the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Jan 2026
Tracing biological, human, and inorganic sources of coarse aerosols via single-particle fluorescence and optical morphology
Abstract. Coarse-mode aerosol particles influence the environment, climate, and human health in diverse ways depending on their type. While mineral dust and sea spray aerosol (SSA) dominate this size range, rarer biological particles can have outsized impacts, for example by initiating ice formation at relatively warm temperatures. Hence, accurate, type-specific characterization of coarse-mode aerosol is essential for understanding their roles in climate and the environment. Using laboratory measurements of single-particle ultraviolet light-induced fluorescence (UV-LIF) spectroscopy and morphology, we provide a new reference dataset for coarse-mode aerosols from common sources, including pollen, dust, bacteria, and microplastics. Comparisons with previously published datasets reveal consistent source-dependent fluorescence features, but also highlights similarities between biological and non-biological particles that can bias classifications based on fluorescence alone.
We present an improved machine learning-based classification algorithm that integrates fluorescence and morphology using laboratory data for training, and evaluate its performance using observations made at Zeppelin Observatory, Svalbard. We apply domain adaptation using field data to improve the identification of combustion-sourced particles, and to better distinguish dust from SSA. The new algorithm reproduces the previously published annual bioaerosol cycle, yields higher concentrations than a fluorescence-only approach, and maintains comparable correlations with biological and combustion tracers. This open-source algorithm provides a basis for quantifying bioaerosols across diverse environments, can revise bioaerosol estimates in previously analyzed observations, and can be refined as additional characterization data become available.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics. The peer-review process was guided by an independent editor, and the authors also have no other competing interests to declare.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(4379 KB)
-
Supplement
(507 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4379 KB) - Metadata XML
-
Supplement
(507 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-59', Anonymous Referee #1, 10 Mar 2026
- AC1: 'Reply on RC1', Aiden Jönsson, 11 Jun 2026
-
RC2: 'Comment on egusphere-2026-59', Anonymous Referee #2, 02 Apr 2026
The study presents laboratory test to what extent single paricle measurements in airflow by using the Multiparameter Bioaerosol Spectrometer could enable classification and quantification of bioaerosols. The study creates extensive reference dataset of fluorescence and optical scattering properties for different aerosols and develops machine learning classification algorithm for discrimination of the following classes: pollen (i.e. Fraxinus excelsior, Alnus glutinosa, Betula pendula, Corylus avellana, Juniperus communis, Pinus sylvestris, Salix caprea), marine bacteria (mixed cultures of cyanobacteria), dust, and synthetic contaminants (e.g. crystalline cellulose, polyethylene, PSL, natural dust samples from 4 geographical regions, combustion product from ship stack).
I do not understand how chemical tracers were used in domain adaptation and how it helped testing possibility that model trained in lab can be used in real life field observation. More detailed discussion on expected performance of the MBS coupled with machine learning algorithm for identification listed bioaerosols in real life field measurements is needed.
If the purpose of the measurements are to enable automatic quantification of different airborne coarse particles then I think the methodological approach limits it notably. Sampling only 0.33 l/min (and analizing half of it) is hardly enough to representatively detect and quantify rare airborne particles such as pollen and fungal spores. If you check the aerobiological studies, atmosphare only in rare events contains more than 5000 pollen grains/m3 and more than 50000 fungal spores/m3 (most often it is even a magnitude less). This way
The authors claim pollen rupture is the reason for detecting large variety of particles shapes and sizes. It should be noted that pollen grains exine is not prone to rupturing in normal conditions. In the atmosphere it actually rarely happens and seen variability should rather be attributed either to irregular shapes of dry pollen grains or their non-homogenous structure (cytoplasm with different water and chemical content. sometimes even air trapped in exine) as it is suggested in a study that tested Rapid-E+ air flow cytometer (Sikoparija et al. 2024). In fact those authors argued that due to wrinkly and inhomogeneous pollen cytoplasm function for size estimation usually derived from measurements PSL does not fit.
I do not agree if the authors with “wet nebulized pollen show more spherical morphologies, that many particles are dried residues of dissolved constituents rather than physical fragments” claim that dry pollen is fragmented in air. Dry pollen is wrinkly which might make it appear as irregular fragment in the scattering signal, but it is still an intact particle that should be discriminated from other coarse particles of similar size.
In my opinion the presented results indicated that MBS measurements have a very limited applicative value for discriminating between tested bioaerosols. From comparison of fluorescence signals between dry and wet nebulized pollens one could see notable differences in signals (and large variability). This indicates that only change in humidity during real life sampling could change signal characteristics to the degree it confuses between different pollen classes. And possibly even between different aerosols. Although model validation results presented in Table 2 indicate models in laboratory conditions sees some differences between classes, in real life field conditions when facing numerous unseen particles the accuracy notably drops.
