Ying Rao1,2,3, Heyang Li2,4, Mingxia Chen5, Qingyan Fu6, Guoshun Zhuang This email address is being protected from spambots. You need JavaScript enabled to view it.1, Kan Huang This email address is being protected from spambots. You need JavaScript enabled to view it.1,7,8 1 Center for Atmospheric Chemistry Study, Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention (LAP3), Department of Environmental Science and Engineering, Fudan University, Shanghai 200433, China
2 Third Institute of Oceanography, Ministry of Natural Resources, Xiamen 361005, China
3 Health Center of Minnan Normal University, Zhangzhou 363000, China
4 Fujian Provincial Key Laboratory of Marine Ecological Conservation and Restoration, Xiamen 361005, China
5 Department of Biological technology and Engineering, HuaQiao University, Xiamen 361021, China
6 Shanghai Environmental Monitoring Center, Shanghai 200030, China
7 Institute of Eco-Chongming (IEC), Shanghai 202162, China
8 Institute of Atmospheric Sciences, Fudan University, Shanghai 200433, China
Received:
January 21, 2020
Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.
Revised:
April 15, 2020
Accepted:
May 21, 2020
Download Citation:
||https://doi.org/10.4209/aaqr.2020.01.0030
Rao, Y., Li, H., Chen, M., Fu, Q., Zhuang, G. and Huang, K. (2020). Characterization of Airborne Microbial Aerosols during a Long-range Transported Dust Event in Eastern China: Bacterial Community, Influencing Factors, and Potential Health Effects. Aerosol Air Qual. Res. 20: 2834–2845. https://doi.org/10.4209/aaqr.2020.01.0030
Cite this article:
Samples of atmospheric microbial aerosols were collected before, during, and after a dust invasion in Shanghai and analyzed using 16S rRNA high-throughput sequencing. The bacterial community structures in the mixed pollutive aerosols and dust were characterized, and the key environmental factors were identified. The dominant phyla were Proteobacteria, Actinomycetes, and Firmicutes, and the relative abundance of Acidobacteria increased significantly during the episode. Additionally, marked differences in the relative abundances of the 22 detected genera were observed between the three sampling stages: The dominant genera were Rubellimicrobium and Paracoccus prior to the arrival of the dust but became Deinococcus and Chroococcidiopsis during the invasion and then Clostridium and Deinococcus afterward. Notably, the relative abundance of Cyanobacteria, which is known to cause hepatotoxicity and promote tumor growth in humans, grew substantially during the event. Finally, statistical analysis revealed the largest environmental factors affecting the bacterial communities to be wind speed and the SO2, SO42–, NO3–, PM10, NH4+, and Ca2+ concentrations.HIGHLIGHTS
ABSTRACT
Keywords:
Airborne bacteria; Bacterial community structure; Influencing factors; Long-range transported dust.
Atmospheric aerosols which contained biological substances such as microorganisms or biomolecules were called bioaerosols. Of which, the part that contained microorganisms was called microbial aerosols. Atmospheric microbial aerosols are closely related to air pollution, biosphere, cloud chemistry, climate, and human health (Delort et al., 2010; Peccia et al., 2010; Fröhlich-Nowoisky et al., 2016). Bioaerosols was unique as circulation of the materials, ecological balance and so many biological phenomena were all relevant to it (Mancinelli and Shulls, 1978). More than 25% of the particles over the earth’s surface was composed of bioaerosols (Jones and Harrison, 2004). Bioaerosols in PM2.5 were estimated in the range of 3–11% by weight (Womiloju et al., 2003; Boreson et al., 2004). The emission intensity of the global microbial aerosols varied from 10–1000 Tg y–1, while the portion of bacterial aerosols was about 0.74–28.1 Tg y–1 (Després et al., 2012). The concentration of bacterial aerosols near the surface of land was usually higher than 1 × 104 cells m–3 (Bauer et al., 2002), while the concentration of bacteria above the ocean was much less of about 100–1000 cells m–3 (Prospero et al., 2005; Griffin et al., 2006). The diurnal variation of the concentrations of bacteria in the boundary layer was significantly influenced by meteorological factors, e.g., wind speed (Sun et al., 2003). One study conducted in Xiamen, China, suggested the concentrations of bacterial aerosols were associated with the weather conditions, showing the lowest during sunny days while the highest during the haze period (Liao, 2013). Similar to the transport of particulate matters, the transport of microbial