Abstract
Objective
To identify putative members of the canine cutaneous core flora. A knowledge of this flora is important if we are to properly understand cutaneous dysbiosis. Cutaneous dysbiosis has been reported in allergic skin conditions, for example.
Methods
The unclipped skin on the axilla, umbilical region, and groin of 15 healthy dogs was swabbed monthly for 5 months using flocked cotton swabs. The samples were taken within a sterile brass guide of the known area. The swabs were submitted for next-generation 16S rRNA and internal transcribed spacer 2 sequencing to identify components of the bacterial and fungal microbiome.
Results
The density of the bacteria and fungi were expressed as IQR/cm2. Coagulase-negative staphylococci, Sphingomonas subsp, Malassezia var, Vishniacozyma victoriae, and Cladosporium var were recovered from 70% of samples. None of these 5 were found on every dog on every occasion. The density of the bacteria was 6.4 × 104 (IQ [interquartile range] 1, 3.2 × 104; IQ3, 1.2 × 105), the density of the filamentous fungi was 1.04 × 103/cm2 (IQ1, 450; IQ3, 2.13 × 105), and the density of malassezial yeast was 1.3 × 122 (IQ1, 90; IQ3, 1.8 × 10).
Conclusions
We propose that these 2 bacterial species and 3 fungal species be considered members of the core cutaneous flora of dogs.
Clinical Relevance
Knowledge of the normal flora will help our understanding of aberrations in flora that might be associated with underlying diseases, such as atopy.
We previously reported the quantification of bacteria and fungi at a single timepoint on the periumbilical skin of 20 healthy dogs.1,2 We showed that there was no difference in healthy dogs in the sample yield from swabbing or swabbing after superficial scraping.1,2 The median density of all bacteria was 1.1 × 105 cm−2 (IQR, 1.22 × 104 to 1.6 × 105).2 We found that Sphingomonas subsp, Corynebacterium kroppenstedtii, and Nocardioides subsp were the most prevalent bacterial species in our study, although none were found on every dog. The median density of the fungi was 1.1 × 105 cm−2 (IQR, 2.76 × 104, 4.10 × 105). Cladosporium var and Vishniacozyma victoriae were found on every dog.
The core flora, at least in terms of the skin microbiome, comprises those microorganisms that are found in most, or all, samples from similar sites in different individuals.3,4 The question as to whether a certain bacterium, Sphingomonas aerolata, for example, is simply a transient or potentially a member of the core skin flora cannot be definitively answered without a longitudinal or temporal study and without environmental comparison.5
The temporal (or dynamic) core microbiome identifies temporally stable, or predictable, taxa across the host species/population detected in 70% of sampling events, for example, Risely,6 citing Bjork et al.7 Although setting the proportion of occurrence (percentage cutoff) that might be considered relevant is at the discretion of the researchers, it is important to consider whether an arbitrarily high cutoff might result in missing important members of the biome that might be low in abundance but still very relevant.6,8 For example, Torres et al9 found that none of the bacteria isolated from the dogs in their study achieved 80% prevalence in all groups. Accordingly, they introduced a 50% cutoff, and, using this, they suggested that Corynebacterium subsp, Porphyromonas subsp, Cutibacterium acnes (synonym Propionibacterium acnes), and, perhaps, Haemophilus subsp be proposed as core members of the canine skin bacterial flora.9 Recently, Whittle et al10 reported a 71-dog study wherein they sampled 4 body sites on 1 occasion. They reported 15 species of interest, each of which achieved 80% prevalence across the 71 dogs and was represented by 2 or more strains. Interestingly, they did not report any coagulase-negative staphylococci.
The study9 referred to above specifically looked at the temporal carriage of bacteria on 40 healthy dogs, sampling the clipped skin of the dorsal neck, axilla, and ventral abdomen on 3 occasions in winter, spring, and summer. Four other dermatological studies11–14 had a temporal component in their design, but none of these made an attempt to identify members of the core flora on healthy dogs. In addition to medication, other factors that can affect the cutaneous biome of dogs are the presence of other pets within the household, the wider environment and cohabiting members of the household, and diet.15,16
The core flora on human skin is considered stable, although there is variation in the composition of the skin microbiome between body sites and between individuals.7,17–19 A similar variation in composition has been reported in dogs.20,21 Environmental sampling was only reported in 1 study22 of the canine cutaneous biome, notwithstanding its relevance.5
This study was designed to identify members of the core flora on the skin of 15 healthy dogs sampled at 3 sites every 4 weeks over a 5-month period, from January through June 2024. In addition, there was monthly environmental sampling of the table in the consulting room.