References
Sikoparija, B., Matavulj, P., Simovic, I., Radisic, P., Brdar, S., Minic, V., Tesendic, D., Kadantsev, E., Palamarchuk, J., and Sofiev, M.: Classification accuracy and compatibility across devices of a new Rapid-E+ flow cytometer, Atmospheric Measurement Techniques, 17, 870 5051–5070, https://doi.org/10.5194/amt-17-5051-2024, 2024.
Citation: https://doi.org/10.5194/egusphere-2026-59-RC2 - AC2: 'Reply on RC2', Aiden Jönsson, 11 Jun 2026
Peer review completion
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2026-59', Anonymous Referee #1, 10 Mar 2026
In this manuscript, the authors present a laboratory characterization study of different biological aerosol particles, dust samples, sea spray aerosol and microplastics with a single particle fluorescence spectrometer (MBS). They use this data, together with data from previously characterized particles and train a novel classification algorithm using fluorescence, but also several parameters related to the morphology of single particles. The differences and similarities between different types of particles are discussed. I especially appreciate that the algorithm is then applied to an already published data set from the Zeppelin Observatory in Svalbard and compared to the original interpretation. Additional soft labels derived from tracer measurements were used to further tune the algorithm. The code and data is/will be available online. In my opinion, the article is highly relevant and suitable for publication in ACP after some revisions.
General comments:
1) I am missing one compressed paragraph (for example towards the end of the discussion) that compares the new algorithm with Freitas’ in detail: What are the advantages? Is it superior? If not, does it need more validation in order to be? If yes, why? etc. This paragraph should clearly justify why this new algorithm is worth using.
2) One main concern lies in the definitions, interpretations and discussions regarding pollen (3.1: and 480 – 487):
Regarding the definition of SPP (Sub-pollen particles): There is no uniform definition of SPP and people use the term loosely. The term originally comes from the medical community and refers only to the solid starch granules from the inside of the pollen grains (Bacsi et al., 2006; Schleh et al., 2010; Sénéchal et al., 2015; Burkart et al., 2021). These solid starch granules are usually between 0.5 and 2.5 µm in size (e.g. Burkart et al., 2021 for birch; Suphioglu et al., 1992; Schleh et al., 2010 for grasses).
Beside soluble cytoplasmic content, these starch granules are the particles that get expelled during rupture at high RH or when immersed in water. However, at least for birch, this only works for freshly collected pollen grains, and not for purchased ones, even if you put them in water for a longer period (Burkart et al., 2021). Since your measured size distributions of the pollen samples (Fig. 1) are mostly above 2 µm, and you used purchased (not fresh) pollen, I think it is unlikely that the majority of the measured particles are starch granules. This should be clarified.
Some authors also use the term SPP to describe aerosolized and dried aqueous extracts of pollen (Gute and Abbatt, 2020; Gute et al., 2020; Mikhailov et al., 2019; Steiner et al., 2015), but these are typically in the range <500 nm. Since your measured size distribution does not fit any of those two definitions, I would suggest simply calling them pollen fragments, as done, for example by Hughes et al., 2020, or to clearly define what you are talking about. In general, I would switch out “pollen” with “pollen fragments” whenever you discuss your results, since you are not actually measuring pollen grains. This could be confusing to readers.
I would also suggest examining one pollen sample under a microscope to see what these fragments > 2 µm look like. It should be easy to identify if these are starch granules (smooth surface, round to ellipsoid in shape, compare with Fig. 2-4 in Burkart et al., 2021), or really fragments of the pollen exine. You could include microscopy images in the Supplement.
This should also be considered in your discussion in 480-487.
3) 425-445:
The discussion about pollen (fragments) needs to be even more critical. I highly doubt that 60-80 % of bioaerosols in the arctic are pollen grains, let alone pollen fragments which aerobiologists do not observe in higher concentrations than intact pollen grains in their Hirst traps (they would be colored pink after treatment with fuchsin, as the pollen grains are, and therefore easy to spot). Some fragments, however, were observed by Hughes et al., 2020 (see Fig. S5 in their Supplement).
You already discuss possible reasons for misclassification. However, this could be better structured. Therefore, I would start with a statement that such high pollen (fragment) concentrations are highly unlikely, and support this with a comparison to earlier studies that you already mention throughout the paragraph (SEM pictures and pollen counts). You state that earlier studies found low concentrations of pollen grains. What is the approximate concentration Johansen and Hafsten found, and how does this compare to your pollen (fragments) concentration?
I would then continue to explain the possible reasons why the particles classified as pollen (fragments) here are too high. You have already discussed several things. For instance, misclassification of combustion particles and fungal spores. However, the correlation of pollen and fungal spores with mannitol and arabitol are not similar, but pollen correlation is significantly higher. This should be clearly emphasized.
