Preliminary insight into the relationship between bioaerosols and urban environment obtained from the COVID-19 self-quarantine period in the Tokyo metropolitan area

Anthropogenic activities and meteorological conditions influence the composition of urban bioaerosols. The COVID-19 epidemic drastically reduced anthropogenic activities in the metropolitan areas of Japan in 2020. This study examined inter-day fluctuations in airborne bacterial and fungal compositions in Tokyo, Japan, between April and June 2020, under exceptionally unusual conditions where the movement of people and economic activity had been greatly reduced. The result showed, as expected, that local meteorological factors, especially wind and precipitation, substantially impacted emissions and changes in bacterial and fungal aerosols. However, we found that variations in the composition of urban bioaerosols could potentially be affected by anthropogenic influences, such as the concentrations of nitrogen oxides, ozone, and particulate matter, and human density. Although such factors are not as influential as wind and precipitation, it was shown for the first time that drastic changes in human activities within an area can affect the dynamics of bioaerosols. This could be a finding that should be considered for public health issues related to air quality in changing urban areas such as fast-growing cities.


Introduction
Bioaerosols, or biological aerosol particles, play a variety of roles in the atmosphere and act as pathogens (Blais-Lecours et al 2017, Humbal et al 2018), ice-nucleating particles in clouds (Hill et al 2017), and intermediaries in biogeochemical cycles via atmospheric transport (Delort et al 2017, Morris andSands 2017).In the atmosphere, microorganisms are present as bioaerosols with diameters of a few micrometers.Their dynamics are often associated with atmospheric particulate matter (PM) pollution (Shammi et al 2021, Bowers et al 2013).In urban areas, the concentrations of PM 10 (particles with diameters smaller than 10 μm) and chemical markers of bioaerosols have been reported to be higher than those in rural areas (Rathnayake et al 2016).A study in Beijing showed that most bioaerosols during haze were of soil origin, and that allergic and pathogenic microorganisms tended to increase with increase in PM pollution (Sun et al 2018).Resuspension of roadside dust due to high traffic volumes could be an essential source of bioaerosols and related allergens (Madhwal et al 2020, Miguel et al 1999).Several studies in megacities have reported correlations between air pollution and bioaerosols (Shammi et al 2021, Sun et al 2018); there might be complex interactions among air pollution, bioaerosols, and allergens.A worldwide increase in allergic diseases has been observed and linked to air pollution (Bartra et al 2007), although the exact mechanism is unknown.Therefore, the characteristics of bioaerosols modified by anthropogenic activities in urban environments are vital for considering health hazards where the population is large or rapidly growing.
Humans may also be a source of bioaerosols (Gollakota et al 2021).The human body contains a variety of bacteria, fungi, and viruses, especially those found on the skin and in the nostrils and hair, which can be released into air.In indoor air, human density and movement are significant factors related to bioaerosol dynamics (Bhangar et al 2016, Adams et al 2015, Hospodsky et al 2012).The relationship between human density/ movement and bioaerosols in outdoor air has received little attention thus far, probably because it is challenging to account for wind and other meteorological variables in an open system.However, a recent study found that urban air contains more human skin bacteria than rural air (Tanaka et al 2020).
It is difficult to assess the influence of a broad spectrum of anthropogenic activities on urban bioaerosols because these activities are in uncontrollable dynamic equilibrium in cities.However, in 2020, the COVID-19 pandemic started, and measures to suppress the spread of the infection were taken around the world, including bans on domestic and international travel and economic activity.On April 7, 2020, a state of emergency was declared in Tokyo, Japan, and, for the next 48 days, individuals were requested to refrain from going outside and engaging in economic activities.The significant shift in urban activities, of an unprecedented scale in modern times, could have had an impact on bioaerosols associated with anthropogenic activities in urban areas.Previously, only simulation studies had been able to assess the impact of such a widespread cessation of urban activity.This was a unique opportunity to observe the influence of anthropogenic elements on bioaerosols in the urban environment.
In this study, we investigated the 52-day variation in outdoor airborne bacteria and fungi in Tokyo, Japan, from April to June 2020, during and after the restriction on anthropogenic activities for the COVID-19 outbreak.Out of the observation period, 33 days corresponded to the restriction period, while 19 days corresponded to the period following the lifting of the restriction.Background factors such as local meteorology are expected to impact bioaerosols in urban areas.The role of anthropogenic variables should have been more apparent than usual, as human activities increased with deregulation.This study aimed to examine the meteorological and anthropogenic factors that affect bacterial and fungal communities in urban air and to shed light on the dynamics of bioaerosols under a range of urban human activities.Our specific focus is to discern the environmental factors linked to the daily variations and basic alpha diversity metrics, such as Shannon index, of bacterial and fungal compositions within the atmosphere.This analytical approach enables us to infer the broader alterations in airborne bacteria and fungi within urban settings during periods of lockdown/restriction. Furthermore, by identifying particular bacterial and fungal genera that exhibit concurrent fluctuations with environmental factors, we aim to elucidate potential bacterial or fungal genera capable of serving as indicators for urban atmospheric conditions.

