the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Oct 2025
Vertebrates impact on bacterial community structure of coastal Arctic snowpacks in the spring
Abstract. Snow covers up to 35 % of the Earth's surface seasonally and forms a microbial habitat despite harsh and variable conditions. While atmospheric deposition is a well-known source of microbial input, the role of vertebrates in shaping snow microbiomes remains underexplored. In Arctic ecosystems, seabirds and terrestrial mammals contribute not only nutrients but also microbial communities. Here, we explore the role of vertebrates in shaping snow microbial biodiversity of Arctic terrestrial snowpacks. The study was conducted on the northern coast of Hornsund Fjord on Spitsbergen. Fourty snow samples were collected in four transects, two established along the gradient from the centre of a seabird (Alle alle) colony towards non-impacted areas and two transects along the coast. We identified a total of 8,521 bacterial OTUs using short-read sequencing of the 16S rRNA gene. Samples clustered into four snow groups based on community composition, but not linked to spatial factors such as distance to colonies. Bird and terrestrial mammal faecal indicators like Catellicoccus or Streptococcus were detected in 17 out of the 40 samples and drove the formation of two distinct clusters. These findings suggest that coastal Arctic snow microbiomes are strongly shaped by biological activity, with wildlife acting as key microbial vectors.
- Preprint
(1864 KB) - Metadata XML
-
Supplement
(1027 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-5033', Anonymous Referee #1, 18 Nov 2025
-
AC1: 'Reply on RC1', Sławomir Sułowicz, 22 Dec 2025
Dear Reviewers,
Thank you for your comments. We received an editorial message stating: “Please note that your revised manuscript should not be prepared at this stage”. Nevertheless, some of the changes have already been incorporated into our manuscript, as we believe that regardless of the final outcome of the review process in this journal, they are valuable and we want to keep them. Therefore, in the case of some comments, we have provided corrected parts of the text.
We revised the manuscript according to the comments and suggestions. We have put a lot of effort into this, and we believe that it has allowed us to improve the manuscript. The most significant change is taking into account the suggestions of both reviewers regarding the re-analysis of NGS data using ASV analysis with DADA2. However, as one reviewer expected, the results did not change by altering the clustering approach, and the clusters identified in the first version of the manuscript are reflected in the ASV data and new analyses, without altering the major conclusions of the manuscript. We included 'response to the reviewers' comments' and in answers we presented how the given comment was addressed.
Reviewer #1: General Comments
The manuscript of Sulowicz et al describes a dataset of 40 bacterial community samples collected from Arctic surface snow in 4 transects. The authors clustered the bacterial community data and concluded that some communities were driven by the influence of bird colonies while one transect was influenced by mammal activity.
While the authors describe a unique dataset of scientific value, the methods for data analysis are outdated and the interpretation and conclusions are overstated. With a thorough re-analysis of the dataset with state-of-the-art methods and (taxonomy) databases and interpretation of the results as they are, the manuscript should be publishable. I suggest the following major revisions to get there:
COMMENT 1. The introduction is a bit lengthy and partly redundant. Consider shortening, e.g. antibiotic resistance is not a subject of this study so remove, also shorten references to snow-free works as snow communities are studied here.
ANSWER 1. We shortened the Introduction by removing unnecessary general information and focusing more specifically on microbial communities in seasonal snow cover.
COMMENT 2. At the end of the introduction, the authors claim that their study ‘aims to clarify how vertebrate-driven microorganisms influence snow biodiversity and its cascading effects on Arctic terrestrial ad coastal ecosystems under climate change.’ I don’t see how this is possible with the collected observational (DNA) amplicon data. There is no causal link between the vertebrate influence, bacterial communities, activity of these communities and climate change.
ANSWER 2. We deleted this sentence.
COMMENT 3. Introduction: remove the climate change aspect, this is not tackled by this manuscript (and cannot with this data).
ANSWER 3. We removed the climate change aspect from the Introduction.
COMMENT 4. Section Experimental Procedures: line 118: which species richness is referred to here? Mammals, vertebrates, bacteria?
ANSWER 4. It should be “Animal species richness”. We added the missing word.
COMMENT 5. Section Experimental Procedures, 2nd paragraph partially overlaps with the introduction. The part with the references would better fit in the introduction.
ANSWER 5. We moved this paragraph from Section 2.1 to the Introduction and included the rest of the text in the description of subchapter “Sample collection, DNA isolation, and NGS sequencing.”
COMMENT 6. Bioinformatics: OTU analyses is very dated. The data needs to be reanalyzed with a state of the art method such as ASV analysis with DADA2.
ANSWER 6. We reran all the analysis using DADA2. The new sequence analysis section reads as follows. We will additionally provide the scripts used. We would like to note that the OTU vs ASV method did not change the major conclusions of the manuscript, but we agree that DADA2 is now considered a standard tool and have therefore incorporated this in our analysis.