Methods
The Royal College of Veterinary Surgeons Ethics Committee reviewed and approved the methods and text of the consent document (Royal College of Veterinary Surgeons Ethical Permission 2023-089-Harvey2). Signed consent documents were collected from all owners.
All dogs were owned by staff members at the Willows Veterinary Centre, Shirley, Solihull, UK. Examination of the dog’s records allowed confirmation that they were healthy and had not been administered any medications other than routine vaccinations and systemic antiparasitic products and, further, that they had not been diagnosed with any chronic allergic skin disease and had not received topical or systemic antimicrobial treatments within the previous 6 months.
The methods have been described previously.1,2 Briefly, dogs were gently restrained in lateral recumbency, and flocked swab samples were taken within a brass guide in a standard manner on the unclipped skin of the axilla, umbilical region, and groin. The 3 sites were chosen because hair clipping was not required to swab the skin surface, thus satisfying the ethical conditions of the study.
To minimize contamination of the skin samples, a number of steps were taken. First, owners were given an appointment, such that they could then be escorted to the sampling site without waiting. Second, examination tables were cleaned before sampling with a disinfectant expected to denature any residual DNA (Virkon; VioVet). Third, sterile, single-use gloves were worn. Fourth, the guides were brass No. 9 door furniture letters, and these were cold sterilized between patients and samples with Reprodis HLD4l (Hyperdrug, UK). In addition, we took swab samples from the consulting-room table every month prior to sampling in an attempt to identify environmental contaminants.
Swab samples were taken by using sterile, dry, DNA-free HydraFlock (catalog No. 25-3406-H; Puritan). The swabs were immediately immersed into vials containing a sterile DNA preservative (DNA/RNA Shield; catalog No. R1108; Zymo Research Corp) and immediately frozen. Samples were shipped en masse to the MiDOG LLC testing facility.
Using methods previously described,1,2 genomic DNA was purified using the ZymoBIOMICSTM-96 DNA Kit (catalog No. 79 D4304; Zymo Research Corp). Sample library preparation and data analysis for fungal profiling were performed by MiDOG LLC using a Quick-16S NGS Library Prep Kit (catalog No. D6400; Zymo Research Corp) with minor modifications.1
As internal controls to ensure the accuracy and cleanliness of the data generated and to control for any potential contamination of the equipment, sequencing buffers, and other material, several negative controls were run for both the extraction process and the library preparation. These included an “extraction negative control,” which was the storage buffer (catalog No. R1100-50; DNA/RNA Shield), which was lysed, extracted, library prepped, and sequenced in parallel with experimental samples. Further, a “library preparation negative control” and a “no template control” for the library preparation were run. The workflow is automated using a Hamilton Star liquid-handling robot (Hamilton Company) to minimize human error during the sampling process.
To control for contamination, both cellular and DNA mock communities were used as positive controls (ZymoBIOMICS Microbial Community Standard; catalog Nos. D6300 and D6305; Zymo Research Corp) to account for any bias in the workflow starting from the DNA extraction process. ZymoBIOMICS Microbial Community Standard (Zymo Research Corp) was used as a positive control to monitor the performance of all steps of the next-generation sequencing workflow, including the bioinformatic analysis. Primer sequences targeted the internal transcribed spacer 2 region for mycobiome analysis as previously described.23 Libraries were sequenced using a HiSeq 1500 sequencer (Illumina) for a sequencing depth of 7 to 8 million reads, generating at least 10,000 reads/sample. Reads were filtered through Dada2 (R package, version 3.4; R Foundation for Statistical Computing). Taxonomy prediction was performed with Centrifuge24 (version 24; Zymo Research) combined with a custom reference database curated, in part, from draft or complete genomic sequences available from the National Center for Biotechnology Information’s GenBank.
Phylotypes were computed as percentage of proportions based on the total number of sequences in each sample. The species-level resolution of the sequencing approach used here has previously been demonstrated by shotgun sequencing.24 Absolute microbial quantification was achieved using real-time PCR targeting the internal transcribed spacer 2 region for fungi and the V1 to V3 16S rRNA for bacteria.