aerosols in the atmosphere was passive (Sun et al., 2003). The residence time of bacteria in the atmosphere was about one week (Burrows et al., 2009). Dust storm is one of the important natural phenomena in the earth system (Goudie and Middleton, 2006; Ravi et al., 2011; Shao et al., 2011). Some studies found that bacterial aerosols from deserts or arid areas could be transported intercontinentally (Polymenakou et al., 2008; Lim et al., 2011; Hara and Zhang, 2012; Barberán et al., 2014) and the bacterial community of the downwind areas were significantly changed (Maki et al., 2013). During the long-range transport, the aeolian dust mixed with reactive gases and polluted aerosols, and it was common that bacteria adhered on the particles (Yamaguchi et al., 2012). The abundances of bacterial aerosols collected at 1000 km downstream far from the dust source area during the dust episode were 1–10 times more than that during the non-dust period (Jeon et al., 2011). Significant differences of the bacterial community were found between dust samples and non-dust samples. Aquabacterium sp., Flavobacteriales, and Prevotellaceae were dominant in dust samples while Propionibacterium sp., Bacillus sp., and Acinetobacter sp. were dominant in non-dust samples (Lee et al., 2009). It was found that dust plumes posed threat on human health not only in the dust source areas but also the downwind areas (Griffin, 2007; Yamaguchi et al., 2012; Goudie, 2014). The diffusion of bioaerosols in atmosphere may harm human health, including infectious diseases, allergy, and occupational hazard (Douwes et al., 2000; Den Boer et al., 2002). Some bacteria carried by dust were pathogenic such as Neisseria meningitidis and Clostridium perfringens. Meningococcal meningitis caused by Neisseria meningitidis was epidemic throughout sub-Saharan Africa as Neisseria meningitidis can adhere on dust particles and activate by high iron contents in dust (Thomson et al., 2009). Clostridium perfringens has been found as the primary pathogen of food poisoning (Grass et al., 2013) and has also been observed in dust which may cause disease by inhalation (Leski et al., 2011). In this study, one dust event in Shanghai during February 20–21, 2016, was monitored based on the air mass backward trajectory modeling and the chemical tracer of aerosols. The 16S rRNA gene analysis method was applied to investigate the structure of bacterial community in atmospheric aerosols before, during, and after the dust. The differences of the bacterial community structure during different periods were revealed. Factors influencing the bacterial community and potential threat on human health were discussed. It should be noted that the assessment of the magnitude of dust impact on human health was beyond the scope of this study. Microbial aerosol samples were collected on the roof of the 4th Teaching Building of Fudan University in Shanghai, China (121°29′E, 31°14′N; 20 m above ground level). No strong point sources were located around this site. This site has been regarded as a representative of the urban environment of Shanghai, standing for the mixing of traffic, residential, and industrial sources. Microbial aerosol samples were obtained on the 37-mm sterile filter membrane (FMCE) by using the sampler (XMX-CV; Dycor, Canada) at a flow rate of 530 L min–1. All the samples obtained were kept under –20°C until DNA extraction. The sampling information is shown in Table 1. Atmospheric PM2.5 samples were synchronously collected with the microbial aerosols at the same site. Aerosol samples were collected on Whatman Grade 41 filters (Whatman Inc., Maidstone, UK) by a medium-volume sampler (flow rate: 77.59 L min–1; TSP/PM10/PM2.5-2; Beijing Geological Instrument-Dickel Co., Ltd., China). All the samples were put in polyethylene plastic bags immediately after sampling and then reserved in a refrigerator. The filters were weighed before and after sampling using an analytical balance (reading precision: 10 µg; 2004 MP; Sartorius) after stabilizing in constant temperature (20 ± 1°C) and humidity (40 ± 2%) for 48 hours. All the procedures were strictly quality controlled to avoid any possible contamination of the samples. Meteorological parameters were measured by an automated weather station (Vaisala). The frozen filter membrane was cut into pieces and put into a Lysing Matrix E tube, and then the total environmental genomic DNA was extracted using a FastDNA® Spin Kit for Soil (MP Biomedical, USA) according to the standard protocol. The purity and concentration of the DNA was examined