“Paired-end 16S rRNA gene amplicon reads were processed in R using DADA2 (Callahan et al., 2016) to generate amplicon sequence variants (ASVs). Briefly, primers (341F 3’-CCTACGGGNGGCWGCAG-5’ and 785R 3’-GACTACHVGGGTATCTAATCC-5’; (Klindworth et al., 2013)) targeting the V3-V4 region of the bacterial 16S rRNA gene were cut from raw reads with cutadapt (Martin, 2011) in paired-end mode with a minimum post-trim length of 50 bp. Forward and reverse reads were cut to 270 and 260 bp based on observation of quality profiles and reads further filtered using truncQ = 12 and maxEE = 1 generating in total 1.4 million high quality reads. Error models, dereplication, ASV inference (pool = FALSE) and merging were done with default options. Merged reads with a size between 400–435 bp based on the expected amplicon size were retained which corresponded to 8,611 ASVs and chimeras removed using the consensus method, resulting in 7,640 ASVs. Taxonomy was assigned to these ASVs using SILVA nr99 training set v138.2 (Chuvochina et al., 2025). Contaminants were identified using the R package decontam in prevalence mode and a threshold of 0.2. We performed a stringent prevalence (at least present in 3 samples) and abundance (at least 5 counts per ASV over all samples) filtering to further reduce the influence of spurious low abundance ASVs (Bokulich et al., 2013). After removal of ASVs annotated to chloroplast and mitochondrial sequences this analysis resulted in a set of 854 ASVs.”
COMMENT 7. Bioinformatics: the default taxonomy similarity percentage is very low (80%), youcannot report at genus level, the OTUs might even be from different families than what you report!
ANSWER 7. As stated in the new methods suggestion above, we reran the analysis of the sequencesusing DADA2. The similarity percentage used was 99%, with extra filtering to ensure that the taxonomy is robust.
COMMENT 8. Statistics: Please clearly describe what is used as input for the different methods: counts, normalized counts, relative abundances, …?
ANSWER 8. We have added additional explanations in the Experimental Procedures section. For example “Alpha-diversity richness indices (total number of OTUs, Chao1 and Chao1 bias-corrected, Faith’s phylogenetic diversity PD) were calculated based on ASV counts (presence–absence), whereas diversity indices accounting for evenness (Shannon entropy and Simpson’s index) were calculated using ASV read abundances. (Faith and Baker, 2006; Malandrakis et al., 2019; Qiagen, 2024).” or “Beta-diversity analyses were performed using generalized UniFrac distances d(0.5) calculated from ASV tables based on read abundances.”
COMMENT 9. Statistics: a significant p-value of a Permanova indicates that either the variances or the centroids of the clusters are different. If the authors want to make a statement about differences, they have to test for equality of variances and apply a post-hoc test for differences.
ANSWER 9. We agree that PERMANOVA significance may reflect differences in either centroids or dispersion. The PERMANOVA analysis was conducted using the CLC Microbial Genomics Module, which does not provide a separate PERMDISP/betadisper test for homogeneity of multivariate dispersions. Therefore, we have carefully rephrased the Results and Discussion to interpret PERMANOVA outcomes as overall multivariate differences in community composition rather than strictly centroid differences. Additionally, Pairwise PERMANOVA results were provided to illustrate which group contrasts contributed most strongly to the overall multivariate pattern.
However, based on the analysis of ASV data, line factor is now not significant and only differences are detected between clusters. Now this part is described as: “When transect line was considered as a grouping factor, PERMANOVA did not revealed significant overall multivariate differences in bacterial community composition among lines (P =0.089). However, based on the 2D (Fig. 2) and 3D view (Fig. S2) of principal coordinate analysis (PCoA) of bacterial 16S rRNA profiles samples formed four clusters characterized by distinct multivariate community profiles, a pattern supported by PERMANOVA results (P < 0.001 for each pair of clusters, except for the comparison of clusters 1 and 2, where P = 0.002).”
COMMENT 10. Physicochemical analysis. Line 232 ff. Which environmental parameters were analyzed? What is PAST software? What values were log-transformed? Why? Please provide data in supplement or permanent repository. I don’t understand the 2 additional microbial parameters, 56 genera plus total number of reads plus total number of OTUs?
ANSWER 10. We thank the reviewer for pointing out the need for clarification.
PAST is a freely available statistical software package widely used in ecological and environmental studies for multivariate analyses. In this study, it was applied only to perform CCA in order to examine relationships between microbial community composition and measured environmental variables. No other statistical analyses were conducted in PAST.
The environmental parameters included in the CCA were described in the manuscript L.232-239. “Among 20 measured environmental factors describing snow physicochemical properties, after preliminary analysis, twelve non-correlated factors (pH, conductivity, Non-Purgeable Organic Carbon NPOC, cations: NH4+, Ca2+, Mg2+ and anions: HCO3-, Br-, Cl-, NO3-, PO43-, SO42-) were used. As microbial parameters, 58 values were used (read number of 56 genera present in each cluster, which is over 1% of total read content, total number of reads, and OTU).”
For clarification purposes, out of all environmental factors, we selected 12 factors for analysis, which, in accordance with the CCA analysis recommendation, were uncorrelated with one another. Microbial community data were analysed at the genus level and restricted to 56 genera exceeding (threshold, ≥1% relative abundance in at least one cluster], in order to reduce sparsity and improve robustness of the ordination. In addition, two sequencing-derived technical variables—total number of sequencing reads per sample and total number of detected OTUs per sample—were included as auxiliary variables to account for variation related to sequencing depth and observed richness. These variables were not interpreted as ecological drivers.
