to be related to the site conditions (street or park), but that 2) across all trees per species, the mean does not importantly deviate from database values. This means that, in our results overall, there is no consistent bias towards much higher or lower emissions despite the large tree-to-tree variability. Although at first read they may seem in opposition, those are not, in fact, contradictory or mutually excluding, and are both important. However, to avoid confusion and to relate the two conclusions to each other, we will word these conclusions more clearly in the discussion and conclusions sections – thank you for pointing out the apparent inconsistency. We will more explicitly note why, in some cases, we think using non-urban databases is generally fine in cities similar to our study cities, and why, in other cases, it may be more important to note the variability within urban environment.
Technical corrections:
1. We assume the referee means O3, where the 3 was accidentally superscript instead of subscript, and we will correct this, thank you for pointing out the typo.
2. This would indeed be very helpful; we will add a table of abbreviations.
Author response comment to Referee #2 (citation )
Thank you for your time and effort put in reviewing the manuscript and pointing out the potential sources of confusion and inconsistencies. The comments will greatly help us to make the manuscript more readable and to deepen the data analysis regarding the potential sources of stress and their effects on BVOC emissions. We answer the major and minor below, so that the answer numbers refer to the comment numbers.
Major comments:
1. Even though we discuss the potential for higher emission potentials with “urban stress” in the introduction, our study was not designed to test this as a formal hypothesis (accordingly we did not formulate such hypothesis), but rather to first test whether the urban tree BVOC emissions are in line with database values or not (regardless of reason for higher / lower emissions). We also discuss in the introduction the potential error coming from BVOC emission estimates that are not species-specific but rather at a genus level. To avoid confusion regarding the study questions, we will modify the introduction to emphasise the stress discussion less in comparison to the other potential sources of deviation.
Yet we agree that adding other measures of physiological stress indicators, such as Fv/Fm or photosynthetic capacity would have been interesting to deeper explore the data and something to consider in further studies. Based on this comment and the referee #1 comments 3 and 4, we did perform a correlation analysis to better pinpoint which potential sources of stress could correlate with higher emission potentials of BVOCs (see answer to Referee #1 comment 4 and Figure R1 in supplement). The analysis showed species-specific relations between BVOC emission potentials and O3 concentrations, high temperatures or sun exposure.
2. Indeed, our conclusion was that 1) we see large tree-to-tree variation in BVOC emissions by urban trees, a part of which seems to be related to the site (street or park) conditions, but that 2) across all trees per species, the mean does not deviate importantly from database values. Therefore, there is no consistent bias towards much higher or lower emissions despite the large tree-to-tree variability. To avoid confusion (referee #1 had the same comment) and to better relate the two conclusions to each other, we will address the seeming inconsistency and word these conclusions more clearly in the discussion and conclusions sections.
To illustrate the differences from the database values by site type, we created an alternative version of Figure 3 (see supplement). In certain cases, such as in Montreal, PC isoprene emissions in Period II, TC monoterpene emissions in Period II and TC isoprene emissions in Period I, and in Helsinki BP monoterpene emissions and UG sesquiterpenoid emissions, the deviations from the database values were different between the site types (often driven by large emissions by one tree individual). Given the number of repetitions, we are hesitant to do further statistical testing with these data without pooling the park and street, but the figure will allow discussing the effect of park vs street on the deviations from the database values (and potential cases where database values are not useful), along with the impacts of potential sources of stress that could play into the larger emission potentials of certain individual trees.
3. Thank you for the suggestion; we will do this to help the readability of the paper.
Minor comments:
L. 155: We will add in the methods that high-latitude cities in the northern temperate/boreal zones are a type of city where little research has been done on the study questions, and the two cities were chosen as examples of mid-sized cities representative of this city type. The two cities being on two separate continents also gave the study some geographical width and a larger selection of tree species used in urban areas. We will also precise that the species selected are common within the climate zone, as suggested.
L. 153: The flow rate through the shoot enclosure was always 2 L/min, providing a change of the enclosure air approximately every 3 minutes. The flow rate of 0.08 L/min was used in the side streams of incoming and outgoing air that were sampled into the adsorbent tubes. To avoid confusion in this regard, we will clarify the methods section about the flow rates.
L. 211: The incoming ambient air was sampled during every measurement (we drew adsorbent tube samples as a side stream of both the incoming and outgoing air always). Thus, the local ambient background per tree was always subtracted in calculating the emission rate (concentration in “out” sample – concentration in “in” sample). The compound-wise variability in the incoming ambient air sample can be approximated from the Table S3, where the sapling system detection limit was calculated as the “in” sample mean concentration + 3*standard deviation. We will review the methods section to ensure the clarity of this procedure and avoid confusion.
The daily empty bag sampling was done especially to control for any impurities within the sampling system. This build-up of impurities within tubing, pumps or filters is a slower process than the variation in ambient background between sites, so we regarded once a day sufficient.
Methods S2: Thank you for pointing out this inconsistency, we will correct this and the following sentence.
Data sets
BVOC emission potentials for street and park trees of five common species is Montreal (Canada) and Helsinki (Finland) K. Rissanen et al. https://doi.org/10.5281/zenodo.15379393
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 317 | 33 | 12 | 362 | 24 | 14 | 23 |
- HTML: 317
- PDF: 33
- XML: 12
- Total: 362
- Supplement: 24
- BibTeX: 14
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1