Aerosol sampling
From April 22, 2020, to June 13, 2020, 52 days of filter sampling were conducted on the roof of Tokyo Gakugei University at Koganei in Tokyo, Japan (35°42'18.91″N,139°29'25.85″E,approximately 13 m above the ground).The collection filters were polycarbonate membrane filters (Isopore™, Millipore, 47 mm diameter, 0.2 μm pore size) set in filter holders (XX4304700, Millipore).The filters and filter holders were pretreated by autoclave sterilization (121 °C, 30 min).Collection was conducted at a flow rate of 10 l min −1 (20 °C, 1 atm) for 23 h daily from 12:00 am to 11:00 am.Three sets of the same equipment were prepared for each sample, and samples were collected in triplicates to account for collection and analytical variability.The collected samples were quickly frozen and stored at −20 °C until analysis.Details, such as the collection time for each sample, are shown in the supplementary data.

Environmental factors (meteorological and anthropogenic)
Six meteorological factors (temperature, relative humidity, pressure, total precipitation, sunshine duration, and wind speed) were obtained from the nearest meteorological observation sites managed by the Japan Meteorological Agency (JMA) and from the Atmospheric Environmental Regional Observation System (AEROS) stations managed by the Tokyo Metropolitan Government.The concentrations of suspended particulate matter (SPM, i.e., PM 10 in the Japanese environmental standard), PM 2.5, NO, NO 2 , and O x (i.e., oxidants in the Japanese environmental standard, which mainly consist of O 3 ) observed at five AEROS stations around the sampling points were averaged and used as indices of the air quality.The sum of NO and NO 2 was treated as NO x , which was also used as a parameter for the quasi-total amount of nitrogen oxides in the atmosphere.Number of people near railway stations based on GPS data from mobile phones, which were published on the Internet (Agoop Corporation 2020), were employed to assess urban activity.Six stations in Tokyo, chosen for their heavy use or proximity to tourist sites, were compared (figure S2).Two distinct patterns emerged: stations with noticeable weekday and holiday fluctuations and those with consistent increases.It was deduced that the former catered to commuters while the latter served travelers.In this study, data near Shinjuku Station (19 km east-southeast from Koganei) and Kichijoji Station (8 km east from Koganei), closest to the observation sites, were utilized as indices of commuter and traveler activity, respectively.Due to the similarities in variations among distant stations (figure S2), the fluctuations in number of people near Shinjuku and Kichijoji stations were assumed to reflect commuting and traveling in the urban area.For these reasons, the present study does not look at these stations as a source of bioaerosol but rather as a proxy of different urban activities in the entire urban area: commuting and traveling.We also used vehicle traffic volumes on the Metropolitan Expressway as published by the Ministry of Land, Infrastructure, Transport, and Tourism (Ministry of Land, Infrastructure, Transport, and Tourism 2020).Details of the location, data collection, and processing are provided in the supplementary data.