Each variables (except for the pH value, which is already a logarithmic value) were log10-transformed prior to CCA to stabilise variances and limit the influence of extreme values on the ordination.
All data used as input for the CCA, including physicochemical parameters and sequencing summary statistics, will be provided in the Supplementary Material.
The current description, updated for ASV data, looks like this:
“Canonical correspondence analysis was used to identify the most important environmental factors affecting the snow clusters' microbial parameters. The CCA was performed using PAST software (Hammer et al., 2001). Among 20 measured environmental variables describing snow physicochemical properties, twelve non-correlated parameters were selected following preliminary screening, in accordance with CCA assumptions: pH, conductivity, non-purgeable organic carbon (NPOC), cations (NH₄⁺, Ca²⁺, Mg²⁺) and anions (HCO₃⁻, Br⁻, Cl⁻, NO₃⁻, PO₄³⁻, SO₄²⁻). Microbial community data were analysed at the genus level and restricted to 50 bacterial genera exceeding 1% relative abundance in at least one cluster, in order to reduce data sparsity and improve robustness of the ordination. In addition, two sequencing-derived technical variables—the total number of sequencing reads per sample and the total number of detected ASVs per sample—were included as auxiliary variables to account for variation related to sequencing depth and observed richness; these variables were not interpreted as ecological drivers. Prior to analysis, all variables except pH (which is inherently logarithmic) were log10-transformed to stabilise variances and reduce the influence of extreme values on the ordination.”
COMMENT 11. Results, Bacterial community structure: could chloroplasts be distinguished from Cyanobacteria and were they excluded?
ANSWER11. We removed chloroplast sequences prior to analysis. This is now described in the methods: After removal of ASVs annotated to chloroplast and mitochondrial sequences this analysis resulted in a set of 854 ASVs.
COMMENT 12. Results: The markers Streptococcus and Catellicoccus are not consistent within a cluster. Please justify how you generated the clusters with respect to the markers and the PCoA clustering. While the origin of the first axis was used to separate clusters 2 and 3 from the rest, this was not the case for the y-axis. The clustering seems somewhat arbitrary from the PCoA.
ANSWER 12. We acknowledge that individual markers like Streptococcus and Catellicoccus show within-cluster variation. However, clusters were defined based on overall community similarity, which was evaluated taking into account both the results of numerous working analyses using hierarchical clustering at various taxonomic levels (e.g. genus-level, phylum-level, OTU-level), the content of individual groups of microorganisms in the microbial community (presented in a bar chart), as well as PCoA analysis. That's right, the first axis separated clusters 2 and 3 from the others. And it may seem that the snow cluster is not divided into two clusters. However, as indicated in the text, cluster number one (Fig. 2, marked as gold), was distinguished based on the PCoA 3D view [not only along PCoA2 (16.60%) but mainly along PCoA3 (11.54%)] and was characterized by high bacterial diversity shown in the bar chart and alpha diversity figures. While marker distribution may vary within clusters, the overall microbial community structure captured by beta diversity was significantly different between clusters (PERMANOVA: P<0.001), confirming the correctness of separating 4 clusters.
Moreover, based on the ASVs analysis, we now observe exactly the same division into clusters. This separation is even clearer than before due to PCoA2 (15.12%) and PCoA3 (10.04%). Additionally, our assignment is also consistent with a new analysis that we performed - hierarchical clustering of samples (Hellinger+Bray). Taking all these results into account, we believe that distinguishing four clusters is justified.
COMMENT 13. Move section 3.3 to the supplement, it is not contributing much.
ANSWER 13. We have modified the results significantly and moved the figures and discussion relative to diversity indices to the supplementary material. We added a sentence about differences in diversity indices to the main text in the results section to describe the clusters of the snow samples.
COMMENT 14. Section 3.4: this is an overinterpretation of CCA. CCA is correlation-based and cannot derive causal effects of environmental parameters on species abundance.
ANSWER 14. We have modified the language to clarify that this is correlation-based
COMMENT 15. Discussion: the discussion overstates the findings and causality. While some samples were dominated by DNA commonly found in bird guts/feces, this does not mean that these organisms are active and seeding or driving microbial community composition. DNA amplicon sequencing captures DNA from dead and alive organisms as well as free (environmental) DNA. It might be that there is absolutely no activity from these gut organisms. Since the majority of gut microorganisms are anaerobic or microaerophile and adapted to warm temperatures, it is actually unlikely that they are still alive on the ice. 16S rRNA sequencing (measuring active rRNA) or other methods are needed to make a statement about activity. Same is true for the mammal-influenced cluster of course.
ANSWER 15. We did not state anywhere in the text that these organisms are active and have added a sentence to the discussion to make this even clearer. This was also mentioned in the introduction. We agree that these organisms might not be active and that RNA-based approaches would be needed to infer this. We have edited the discussion to tone down any potential confusion. We added a small section to describe the impact of non-successful colonizers on microbial communities: We determined that wild animals significantly contribute to seeding the surface snow microbiomes, with bacteria of animal origin identified in almost half of our samples (clusters two and three). Successful colonization into snow communities by these microorganisms is probably unlikely due to the significant differences in ecosystem characteristics between snow and the host environment that may limit their survival. However, studies have shown that organisms that fail to establish long-term populations can still impact the recipient community regardless of their ability to proliferate or spread or example, the genetic pool of ecosystems can be impacted through the release of free DNA, which can then be taken up by the local community.