Parts of this paragraph already read more as a discussion than results, and lines 514-527 in the discussion section partly repeat the earlier discussion but also introduces new ideas, such how the validation could be improved (for instance through comparison with pollen and fungal spore counts). I recommend merging those parts and put all in the discussion.
4) The last thing that is unclear for me is the following: You used soft labels in order to tune the algorithm. Since prior to tuning, the results seemed unreasonable, do you expect that you need these labels and therefore tracer measurements in every future campaign, or was this sufficient as a one-time adjustment to get meaningful classes for future measurements? Please comment on this.
Specific comments:
- The Supplement could be better structured according to their appearance in the main text.
- Titel: Wouldn’t “anthropogenic” be a better word than “human”?
- 6: I suggest to add the instrument here.
- 30 – 31: Can the concentration range you give for bacteria also be found in Huang et al., 2021? If yes, make that clear, if not, find a reference supporting this range.
- 33 & 35: Since “online aerosol mass spectrometry” includes “Single-particle mass spectrometry”, I suggest rephrasing the former, or explain both under the parent phrase “online aerosol mass spectrometry”.
- 34: change “only one size range” to “only a narrow size range”.
- 60: While it may be true, that microplastics contain PAHs, they are not produced during production. The main reason for their fluorescence is either because they contain aromatic structures that are expected to fluoresce (e.g. PET, PS) (Gratzl et al., 2024), the production of trace amounts of α-β-unsaturated carbonyls during synthesis or storage (e.g. PE, PP) (Grabmayer et al., 2014; Htun and Klein, 2010), or due to additives (Morgana et al., 2024). I also suggest including Beres et al., 2024 in the references, which also used autofluorescence to detect airborne microplastics.
- 89: Add the names of the channels (A - H). I think it would also make it easier for the readers if the central wavelengths were written here.
- 95: Is the background only measured with one single light pulse?
- 124: How was it ground into particles?
- 125-132: How old were the pollen samples? How were they stored?
- 137: How was the PE ground? How was the UV-aging exactly achieved? What light source / what wavelength was chosen for the aging? How long was the exposure?
- 140-145: Are these commercial dust samples or did you collect them?
- 156: Were background measurements taken between the samples?
- Table 1: Include the size of the PSL.
- 174: here you call them objectives, but later on you refer to task 1, task 2, … Maybe stick to task.
- 215: eBC data: Since there is a minutely resolution available, why did the author only compare 24 h data, and not for example 1 h data?
- 216: You should include for what aerosol classes the tracers are for in Table S3 and give some examples here in the main text.
- 264: You could add the central emission wavelength of the B channel here.
- 265: The authors may correct me if I’m wrong, but I did not find evidence that bacteria fluorescence peaks in this channel based on the two references provided. While I agree that tryptophan, the main fluorophore responsible for bacterial fluorescence peaks in the B channel, whole cells are not discussed in these publications. I suggest to reference Kinahan et al., 2019; Savage et al., 2017 and/or Hernandez et al., 2016). They do not use the exact same excitation/emission wavelength as the MBS but measure actual cells.
- 293: What is the approximate size range of the combustion particles?
- 296: regarding discrimination: Studies found a high correlation of eBC concentrations with a WIBS channel using a different excitation wavelength than the MBS (Gratzl et al., 2025; Gao et al., 2024; Beck et al., 2024).
- 301: Why was 3 µm used?
- 310: The explanation of what PSL particles are should come in the methods, not here.
- 310-328. The dust and SSA results should also be compared to the other particle types and not just to each other (e.g. low to zero HFAP contribution).
- 4: A color legend for the different classes would be nice.
- 340: Why was 2.5 µm chosen as the threshold?
- 341: The terms “Pollution model” and “pollution-likelyhood” should appear already in Section 2.5. Same for “dust model” at line 348.
- 343, 348: What is the confidence threshold?
- 396: Since you reference to the Supplement figure here that includes “tuned” and “untuned” data, you should introduce these two terms already earlier in the section.
- 398: Refer to Fig. 7A after “biological aerosol influence”.
- 7a: fPBAP (this study) is the tuned one, right? This should be in the legend.
- 404: Which tracers were used as soft lables?
- 407: Why 20%?
- 411: Interferents in HFPs are not directly visible in Fig. 7A. You could include the interferents concentration in Fig 7A if it does not become too messy. Alternatively, explain again that interferents are HFPs minus fPBAPs.
- 412: There is no “pollution class” in Fig. 7e. Do you mean combustion? Be consistent! What does “lead” mean in lead eBC?
- 413: I would add “on fPBAP” after “reduces the influence of combustion particles”.
- 415: For what type is it validated?
- 424: State the correlation coefficients here.
- 446: Name the SSA and dust tracers.