by NanoDrop® ND-2000 UV-Vis Spectrophotometer (NanoDrop Technologies, USA), TBS-380 Mini-Fluorometer (Turner Biosystems, USA), and 1% agarose gels. The DNA was stored at –20°C until target-gene amplification. The V3–V4 region of the 16S rRNA gene was PCR amplified using the universal primers 338F (5ʹ-ACTCCTACGGGAGGCAGCAG-3ʹ) and 806R (5ʹ-GGACTACHVGGGTWTCTAAT-3ʹ) (Srinivasan et al., 2012). The 16S rRNA gene amplification was performed in 20-µL reaction volumes containing 1U TransStart FastPfu DNA Polymerase (TransGen Biotech Co., Ltd., China), 4 µL 5× FastPfu Buffer, 2 µL dNTPs (2.5 mM), 0.8 µL each primer (5 mM), and 1 µL DNA template (10 ng µL–1). The PCR reaction was conducted with a temperature program of 27 cycles of 30 s at 95°C, 30 s at 55°C, 45 s at 72°C. The PCR products with length of approximately 450 bp were excised from a 2% agarose gel and purified using the DNA Gel Purification Kit (Axygen Biosciences, USA). Purified amplicons were quantified using QuantiFluor™ ST (Promega, USA) according to the manufacturer’s protocol. Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 300) on an Illumina MiSeq platform (Illumina, USA) according to the standard protocols by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). Sequences of each sample were separated according to barcodes (exactly matching) and primers (allowing 2 nucleotide mismatching). Paired-end clean reads were merged using FLASH (v1.2.11; https://ccb.jhu.edu/software/FLASH/). Low-quality sequences (< 300 bp in length, < 20 in quality score, containing ambiguous characters and mismatch primer) were removed from raw sequences according to Trimmomatic quality controlled process (v0.33; http://www.usadellab.org/cms/?page=trimmomatic). The high-quality sequences were clustered into operational taxonomic units (OTUs) at a 97% nucleic acid similarity using QIIME software (v1.8.0; http://qiime.org/scripts/assign_taxonomy.html). The singleton OTU were removed using USEARCH (http://www.drive5.com/usearch/manual/chimera_formation.html) after OTU cluster, and then the chimera sequences were detected and removed using the UCHIME de novo algorithm (http://www.drive5.com/usearch/manual/uchime_algo.html). The taxonomy of each 16S rRNA gene sequence was analyzed by the RDP classifier algorithm (http://rdp.cme.msu.edu/) against the Silva (SSU123) 16S rRNA database using confidence threshold of 70%. The paired-end raw Illumina sequencing data were deposited into NCBI Sequence Read Archive (SRA; http://www.ncbi.nlm.nih.gov/bioproject?LinkName=biosample_bioproject&from_uid=3273385) under accession number of SRP219061. Alpha diversity is applied in analyzing complexity of species diversity for each sample. The indices of Chao1, Shannon, and Simpson were selected to identify the community richness and diversity. All indices in this study were calculated using QIIME (v1.9.1) displayed with R software (v2.15.3). Concentrations of 10 anions (F–, CH3COO–, HCOO–, MSA, Cl–, NO2–, NO3–, SO42–, C2O42–, PO43–) and 5 cations (Na+, NH4+, K+, Mg2+, Ca2+) in aqueous extracts of the particle samples were determined by ion chromatography (IC; ICS-3000; Dionex, USA). The recovery of ions was in the range of 80–120% by adding standard reference material of each ion component into the filtrates for ion chromatography analysis. Reproducibility test showed that relative standard deviation was less than 5%. The ion concentrations of the sample blanks were below detection limits or under 0.02 µg m–3 and had been deducted from the observation values. A quarter of the sample filters were digested at 190°C for 1 h with 8 mL HNO3 and 2 mL HF. After cooling, the solutions were diluted to 30 mL with distilled-deionized water. Blank filters were in parallel processing in order to reduce the error. A total of 21 elements (Al, As, Ba, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, P, Pb, S, Sn, Sr, Ti and Zn) were determined by inductively coupled plasma optical emission spectrometry (ICP-OES; SPECTRO, Germany). The Kruskal-Wallis test was used to reveal the difference of community structure among the collected samples. As the number of samples was small in this study, the Kruskal-Wallis test was a suitable tool without assuming the data to follow the normal distribution. The Mantel test and redundancy analysis were used for analyzing the relationship between species and environmental factors. The Mantel test is a non-parametric method which is often used to test the relationship between the distance matrix of community and the distance matrix of environmental factors in ecology. Redundancy analysis is environmentally constrained PCA analysis, which is used to analyze the characteristics of species with certain environmental factors. On February 20 and 21, 2016, the daily mean PM10 concentrations in Shanghai reached 114.5 and 113.3 µg m–3, respectively. In the meantime, the mean PM2.5/PM10 ratio was as low as 0.66 and 0.40 compared to that of 0.82 during the previous three days (February 17–19), clearly indicating the enhancement of coarse particles. The air mass backward trajectory analysis on these two days showed that Shanghai was mainly affected by the mixed air masses from northwest China (Fig. 1). According to the report from China Meteorological Administration, dust originated in southern Xinjiang, central and western Inner Mongolia, and Gansu, Ningxia, and Qinghai Provinces. Air quality over these areas were in light to medium pollution levels and some were even encountered heavy pollution during this period. Hence, it was clearly that air quality in Shanghai on February 20–21 was mainly affected by the long-range transported dust from northwestern China. Chemical analysis of aerosol components was further used to confirm this long-range transport dust event. The elemental Ca/Al ratio in atmospheric particulate matters had been proven to be useful for distinguishing the source regions of dust (Sun et al., 2004; Shen et al., 2007; Huang et al., 2010). Table 2 shows the typical Ca/Al ratios of the dust source regions in China and the measured Ca/Al ratios of dust samples collected in this study. The Ca/Al ratios in the dust samples, i.e., sh4, sh5, and sh6 (see sampling information in Table 1), ranged from 1.15 to 1.23, which were close to the ratio of Ca/Al from the dust source area in northwest China and corroborated the backward trajectory analysis. The high-throughput sequencing analysis results showed that the eight samples collected during different sampling periods showed good clustering effect at the OTU level (Fig. 2). The clustering results were in good agreement with the sample groups in Table 1. Due to the influence of dust, the samples in the three sampling groups had their respective similar bacterial structures, while the bacterial structures among the three sampling groups were significantly different. Compared to the samples during the non-dust periods (Group1 and Group3), the dust samples (Group2) had the shortest clustering distance among all three groups, indicating that the bacterial structure of the three samples in Group2 were the most similar. The number of mutual OTUs in all these three dust samples reached 955, accounting for 55.72% of the total OTUs of the three dust samples. The mutual OTU accounted for 69.96%, 73.24%, and 72.18% of the three dust samples, respectively, indicating the dominant OTU of the dust samples were similar. Yamaguchi et al. (2016) also found the dominant bacteria in twelve dust samples collected in Beijing were similar, and the dominant phylum were Actinobacteria, Firmicutes, and Proteobacteria. Alpha diversity index was further used to analyze the diversity of bacterial structure. The values of Coverage, Ace, Chao, Shannon, and Simpson in each group are listed in Table 3. It is shown that the gene coverage (Coverage) in each group was more than 99%, indicating the majority of the sample sequences were detected. Index of richness (Chaos and Ace) of each group were at high levels, as well as for the index of diversity (Shannon and Simpson). As shown in Table 4, the significance test suggested there were significant differences of diversity between the dust samples (Group2) and the other two groups of non-dust samples (Group1 and Group3). As a comparison, no significant differences were found between the non-dust samples. Fig. 3 shows the composition of bacterial community structure of all samples at the phylum level, showing some similarity among the three groups. Sequences in all samples were affiliated with 5 known bacterial phyla, with confidence score thresholds of 94–97%. The most abundant bacterial phylum found from Group1–Group3 was Proteobacteria (39.28%, 30.40%, and 38.22%), followed by Actinobacteria (25.02%, 28.38%, and 21.76%) and Firmicutes (12.39%, 12.36%, and 21.47%). This was similar with a study in Japan (Yamaguchi et al., 2014). The results showed that the dominant phylum were similar during the three different sampling periods. One study on a dust event observed in Japan showed that although dust acted as good carriers for abundant bacteria (around 104–105 cells m–3), the dominant bacteria at