COMMENT 16. Discussion and Methods: the oral microbiome members, especially the human ones might be contamination. How did you control for contamination? I did not see negative and positive controls being mentioned in the methods. These are essential to control contamination.
ANSWER 16. We used the decontam program to account for contamination.
COMMENT 17. The discussion is lengthy and sometimes deviates a lot from the study area.
ANSWER 17. We have streamlined the discussion.
COMMENT 18. Conclusions repeat the overstated findings.
ANSWER 18. We have edited the conclusion. It now reads as follows: “Our study shows that vertebrates influence the bacterial community structure of coastal Arctic snowpacks in the spring. In addition to identifying cold-adapted organisms typically associated with psychrophilic environments, we also detected microbes of vertebrate origin. These were mainly linked to the gut microbiome of birds and terrestrial mammals, including humans. Although further studies are needed to determine whether these organisms are active and able to survive in the snow, sequences classified to Catellicoccus, Streptococcus and Actinomyces genera indicated vertebrate activity and nutrient enrichment. Snow chemistry, especially salinity, pH, and nutrient concentrations, was impacted by both vertebrates and marine aerosols and likely influenced the snow microbiome. Our study indicated that compared to previously considered areas, animals are an important source of microorganisms to the snow surface. Overall, our findings indicated that in addition to transport factors such as atmospheric deposition and wind-transported materials from snow-free terrestrial environments, microbial composition in Arctic coastal snowpacks is more dependent on wildlife than previously assumed. Both abiotic and biotic transport factors should be considered when assessing microbial ecosystem dynamics in polar regions.”
Minor comments:
- Please check spelling and grammar - DONE
- Line 206: introduce the abbreviation MCE. - DONE
- Line 219: introduce the abbreviation TOC. - DONE
- Lines 227-228: introduce the abbreviations LOD and LOQ. - DONE
- 2: some labels are missing or not readable. Please fix. – ANWER: Based on the new ASV results, we are preparing new figures. We remember the reviewers' comments on the figures and take them into account.
- 3B: Please label all of the samples such that they can me mapped to Fig. 2.- ANWER: As above, based on the new ASV results, we are preparing new figures. We remember the reviewers' comments on the figures and take them into account.
- Line 268: the plural of ‘taxon’ is ‘taxa’. - DONE
- Line 281: a Venn diagram is not an analysis, it is a visualization. - DONE
- 4, legend: unclear what data ‘normalized distribution …’ refers to. – DONE - Normalized distribution of the 50 most abundant genera (based on read abundances)
- Actinobacteria have been renamed to Actinomycetota, please use recent taxonomy.ANSWER: Thank you for pointing this out. We are already using the new terminology in our current analyses. We have also updated the text of the manuscript.
Citation: https://doi.org/10.5194/egusphere-2025-5033-AC1
-
AC1: 'Reply on RC1', Sławomir Sułowicz, 22 Dec 2025
-
RC2: 'Comment on egusphere-2025-5033', Jakob Zopfi, 04 Dec 2025
Review: Biogeosciences Nov. 2025; Sulovicz et al.,
«Vertebrates impact on bacterial community structure of coastal Arctic snowpacks in the spring»
General comments:
Sulovicz and colleagues present novel and interesting data on microbial community structures and taxonomic compositions in spring snow cover on Svalbard. Samples have been collected along four transect lines in the coastal area of Hansvika in the Hornsund (Svalbard). The data are clear and convincing. Some snow samples have a clearly different community structure (uneven community composition, lower diversity, presence of indicator taxa) than other samples (even community structure, i.e. not dominated by specific taxa, high richness), indicating that the former group of samples was influenced by vertebrate (mostly bird feces). Although is a local case study, the observed patterns/mechanisms are likely transferable to other arctic or alpine environments.
Weaknesses of the study involve data presentation and choice of clustering method. However, the results are clear, and will not alter by changing the clustering approach. Although it would enable a better comparison with other studies and allow a finer taxonomic resolution.
Data presentation (Results) contains redundant information and redundant figures (Fig. 3 a,b, Fig. 5 c, d.). My personal impression is, that also the taxonomic description of the different cluster is somewhat redundant/repetitive. One solution could be i) to describe first the cluster formation by PCoA and PERMANOVA as it is now (in 3.1.), but, without describing which taxonomic groups characterizing the different clusters. ii) Then bring the alpha diversity characteristics for the different clusters (not 6 alpha diversity measures needed. 1 Richness indicator and 1-2 diversity indicators should be sufficient, e.g. Shannon and PD). iii) Thereafter, dive into the taxonomic composition of the four different clusters. This structure should eliminate repetitions, shorten the text and avoid the need for hopping back and forth between taxa descriptions in 3.1 and the later chapters.
Minor or Specific comments:
In title: delete “the” in front of spring, and put “structures” in plural
Sampling: How thick was the snow cover at the time of sampling? Were bird or fox droppings visible close to the sampling sites?
I assume that the snow cover thickness in Lines 140ff are the winter maximum values. Could the microbial community structure in the snow be inoculated from below, from the soil if the snow cover is thinning when melting intensifies in May? You don’t happen to have soil microbial community structure data? Personally, I’d find it interesting to see whether the same key taxa (e.g. indicating faecal contamination by birds) are also found in the soil itself.