Technical comments:
- 24: < - 20 °C (the minus sign is missing)
- 161: “of” missing: inlet OF the MBS.
- 220: “and the” is written twice.
- 240: you just refer to the SI here. Which part?
- 326: Citation should be in brackets.
- 364: Missing full stop after “classes”.
- 405: Where in the SI?
- 461: Where in the SI?
- 465: should be “repeated”.
Supplement:
- Page 8, Fig. S3: It would make more sense to add the starting wavelength of each channel instead of random 100 nm steps on the lower x-axis. Add the excitation wavelength in the caption.
- S4: a and b are switched!
- Table S3: You should include what these substances are tracers for in the table.
- S5: One bar is always orange instead of red.
References:
Bacsi, A., Choudhury, B. K., Dharajiya, N., Sur, S., & Boldogh, I. (2006). Subpollen particles: Carriers of allergenic proteins and oxidases. Journal of Allergy and Clinical Immunology, 118, 844-850. doi:https://doi.org/10.1016/j.jaci.2006.07.006
Beck, I., Moallemi, A., Heutte, B., Pernov, J. B., Bergner, N., Rolo, M., . . . Schmale, J. (May 2024). Characteristics and sources of fluorescent aerosols in the central Arctic Ocean. Elementa: Science of the Anthropocene, 12, 00125. doi:10.1525/elementa.2023.00125
Beres, N. D., Burkart, J., Graf, E., Zeder, Y., Dailey, L. A., & Weinzierl, B. (2024). Merging holography, fluorescence, and machine learning for in situ continuous characterization and classification of airborne microplastics. Atmospheric Measurement Techniques, 17, 6945–6964. doi:10.5194/amt-17-6945-2024
Burkart, J., Gratzl, J., Seifried, T. M., Bieber, P., & Grothe, H. (2021). Isolation of subpollen particles (SPPs) of birch: SPPs are potential carriers of ice nucleating macromolecules. Biogeosciences, 18, 5751–5765. doi:10.5194/bg-18-5751-2021
Gao, K., Vogel, F., Foskinis, R., Vratolis, S., Gini, M. I., Granakis, K., . . . Nenes, A. (2024). Biological and dust aerosols as sources of ice-nucleating particles in the eastern Mediterranean: source apportionment, atmospheric processing and parameterization. Atmospheric Chemistry and Physics, 24, 9939–9974. doi:10.5194/acp-24-9939-2024
Grabmayer, K., Wallner, G. M., Beißmann, S., Braun, U., Steffen, R., Nitsche, D., . . . Lang, R. W. (2014). Accelerated aging of polyethylene materials at high oxygen pressure characterized by photoluminescence spectroscopy and established aging characterization methods. Polymer Degradation and Stability, 109, 40-49. doi:https://doi.org/10.1016/j.polymdegradstab.2014.06.021
Gratzl, J., Böhmländer, A., Pätsi, S., Pogner, C.-E., Gorfer, M., Brus, D., . . . Grothe, H. (2025). Locally emitted fungal spores serve as high-temperature ice nucleating particles in the European sub-Arctic. Atmospheric Chemistry and Physics, 25, 12007–12035. doi:10.5194/acp-25-12007-2025
Gratzl, J., Seifried, T. M., Stolzenburg, D., & Grothe, H. (2024). A fluorescence approach for an online measurement technique of atmospheric microplastics. Environ. Sci.: Atmos., 4(6), 601-610. doi:10.1039/D4EA00010B
Gute, E., & Abbatt, J. P. (2020). Ice nucleating behavior of different tree pollen in the immersion mode. Atmospheric Environment, 231, 117488. doi:https://doi.org/10.1016/j.atmosenv.2020.117488
Gute, E., David, R. O., Kanji, Z. A., & Abbatt, J. P. (December 2020). Ice Nucleation Ability of Tree Pollen Altered by Atmospheric Processing. ACS Earth Space Chem., 4, 2312–2319. doi:10.1021/acsearthspacechem.0c00218
Hernandez, M., Perring, A. E., McCabe, K., Kok, G., Granger, G., & Baumgardner, D. (2016). Chamber catalogues of optical and fluorescent signatures distinguish bioaerosol classes. Atmospheric Measurement Techniques, 9, 3283–3292. doi:10.5194/amt-9-3283-2016
Htun, T., & Klein, U. K. (2010). Laser-induced fluorescence decays of polyethylene films. Journal of Luminescence, 130, 1275-1279. doi:https://doi.org/10.1016/j.jlumin.2010.02.038
Hughes, D. D., Mampage, C. B., Jones, L. M., Liu, Z., & Stone, E. A. (June 2020). Characterization of Atmospheric Pollen Fragments during Springtime Thunderstorms. Environ. Sci. Technol. Lett., 7, 409–414. doi:10.1021/acs.estlett.0c00213
Kinahan, S. M., Tezak, M. S., Siegrist, C. M., Lucero, G., Servantes, B. L., Santarpia, J. L., . . . Pan, Y.-L. (2019). Changes of fluorescence spectra and viability from aging aerosolized E. coli cells under various laboratory-controlled conditions in an advanced rotating drum. Aerosol Science and Technology, 53, 1261–1276. doi:10.1080/02786826.2019.1653446