the phylum level did not change significantly in the downwind areas (Yamaguchi et al., 2012). The abundance of Actinobacteria in Group2 was the highest among all three groups, indicating the dust tended to bring higher amounts of Actinobacteria cells to the downwind areas (Yamaguchi et al., 2016). Previous studies also found that Actinobacteria was the dominant type in both Asian and African dust (Griffin et al., 2001, 2003; Yamaguchi et al., 2012) and was confirmed with the study on a long-range transported dust event in Japan (Maki et al., 2015). A total of 598, 600, and 491 genera were detected in Group1, Group2, and Group3, respectively. The dominant genera with relatively high abundances in three groups are shown in Fig. 4 and showed changes to some extent as below. Rubellimicrobium, Paracoccus, Sphingomonas, Deinococcus, and Chroococcidiopsis were the top 5 dominant genus in Group1. Deinococcus, Chroococcidiopsis, Cyanobacteria, Sphingomonas, and Rubellimicrobium were the top 5 dominant genera in Group2. As for Group3, Clostridium, Deinococcus, Paracoccus, Rubellimicrobium, and Sphingomonas were the most abundant genera. It should be noted the relative abundance of Cyanobacteria increased obviously during dust (Group 2) and then decreased after dust (Group 3), which indicated that dust acted as an efficient medium for the accumulation of Cyanobacteria. Some bacteria such as Clostridium were found increased dramatically after the dust (5.71% in Group3) compared to the samples during the dust (2.40% in Group2). This was consistent with a survey about heavy dust storm in Beijing (Yamaguchi et al., 2016). The highest abundance of Clostridium was found of the samples in Group3 (after dust). This hysteresis effect might be due to the strong adaptability of the bacteria to the harsh environment or the local dust resuspended by the strong wind. In addition, the results of the Kruskal-Wallis test showed there were significant differences in 22 bacteria among the three groups at the genus level and some of them have been identified as pathogens or potential pathogens such as Moraxella and Bryocella (Table S1). Moraxella was found increased during the dust and reached the highest abundance after the dust. The health effect of Moraxella will be discussed in Section 3.4. The Mantel test, redundancy analysis (RDA), and other statistical methods were applied to study the effect of environmental factors on the airborne bacteria. The considered environmental factors included meteorological parameters such as wind speed, temperature, and relative humidity and atmospheric pollutants such as SO2, O3, CO, NO2, PM2.5, PM10, and major aerosol chemical components. All data are listed in Table 5. The results based on the Mantel test analysis are shown in Table 6. It is shown that wind speed, PM10, and Ca2+ correlated moderately with the airborne bacteria, probably suggesting the positive effect of dust on enhancing the bacteria. In addition, SO2, SO42–, and NO3– were also moderately correlated with bacteria at the genus level. These species were indicative of anthropogenic emissions, suggesting human activities were also beneficial for the enhancement of airborne bacteria. In order to eliminate the potential influence of the internal interactions among environmental factors on the bacterial community, the partial Mantel test analysis of each environmental factor control matrix and bacterial community was made (Table 7). The results showed that the correlation between the remaining environmental factors and the bacterial community still existed. The redundancy analysis was applied to investigate the relationship between the environmental factors and the bacterial community structure. As shown in Fig. 5, during the dust period, the sh6 sample was affected mostly by PM10 while the sh4 and sh5 samples were more affected by wind speed. This was consistent with the results that dust particles tended to bring more bacteria. Also, the dusty days were usually accompanied by higher wind speed. At the genus level, the top 10 bacteria that were significantly affected by the seven environmental factors in Table 7 included Deinococcus, Rubellimicrobium, Sphingomonas, Paracoccus, Chroococcidiopsis, Cyanobacteria, Clostridium, Kocuria, Blastococcus, and Methylobacterium. The relative importance of the seven factors followed the order of wind speed, SO2, SO42–, NO3–, PM10, NH4+, and Ca2+. As for the identified top 10 bacteria, most of them can survive and associate with dust due to their strong adaptability to the environmental. For