DNA extraction, PCR, and sequencing
DNA was extracted from the filters using the DNeasy PowerSoil Pro Kit (Qiagen, Hilden, Germany).After 10 min of bead-beating at 25 Hz with TissueLyser II, the tube was incubated at 56 °C for 10 min before the supernatant was transferred to a new tube and mixed with solution CD2.To increase the DNA concentration, the purified DNA was eluted using 50 μl of elution buffer (solution C6).For other processes, DNA extraction was performed according to the manufacturer's instructions.DNA extraction from blank filters (sterilized and unused) was also performed using the same procedure; however, amplification in the subsequent PCR, as described below, was not observed.The prokaryotic 16S rRNA v4 and Fungal ITS1 regions were PCR-amplified and used for bacterial and fungal DNA sequence analyses on the Illumina MiSeq platform.In total, 16,597,611 sequencing reads were obtained.The raw sequencing data were converted, demultiplexed, and denoised, and molecular identification of the obtained amplicon sequence variants (ASVs) was performed based on the naive Bayesian classifier method using the SILVA v.132 databases.In total, 11848 prokaryotes (bacterial or archaeal) ASVs and 3313 fungal ASVs were detected.
The filtered matrix was then rarefied to 1000 reads per sample.Samples with fewer than 1000 reads were discarded during this process.Out of 52 days, three replicates (156 samples) were sequenced, and the rarefied matrices of the 16S and ITS1 regions included 154 and 139 samples, respectively.The relationship between community composition and environmental factors was analyzed using non-metric multidimensional scaling (NMDS) and permutational analysis of variance (PERMANOVA), with 999 permutations with the vegan 'adonis2' function.The alpha diversities of 154 bacterial and 139 fungal samples were calculated using the Shannon diversity index.Statistical analyses were performed using the vegan 2.5.6 package of R 3.6.1.The detailed conditions of the DNA extraction, PCR, sequencing, and post-sequencing processes are described in the supplementary data.

Results and discussion
3.1.Overview of the environmental factors around the sampling location Figure 1 shows the variation in environmental factors during the observation period.There was an increasing trend in relative humidity (possibly seasonal change origin), NO concentration, NO x concentration, number of commuters, number of travelers, and traffic volume (Mann-Kendall trend test; p = 0.031, 0.016, 0.044, and 3.6 × 10 -12 , 2.1 × 10 -12 , 1.4 × 10 -5 , respectively), but not in other factors.The state of emergency in Tokyo was terminated on May 25, 2020, and it is plausible that these factors linked to anthropogenic activities (i.e., except relative humidity) showed an increasing trend.However, the traffic volume and the number of commuters and travelers had already started increasing before this termination.A possible explanation for this might be that people became careless about refraining from certain activities during the state of emergency period, and people might gradually restart going out by cars and trains during the observation period.No decreasing trend was observed for any of the examined environmental factors.

Overview of the community compositions during the observation period
The bacterial community consisted mainly of Proteobacteria, Firmicutes, and Actinobacteria, while the fungal community consisted mainly of Basidiomycota and Ascomycota (figure S3).These phyla are usually found in bioaerosols in different environments (Bowers et S3).To identify the environmental factors associated with daily variation, we statistically analyzed the bacterial and fungal assemblages in terms of dissimilarity, diversity, and the occurrence of specific genera.

Environmental factors linked to dissimilarity
The dissimilarity in community composition at the order and genus levels is shown in figure 2. As the bioaerosol-transporting air parcels always vary daily, the dissimilarity here is analogous to the daily compositional variations in airborne bacteria and fungi.The covariance relationship between daily variations and environmental factors was depicted by inserting vectors into the plots.There was no clear pattern in the distribution of the daily dissimilarities after 52 days.However, there were significant covariate relationships between the daily dissimilarity and environmental factors.Although significant environmental factors differed slightly with the resolution of the taxonomy (figures S4 and S5), the overall trend indicated that the composition fluctuated in response to environmental factors with different orientations.For bacteria, the wind speed vector The environmental factors that could explain the bacterial and fungal compositional variations (p < 0.05) were deduced using PERMANOVA (tables S2 and S3).The significant factors changed slightly depending on the taxonomic resolution.Taking the results of fungi as an example (table S3), when classifying fungi at the phylum level, four factors were detected, whereas when classified at the genus level, ten factors were detected.This demonstrates that the results of the PERMANOVA vary depending on the taxonomic level at which the aggregation is performed.Therefore, we examined the factors that were commonly significant at all taxonomic levels.For bacterial compositional variation, wind speed, relative humidity, precipitation, sunshine duration, NO 2 concentration, NO x concentration, the number of commuters, and the number of travelers were commonly significant.Relative humidity, precipitation, temperature, and O x concentration were found to be significant for fungal compositional variation at all taxonomic levels.The results indicate that daily variations in bacteria are more susceptible to anthropogenic influences than those in fungi.This difference is likely due to the distinct sources of origin for fungi and bacteria.Fungi found in the atmosphere are mainly plant pathogens and are linked to plant-related sources (Chen et al 2021), while bacteria are believed to have a wider range of sources, including humans and animals.