Mikhailov, E. F., Ivanova, O. A., Nebosko, E. Y., Vlasenko, S. S., & Ryshkevich, T. I. (2019). Subpollen Particles as Atmospheric Cloud Condensation Nuclei. Izvestiya, Atmospheric and Oceanic Physics, 55, 357–364. doi:10.1134/S000143381904008X
Morgana, S., Casentini, B., Tirelli, V., Grasso, F., & Amalfitano, S. (2024). Fluorescence-based detection: A review of current and emerging techniques to unveil micro/ nanoplastics in environmental samples. TrAC Trends in Analytical Chemistry, 172, 117559. doi:https://doi.org/10.1016/j.trac.2024.117559
Savage, N. J., Krentz, C. E., Könemann, T., Han, T. T., Mainelis, G., Pöhlker, C., & Huffman, J. A. (2017). Systematic characterization and fluorescence threshold strategies for the wideband integrated bioaerosol sensor (WIBS) using size-resolved biological and interfering particles. Atmospheric Measurement Techniques, 10, 4279–4302. doi:10.5194/amt-10-4279-2017
Schleh, C., Erpenbeck, V. J., Winkler, C., Lauenstein, H. D., Nassimi, M., Braun, A., . . . Hohlfeld, J. M. (2010). Allergen particle binding by human primary bronchial epithelial cells is modulated by surfactant protein D. Respiratory Research, 11, 83. doi:10.1186/1465-9921-11-83
Sénéchal, H., Visez, N., Charpin, D., Shahali, Y., Peltre, G., Biolley, J.-P., . . . others. (2015). A review of the effects of major atmospheric pollutants on pollen grains, pollen content, and allergenicity. The Scientific World Journal, 2015, 940243.
Steiner, A. L., Brooks, S. D., Deng, C., Thornton, D. C., Pendleton, M. W., & Bryant, V. (2015). Pollen as atmospheric cloud condensation nuclei. Geophysical research letters, 42, 3596–3602.
Steiner, A. L., Brooks, S. D., Deng, C., Thornton, D. C., Pendleton, M. W., & Bryant, V. (2015). Pollen as atmospheric cloud condensation nuclei. Geophysical research letters, 42, 3596–3602.
Suphioglu, C., Singh, M. B., Taylor, P., Knox, R. B., Bellomo, R., Holmes, P., & Puy, R. (1992). Mechanism of grass-pollen-induced asthma. The Lancet, 339, 569-572. doi:https://doi.org/10.1016/0140-6736(92)90864-Y
Citation: https://doi.org/10.5194/egusphere-2026-59-RC1 - AC1: 'Reply on RC1', Aiden Jönsson, 11 Jun 2026
-
RC2: 'Comment on egusphere-2026-59', Anonymous Referee #2, 02 Apr 2026
The study presents laboratory test to what extent single paricle measurements in airflow by using the Multiparameter Bioaerosol Spectrometer could enable classification and quantification of bioaerosols. The study creates extensive reference dataset of fluorescence and optical scattering properties for different aerosols and develops machine learning classification algorithm for discrimination of the following classes: pollen (i.e. Fraxinus excelsior, Alnus glutinosa, Betula pendula, Corylus avellana, Juniperus communis, Pinus sylvestris, Salix caprea), marine bacteria (mixed cultures of cyanobacteria), dust, and synthetic contaminants (e.g. crystalline cellulose, polyethylene, PSL, natural dust samples from 4 geographical regions, combustion product from ship stack).
I do not understand how chemical tracers were used in domain adaptation and how it helped testing possibility that model trained in lab can be used in real life field observation. More detailed discussion on expected performance of the MBS coupled with machine learning algorithm for identification listed bioaerosols in real life field measurements is needed.
If the purpose of the measurements are to enable automatic quantification of different airborne coarse particles then I think the methodological approach limits it notably. Sampling only 0.33 l/min (and analizing half of it) is hardly enough to representatively detect and quantify rare airborne particles such as pollen and fungal spores. If you check the aerobiological studies, atmosphare only in rare events contains more than 5000 pollen grains/m3 and more than 50000 fungal spores/m3 (most often it is even a magnitude less). This way
The authors claim pollen rupture is the reason for detecting large variety of particles shapes and sizes. It should be noted that pollen grains exine is not prone to rupturing in normal conditions. In the atmosphere it actually rarely happens and seen variability should rather be attributed either to irregular shapes of dry pollen grains or their non-homogenous structure (cytoplasm with different water and chemical content. sometimes even air trapped in exine) as it is suggested in a study that tested Rapid-E+ air flow cytometer (Sikoparija et al. 2024). In fact those authors argued that due to wrinkly and inhomogeneous pollen cytoplasm function for size estimation usually derived from measurements PSL does not fit.
I do not agree if the authors with “wet nebulized pollen show more spherical morphologies, that many particles are dried residues of dissolved constituents rather than physical fragments” claim that dry pollen is fragmented in air. Dry pollen is wrinkly which might make it appear as irregular fragment in the scattering signal, but it is still an intact particle that should be discriminated from other coarse particles of similar size.
In my opinion the presented results indicated that MBS measurements have a very limited applicative value for discriminating between tested bioaerosols. From comparison of fluorescence signals between dry and wet nebulized pollens one could see notable differences in signals (and large variability). This indicates that only change in humidity during real life sampling could change signal characteristics to the degree it confuses between different pollen classes. And possibly even between different aerosols. Although model validation results presented in Table 2 indicate models in laboratory conditions sees some differences between classes, in real life field conditions when facing numerous unseen particles the accuracy notably drops.
References
Sikoparija, B., Matavulj, P., Simovic, I., Radisic, P., Brdar, S., Minic, V., Tesendic, D., Kadantsev, E., Palamarchuk, J., and Sofiev, M.: Classification accuracy and compatibility across devices of a new Rapid-E+ flow cytometer, Atmospheric Measurement Techniques, 17, 870 5051–5070, https://doi.org/10.5194/amt-17-5051-2024, 2024.
Citation: https://doi.org/10.5194/egusphere-2026-59-RC2 - AC2: 'Reply on RC2', Aiden Jönsson, 11 Jun 2026
Peer review completion
Journal article(s) based on this preprint
Model code and software
TRUFFLE: Trained Recognition of Unique Fluorescence- & Form-based Labels for Environmental aerosols Aiden Jönsson https://github.com/aidenrobert/truffle/
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,665 | 1,210 | 147 | 3,022 | 283 | 124 | 242 |
- HTML: 1,665
- PDF: 1,210
- XML: 147
- Total: 3,022
- Supplement: 283
- BibTeX: 124
- EndNote: 242
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Jinglan Fu
Gabriel Pereira Freitas
Ian Crawford
Pavla Dagsson-Waldhauserová
Radovan Krejci
Yutaka Tobo
Karl Espen Yttri
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4379 KB) - Metadata XML
-
Supplement
(507 KB) - BibTeX
- EndNote
- Final revised paper
In this manuscript, the authors present a laboratory characterization study of different biological aerosol particles, dust samples, sea spray aerosol and microplastics with a single particle fluorescence spectrometer (MBS). They use this data, together with data from previously characterized particles and train a novel classification algorithm using fluorescence, but also several parameters related to the morphology of single particles. The differences and similarities between different types of particles are discussed. I especially appreciate that the algorithm is then applied to an already published data set from the Zeppelin Observatory in Svalbard and compared to the original interpretation. Additional soft labels derived from tracer measurements were used to further tune the algorithm. The code and data is/will be available online. In my opinion, the article is highly relevant and suitable for publication in ACP after some revisions.
General comments:
1) I am missing one compressed paragraph (for example towards the end of the discussion) that compares the new algorithm with Freitas’ in detail: What are the advantages? Is it superior? If not, does it need more validation in order to be? If yes, why? etc. This paragraph should clearly justify why this new algorithm is worth using.
2) One main concern lies in the definitions, interpretations and discussions regarding pollen (3.1: and 480 – 487):
Regarding the definition of SPP (Sub-pollen particles): There is no uniform definition of SPP and people use the term loosely. The term originally comes from the medical community and refers only to the solid starch granules from the inside of the pollen grains (Bacsi et al., 2006; Schleh et al., 2010; Sénéchal et al., 2015; Burkart et al., 2021). These solid starch granules are usually between 0.5 and 2.5 µm in size (e.g. Burkart et al., 2021 for birch; Suphioglu et al., 1992; Schleh et al., 2010 for grasses).
Beside soluble cytoplasmic content, these starch granules are the particles that get expelled during rupture at high RH or when immersed in water. However, at least for birch, this only works for freshly collected pollen grains, and not for purchased ones, even if you put them in water for a longer period (Burkart et al., 2021). Since your measured size distributions of the pollen samples (Fig. 1) are mostly above 2 µm, and you used purchased (not fresh) pollen, I think it is unlikely that the majority of the measured particles are starch granules. This should be clarified.