Line 179: Unclear how OTUs have been formed. Have they been produced through clustering sequences by similarity? If so, then provide the similarity threshold (e.g. 97% or 99% sequence similarity).
Why was not an ASV (amplified sequence variant, amplicon sequence variant) approach chosen? As opposed to OTU clustering where several similar sequences are combined into a consensus sequence, ASVs represent exact sequences, which are indeed present in a sample. They have a higher taxonomic resolution because minor sequence differences are not blurred by clustering, and can be better compared with other studies.
(Callahan, B., McMurdie, P., Rosen, M. et al. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13, 581–583 (2016). https://doi.org/10.1038/nmeth.3869).
Line 216: format chemical formula (subscript)
Line 256ff: What determines the cluster number name (1,2,3,4), respectively, what determines the unconventional presentation order of the clusters, starting with cluster 3, followed by 2, then 4, and 1)? Wouldn’t it make more sense to call the “most important” cluster 1, and so forth, and present them consistently in the same (1,2,3,4) order throughout in the results?
Line 260: instead of referring to Fig3B, it would be more quantitative to refer to Fig. 5
Line 260: the statement (“marked as gold”) seems to refer to Figure 3, but should be Figure 2.
Line 291f: Why are there 3 different richness indicators presented? They show all the same thing and are not used in the further discussion of the data. E.g. Figure 5 C and D are virtually identical, and very close to Figure 5 A. Hence, presenting observed OTU Richness is sufficient.
To some degree the same statement also applies to the different diversity indexes (PD, Shannon, Simpson) although they show different aspects of diversity. However, this information is not exploited in the discussion – and, thus, probably not necessary to show all of them.
Figure 3: could go into the supplementary information. The graphs are good for checking data consistency but more detailed information cannot be extracted from these plots. One plot is sufficient in my opinion (e.g. 3b).
Figure 4: Over- and under representation of taxa is not well visible with a red-blue colour gradient. I suggest increasing the contrast by using a gradient between three colours. Over and under representation would be more easily recognizable: Gradient from e.g. Blue (<-3) to e.g. White (0) and from there to e.g. red (>4) – or, another other nice cold colour to white, and white to warm colour gradient combination.
Figure 4: On which type of normalization is the heatmap based? Centered log ratio normalization?
Line 444: mammals (plural)
Line 406f: Tone down statement. Release of NO3 by N2-fixing cyanobacteria is not to be expected - as the N is bound in Glutamine and other amino acids – unless the cells lyse and are mineralized. Whether the NO3 in the snow is indeed from N2 fixation or from oxidized NH4+ of faecal origin could be tested by del15N measurement of NO3.
Data availability
Illumina raw sequencing data have been made available by deposition on NCBI. Submission of other data used in this manuscript to Pangaea (curated) or Zenodo (not curated) public databases e.g. for environmental data, scripts, treated sequence data and OTU tables, is also encouraged in order to comply with FAIR principles.
Supplementary information:
Figure S1: Although there is a map in the main manuscript, it would be good to recall in the figure caption meaning of A, B, C, and D - the locations of the different transects. Maybe something in the line of:
…..from transect lines on the slope and the base of Fugleberget (A, B), the wetland plain (C), and close to the sea shore of Hansvika (D).
Figure S2 and S4: Mention both, the cluster numbers and their respective colours, in the caption.
Citation: https://doi.org/10.5194/egusphere-2025-5033-RC2 -
AC2: 'Reply on RC2', Sławomir Sułowicz, 22 Dec 2025
Dear Reviewers,
Thank you for your comments. We received an editorial message stating: “Please note that your revised manuscript should not be prepared at this stage”. Nevertheless, some of the changes have already been incorporated into our manuscript, as we believe that regardless of the final outcome of the review process in this journal, they are valuable and we want to keep them. Therefore, in the case of some comments, we have provided corrected parts of the text.
We revised the manuscript according to the comments and suggestions. We have put a lot of effort into this, and we believe that it has allowed us to improve the manuscript. The most significant change is taking into account the suggestions of both reviewers regarding the re-analysis of NGS data using ASV analysis with DADA2. However, as one reviewer expected, the results did not change by altering the clustering approach, and the clusters identified in the first version of the manuscript are reflected in the ASV data and new analyses, without altering the major conclusions of the manuscript. We included 'response to the reviewers' comments' and in answers we presented how the given comment was addressed.
Reviewer #2: General comments:
Sulovicz and colleagues present novel and interesting data on microbial community structures and taxonomic compositions in spring snow cover on Svalbard. Samples have been collected along four transect lines in the coastal area of Hansvika in the Hornsund (Svalbard). The data are clear and convincing. Some snow samples have a clearly different community structure (uneven community composition, lower diversity, presence of indicator taxa) than other samples (even community structure, i.e. not dominated by specific taxa, high richness), indicating that the former group of samples was influenced by vertebrate (mostly bird feces). Although is a local case study, the observed patterns/mechanisms are likely transferable to other arctic or alpine environments.
Weaknesses of the study involve data presentation and choice of clustering method. However, the results are clear, and will not alter by changing the clustering approach. Although it would enable a better comparison with other studies and allow a finer taxonomic resolution.