Environmental factors linked to diversity
Wind speed was positively correlated with bacterial alpha diversity, while relative humidity was negatively correlated (r = 0.4322 and −0.3906, table 1).A previous study conducted at a 458 m-high tower in Tokyo reported similar results, which concluded that wind supplied a variety of bacteria from the ground surface to the atmosphere, while humidity suppressed the suspension from the ground surface (Uetake et al 2019).Precipitation was positively correlated with the fungal alpha diversity (r = 0.2981, table 1), although no such correlation was observed with relative humidity.Some fungal species of Ascomycota and Basidiomycota release spores upon precipitation and humidity (Elbert et al 2007).These increases, along with alpha diversity, suggest that these meteorological factors could be initiators of bioaerosol generation.The concentrations of NO 2 and NO x exhibited a negative correlation with bacterial alpha diversity (r = −0.3468and −0.3712), while the concentration of PM 2.5 displayed a negative correlation with fungal alpha diversity (r = −0.4127).Furthermore, the concentration of SPM showed a negative correlation with both bacterial and fungal alpha diversity (r = −0.2806and −0.3499).NO concentration, which tended to increase during the observation period, did not display a significant correlation with microbial diversity.Although several studies have reported that air pollutants are associated with airborne bacteria and fungi, the process is still not fully understood (Ruiz-Gil et al 2020, Kathiriya et al 2020).

Environmental factors linked to specific genera
The identified genera that increased in response to the environmental factors are shown in table S4.Notably, Tumebacillus increased with increasing PM 2.5 .Tumebacillus has been reported to increase significantly during severe haze in Beijing (Yan et al 2018), thus possibly being a microorganism concurrent with PM 2.5 .Alternaria and Erysiphe increased with the PM 2.5 , SPM, and O x concentrations.They are often reported to be associated with the atmospheric environment: Alternaria is reportedly transported long distances in the atmosphere (Skjøth et al 2016) and is predominant in polluted air (Yan et al 2016, Chen et al 2021); Erysiphe is reportedly found in the atmosphere throughout the year and to have a long survival time (Chen et al 2021, Carisse et al 2009).These fungal genera were also associated with temperature and O x concentration.Although the mechanism by which they increase has not been determined, Alternaria and Erysiphe may be representative airborne fungi and essential groups that reflect air quality.It also should be noted that Abundisporus was found to increase in response to NO concentrations.To the best of our knowledge, there are no prior reports focusing on Abundisporus as a bioaerosol.Additionally, Abundisporus showed an increase in response to not only NO but also temperature, atmospheric pressure, and sunlight duration, suggesting a potential relationship with boundary layer development.However, further data are required to verify this hypothesis.Five bacterial genera (Marmoricola, Methylosinus, Nocardioides, Sphyingomonas, Conexibacter) increased with increasing wind speed, suggesting that these genera were specifically suspended by the wind.In contrast to bacteria, more genera were identified for fungi, but there was no clear distinction between the anthropogenic and meteorological factors.
Many fungal genera increased in abundance with increasing humidity.It is plausible that high humidity favors fungal aerosolization, as shown by alpha diversity analysis.