Results
There were 9 neutered males and 6 neutered females. All 15 dogs attended, as required, at 4-week intervals. All were sampled at the 3 sites described above every 4 weeks over a 5-month period, from January through June 2024. No fungi were isolated from any of the environmental control samples. Streptococcus canis and C kroppenstedtii were found once on different environmental samples.
There was no statistical difference between samples from the axilla, umbilicus, and groin, and, accordingly, results were pooled. Thus, each dog yielded 1 collective sample a month, giving 75 samples in total.
The density of all bacteria was 6.4 × 104/cm2 (IQ [interquartile range] 1, 3.2 × 104; IQ3, 1.2 × 105), the density of the filamentous fungi was 1.04 × 103/cm2 (IQ1, 450; IQ3, 2.13 × 105), and the density of malassezial yeast was 1.3 × 122 (IQ1, 90; IQ3, 1.8 × 101; Table 1).
The mean density (per cm2) of the bacteria, malassezial yeast, and fungi from all samples.
| Species | Density (mean [IQ (inerquartile range) 1:IQ3/cm2]) |
|---|---|
| All bacteria | 6.4 × 104 (3.2 × 104:1.2 × 105) |
| All Malassezia var | 1.3 × 102 (90:1.8 × 102) |
| All filamentous fungi | 1.04 × 102 (4.5 × 102:2.13 × 105) |
IQ1 = Interquartile range 1. IQ3 = Interquartile range 3.
Sphingomonas subsp, coagulase-negative staph-ylococci, Malassezia var, Vishniacozyma victoriae, and Cladosporium var were found in over 70% samples, although none were found on every dog on every occasion (Tables 1 and 2). Other bacteria, found at between 33% and 46% prevalence, included Sphingomonas aerolata and Sphingomonas aurantiaca-faenia, Corynebacterium kroppenstedtii, Porphyromonas cangingivalis, Streptococcus canis, Rothia kristinae, Nocardioides subsp, and Staphylococcus pseudintermedius.
The prevalence and density from the 75 samples of the principal species recovered over the 5 months of the study.
| Bacteria | Prevalence | Density (mean [IQ1, IQ3/cm2]) |
|---|---|---|
| Sphingomonas subsp | 59 | 8.1 × 102 (IQ1, 2.8 × 102; IQ3, 1.8 × 103) |
| Coagulase-negative staphylococci | 55 | 2.2 × 103 (IQ1, 79; IQ3, 7.4 × 104) |
| S aerolata | 39 | 5.1 × 102 (IQ1, 1.9 × 102; IQ3, 9 × 103) |
| S aurantiaca-faeni | 39 | 4.4 × 102 (IQ1, 1.9 × 102; IQ3, 6.8 × 102) |
| C kroppenstedtii | 35 | 6.2 × 102 (IQ1, 1.9 × 102; IQ3, 1.4 × 103) |
| P canigingivatis | 33 | 7.1 × 102 (IQ1, 1.8 × 102; IQ3, 2.3 × 103) |
| S canis | 31 | 5.7 × 102 (IQ1, 1.8 × 102; IQ3, 12 × 103) |
| R kristinae | 27 | 6.2 × 102 (IQ1, 1.9 × 102; IQ3, 1.4 × 103) |
| Nocardioides subsp | 27 | 1.8 × 103 (IQ1, 5 × 102; IQ3, 3.4 × 103) |
| S pseudintermedius | 25 | 1.4 × 103 (IQ1, 7.4 × 102; IQ3, 2.1 × 103) |
| Fungi | ||
| Malassezia var | 65 | 1.3 × 102 (90:1.8 × 102) |
| V victoriae | 58 | 2.6 × 102 (1.1 × 102:5.6 × 102) |
| Cladosporium var | 56 | 1.7 × 102 (1.0 × 102:3.2 × 102) |
Discussion
The density of the members of the cutaneous biome was of the same magnitude as previously reported.1,2 It is not surprising that we did not record 100% recovery with any of these putative members of the core flora. It has been reported that the relative abundance of even highly abundant members of the core flora may vary across subjects by multiple orders of magnitude.19 The most common external reason for a variation in a temporal core is the effect of season.6 It is for this reason that it been suggested that investigators studying the canine skin have a regard for changes in season when planning a temporal study.20 We chose January to June for our study as in the UK that avoids the compounding effect of major changes in heat and humidity.