Data presentation (Results) contains redundant information and redundant figures (Fig. 3 a,b, Fig. 5 c, d.). My personal impression is, that also the taxonomic description of the different cluster is somewhat redundant/repetitive. One solution could be i) to describe first the cluster formation by PCoA and PERMANOVA as it is now (in 3.1.), but, without describing which taxonomic groups characterizing the different clusters. ii) Then bring the alpha diversity characteristics for the different clusters (not 6 alpha diversity measures needed. 1 Richness indicator and 1-2 diversity indicators should be sufficient, e.g. Shannon and PD). iii) Thereafter, dive into the taxonomic composition of the four different clusters. This structure should eliminate repetitions, shorten the text and avoid the need for hopping back and forth between taxa descriptions in 3.1 and the later chapters.
ANSWER: We thank the reviewer for this constructive comment regarding redundancy in data presentation and the structure of the Results section. We agree that, in the previous version, some information may have appeared repetitive. However, we would like to clarify that the presentation of multiple complementary analyses was intentional, as their purpose was to demonstrate that different analytical approaches consistently support the division of samples into four clusters. In the revised version of the manuscript, we have improved clarity and reduced unnecessary repetition by reorganizing the Results section, while retaining the key analyses that jointly justify the four-cluster solution. Additionally, we moved alpha diversity characteristics to Supplementary Materials.
Minor or Specific comments:
COMMENT 1. In title: delete “the” in front of spring, and put “structures” in plural
ANSWER 1. We have corrected the title.
COMMENT 2. Sampling: How thick was the snow cover at the time of sampling? Were bird or fox droppings visible close to the sampling sites? I assume that the snow cover thickness in Lines 140ff are the winter maximum values. Could the microbial community structure in the snow be inoculated from below, from the soil if the snow cover is thinning when melting intensifies in May? You don’t happen to have soil microbial community structure data? Personally, I’d find it interesting to see whether the same key taxa (e.g. indicating faecal contamination by birds) are also found in the soil itself.
ANSWER 2 We clarified the sampling depth in the methods description. We collected a 10 cm deep layer of surface snow, since the birds had only returned 2 weeks prior to our field campaign. At the time of sampling in early May, the snow cover was still continuous and relatively thick across all transects. In addition, snow depth varied across the sampling sites, and collecting bulk samples would have made normalization difficult. Snow depth varied with topography and wind redistribution and ranged from approximately 58 cm at coastal sites to over 100 cm at the foot of the Fugleberget slope. The values reported in Lines 140ff refer to the winter maximum; however, during sampling the snowpack had not yet thinned to the extent that contact with underlying soil occurred. Snowmelt had only recently begun, and vertical percolation was limited.
Signs of vertebrate activity were detected in the study area, including reindeer herds, polar fox tracks, feathers, bones and surface discoloration consistent with biological material deposition. Fresh bird or fox droppings were not consistently present at individual sampling points and were not used as a criterion for site selection, as sampling was designed to capture broader gradients of vertebrate influence rather than immediate proximity to faecal deposits. Given the substantial snow thickness and the early stage of the melt season, direct microbial inoculation from underlying soil is considered unlikely. The vertebrate-associated microbial signatures detected in snow are therefore more plausibly explained by surface deposition and atmospheric or wind-driven inputs rather than upward transfer from soil.
While we agree that soil is a potential colonization source to basal snow samples, we really wanted to focus on surface sampling to test for the effect of seabirds and animals on seasonal snow. Given that these colonies re-establish every year, the soil has likely been enriched in bird droppings over years, if not decades and we were worried that this might complicate our study design. As a result, soil microbial community data were not collected within the scope of this study. We agree that a direct comparison between snow and soil communities would be valuable and now note this explicitly as an important direction for future work.
COMMENT 3. Line 179: Unclear how OTUs have been formed. Have they been produced through clustering sequences by similarity? If so, then provide the similarity threshold (e.g. 97% or 99% sequence similarity).
ANSWER 3. We thank the reviewer for pointing out that this step required clearer description. We reran all the sequence analysis using DADA2 to generate ASV tables and used a similarity threshold of 99%. The methods now read as follows: Paired-end 16S rRNA gene amplicon reads were processed in R using DADA2 (Callahan et al., 2016) to generate amplicon sequence variants (ASVs). Briefly, primers (341F 3’-CCTACGGGNGGCWGCAG-5’ and 785R 3’-GACTACHVGGGTATCTAATCC-5’; (Klindworth et al., 2013)) targeting the V3-V4 region of the bacterial 16S rRNA gene were cut from raw reads with cutadapt (Martin, 2011)in paired-end mode with a minimum post-trim length of 50 bp. Forward and reverse reads were cut to 270 and 260 bp based on observation of quality profiles and reads further filtered using truncQ = 12 and maxEE = 1 generating in total 1.4 million high quality reads. Error models, dereplication, ASV inference (pool = FALSE) and merging were done with default options. Merged reads with a size between 400–435 bp based on the expected amplicon size were retained which corresponded to 8,611 ASVs and chimeras removed using the consensus method, resulting in 7,640 ASVs. Taxonomy was assigned to these ASVs using SILVA nr99 training set v138.2 (Chuvochina et al., 2025). Contaminants were identified using the R package decontam in prevalence mode and a threshold of 0.2. We performed a stringent prevalence (at least present in 3 samples) and abundance (at least 5 counts per ASV over all samples) filtering to further reduce the influence of spurious low abundance ASVs (Bokulich et al., 2013). After removal of ASVs annotated to chloroplast and mitochondrial sequences this analysis resulted in a set of 854 ASVs.