Some authors also use the term SPP to describe aerosolized and dried aqueous extracts of pollen (Gute and Abbatt, 2020; Gute et al., 2020; Mikhailov et al., 2019; Steiner et al., 2015), but these are typically in the range <500 nm. Since your measured size distribution does not fit any of those two definitions, I would suggest simply calling them pollen fragments, as done, for example by Hughes et al., 2020, or to clearly define what you are talking about. In general, I would switch out “pollen” with “pollen fragments” whenever you discuss your results, since you are not actually measuring pollen grains. This could be confusing to readers.
I would also suggest examining one pollen sample under a microscope to see what these fragments > 2 µm look like. It should be easy to identify if these are starch granules (smooth surface, round to ellipsoid in shape, compare with Fig. 2-4 in Burkart et al., 2021), or really fragments of the pollen exine. You could include microscopy images in the Supplement.
This should also be considered in your discussion in 480-487.
3) 425-445:
The discussion about pollen (fragments) needs to be even more critical. I highly doubt that 60-80 % of bioaerosols in the arctic are pollen grains, let alone pollen fragments which aerobiologists do not observe in higher concentrations than intact pollen grains in their Hirst traps (they would be colored pink after treatment with fuchsin, as the pollen grains are, and therefore easy to spot). Some fragments, however, were observed by Hughes et al., 2020 (see Fig. S5 in their Supplement).
You already discuss possible reasons for misclassification. However, this could be better structured. Therefore, I would start with a statement that such high pollen (fragment) concentrations are highly unlikely, and support this with a comparison to earlier studies that you already mention throughout the paragraph (SEM pictures and pollen counts). You state that earlier studies found low concentrations of pollen grains. What is the approximate concentration Johansen and Hafsten found, and how does this compare to your pollen (fragments) concentration?
I would then continue to explain the possible reasons why the particles classified as pollen (fragments) here are too high. You have already discussed several things. For instance, misclassification of combustion particles and fungal spores. However, the correlation of pollen and fungal spores with mannitol and arabitol are not similar, but pollen correlation is significantly higher. This should be clearly emphasized.
Parts of this paragraph already read more as a discussion than results, and lines 514-527 in the discussion section partly repeat the earlier discussion but also introduces new ideas, such how the validation could be improved (for instance through comparison with pollen and fungal spore counts). I recommend merging those parts and put all in the discussion.
4) The last thing that is unclear for me is the following: You used soft labels in order to tune the algorithm. Since prior to tuning, the results seemed unreasonable, do you expect that you need these labels and therefore tracer measurements in every future campaign, or was this sufficient as a one-time adjustment to get meaningful classes for future measurements? Please comment on this.
Specific comments:
Technical comments:
Supplement:
References:
Bacsi, A., Choudhury, B. K., Dharajiya, N., Sur, S., & Boldogh, I. (2006). Subpollen particles: Carriers of allergenic proteins and oxidases. Journal of Allergy and Clinical Immunology, 118, 844-850. doi:https://doi.org/10.1016/j.jaci.2006.07.006
Beck, I., Moallemi, A., Heutte, B., Pernov, J. B., Bergner, N., Rolo, M., . . . Schmale, J. (May 2024). Characteristics and sources of fluorescent aerosols in the central Arctic Ocean. Elementa: Science of the Anthropocene, 12, 00125. doi:10.1525/elementa.2023.00125
Beres, N. D., Burkart, J., Graf, E., Zeder, Y., Dailey, L. A., & Weinzierl, B. (2024). Merging holography, fluorescence, and machine learning for in situ continuous characterization and classification of airborne microplastics. Atmospheric Measurement Techniques, 17, 6945–6964. doi:10.5194/amt-17-6945-2024
Burkart, J., Gratzl, J., Seifried, T. M., Bieber, P., & Grothe, H. (2021). Isolation of subpollen particles (SPPs) of birch: SPPs are potential carriers of ice nucleating macromolecules. Biogeosciences, 18, 5751–5765. doi:10.5194/bg-18-5751-2021
