From the journal Environmental Science: Atmospheres Peer review history

01-Jun-2022

Dear Dr Zieger:

Manuscript ID: EA-ART-04-2022-000047
TITLE: Emission of Primary Bioaerosol Particles from Baltic Seawater

Thank you for your submission to Environmental Science: Atmospheres, published by the Royal Society of Chemistry. I sent your manuscript to reviewers and I have now received their reports which are copied below.

I have carefully evaluated your manuscript and the reviewers’ reports, and the reports indicate that major revisions are necessary.

Please submit a revised manuscript which addresses all of the reviewers’ comments. Further peer review of your revised manuscript may be needed. When you submit your revised manuscript please include a point by point response to the reviewers’ comments and highlight the changes you have made. Full details of the files you need to submit are listed at the end of this email.

Please submit your revised manuscript as soon as possible using this link:

*** PLEASE NOTE: This is a two-step process. After clicking on the link, you will be directed to a webpage to confirm. ***

https://mc.manuscriptcentral.com/esatmos?link_removed

(This link goes straight to your account, without the need to log on to the system. For your account security you should not share this link with others.)

Alternatively, you can login to your account (https://mc.manuscriptcentral.com/esatmos) where you will need your case-sensitive USER ID and password.

You should submit your revised manuscript as soon as possible; please note you will receive a series of automatic reminders. If your revisions will take a significant length of time, please contact me. If I do not hear from you, I may withdraw your manuscript from consideration and you will have to resubmit. Any resubmission will receive a new submission date.

The Royal Society of Chemistry requires all submitting authors to provide their ORCID iD when they submit a revised manuscript. This is quick and easy to do as part of the revised manuscript submission process. We will publish this information with the article, and you may choose to have your ORCID record updated automatically with details of the publication.

Please also encourage your co-authors to sign up for their own ORCID account and associate it with their account on our manuscript submission system. For further information see: https://www.rsc.org/journals-books-databases/journal-authors-reviewers/processes-policies/#attribution-id

Environmental Science: Atmospheres strongly encourages authors of research articles to include an ‘Author contributions’ section in their manuscript, for publication in the final article. This should appear immediately above the ‘Conflict of interest’ and ‘Acknowledgement’ sections. I strongly recommend you use CRediT (the Contributor Roles Taxonomy from CASRAI, https://casrai.org/credit/) for standardised contribution descriptions. All authors should have agreed to their individual contributions ahead of submission and these should accurately reflect contributions to the work. Please refer to our general author guidelines http://www.rsc.org/journals-books-databases/journal-authors-reviewers/author-responsibilities/ for more information.

I look forward to receiving your revised manuscript.

Yours sincerely,
Dr Lin Wang

Associate Editor, Environmental Science: Atmospheres



Environmental Science: Atmospheres is accompanied by companion journals Environmental Science: Nano, Environmental Science: Processes and Impacts, and Environmental Science: Water Research; publishing high-impact work across all aspects of environmental science and engineering. Find out more at: http://rsc.li/envsci

************

Reviewer 1

The authors present an interesting and timely study investigating the release of PBAP from the sea surface via bubble breaking which may impact cloud evolution. They achieve this via real-time measurements of chamber aerosolized seawater samples using a shipborne MBS UV-LIF spectrometer, where the single particle data is then classified using unsupervised machine learning methods. The data products generated by this classification scheme underpin the majority of the work presented in this manuscript, thus great attention and care must be paid to the application of the model. It is this aspect of the work undertaken that I have the most concern with as it forms the foundations of the subsequent analysis. I believe that while the presented results are interesting, the authors must include greater detail on why they have chosen the classification scheme employed and they also must provide stronger evidence that the derived data products are actually representative of the broad classes which they are assigned, or at a minimum caveat this section appropriately. As such, I recommend that this paper undergoes major revisions to improve this aspect of the underpinning work. This is a real shame as the bacterial composition analysis showing that two distinct populations exist and that this is reflected in the real-time MBS data, suggesting preferential aerosolization is a great result. I do strongly encourage the authors to revisit the MBS classifications using better performing methods to improve confidence in these results.
<br>
<br>
General comments
<br>
The authors have chosen to use k-means clustering on only the single particle fluorescent spectra, excluding size and shape parameters. Generally speaking, k-means clustering, while fast, performs poorly for clustering UV-LIF data (Robinson et al., 2013, Crawford et al., 2015, Ruske et al., 2018). I would like the authors to justify their choice of k-means clustering, especially since they cite papers (Ruske et al., 2017, Huffman et al., 2020) which highlight its poor performance. Can the authors also comment on why they have not chosen to use better performing schemes recommended in these publications, such as hierarchical agglomerative clustering?
<br>
<br>
The choice to exclude size, and to a lesser extent shape parameters, is odd as they have been shown to dominate variable importance (Ruske et al., 2017). Can the authors expand on why they haven’t included these parameters? Crawford et al., (2020) also demonstrated their use for disentangling particles into different classes
<br>
<br>
The authors performed k-means clustering on the absolute fluorescent spectra. Clustering performance of UV-LIF data has been shown to be highly dependent on data pre-processing prior to clustering (Crawford et al., 2015, Rusk et al., 2017-18). Typically non-fluorescent particles would be removed from the population to perform clustering on, then the remaining data would be normalised in some way, rather than using the absolute intensities. It was demonstrated that failing to do this step would generally negatively impact performance across a range of clustering algorithms. Can the authors comment on why this critical normalisation step has been omitted and expand on how this may impact their results?
<br>
<br>
I am concerned by the semi-arbitrary way in which the number of clusters has been derived. The CH index failing to provide a clear result or a selection of similarly performing results can either indicate that the clustering algorithm is not suitable to disentangle the populations, or that the population cannot be readily disentangled as it is either homogeneous or contains too many small populations which are erroneously merged as they don’t form strong centroids. My feeling is that this is a shortcoming of k-means which favours producing clusters of approximately equal size, which is unlikely to be representative of the ambient population measured here. Can the authors show the CH index result for the last 20 clusters? This may help understand what is occurring here.
<br>
<br>
Crawford et al., (2017) showed Ward clustering to produce a distinct CH index result, where the antarctic marine influenced population at Halley research station is likely to be similar to that presented in this manuscript. They showed that PBAP accounted for less than 2.5% of the fluorescent population with the remainder likely to be representative of the FSSA class discussed here, demonstrating that an appropriate clustering strategy and CH index should disentangle the unbalanced populations. I suggest that the authors apply the method outlined in Crawford et al., (2017) to their data to see if the results are clearer.
<br>
<br>
As the CH index does not cleanly resolve representative clusters, the authors have manually selected a range of clustering results and compared the returned populations to anticipated populations until a clear PBAP class was observed. I would be much happier if the end classification was based on a statistical process rather than guided by expectation. I am unaware of any other UV-LIF studies which have performed a subjective analysis to find the optimal cluster solution. This is not to say that this is invalid, but some understanding of the performance of this approach using labelled laboratory samples would improve confidence in the results presented.
<br>
<br>


Specific comments
<br>
<br>
Pg 2, ln 32: More references on UV-LIF instruments are needed here.
<br>
<br>
Pg 3, ln 3: Isn’t this generally related to the D50 transmission efficiency?
<br>
<br>
Pg 3, ln 10: More detail on the CMOS shape parameters - why weren’t more parameters used?
<br>
<br>
Pg 3, ln 31: Can the authors comment on the use of a 3γ threshold. Greater thresholds (e.g., 9γ) have been shown to be useful to reject interferent particles while retaining PBAP (Savage et al., 2017). An increased threshold may remove the less interesting particles and simplify analysis.
<br>
<br>
Pg 3, ln 42: The MBS AsymLR and WIBS AF are not directly comparable in this way, thus the value of 20 presented here as a sphericity threshold isn’t valid.
<br>
<br>
Pg 4, ln 13: Refs 43 and 44 explicitly do not use k-means, citing its poor performance. Ref 44 used Ward linkage HAC which performs much better than k-means.
<br>
<br>
Pg 4, ln 17: Please explain why size and shape parameters were not included.
<br>
<br>
Pg4, Fig.1: Panel A would be easier to interpret as box and whisker plots as direct comparisons between channels would be more apparent.
<br>
<br>
Pg4, ln 27-42: These are almost certainly the same particle type, but they have been split into 3 clusters due to k-means preference to produce equal sized clusters. The concentration of these classes will be highly dependent on small changes in the instrument baseline and this should be caveated.
<br>
<br>
Pg 4, ln 48: I advise caution here as the spectral response of such fluorophores may be highly dependent on RH and pH.
<br>
<br>
In section 3.1, it would be beneficial to the reader to introduce the laboratory samples which are used to justify the FSSA/HFSSA cluster classification before the discussion assigning the clusters to these classes, or at the very least mention that there are such samples used to guide the classification in the opening section (pg 4, ln 19).
<br>
<br>
Pg 5, ln 11: Do they use a similar excitation wavelength to the MBS? Please clarify this in the text to validate the comparison.
<br>
<br>
Pg 5, ln 91: Can you expand on how the choice of threshold may impact this result. Would a 5γ threshold remove this class entirely?
<br>
<br>
References:
<br>
<br>
Robinson et al., (2013), DOI: 10.5194/amt-6-337-2013
<br>
Crawford et al., (2015), DOI: 10.5194/amt-8-4979-2015
<br>
Ruske et al., (2017), DOI: 10.5194/amt-10-695-2017
<br>
Ruske et al., (2018), DOI: 10.5194/amt-11-6203-2018
<br>
Crawford et al., (2020), DOI: 10.3390/atmos11101039
<br>
Huffman et al., (2020), DOI: 10.1080/02786826.2019.1664724
<br>
Savage et al., (2017), DOI: 10.5194/amt-10-4279-2017

Reviewer 2

This manuscript reports the measurement and analysis of fluorescent sea spray using novel instrumentation and a portable sea spray generator during a field trial lasting 11 days. The manuscript is well-written with clear prose, well-designed graphics, and a cogent exposition of the experiment and data analysis. I have some minor comments regarding the generation of sea spray, which has implications for the application of the authors' conclusions to SSA generated by breaking ocean surface waves, and some of the authors remarks. These comments would no be an impediment to publication in a regular journal.

With regard to impact, Environmental Science: Atmospheres maintains high standards and this is where I see a potential problem. The flux of SSA containing bioparticles from the sea surface has important implications for ice nucleating SSA and therefore implications for the mixing state of clouds over the Arctic and great southern ocean, for example. It is stated in the abstract that "the surface flux and concentration of primary biological particles that are emitted from the seawater through bubble bursting" are determined but estimates of primary biological particle flux are not actually reported (although concentrations are). This is important because it makes it difficult to use the results of the study in GCM's, for example. Moreover, it is noted in their reference 29 (Salter et al., On the seawater temperature dependence of the sea spray aerosol generated by a continuous plunging jet, 2014) that "Second, the ratio of important factors (e.g., gravity and surface tension) in the plunging jet experiments should be as close as possible to the prototype of breaking waves, such that dynamic similarity between plunging jets and breaking waves can be achieved. Unfortunately, this is impossible with geometrically similar models because of the large number of relevant dimensionless parameters. This leads to scale effects and means that laboratory scale models cannot be directly compared with the prototype of breaking waves." The implication of this problem is that even if the flux of primary biological particles from the plunging-jet tank were reported, there is no means of translating these numbers to the wind-driven, open ocean. If is therefore difficult to see how the results of the study can be translated to a global context and also raises the question of whether or not the results would hold using a more realistic breaking wave proxy.

Given the novel nature of the measurements and the importance of open-ocean studies, I support the publication of the manuscript if the authors can forge a stronger link between their results and their application to GCM's or some other path forward demonstrating significant impact.

Minor comments.
1. P8, L31. Three microns is not small enough for the SSA to be considered film drops; the division between film and jet drops is more usually considered to occur at 1 micron.
2. It is stated that changes in the morphology of particles observed are most likely due to changes in the biogeochemical properties of the surface water. It is clear that there is a correlation between water biogeochemical properties , but correlation does not necessarily mean causation. Can you make a stronger case here?

We thank both reviewers for their constructive and helpful comments that helped to improve our manuscript. We have incorporated all comments. Based on the comments of reviewer 1, we have removed the classification of the fluorescent particles using the k-means cluster algorithm and are now using a simple but robust classification scheme (or decision tree) based on the measured fluorescence spectra (similar to the concept used by the WIBS instrument community as described in [Perring et al., 2015]). The overall results are not largely affected by this change (mainly the concentrations change) and our main findings stay the same. More details are given below. Within the PDF, our response is given in blue, here we respond to the comments directly and the figure numbering refers to the revised manuscript.

*** Response to reviewer 1 ***
We thank reviewer 1 for their very useful comments. We agree that the unsupervised learning algorithm still has too many uncertainties with regards to choosing the right number of clusters, the correct separations of the clusters and the choice of the input parameters. Unfortunately, for our work no training data set is yet available to use other machine learning algorithms (and it is also not really needed since we describe results from a controlled laboratory experiment and do not show ambient air measurements), as such we decided to replace the k-means clustering by a straight-forward decision tree classification of the fluorescent particles using the measured emission spectra (similarly as done by the WIBS instrument but using all 8 channels of the MBS). Te results stay almost unchanged, only the absolute concentrations change by about a factor of approximately 4 (a figure is included in the PDF of the reply letter). We believe that this approach is much more robust and gives more confidence in our results. We added additional information on the new classification scheme to the revised manuscript and the SI (incl. a schematic that describes the new approach), along with revised concentration, contribution and correlation numbers throughout the manuscript.

* Reply to comment 1 *
Originally, we have chosen an unsupervised algorithm (k-means clustering) as it does not require a training data set, which was not available for our study. k-means clustering can be useful tool when no training date set is available. In our work, we are dealing purely with sea spray particles (excluding other aerosol particle types) and as such we can rely on previous findings by [Santander et al., 2021] and our own laboratory study (see Fig S7 in the SI) showing the spectral behaviour of sea water, filtered sea water and bacteria isolates at our excitation and emission wavelengths. As such, no sophisticated clustering algorithm is really needed and we can use a simple decision tree as implemented in the revised version. We have added the following text, Table 1 and Figures 1, 2 and 3 (see PDF reply) to the revised manuscript:

“Furthermore, each individual channel signal from the MBS detector is subtracted by 3 times the background standard deviation (3γ). Negative values are clipped to 0. Here, CP particles with fluorescent signal (FL) greater than this threshold at one of the 8 fluorescence channels were classified as fluorescent particles (FP) and particles with fluorescence signal greater than 9γ are classified as highly fluorescent, further details follow in section 3.1”.

“Throughout the campaign a constant flux of large FP (optical diameter >0.8 μm) was observed. Not all FP should be considered as having primary biological origin since interference can come from e.g. sea spray aerosols enriched in organic matter [Pöhlker et al., 2012]. To further differentiate the classes of FP detected by the MBS, a group classification similar to [Perring et al., 2015] is performed. According to Table 1, a particle that exceeds the 9 γ fluorescence threshold at different channels, is labelled by the corresponding letters. For example, a particle that exceeds this threshold at channels 2 and 3 is grouped as a BC particle. This follows the work by [Savage et al., 2017], where they showed that a 9γ threshold is useful to distinguish PBAP from interfering low fluorescent particles. Particles that exhibit fluorescence lower than 9γ at all channels are collectively called fluorescent sea spray aerosol (FSSA). The most abundant classes of particles present during the campaign are shown in Figure 1 and their corresponding mean fluorescence signal is given in Figure S5 along with a timeline and period contribution of classes throughout the cruise in Figures S6 and S7. In Figure 1, it is shown that most of the highly fluorescent particles classes within the size range of 1-6μm come from class B particles, whereas the larger particles are dominated by particle classes that exceed the 9γ threshold in more than one channel.”

“Laboratory sea spray experiments were performed after the campaign to evaluate which groups could potentially contain PBAP. SSA produced from Baltic seawater, culture medium prepared from filtered Baltic seawater and a pure inorganic sea salt solution chamber were analyzed (see SI for a more detailed description of the experiments). No highly fluorescent groups with more than 5 particles were observed in the inorganic sea salt solution, while some were found for the culture medium and seawater (Figures S8 and S9). Strikingly, groups whose highest fluorescence signal was present in the B channel are only present in the seawater experiment, which is the only sample where PBAP are expected to be found. Other particle classes (mainly with signals in the C, D, E and F channels) were found in both the sea water and the filtered seawater, showing that those particles were probably sea spray particles coated with water soluble organics.”

“These laboratory findings are in line with results by [Santander et al., 2021] and justify our approach that highly fluorescent particle classes with a maximum signal in the B channel are classified as PBAP, and the remaining particles, as highly fluorescent sea spray aerosol (HFSSA). A visualization of this decision tree is given in Figure 2. The FSSA group generally shows very weak fluorescence in all channels (Figure 3a) since they fall below the 9γ-threshold. They also share a similar size distribution (Figure 3b) and shape parameters (Figure 3c) as the non-fluorescent CP, while comprising 10.4% of all CP and up to 98.8% of the FP. The dominance of FP population by faintly fluorescing particles has been previously observed by [Crawford et al., 2017], where 97.7% of the FP exhibited such fluorescence behaviour. The fluorescence spectra of the HFSSA resembles that of water-soluble organic compounds excited at 280nm [Duarte et al., 2004], although these fluorophores emission wavelengths can be dependant of other parameters such as pH [Duarte et al., 2004]. HFSSA is represented by large particles (mean size, 5.9 μm, see Figure 3b) which are often more asymmetrical and elongated than FSSA (mean asymmetry and peak-to-mean ratio of 30.8 and 3.1, respectively, see Figure 3c). It is probable that these particles are larger sea spray aerosol particles coated with increased amounts of fluorescent organic molecules leading to a very broad and high fluorescence emission signal with no particular defined peak and diverse morphology. [Santander et al., 2021] used excitation emission matrix spectroscopy (EEMS) to show that sea spray aerosol particles from filtered sea water retain this characteristic emission, further strengthening our classification.”

* Reply to comment 2 *
Given the lack of laboratory data, properly weighting the influence of size, shape parameters and fluorescence signals would prove a challenging task. Furthermore, fluorescence is the only property intrinsically linked to particle biochemical properties. Working with sea spray aerosol means particles (both biological and non-biological) are coated constantly with different organics present in the sea surface microlayer. Thus, different water biogeochemical properties could influence particle sizes and morphology on various different scales of time [Kaluarachchi et al., 2021]. Given the lack of controlled laboratory experiments, size and morphology of a particle containing a bacteria could vary wildly.

* Reply to comment 3 *
As described above, we have removed the clustering via k-means within this work. In the original approach, normalization was not performed to retain some size dependency within the algorithm. Given the lack of weights to integrate size and morphology into the clustering, performing the algorithm with absolute fluorescence was shown to better resolve smaller clusters containing highly fluorescent particles instead of clustering them into equally populated groups with barely fluorescent particles at the each detector channel.

* Reply to comment 4 *
We agree with your concerns, and as mentioned above in the response to Comment 1, we decided to replace the k-means clustering by a decision tree classification.

* Reply to comment 5 *
Indeed, [Crawford et al., 2017] had a similar dataset, which also showed how useful UV-LIF methods can be for ambient data sets even when proper calibration experiments (e.g. using prior knowledge/training sets) are not performed. In the original work, we favoured k-means instead of Ward linkage given that [Ruske et al., 2017] showed that both perform similarly for the MBS (k-means even performed slightly better). However, with the new approach and the decision tree, we are not using unsupervised learning methods any more.

* Reply to comment 6 *
We agree. As described in the original manuscript, we performed laboratory experiments (see SI) with real sea water (containing organic substances and PBAP), culture medium prepared with filtered Baltic seawater (removing most of the PBAP) and inorganic sea water (almost no organic substances). Here, the PBAP cluster resolved by the k-means algorithm was shown to only be present in the ambient seawater sample and not in both filtered sea water and inorganic sea salt solution. This laboratory experiment originally guided us in the selection of the number of clusters. However, in the revised version we now use the decision tree and no k-means clustering. We agree that the k-means clustering is not the best method for our work, since the concentrations of PBAP are now higher (although the temporal variations did not change much).

* Reply to specific comment 1 *
We agree and have added the following text and references to the revised manuscript: [Agranovski et al., 2003, Huffman et al., 2020, Kaye et al., 2005, Gabey et al., 2010, Könemann et al., 2019, Kiselev et al., 2013]

“Recently, real-time single-particle analysis instruments have been developed based on ultraviolet light-induced fluorescence (UV-LIF)19−24”.

* Reply to specific comment 2 *
This has been rephrased for clarification and now reads:
“Within this work, particles with optical diameters below 0.8 μm were discarded as they fall below the 50% counting efficiency [Gabey et al., 2011] and the remaining particles were labeled as coarse particles (CP).”

* Reply to specific comment 3 *
Asymmetry and the peak-to-mean ratio shape parameters are more easily interpreted as real morphological characteristics and were shown in correlation plots to be less dependent on other shape and size parameters than others. Thus, we have chosen to focus our discussion on these two. For clarification the following sentence has been rephrased clarity:

“One of these was the so-called asymmetry factor (AF) that represents the degree of asymmetry between the left and right CMOS arrays scattered signal. In principle, high AF values, on a scale 0-100, represent more irregularly shaped rough surfaced particles while lower AF values represent more regularly shaped particles that scatter with greater azimuthal uniformity. Another shape parameter is the so-called peak-to-mean ratio (PTMR) that represents the highest light scatter value recorded within an array divided by the mean signal across the whole array.”

* Reply to specific comment 4 *
We agree and have changed the analysis using the 9γ-threshold for identifying potential PBAP, while particles just below the 9γ-threshold and above the 3γ-threshold are classified as fluorescent sea spray aerosol (FSSA). See Fig. R2 and the answer text for Comment 1.

* Reply to specific comment 5 *
We agree that this comparison between WIBS and MBS can not be made and have removed this sentence.

* Reply to specific comment 6 *
This has been corrected and the sentence removed, as most of the section 3.1 - Classification of fluorescent particles - has been reworked.

* Reply to specific comment 7 *
As mentioned above, we used solely the fluorescence signal, which is more indicative for PBAP and looked at the shape and size parameters independently. However, we no longer use clustering algorithms in the revised version and have removed the respective sentence.

* Reply to specific comment 8 *
We agree and have now use box and whisker plots in the revised version (see now Fig. R5a).

* Reply to specific comment 9 *
We agree, in the original work these clusters 1-3 were not treated independently. They were grouped as part of the same FSSA group. With the new decision tree, these particles would fall into the FSSA group as well since they are below the 9γ but above the 3γ threshold.

* Reply to specific comment 10 *
We agree, the sentence has been reworked to exercise caution:

“The fluorescence spectra of the HFSSA most likely resembles that of water-soluble organic compounds excited at 280nm [Duarte et al., 2004], although these fluorophores emission wavelengths can be dependant of other parameters such as pH [Duarte et al., 2004].”

* Reply to specific comment 11 *
We agree and we have reworked the section so that the laboratory results are shown before the classification between HFSSA and PBAP is introduced. The reworked text is shown in the answer to Comment 1.

* Reply to specific comment 12 *
Yes, the same excitation wavelength was used. Clarification has been added to the text:

“A comparison of our identified groups with the excitation emission matrix spectroscopy (EEMS) spectra with a 280nm excitation wavelength of SSA extracts by [Santander et al., 2021] is shown in Figure S10.”

* Reply to specific comment 13 *
The Figure S12 in the supplementary information has been updated to now include both 3γ and 9γ comparison, along with PBAP and HFSSA. For 3γ the conclusion remains the same, for 9γ the contribution increases with size but reaches 60% at the last size bin (around 15μm), which means that big particles don’t necessarily have enough fluorophore material to reach this threshold. We chose to keep the text as the message holds: big sea spray aerosol particles will most likely surpass the 3γ threshold. With our new decision tree, the FSSA class is completely segregated from PBAP and HFSSA by applying a 9γ threshold. This is backed up by our laboratory experiments.

*** Response to reviewer 2 ***

We thank the reviewer for their helpful comments. Indeed, we had a flux-estimate included in an earlier version of the manuscript but removed it prior to submission because we thought that concentrations and the respective ratios would be sufficient (as they can also be used to derive a
flux). However, we have now brought this aspect back to the revised version of the manuscript. A table and an additional figure containing the contribution of PBAP (and HFSSA) for different sizes of sea spray aerosol is now presented in the supporting information (see Fig. S12 and Table S1 in the SI) which can be used to derive flux estimates together with an existing SSA source function. Currently, there are many different sea spray source functions proposed in the literature and used by various Earth system models [Grythe et al., 2014]. Of course, this can only be regarded as a first estimate and future estimations will need to include other factors such as salinity, the biogeochemical activity of the water/surface layer and only considered fluorescent PBAP. As discussed above and in the manuscript, the limitations are given by the fact that potential non-fluorescent or sub-micron PBAP are not detected by our method and might be different for other (e.g. high salinity) seawater. In addition to the new figure and table, we added the following text to the revised manuscript and refined the conclusions (by moving some parts back into the result section and focusing on the main findings):

“The herein calculated concentrations are experiment specific and characteristic of a combination of seawater physio-chemical properties, such as temperature and dissolved matter. Although the biogeochemical properties of the SML inside the SSSC might differ from the SML present at the air-sea interface, our experiment is still likely to provide a reasonably good model for the emission of PBAP from fresh seawater. The entrainment of air within the SSSC has been characterized and shown to reflect natural entrainment and bubble area density due to wind sheer [Salter et al., 2014]. The PBAP concentrations presented here are a first estimation but are in a similar range as PBAP emissions found within sea spray aerosol using filter based methods[Mayol et al., 2017]. Furthermore, it is estimated that INP constitute approximately 1 in every 105-106 particles in the atmosphere [DeMott et al., 2010, DeMott et al., 2016]. Given that we expect PBAP to be efficient INP, [Tobo et al., 2013, Wilson et al., 2015, DeMott et al., 2016] we also expect PBAP populations to be present at similar concentrations. Indeed, with our experiment we identified PBAP to be approximately 1 in every 104 particles above 0.8 μm which lies in the same range as expected INP concentrations in marine environments (keeping in mind that the number size distribution of SSA particles is dominated by sub-micron particles which are not detected by the MBS, see Figure S4).”

“It is difficult to directly determine the PBAP emission flux from our observations due to the general challenge of up-scaling laboratory sea spray chamber experiments. However, it is possible to give a first estimate of the PBAP flux using a selected sea spray source function (see e.g. [Salter et al., 2015] or [Grythe et al., 2014]) in combination with the here presented size-resolved contribution of PBAP (or HFSSA) to coarse mode aerosol (see Figure S12 and Table S1 in the SI).”

* Reply to minor comment 1 *
We agree and have removed this sentence during the reformulation of the particle classification section.

* Reply to minor comment 2 *
Work by [Kaluarachchi et al., 2021] shows how biogeochemical changes in the water impact the morphology of sea spray aerosol particles. Completely disentangling the influences of each individual seawater parameter in the morphology of the particles proves hard, given the high degree of complexity of the system. [Flores et al., 2021] showed that the coarse mode sea spray aerosol concentration might be linked to the diel cycle of microorganisms. These are evidences that the biogeochemical properties of the water change both morphology and concentration of particles. This is also what we have seen within this work. Some sentences have been improved and text from the conclusion reallocated to better illustrate this relation:

“Interestingly, there is a clear difference in particle classes composition between these two periods, as shown in Figures S6 and S7 in the SI, further illustrating this differentiation. This sudden change in particle shape could have been due to changes in the water composition leading to differences in the chemical and biological sea spray composition. Such relation has been previously reported by [Kaluarachchi et al., 2021].”

“Clear particle morphology changes were observed throughout the experiment that coincided with changes in sampling locations and differences in the physical and chemical properties of the sampled seawater. In agreement with previous studies, our results imply that the changes in seawater composition will also lead to differences in the biological and organic composition of the emitted SSA particles [Hasenecz et al., 2020] with potential impacts on the particle morphology [Kaluarachchi et al., 2021] that in turn could potentially influence their ice nucleating propensity [Roy et al., 2021].”

“The distinction into two discrete sampling periods is observed by changes in the fluorescent particle classes detected by the MBS but also supported through offline analysis of the bacterial community composition of particle and seawater samples.”

*** References ***

Agranovski, V., Ristovski, Z., Hargreaves, M., Blackall, P. J., and Morawska,
L. (2003). Real-time measurement of bacterial aerosols with the UVAPS: Performance evaluation.
Journal of Aerosol Science, 34(3):301–317.

Crawford, I., Gallagher, M., Bower, K., Choularton, T., Flynn, M., Ruske,
S., Listowski, C., Brough, N., Lachlan-Cope, T., Fleming, Z., Foot, V., and Stanley, W. (2017).
Real Time Detection of Airborne Bioparticles in Antarctica. Atmospheric Chemistry and Physics
Discussions, pages 1–21.

DeMott, P. J., Hill, T. C., McCluskey, C. S., Prather, K. A., Collins, D. B.,
Sullivan, R. C., Ruppel, M. J., Mason, R. H., Irish, V. E., Lee, T., Hwang, C. Y., Rhee, T. S., Snider,
J. R., McMeeking, G. R., Dhaniyala, S., Lewis, E. R., Wentzell, J. J., Abbatt, J., Lee, C., Sultana,
C. M., Ault, A. P., Axson, J. L., Martinez, M. D., Venero, I., Santos-Figueroa, G., Stokes, M. D.,
Deane, G. B., Mayol-Bracero, O. L., Grassian, V. H., Bertram, T. H., Bertram, A. K., Moffett, B. F.,
and Franc, G. D. (2016). Sea spray aerosol as a unique source of ice nucleating particles. Proceedings
of the National Academy of Sciences of the United States of America, 113(21):5797–5803.

DeMott, P. J., Prenni, A. J., Liu, X., Kreidenweis, S. M., Petters, M. D., Twohy,
C. H., Richardson, M. S., Eidhammer, T., and Rogers, D. C. (2010). Predicting global atmospheric
ice nuclei distributions and their impacts on climate. Proceedings of the National Academy of Sciences
of the United States of America, 107(25):11217–11222.

Duarte, R. M., Pio, C. A., and Duarte, A. C. (2004). Synchronous scan and
excitation-emission matrix fluorescence spectroscopy of water-soluble organic compounds in atmospheric
aerosols. Journal of Atmospheric Chemistry, 48(2):157–171.

Flores, J. M., Bourdin, G., Kostinski, A. B., Altaratz, O., Dagan, G., Lombard,
F., Häentjens, N., Boss, E., Sullivan, M. B., Gorsky, G., Lang-Yona, N., Trainic, M., Romac,
S., Voolstra, C. R., Rudich, Y., Vardi, A., and Koren, I. (2021). Diel cycle of sea spray aerosol
concentration. Nature Communications, 12(1).

Gabey, A. M., Gallagher, M. W., Whitehead, J., Dorsey, J. R., Kaye, P. H., and
Stanley, W. R. (2010). Measurements and comparison of primary biological aerosol above and below
a tropical forest canopy using a dual channel fluorescence spectrometer. Atmospheric Chemistry and
Physics, 10(10):4453–4466.

Gabey, A. M., Stanley, W. R., Gallagher, M. W., and Kaye, P. H. (2011). The
fluorescence properties of aerosol larger than 0.8 μ in urban and tropical rainforest locations. Atmospheric
Chemistry and Physics, 11(11):5491–5504.

Grythe, H., Ström, J., Krejci, R., Quinn, P., and Stohl, A. (2014). A review of seaspray
aerosol source functions using a large global set of sea salt aerosol concentration measurements.
Atmospheric Chemistry and Physics, 14(3):1277–1297.

Hasenecz, E. S., Jayarathne, T., Pendergraft, M. A., Santander, M. V., Mayer,
K. J., Sauer, J., Lee, C., Gibson, W. S., Kruse, S. M., Malfatti, F., Prather, K. A., and Stone, E. A.
(2020). Marine Bacteria Affect Saccharide Enrichment in Sea Spray Aerosol during a Phytoplankton
Bloom. ACS Earth and Space Chemistry, 4(9):1638–1649.

Huffman, J. A., Perring, A. E., Savage, N. J., Clot, B., Crouzy, B., Tummon,
F., Shoshanim, O., Damit, B., Schneider, J., Sivaprakasam, V., Zawadowicz, M. A., Crawford, I.,
Gallagher, M., Topping, D., Doughty, D. C., Hill, S. C., and Pan, Y. (2020). Real-time sensing of
bioaerosols: Review and current perspectives. Aerosol Science and Technology, 54(5):465–495.

Kaluarachchi, C. P., Or, V. W., Lan, Y., Madawala, C. K., Hasenecz, E. S.,
Crocker, D. R., Morris, C. K., Lee, H. D., Mayer, K. J., Sauer, J. S., Lee, C., Dorce, G., Malfatti,
F., Stone, E. A., Cappa, C. D., Grassian, V. H., Prather, K. A., and Tivanski, A. V. (2021). Size-
Dependent Morphology, Composition, Phase State, and Water Uptake of Nascent Submicrometer
Sea Spray Aerosols during a Phytoplankton Bloom. ACS Earth and Space Chemistry.

Kaye, P. H., Stanley, W. R., Hirst, E., Foot, E., Baxter, K. L., and Barrington, S. J.
(2005). Single particle multichannel bio-aerosol fluorescence sensor. Optics Express, 13(10):3583.

Kiselev, D., Bonacina, L., and Wolf, J. P. (2013). A flash-lamp based device for
fluorescence detection and identification of individual pollen grains. Review of Scientific Instruments,
84(3).

Könemann, T., Savage, N., Klimach, T., Walter, D., Fröhlich-Nowoisky, J.,
Su, H., Pöschl, U., Alex Huffman, J., and Pöhlker, C. (2019). Spectral Intensity Bioaerosol Sensor
(SIBS): An instrument for spectrally resolved fluorescence detection of single particles in real time.
Atmospheric Measurement Techniques, 12(2):1337–1363.

Mayol, E., Arrieta, J. M., Jim´enez, M. A., Mart´ınez-Asensio, A., Garcias-Bonet,
N., Dachs, J., Gonz´alez-Gaya, B., Royer, S. J., Ben´ıtez-Barrios, V. M., Fraile-Nuez, E., and Duarte,
C. M. (2017). Long-range transport of airborne microbes over the global tropical and subtropical
ocean. Nature Communications, 8(1):1–8.

Perring, A. E., Schwarz, J. P., Baumgardner, D., Hernandez, M. T., Spracklen,
D. V., Heald, C. L., Gao, R. S., Kok, G., McMeeking, G. R., McQuaid, J. B., and Fahey, D. W.
(2015). Airborne observations of regional variation in fluorescent aerosol across the United States.
Journal of Geophysical Research: Atmospheres, 120(3):1153–1170.

Pöhlker, C., Huffman, J. A., and Pöschl, U. (2012). Autofluorescence of atmospheric
bioaerosols - Fluorescent biomolecules and potential interferences. Atmospheric Measurement
Techniques, 5(1):37–71.

Roy, P., Mael, L. E., Hill, T. C., Mehndiratta, L., Peiker, G., House, M. L., Demott,
P. J., Grassian, V. H., and Dutcher, C. S. (2021). Ice Nucleating Activity and Residual Particle
Morphology of Bulk Seawater and Sea Surface Microlayer. ACS Earth and Space Chemistry,
5(8):1916–1928.

Ruske, S., Topping, D. O., Foot, V. E., Kaye, P. H., Stanley, W. R., Crawford, I.,
Morse, A. P., and Gallagher, M. W. (2017). Evaluation of machine learning algorithms for classification
of primary biological aerosol using a new UV-LIF spectrometer. Atmospheric Measurement
Techniques, 10(2):695–708.

Salter, M. E., Nilsson, E. D., Butcher, A., and Bilde, M. (2014). On the seawater
temperature dependence of the sea spray aerosol generated by a continuous plunging jet. Journal
of Geophysical Research, 119(14):9052–9072.

Salter, M. E., Zieger, P., Acosta Navarro, J. C., Grythe, H., Kirkeväg, A., Rosati,
B., Riipinen, I., and Nilsson, E. D. (2015). An empirically derived inorganic sea spray source function
incorporating sea surface temperature. Atmospheric Chemistry and Physics, 15(19):11047–11066.

Santander, M. V., Mitts, B. A., Pendergraft, M. A., Dinasquet, J., Lee, C.,
Moore, A. N., Cancelada, L. B., Kimble, K. A., Malfatti, F., and Prather, K. A. (2021). Tandem
Fluorescence Measurements of Organic Matter and Bacteria Released in Sea Spray Aerosols.
Environmental Science & Technology.

Savage, N. J., Krentz, C. E., Könemann, T., Han, T. T., Mainelis, G., Pöhlker,
C., and Alex Huffman, J. (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(11):4279–4302.

Tobo, Y., Prenni, A. J., Demott, P. J., Huffman, J. A., McCluskey, C. S., Tian, G.,
Pöhlker, C., Pöschl, U., and Kreidenweis, S. M. (2013). Biological aerosol particles as a key determinant
of ice nuclei populations in a forest ecosystem. Journal of Geophysical Research Atmospheres,
118(17):100–10.

Wilson, T. W., Ladino, L. A., Alpert, P. A., Breckels, M. N., Brooks, I. M.,
Browse, J., Burrows, S. M., Carslaw, K. S., Huffman, J. A., Judd, C., Kilthau, W. P., Mason,
R. H., McFiggans, G., Miller, L. A., Najera, J. J., Polishchuk, E., Rae, S., Schiller, C. L., Si,
M., Temprado, J. V., Whale, T. F., Wong, J. P., Wurl, O., Yakobi-Hancock, J. D., Abbatt, J. P.,
Aller, J. Y., Bertram, A. K., Knopf, D. A., and Murray, B. J. (2015). A marine biogenic source of
atmospheric ice-nucleating particles. Nature, 525(7568):234–238.

07-Aug-2022

Dear Dr Zieger:

Manuscript ID: EA-ART-04-2022-000047.R1
TITLE: Emission of Primary Bioaerosol Particles from Baltic Seawater

Thank you for submitting your revised manuscript to Environmental Science: Atmospheres. I am pleased to accept your manuscript for publication in its current form. I have copied any final comments from the reviewer(s) below.

You will shortly receive a separate email from us requesting you to submit a licence to publish for your article, so that we can proceed with the preparation and publication of your manuscript.

You can highlight your article and the work of your group on the back cover of Environmental Science: Atmospheres. If you are interested in this opportunity please contact the editorial office for more information.

Promote your research, accelerate its impact – find out more about our article promotion services here: https://rsc.li/promoteyourresearch.

We will publicise your paper on our Twitter account @EnvSciRSC – to aid our publicity of your work please fill out this form: https://form.jotform.com/211263048265047

How was your experience with us? Let us know your feedback by completing our short 5 minute survey: https://www.smartsurvey.co.uk/s/RSC-author-satisfaction-energyenvironment/

By publishing your article in Environmental Science: Atmospheres, you are supporting the Royal Society of Chemistry to help the chemical science community make the world a better place.

With best wishes,

Dr Lin Wang
Associate Editor, Environmental Science: Atmospheres

Environmental Science: Atmospheres is accompanied by companion journals Environmental Science: Nano, Environmental Science: Processes and Impacts, and Environmental Science: Water Research; publishing high-impact work across all aspects of environmental science and engineering. Find out more at: http://rsc.li/envsci

Reviewer 1

I am happy with the changes made to the MBS classification scheme employed in this study. The derived results are much more robust and defensible than the k-means approach in the initial submission which I felt was a significant weakness in the underpinning analysis. Generally the authors have taken my suggestions on board and have made the recommended improvements. As such, I am now happy to accept the revised manuscript for publication.