Conclusions
To summarize this study, we have listed the parameters of the bioaerosol composition (i.e., their variability, diversity, and increase in specific genera) related to each environmental factor (table 2).Environmental factors that are linked to two or more bioaerosol characteristics are more likely to have a significant impact on bioaerosols.Only wind speed and precipitation were associated with all three bioaerosol parameters.Wind speed can release highly diverse bacterial communities from the ground surface into the atmosphere, and precipitation can release highly diverse fungal communities by stimulating fungal spore release mechanisms.From a common-sense perspective, it might be expected that high winds lead to an increase in fungal genera in the atmosphere.However, fungal genera related to wind speed were not detected.One possible explanation is that fungi, unlike bacteria, possess an effective mechanism for actively releasing spores in response to precipitation, which significantly diminishes the influence of wind speed.Furthermore, since this study focused on the detection at the genus level, it is conceivable that fungi may be detected when analyzed at coarser taxonomic levels.In any case, based solely on the results of this study, it is highly likely that precipitation is a factor related to the release process of fungi, and wind speed is a factor associated with the suspension process of bacteria in the atmosphere.
On the other hand, nitrogen oxides and ozone are generally not considered to be associated with bioaerosol emissions, because chemical reactions generate the nitrogen oxides and ozone.Ozone and nitrogen oxides are known to react with proteins and other organic matter in the atmosphere (Liu et al 2017), and similar processes may influence the variation or diversity of bioaerosols in urban areas.It is necessary to clarify the process of the covariation between air pollution and bioaerosols in the future because air pollution can modify the health risk levels of bioaerosols through the process.
For the first time, the number of people going outside was detected as a factor affecting bacterial aerosol variations in the outdoor air.Interestingly, fungal aerosol variations were not associated with human movement or density, and this was a significant difference from bacteria.Presumably, the source of the fungi is in places unrelated to human bodies, such as trees and soil, and is therefore basically dependent on meteorological conditions.The notable aspect of the current study was the unique situation of the urban area during the observation period: people density outdoors, traffic volume, and the concentrations of nitrogen oxides increased owing to the deregulation of urban activity controls to prevent the COVID-19 epidemic.In addition, analysis of samples for continuous 52 days revealed the variation in bacterial and fungal aerosols without the influence of seasonal variations in environmental factors.At the preliminary stage, this condition allowed us to demonstrate the background relationship between humans and bioaerosols in an urban area.
Although the variation in bacterial composition was influenced by the number of commuters and travelers, we did not find any bacterial genera that were linked to the number of people.Precipitation and sunshine duration were also related to variations in the bacterial composition (table 2).Historically, precipitation and sunshine duration have been considered significant contributors to the concentration of bacteria in the atmosphere (Jones andHarrison 2004, Joung et al 2017).Similarly, the number of people outdoors will need to be considered as a factor contributing to the concentration of bacterial aerosols, which we did not measure in this study.
The findings of this study demonstrate for the first time the potential importance of outdoor population density as a background factor in bioaerosol dynamics in urban areas.In addition, air pollution affects both bacterial and fungal composition.In future studies on urban bioaerosols, it is important to consider anthropogenic activities affecting the processes of bioaerosols in the urban atmosphere, which will help improve public health in urban areas and provide a basis for responding to atmospheric environmental problems or pandemics.

Figure 1 .
Figure 1.Variation in daily average values of (1) the number of people at the stations and traffic volume on the Metropolitan Expressway, (2) gaseous pollutant concentrations, (3) particulate pollutant concentrations and (4 and 5) meteorological conditions during the observation period.The vertical dotted lines in the figure indicate May 25, 2020, when the emergency declaration was lifted.

Figure 2 .
Figure 2. Dissimilarity in bacterial and fungal communities by order and genus phylogenetic classifications: (a) bacteria and (b) fungi at the order level and (c) bacteria and (d) fungi at the genus level.The orientation of the vectors shows the trend toward increasing environmental factors, which are covariant with the dissimilarity in community composition (p < 0.05).
al 2013, Xu et al 2017, Fröhlich-Nowoisky et al 2016), indicating no noticeable difference in composition from the previously published literature.Both bacterial and fungal community compositions showed slight daily variations (figure

Table 2 .
Summary of the environmental factors related to bioaerosol variation (i.e., dissimilarity), diversity, and specific genera occurrence.The plus sign means increase, and the minus sign means decrease.