instance, Deinococcus has strong resistance to radiation, ultraviolet radiation, peroxides, and other DNA damage agents (Battista et al., 1999). Studies have shown that this unique characteristic may be related to its special genome structure and special membrane proteins (White et al., 1999; Rajpurohit and Misra, 2010; Tian et al., 2010). Cyanobacteria has been found widely in the environment (Cao et al., 2015) and has developed self-protection system during evolution (e.g., screening pigments), which facilitate it exist under extreme conditions such as cold, hot, drought or poor nutrition environment (Sinha et al., 2001; El-Bestawy et al., 2007). Sphingomonas is able to tolerate harsh environment especially the nutrition-deficient environment through the effective adjustment of its growth rate (Eguchi et al., 1996; Joux et al., 1999). Chroococcidiopsis is also widely distributed in the world as some species of it can survive drought and strong radiation (Billi et al., 2000). It should be noted that some opportunistic pathogen was found increasing during and after the dust. For example, the relative abundance of Moraxella was found increasing quickly during the dust and continue to increase after the dust. Moraxella is one of the normal flora on the skin and mucosal surface of human. Infection of Moraxella mainly occurs in people with hypoimmunity such as children, patients with tumor, chemotherapy or radiotherapy. In case of infection, Moraxella may cause arthritis, meningitis, pneumonia, endocarditis, sepsis, and keratitis (Barash and Chou, 2017; Duployez et al., 2017; Franco et al., 2017; Lee et al., 2017; Khalife et al., 2019). It was also found that the relative abundance of Cyanobacteria in dust samples was higher than that of non-dust samples, indicating that certain amounts of Cyanobacteria could be brought to the downwind areas via the long-range transport of dust. It has been found out that cyanotoxins in PM2.5 were harmful for humans by inhalation (Metcalf et al., 2012). Cyanobacterial toxins mainly cause extensive necrosis in the epithelial cells of the nasal cavity and respiratory tract by respiratory exposure. In addition, cyanobacterial toxins carried by the dust would eventually be subject to deposit and part of it may enter the water bodies of the downstream areas. Under this circumstance, the sources of drinking waters could be also polluted. Overall, the increased relative abundance of opportunistic pathogen such as Moraxella and the toxins producing bacteria such as Cyanobacteria during the transported dust may potentially bring threat on human beings. In this study, aerosol samples were collected in Shanghai before, during, and after a long-range transported dust event. Then, the airborne bacterial communities were analyzed by using high-throughput sequencing. The major conclusions are summarized below. The authors acknowledge support of National Programme on Global Change and Air-Sea Interaction (Grant No. GASI-03-01-03-01), the National Science Foundation of China (Grant Nos. 91644105 and 41506179), the National Key R&D Program of China (2018YFC0213105), and the Natural Science Foundation of Shanghai (18230722600 and 19ZR1421100). The paired-end raw Illumina sequencing data were deposited into NCBI Sequence Read Archive (SRA; http://www.ncbi.nlm.nih.gov/bioproject?LinkName=biosamplebioproject&from_uid=3273385) under accession number of SRP219061.INTRODUCTION
METHODOLOGY
Aerosol Sampling
Microbial Aerosol Sampling
Airborne Aerosol Sampling
DNA Extraction and PCR Amplification
Illumina Sequencing and Processing of Sequencing Data
Aerosol Chemical Analysis
Data Analysis
RESULTS AND DISCUSSION
Dust Events and Transport PathwaysFig. 1. Three days’ air mass backward trajectory starting at Shanghai on February 20 and 21, 2016.
Microbial Diversity and Community StructuresFig. 2. Hierarchical clustering tree of samples during the whole study period. The scale represents the relative distance among samples and is unitless.
Fig. 3. Composition of bacterial community structure at the phylum level before (sh1sh2sh3), during (sh4sh5sh6), and after (sh7sh8) the dust.
Fig. 4. Composition of bacterial community structure at the genus level.
Relationship between Environmental Factors and Airborne Bacteria
Fig. 5. RDA analysis of environmental factors and airborne bacterial communities.
The Risk of Airborne Bacteria on Human Health during the Long-range Transported Dust Events
CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
Aerosol Air Qual. Res. 20 :2834 -2845 . https://doi.org/{field 40}