In addition to trying to identify a core biome, it has been suggested that investigators have a regard for “keystone” species.8 These are relatively low-abundance organisms that might, nonetheless, have an effect on the community. With this in mind, if we consider S pseudintermedius a putative keystone species, perhaps we should consider also C kroppenstedtii and R kristinae (synonym Kocuria kristinae) as they were also recovered in similar abundance here and in our previous study.2 Both C kroppenstedti and R kristinae are considered commensals on human skin (and in the oral cavity).25–27
Gram-positive, coagulase-negative cocci predominated in culture-based studies3,4 of the skin biome of both dogs and humans. However, both culture-based studies and those based on 16S mRNA new technology9,10,22,28–30 of the bacteria on normal canine skin have consistently found gram-negative bacteria on normal canine skin. Interestingly, in one of these studies,30 a hair and microdissection study on 8 dogs, the gram-negative bacteria were statistically significantly found on the skin surface and proximal hair shaft rather than on the distal hair shaft. This distribution rather speaks against them being a contaminant. Gram-negative bacteria have also been found on normal human skin, albeit in limited areas.31,32
Sphingomonas subsp are gram-negative, aerobic bacilli that have been reported on the skin of lifeguards and the faces of elderly Thai males, for example.33,34 They have also been reported from canine studies of the cutaneous biome. Indeed, of the 13 studies9,10,12–16,20–23,28,29 reporting 16S rRNS analysis of the canine biome, all but 1 found Sphingomonas subsp, although sometimes in low numbers. One study14 found Sphingomonas subsp on the same dogs at both sampling times 30 days apart, although the species recovered were not reported.
The authors in 1 of the 2 studies33 cited above, wherein Sphingomonas subsp were found on human skin, noted that the skin of these groups of people was dry and had low sebaceous activity—perhaps not dissimilar to the ventral abdomen of a dog.35 Support for this being a relatively lipid-free environment is that malassezial yeast are less often recovered from these areas.36
Finally, sphingomonal bacteria have been correlated with the presence of a pet dog in the home.37 It had been assumed that the dog was essentially a vector, sampling the outdoor environment,37 but equally it might reflect Sphingomonas subsp being a part of the core flora.
We previously reported that Cladosporium var and V victoriae were found on every dog.1 With these 2 species of fungi and Malassezia var having a prevalence of over 70% in this study, it would appear that they also should be considered part of the canine cutaneous biome.
The major limitation of this study is the relatively low number of dogs and the sampling being confined to the ventral aspect of the dogs as bacterial carriage, for example, varies with site.20,21 Other factors known to have an effect and that were not controlled are diet, the presence of other animals within the household, and lifestyle.15,16 Further studies should explore whether our conclusions are valid and, if so, what the implications are for our understanding of the canine cutaneous biome and dysbiosis.
In conclusion, we propose, on the basis of a 70% prevalence in a temporal study, that coagulase-negative staphylococci, Sphingomonas subsp, Malassezia var, V victoriae, and Cladosporium var be considered as candidates for the core biome on the skin of healthy dogs.
Acknowledgments
None reported.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the composition of this manuscript.
Funding
MiDOG provided swabs and paid for the transportation of said swabs, the microbiome analysis, and the subsequent data analysis.
ORCID
R. Harvey https://orcid.org/0000-0002-7176-5487
J. Krumbeck https://orcid.org/0000-0002-4865-7170
References
- 1.↑
Harvey RG, Duclos D, Krumbeck J, Tang S. Quantifying the mycobiome and its major constituents on the skin of 20 normal dogs. Am J Vet Res. 2023;84(10):1–5. doi:10.2460/ajvr.23.04.0071
- 2.↑
Harvey RG, Duclos D, Krumbeck J, Tang S. Quantification of the bacterial flora and its major constituents on the abdominal skin of clinically healthy dogs. Am J Vet Res. 2023;84(10):1–6. doi:10.2460/ajvr.23.04.0072
- 3.↑
Weese JS. The canine and feline skin microbiome in health and disease. Vet Dermatol. 2013;24(1):137–145.e31. doi:10.1111/j.1365-3164.2012.01076.x
- 4.↑
Dréno B, Araviiskaia E, Berardesca E, et al. Microbiome in healthy skin, update for dermatologists. J Eur Acad Dermatol Venereol. 2016;30(12):2038–2047. doi:10.1111/jdv.13965
- 5.↑
Cosseau C, Romano-Bertrand S, Duplan H, et al. Proteobacteria from the human skin microbiota: species-level diversity and hypotheses. One Health. 2016;2:33–41. doi:10.1016/j.onehlt.2016.02.002
- 6.↑
Risely A. Applying the core microbiome to understand host-microbe systems. J Anim Ecol. 2020;89(7):1549–1558. doi:10.1111/1365-2656.13229
- 7.↑
Bjork JR, O’Hara RB, Ribes M, Coma R, Montoya JM. The dynamic core microbiome: structure, dynamics and stability. bioRxiv. 2018;137885. doi:10.1101/137885
- 8.↑
de Cena JA, Zhang J, Deng D, Damé-Teixeira N, Do T. Low-abundant microorganisms: the human microbiome’s dark matter, a scoping review. Front Cell Infect Microbiol. 2021;11:689197. doi:10.3389/fcimb.2021.689197
- 9.↑
Torres S, Clayton JB, Danzeisen JL, et al. Diverse bacterial communities exist on canine skin and are impacted by cohabitation and time. PeerJ. 2017;5:e3075. doi:10.7717/peerj.3075
- 10.↑
Whittle MJ, Castillo-Fernandez J, Amos GCA, Watson P. Metagenomic characterisation of canine skin reveals a core healthy skin microbiome. Sci Rep. 2024;14(1):20104. doi:10.1038/s41598-024-63999-5
- 11.↑
Widmer G, Ferrer L, Favrot C, Paps J, Huynh K, Olivry T. Glucocorticosteroids and ciclosporin do not significantly impact canine cutaneous microbiota. BMC Vet Res. 2018;14(1):51. doi:10.1186/s12917-018-1370-y
- 12.↑
Bradley CW, Morris DO, Rankin SC, et al. Longitudinal evaluation of the skin microbiome and association with microenvironment and treatment in canine atopic dermatitis. J Invest Dermatol. 2016;136(6):1182–1190. doi:10.1016/j.jid.2016.01.023
- 13.
Chermprapai S, Ederveen THA, Broere F, et al. The bacterial and fungal microbiome of the skin of healthy dogs and dogs with atopic dermatitis and the impact of topical antimicrobial therapy, an exploratory study. Vet Microbiol. 2019;229:90–99. doi:10.1016/j.vetmic.2018.12.022
- 14.↑
Leverett K, Manjarín R, Laird E, et al. Fresh food consumption increases microbiome diversity and promotes changes in bacteria composition on the skin of pet dogs compared to dry foods. Animals (Basel). 2022;12(15):1881 doi:10.3390/ani12151881
- 15.↑
Song SJ, Lauber C, Costello EK, et al. Cohabiting family members share microbiota with one another and with their dogs. Elife. 2013;2:e00458. doi:10.7554/eLife.00458
- 16.↑
Lehtimäki J, Sinkko H, Hielm-Björkman A, Laatikainen T, Ruokolainen L, Lohi H. Simultaneous allergic traits in dogs and their owners are associated with living environment, lifestyle and microbial exposures. Sci Rep. 2020;10(1):21954. doi:10.1038/s41598-020-79055-x
- 17.↑
Sharon I, Quijada NM, Pasolli E, et al. The core human microbiome: does it exist and how can we find it? A critical review of the concept. Nutrients. 2022;14(14):2872. doi:10.3390/nu14142872
- 18.
Li K, Bihan M, Methé BA. Analyses of the stability and core taxonomic memberships of the human microbiome. PLoS One. 2013;8(5):e63139. doi:10.1371/journal.pone.0063139
- 19.↑
Huse SM, Ye Y, Zhou Y, Fodor AA. A core human microbiome as viewed through 16S rRNA sequence clusters. PLoS One. 2012;7(6):e34242. doi:10.1371/journal.pone.0034242
- 20.↑
Older CE, Rodrigues Hoffmann A. Considerations for performing companion animal skin microbiome studies. Vet Dermatol. 2024;35(4):367–374. doi:10.1111/vde.13250
- 21.↑
Sánchez A, Altet L, Ferrer L, et al. Individual signatures define canine skin microbiota composition and variability. Front Vet Sci. 2017;4:119. doi:10.3389/fvets.2017.00119
- 22.↑
Cuscó A, Sánchez A, Altet L, et al.Individual signatures define canine skin microbiota composition and variability. Front Vet Sci. 2017;4:6. doi:10.3389/fvets.2017.00006. Published correction appears in Front Vet Sci. 2017;4:119.
- 23.↑
Tang S, Prem A, Tjokrosurjo J, et al. The canine skin and ear microbiome: a comprehensive survey of pathogens implicated in canine skin and ear infections using a novel next-generation-sequencing-based assay. Vet Microbiol. 2020;247:108764. doi:10.1016/j.vetmic.2020.108764
- 24.↑
Kim D, Song L, Breitwieser FP, Salzberg SL. Centrifuge: rapid and sensitive classification of metagenomic sequences. Genome Res. 2016;26(12):1721–1729. doi:10.1101/gr.210641.116
- 25.↑
Jensen MG, Svraka L, Baez E, Lund M, Poehlein A, Brüggemann H. Species- and strain-level diversity of corynebacteria isolated from human facial skin. BMC Microbiol. 2023;23(1):366. doi:10.1186/s12866-023-03129-9
- 26.
Chaubey P, Mohta V, Mishra M, Wagh A, Sapre S. Kocuria kristinae - an unusual opportunistic pathogen in head and neck malignancy: case series from Central India’s regional cancer care centre. Indian J Pathol Microbiol. 2023;66(1):14–18. doi:10.4103/ijpm.ijpm_776_21
- 27.↑
Zaura E, Keijser BJ, Huse SM, Crielaard W. Defining the healthy “core microbiome” of oral microbial communities. BMC Microbiol. 2009;9:259. doi:10.1186/1471-2180-9-259
- 28.↑
Rodrigues Hoffmann A, Patterson AP, Diesel A, et al. The skin microbiome in healthy and allergic dogs. PLoS One. 2014;9(1):e83197. doi:10.1371/journal.pone.0083197
- 29.↑
García-Fonticoba R, Ferrer L, Francino O, Cuscó A. The microbiota of the surface, dermis and subcutaneous tissue of dog skin. Anim Microbiome. 2020;2(1):34. doi:10.1186/s42523-020-00050-8
- 30.↑
Harvey RG, Lloyd DH. The distribution of bacteria (other than staphylococci and Propionibacterium acnes) on the hair, at the skin surface and within the hair follicles of dogs. Vet Dermatol. 1995;6(2):79–84. doi:10.1111/j.1365-3164.1995.tb00047.x
- 31.↑
Myles IA, Reckhow JD, Williams KW, Sastalla I, Frank KM, Datta SK. A method for culturing gram-negative skin microbiota. BMC Microbiol. 2016;16:60. doi:10.1186/s12866-016-0684-9
- 32.↑
Grice EA, Kong HH, Conlan S, et al. Topographical and temporal diversity. Science. 2009;324(5931):1190–1192. doi:10.1126/science.1171700
- 33.↑
Wilantho A, Deekaew P, Srisuttiyakorn C, Tongsima S, Somboonna N. Diversity of bacterial communities on the facial skin of different age-group Thai males. PeerJ. 2017;5:e4084. doi:10.7717/peerj.4084
- 34.↑
Harel N, Ogen-Shtern N, Reshef L, Biran D, Ron EZ, Gophna U. Skin microbiome bacteria enriched following long sun exposure can reduce oxidative damage. Res Microbiol. 2023;174(8):104138. doi:10.1016/j.resmic.2023.104138
- 35.↑
Oh WS, Oh TH. Mapping of the dog skin based on biophysical measurements. Vet Dermatol. 2010;21(4):367–372. doi:10.1111/j.1365-3164.2009.00841.x
- 36.↑
Guillot J, Bond R. Malassezia yeasts in veterinary dermatology: an updated overview. Front Cell Infect Microbiol. 2020;10:79. doi:10.3389/fcimb.2020.00079
- 37.↑
Amin H, Šantl-Temkiv T, Cramer C, et al. Indoor airborne microbiome and endotoxin: meteorological events and occupant characteristics are important determinants. Environ Sci Technol. 2023;57(32):11750–11766. doi:10.1021/acs.est.3c01616