Gao, K., Vogel, F., Foskinis, R., Vratolis, S., Gini, M. I., Granakis, K., . . . Nenes, A. (2024). Biological and dust aerosols as sources of ice-nucleating particles in the eastern Mediterranean: source apportionment, atmospheric processing and parameterization. Atmospheric Chemistry and Physics, 24, 9939–9974. doi:10.5194/acp-24-9939-2024
Grabmayer, K., Wallner, G. M., Beißmann, S., Braun, U., Steffen, R., Nitsche, D., . . . Lang, R. W. (2014). Accelerated aging of polyethylene materials at high oxygen pressure characterized by photoluminescence spectroscopy and established aging characterization methods. Polymer Degradation and Stability, 109, 40-49. doi:https://doi.org/10.1016/j.polymdegradstab.2014.06.021
Gratzl, J., Böhmländer, A., Pätsi, S., Pogner, C.-E., Gorfer, M., Brus, D., . . . Grothe, H. (2025). Locally emitted fungal spores serve as high-temperature ice nucleating particles in the European sub-Arctic. Atmospheric Chemistry and Physics, 25, 12007–12035. doi:10.5194/acp-25-12007-2025
Gratzl, J., Seifried, T. M., Stolzenburg, D., & Grothe, H. (2024). A fluorescence approach for an online measurement technique of atmospheric microplastics. Environ. Sci.: Atmos., 4(6), 601-610. doi:10.1039/D4EA00010B
Gute, E., & Abbatt, J. P. (2020). Ice nucleating behavior of different tree pollen in the immersion mode. Atmospheric Environment, 231, 117488. doi:https://doi.org/10.1016/j.atmosenv.2020.117488
Gute, E., David, R. O., Kanji, Z. A., & Abbatt, J. P. (December 2020). Ice Nucleation Ability of Tree Pollen Altered by Atmospheric Processing. ACS Earth Space Chem., 4, 2312–2319. doi:10.1021/acsearthspacechem.0c00218
Hernandez, M., Perring, A. E., McCabe, K., Kok, G., Granger, G., & Baumgardner, D. (2016). Chamber catalogues of optical and fluorescent signatures distinguish bioaerosol classes. Atmospheric Measurement Techniques, 9, 3283–3292. doi:10.5194/amt-9-3283-2016
Htun, T., & Klein, U. K. (2010). Laser-induced fluorescence decays of polyethylene films. Journal of Luminescence, 130, 1275-1279. doi:https://doi.org/10.1016/j.jlumin.2010.02.038
Hughes, D. D., Mampage, C. B., Jones, L. M., Liu, Z., & Stone, E. A. (June 2020). Characterization of Atmospheric Pollen Fragments during Springtime Thunderstorms. Environ. Sci. Technol. Lett., 7, 409–414. doi:10.1021/acs.estlett.0c00213
Kinahan, S. M., Tezak, M. S., Siegrist, C. M., Lucero, G., Servantes, B. L., Santarpia, J. L., . . . Pan, Y.-L. (2019). Changes of fluorescence spectra and viability from aging aerosolized E. coli cells under various laboratory-controlled conditions in an advanced rotating drum. Aerosol Science and Technology, 53, 1261–1276. doi:10.1080/02786826.2019.1653446
Mikhailov, E. F., Ivanova, O. A., Nebosko, E. Y., Vlasenko, S. S., & Ryshkevich, T. I. (2019). Subpollen Particles as Atmospheric Cloud Condensation Nuclei. Izvestiya, Atmospheric and Oceanic Physics, 55, 357–364. doi:10.1134/S000143381904008X
Morgana, S., Casentini, B., Tirelli, V., Grasso, F., & Amalfitano, S. (2024). Fluorescence-based detection: A review of current and emerging techniques to unveil micro/ nanoplastics in environmental samples. TrAC Trends in Analytical Chemistry, 172, 117559. doi:https://doi.org/10.1016/j.trac.2024.117559
Savage, N. J., Krentz, C. E., Könemann, T., Han, T. T., Mainelis, G., Pöhlker, C., & Huffman, J. A. (2017). Systematic characterization and fluorescence threshold strategies for the wideband integrated bioaerosol sensor (WIBS) using size-resolved biological and interfering particles. Atmospheric Measurement Techniques, 10, 4279–4302. doi:10.5194/amt-10-4279-2017
Schleh, C., Erpenbeck, V. J., Winkler, C., Lauenstein, H. D., Nassimi, M., Braun, A., . . . Hohlfeld, J. M. (2010). Allergen particle binding by human primary bronchial epithelial cells is modulated by surfactant protein D. Respiratory Research, 11, 83. doi:10.1186/1465-9921-11-83
Sénéchal, H., Visez, N., Charpin, D., Shahali, Y., Peltre, G., Biolley, J.-P., . . . others. (2015). A review of the effects of major atmospheric pollutants on pollen grains, pollen content, and allergenicity. The Scientific World Journal, 2015, 940243.
Steiner, A. L., Brooks, S. D., Deng, C., Thornton, D. C., Pendleton, M. W., & Bryant, V. (2015). Pollen as atmospheric cloud condensation nuclei. Geophysical research letters, 42, 3596–3602.
Steiner, A. L., Brooks, S. D., Deng, C., Thornton, D. C., Pendleton, M. W., & Bryant, V. (2015). Pollen as atmospheric cloud condensation nuclei. Geophysical research letters, 42, 3596–3602.
Suphioglu, C., Singh, M. B., Taylor, P., Knox, R. B., Bellomo, R., Holmes, P., & Puy, R. (1992). Mechanism of grass-pollen-induced asthma. The Lancet, 339, 569-572. doi:https://doi.org/10.1016/0140-6736(92)90864-Y