COMMENT 4. Why was not an ASV (amplified sequence variant, amplicon sequence variant) approach chosen? As opposed to OTU clustering where several similar sequences are combined into a consensus sequence, ASVs represent exact sequences, which are indeed present in a sample. They have a higher taxonomic resolution because minor sequence differences are not blurred by clustering and can be better compared with other studies.
(Callahan, B., McMurdie, P., Rosen, M. et al. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13, 581–583 (2016). https://doi.org/10.1038/nmeth.3869).
ANSWER 4. When comparing the OTU results of those generated through ASV analysis, the main conclusions were very similar. However, we decided to go forward with the DADA2 ASV approach. See above for the new methods description
COMMENT 5. Line 216: format chemical formula (subscript)
ANSWER 5. Thank you, we have corrected it.
COMMENT 6. Line 256ff: What determines the cluster number name (1,2,3,4), respectively, what determines the unconventional presentation order of the clusters, starting with cluster 3, followed by 2, then 4, and 1)? Wouldn’t it make more sense to call the “most important” cluster 1, and so forth, and present them consistently in the same (1,2,3,4) order throughout in the results?
ANSWER 6. Cluster numbering followed the order in which clusters emerged from the exploratory PCoA-based grouping and subsequent analyses, rather than reflecting their relative importance or abundance. In the working analyses, clusters 1 and 4 were initially considered as one cluster—cluster no. 1. However, taking into account all the analyses, especially based on the PCoA 3D view and high bacterial diversity shown in the bar chart and alpha diversity figures, we separated them into two clusters. We decided not to change their order due to numerous references in the text and concerns that this would introduce even more confusion to the results that have already been described and reviewed.
COMMENT 7. Line 260: instead of referring to Fig3B, it would be more quantitative to refer to Fig. 5.
ANSWER 7. A cursory analysis of the bar chart indicates greater microbial biodiversity in this cluster, which is why we have included a reference to this figure. But indeed, thank you for pointing out that the alpha index results clearly indicate greater biodiversity in this cluster. We have added a reference to Fig. 5 (now moved to SM).
COMMENT 8. Line 260: the statement (“marked as gold”) seems to refer to Figure 3, but should be Figure 2.
ANSWER 8. We edited the sentence to refer the reader to the main figure (Fig. 2), but at the same time explain the distinction of this cluster based on the results of the 3D view (Fig S2). Edited version: “Cluster number one (Fig. 2, marked as gold), distinguished in the PCoA 3D view (Fig. S2…”
COMMENT 9. Line 291f: Why are there 3 different richness indicators presented? They show all the same thing and are not used in the further discussion of the data. E.g. Figure 5 C and D are virtually identical, and very close to Figure 5 A. Hence, presenting observed OTU Richness is sufficient.
To some degree the same statement also applies to the different diversity indexes (PD, Shannon, Simpson) although they show different aspects of diversity. However, this information is not exploited in the discussion – and, thus, probably not necessary to show all of them.
ANSWER 9. Thank you for these comments. We have moved these figures to the supplementary materials. CLC software allows for the calculation of all these alpha diversity values, which is why all available values were presented. Sometimes statistical changes are only visible in selected indices, which is why it is worth performing all analyses in order to detect any changes. In this case, however, the differences between clusters were statistically significant for all parameters, confirming the validity of the division into four clusters.
COMMENT 10. Figure 3: could go into the supplementary information. The graphs are good for checking data consistency but more detailed information cannot be extracted from these plots. One plot is sufficient in my opinion (e.g. 3b).
ANSWER 10. Thank you for this suggestion. We removed Fig. S3A. Additionally, based on the new ASV results, we are preparing new figures. We remember the reviewers' comments on the figures and take them into account.
COMMENT 11. Figure 4: Over- and under representation of taxa is not well visible with a red-blue colour gradient. I suggest increasing the contrast by using a gradient between three colours. Over and under representation would be more easily recognizable: Gradient from e.g. Blue (<-3) to e.g. White (0) and from there to e.g. red (>4) – or, another other nice cold colour to white, and white to warm colour gradient combination.
ANSWER 11. As was mentioned above, based on the new ASV results, we are preparing new figures. We remember the reviewers' comments on the figures and take them into account. I checked, and CLC software allows us to use three colours in a gradient, with preset values. I agree, that kind of visualisation looks much better. Thank you.
COMMENT 12. Figure 4: On which type of normalization is the heatmap based? Centered log ratio normalization?
ANSWER 12 . CLR normalization was not applied; heatmaps were generated using TMM-normalized log-CPM values with Z-score scaling (Qiagen: User Manual for CLC Microbial Genomics Module 24.0.1, 1–293, 2024).
COMMENT 13. Line 444: mammals (plural)
ANSWER 13. Corrected.
COMMENT 14. Line 406f: Tone down statement. Release of NO3 by N2-fixing cyanobacteria is not to be expected - as the N is bound in Glutamine and other amino acids – unless the cells lyse and are mineralized. Whether the NO3 in the snow is indeed from N2 fixation or from oxidized NH4+ of faecal origin could be tested by del15N measurement of NO3.
ANSWER 14. We thank the reviewer for this important clarification. We agree that N₂ fixation by cyanobacteria results primarily in organic nitrogen incorporated into cellular compounds (e.g. glutamine and other amino acids) and does not directly lead to nitrate release unless subsequent cell lysis and mineralization occur. We have therefore toned down the corresponding statement in the Discussion and revised the text to reflect that the elevated nitrate observed in snow may arise from indirect processes (e.g. mineralization and nitrification) or from alternative sources such as atmospheric deposition or oxidized ammonium of faecal origin. We now explicitly note that resolving nitrate sources would require additional analyses, such as δ¹⁵N measurements, which were beyond the scope of this study:“Cluster four was associated with elevated nitrate concentrations and dominated by cyanobacterial taxa such as Aliterella, Chamaesiphon and Tychonema. Free-living cyanobacteria are known to actively fix atmospheric N₂ in High Arctic tundra ecosystems, including coastal sites on Spitsbergen, with fixed nitrogen initially incorporated into organic biomass (Jungblut et al., 2010). The observed nitrate in snow may therefore reflect indirect processes such as mineralization and subsequent nitrification, or alternative sources including atmospheric deposition or oxidized ammonium of faecal origin (Rousk et al., 2017). Disentangling these potential sources would require additional analyses, such as stable nitrogen isotope measurements (δ¹⁵N), which were beyond the scope of the present study.”
COMMENT 15. Data availability Illumina raw sequencing data have been made available by deposition on NCBI. Submission of other data used in this manuscript to Pangaea (curated) or Zenodo (not curated) public databases e.g. for environmental data, scripts, treated sequence data and OTU tables, is also encouraged in order to comply with FAIR principles.
ANSWER 15. All data used as input for the CCA, including physicochemical parameters and sequencing summary statistics, will be provided in the Supplementary Material. Scripts, treated sequence data and ASV tables are now described in Experimental Procedures section:
“Paired-end 16S rRNA gene amplicon reads were processed in R using DADA2 (Callahan et al., 2016) to generate amplicon sequence variants (ASVs). Briefly, primers (341F 3’-CCTACGGGNGGCWGCAG-5’ and 785R 3’-GACTACHVGGGTATCTAATCC-5’; (Klindworth et al., 2013)) targeting the V3-V4 region of the bacterial 16S rRNA gene were cut from raw reads with cutadapt (Martin, 2011) in paired-end mode with a minimum post-trim length of 50 bp. Forward and reverse reads were cut to 270 and 260 bp based on observation of quality profiles and reads further filtered using truncQ = 12 and maxEE = 1 generating in total 1.4 million high quality reads. Error models, dereplication, ASV inference (pool = FALSE) and merging were done with default options. Merged reads with a size between 400–435 bp based on the expected amplicon size were retained which corresponded to 8,611 ASVs and chimeras removed using the consensus method, resulting in 7,640 ASVs. Taxonomy was assigned to these ASVs using SILVA nr99 training set v138.2 (Chuvochina et al., 2025). Contaminants were identified using the R package decontam in prevalence mode and a threshold of 0.2. We performed a stringent prevalence (at least present in 3 samples) and abundance (at least 5 counts per ASV over all samples) filtering to further reduce the influence of spurious low abundance ASVs (Bokulich et al., 2013). After removal of ASVs annotated to chloroplast and mitochondrial sequences this analysis resulted in a set of 854 ASVs.”
COMMENT 16. Supplementary information: Figure S1: Although there is a map in the main manuscript, it would be good to recall in the figure caption meaning of A, B, C, and D - the locations of the different transects. Maybe something in the line of:
…..from transect lines on the slope and the base of Fugleberget (A, B), the wetland plain (C), and close to the sea shore of Hansvika (D).
ANWER 16. Thank you for this valuable comment. We added this description below the map: “Figure 1. Transect lines on the slope and the base of Fugleberget (A, B), the wetland plain area (C), and close to the sea shore of Hansvika (D) where snow samples were collected (background map Norwegian Polar Institute 2014, bird colonies - Keslinka et al., 2019).”
COMMENT 17. Figure S2 and S4: Mention both, the cluster numbers and their respective colours, in the caption.
ANSWER 17. We have corrected the captions for both figures.
Citation: https://doi.org/10.5194/egusphere-2025-5033-AC2
-
AC2: 'Reply on RC2', Sławomir Sułowicz, 22 Dec 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 282 | 111 | 34 | 427 | 44 | 20 | 21 |
- HTML: 282
- PDF: 111
- XML: 34
- Total: 427
- Supplement: 44
- BibTeX: 20
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript of Sulowicz et al describes a dataset of 40 bacterial community samples collected from Arctic surface snow in 4 transects. The authors clustered the bacterial community data and concluded that some communities were driven by the influence of bird colonies while one transect was influenced by mammal activity.
While the authors describe a unique dataset of scientific value, the methods for data analysis are outdated and the interpretation and conclusions are overstated. With a thorough re-analysis of the dataset with state-of-the-art methods and (taxonomy) databases and interpretation of the results as they are, the manuscript should be publishable. I suggest the following major revisions to get there:
Minor comments: