key: cord-0014260-sjv1zhj8 authors: Robinson, Jake M.; Cando-Dumancela, Christian; Liddicoat, Craig; Weinstein, Philip; Cameron, Ross; Breed, Martin F. title: Vertical Stratification in Urban Green Space Aerobiomes date: 2020-11-25 journal: Environ Health Perspect DOI: 10.1289/ehp7807 sha: bc1635c29ad40e92f6dd755bfecb9f8280a18b6e doc_id: 14260 cord_uid: sjv1zhj8 BACKGROUND: Exposure to a diverse environmental microbiome is thought to play an important role in “educating” the immune system and facilitating competitive exclusion of pathogens to maintain human health. Vegetation and soil are key sources of airborne microbiota––the aerobiome. A limited number of studies have attempted to characterize the dynamics of near surface green space aerobiomes, and no studies to date have investigated these dynamics from a vertical perspective. Vertical stratification in the aerobiome could have important implications for public health and for the design, engineering, and management of urban green spaces. OBJECTIVES: The primary objectives of this study were to: a) assess whether significant vertical stratification in bacterial species richness and evenness (alpha diversity) of the aerobiome occurred in a parkland habitat in Adelaide, South Australia; b) assess whether significant compositional differences (beta diversity) between sampling heights occurred; and c) to preliminarily assess whether there were significant altitudinal differences in potentially pathogenic and beneficial bacterial taxa. METHODS: We combined an innovative columnar sampling method at soil level, 0.0, 0.5, 1.0, and [Formula: see text] , using passive petri dish sampling to collect airborne bacteria. We used a geographic information system (GIS) to select study sites, and we used high-throughput sequencing of the bacterial 16S rRNA gene to assess whether significant vertical stratification of the aerobiome occurred. RESULTS: Our results provide evidence of vertical stratification in both alpha and beta (compositional) diversity of airborne bacterial communities, with diversity decreasing roughly with height. We also found significant vertical stratification in potentially pathogenic and beneficial bacterial taxa. DISCUSSION: Although additional research is needed, our preliminary findings point to potentially different exposure attributes that may be contingent on human height and activity type. Our results lay the foundations for further research into the vertical characteristics of urban green space aerobiomes and their implications for public health and urban planning. https://doi.org/10.1289/EHP7807 Vegetation and soil are known to be key sources of airborne microbiota--i.e., the aerobiome (Joung et al. 2017; Liu et al. 2018 ). Exposure to a diverse suite of microbes from the environment (including the aerobiome) is thought to be important for the development and regulation of the human immune system (Rook et al. 2003 (Rook et al. , 2013 Tasnim et al. 2017) . Furthermore, studies now link the microbiome to a plethora of maladies from Alzheimer's disease (Kowalski and Mulak 2019) and myalgic encephalomyelitis (Hanson and Giloteaux 2017) , through inflammatory bowel (Aschard et al. 2019 ) and skin diseases (Prescott et al. 2017) , to respiratory health (Sokolowska et al. 2018) . Environmental factors are thought to be more important than genetic factors in shaping the composition of the gut microbiome (Rothschild et al. 2018) . Indeed, Browne et al. (2016) showed that spore-forming bacteria (which survive in aerobic conditions) dominated the human gut, comprising 50%-60% of bacterial genera, and displayed greater change in abundance and species over time in comparison with nonspore formers, suggesting that many gut bacteria may come and go from the environment. Gut colonization aside, exposure to airborne microbiota has implications for the human skin and airways. For example, several studies (particularly in agricultural settings) have demonstrated that the composition of the human nasal microbiome is significantly influenced by airborne microbial communities from the surrounding environment (Shukla et al. 2017; Kraemer et al. 2018) . A recent study also showed that the diversity of skin and nasal microbiota increased after exposure to urban green spaces (Selway et al. 2020) . Furthermore, a recent systematic review highlights that despite the relative infancy of aerobiome-human health research, two studies have shown that rural aerobiomes shifted immune function away from allergic (Th2-type) responses (Flies et al. 2020 ). In the indoor environment, studies have also drawn the link between microbial composition and endotoxin levels in dust and immuno-protection (e.g., against asthma) (Gehring et al. 2008; Stein et al. 2016) . Other indoor-based studies show airborne microbes contribute to nasal, oral, and skin microbiomes (Lai et al. 2017; Chen et al. 2019) . Studies have also shown that up to 10 6 microbial cells can be found in a cubic meter of air (Šantl-Temkiv et al. 2018; Tignat-Perrier et al. 2019) . Therefore, there is considerable potential for aerobiomerespiratory system interactions. A limited number of studies have attempted to characterize the community structure and spatiotemporal dynamics of near-surface green space aerobiomes. For example, Mhuireach et al. (2016) compared bioaerosol samples in green spaces and parking lots and found compositional distinctions in bacterial communities between the two land cover types. Furthermore, Mhuireach et al. (2019) explored spatiotemporal controls on the aerobiome and suggested that localized site factors were likely to be important in driving bacterial community structure. However, no known studies have investigated the spatial and compositional factors from a vertical perspective. Support for the existence of aerobiome vertical stratification can be drawn from studies of pollution, allergenic pollen, and fluid dynamics of particulates, where stratification has been shown to occur at various scales. For example, in an indoor agricultural environment and under ventilated conditions, Miles (2008) showed that NH 3 molecule concentrations decreased vertically with increasing distance from source (i.e., the ground). Gao and Nui (2007) found that vertical concentration stratification of particles up to (10:0 lm) occurred under different ventilation conditions. Particles smaller than 2:5 lm were less affected by gravitational factors, and submicron particles with small relaxation times (i.e., the time required for particles to adjust their velocity to new conditions of forces) behaved more like trace gases following main airstreams. Alcázar et al. (1998) found higher concentration of Urtica membranacea pollen at the upper region of their sampling height range of 1:5 m-15 m and higher concentrations of U. urens-Parietaria sp. at lower heights, possibly due to pollen mass and different fluid dynamics. The size range of bacterial cells can vary by eight orders of magnitude (0:013 lm to 750 lm) (Levin and Angert 2015) and can clump together and adhere to larger suspended particles (Tham and Zuraimi 2005; Haas et al. 2013; Gong et al. 2020) . These factors, along with turbulent mixed flow could conceivably influence aerobiome stratification. The existence of aerobiome vertical stratification could have important implications for the design, engineering and management of urban green spaces, particularly those aimed at promoting public health via microbial exposure (Watkins et al. 2020) . For example, do children receive the same exposure to airborne microbiota as taller adults? Do people who lie down or work close to the ground (e.g., gardeners bending over to dig) have different exposure levels to those who remain upright, and what are the downstream implications for health? Developing a refined understanding of this aerobiome-human interface could also have implications for the design and monitoring of nature-based health interventions, for example via green/nature prescribing (Robinson and Breed 2019; Shanahan et al. 2019; Robinson et al. 2020) . Furthermore, protocols for sampling the aerobiome to date have often included a reasonable yet arbitrary sampling height of 2 m (Airaudi and Marchisio et al. 1996; Cordeiro 2010; Mhuireach et al. 2016; Domingue 2017) . Therefore, investigating aerobiome composition at various heights could also provide important methodological insights to fine-tune future study protocols and public health recommendations. In this proof of concept study, we combine innovative columnar aerobiome sampling methods along with remote sensing techniques and high-throughput sequencing of the bacterial 16S rRNA gene. The primary objectives of this study were to: a) assess whether significant vertical stratification in bacterial species richness and evenness (alpha diversity) of the aerobiome occurred; b) assess whether significant compositional differences (beta diversity) between sampling heights occurred; and c) to preliminarily assess whether there were significant altitudinal differences in putative pathogenic and beneficial bacterial taxa. Our study site comprised three vegetated plots totaling 7 hectares (ha) of the southern section of the Adelaide Parklands (Kaurna Warra Pintyanthi), South Australia. The justification for the selected study site was as follows: 1. Its broadly consistent soil geochemistry, because the southern Parklands generally fall within the Upper Outwash Plain soil boundary (coalescing alluvial soil, draining the Eden Fault Block). 2. This area is managed by a single division of the City of Adelaide, minimizing variation in site management and allowing for simpler study logistics. 3. A single study site (i.e., the southern section) in the Parklands provided a degree of control over potential variation in landscape effects on the aerobiome (e.g., dominant vegetation type, distance to coast, elevation, orientation, aspect). 4. Urban Parkland is representative of conditions that both child and adult residents might be exposed to. Following site selection, boundaries of three plots (as polygons) were defined in QGIS 3 (version 3.0.2). These polygons were subsequently converted to shapefiles (.shp), and a random point algorithm was generated. This process provided randomly selected sampling points within each vegetated plot to include in our study (Figure 1 ). The spatial coordinates for each sampling point were recorded and programmed into a handheld global positioning system (GPS) device. This device was operated on site to allow us to identify the relevant locations for setting up the sampling stations. The sampling stations ( Figure 2 ) were constructed using timber (SpecRite 42 mm × 28 mm × 2:7 m screening Merbau). The sampling stations comprised a timber stand with 45 leg braces. Hooks and guy ropes were also installed, ensuring stability in the field. We used standard lab-grade, clear plastic petri dishes (Nest Cell) supported by steel brackets (and attached to the brackets with Velcro tabs) to passively sample the aerobiome per Mhuireach et al. (2016) . The level of stability was tested in two phases. Phase 1: during windy conditions ( ∼ Beaufort scale No: 5) in a yard environment, and Phase 2: in situ, prior to the sampling phase. We installed temperature and relative humidity data loggers (Elitech RC-4HC) at each sampling station. Each logger was programmed to record data at 8-s intervals for the entire sampling period. The dataloggers were calibrated at the start of each sampling day using a mercury thermometer (Gerotherm) and a sling psychrometer (Sper Scientific 736700) taking the range between the two bulbs to determine baseline humidity. The sampling stations were placed into position between 0600 hours and 0800 hours on 4, 5, and 6 November 2019. This procedure ensured that sufficient time was allocated to travel between the sampling locations. From 0800 hours onward and prior to installing the petri dishes for passive sampling, the sampling stations were decontaminated using a 5% Decon 90 solution. The microclimate data loggers were then decontaminated and installed on the sampling stations. The nearest trees (all <10 m height and 20 cm-50 cm in diameter at breast height) were between 2 m and 5 m from the sampling stations. The sampling procedure involved collecting soil samples (actively) and airborne microbiota (passively). Environmental metadata were also collected (e.g., wind speed, temperature, and relative humidity). Soil pH at each site was measured using a digital pH meter (Alotpower). The probe of the pH meter was inserted into the soil and left for a period of 1 min prior to taking a reading, per manufacturer's instructions. Wind speed and direction data for the entire study area were obtained from Adelaide's meteorological weather station at Ngayirdapira (West Terrace): Lat: −34:93, Lon: 138.58, Height: 29:32 m. Wind speed and direction was also recorded at each sampling site on an hourly basis (Mhuireach et al. 2016 ) using the handheld anemometer (Digitech QM-1644). Soil samples. Topsoil samples were collected using a small shovel and stored in 50-mL sterile falcon tubes. The shovel was decontaminated using the 5% Decon 90 solution prior to use. Wearing gloves, we sampled five topsoil samples (depth: 5-7 cm) at equidistant sampling points, 20-30 cm from the central stem of each sampling station (Zarraonaindia et al. 2015) . The soil samples were subsequently pooled and then homogenized, passed through a 1-mm pore sieve, and placed in new sterile 50-mL Falcon tubes. The sample tubes were labeled using a predefined labeling system. We included field controls of soil samples by opening 50-mL sterile falcon tubes for 60 s at each site (Mbareche et al. 2019) . All soil and field control samples were immediately chilled by placing in an ice box in the field and then storing at −80 C in the lab prior to DNA extraction and sequencing (Zarraonaindia et al. 2015) . In total, we collected 15 soil subsamples per sampling day across the three sampling stations for each of the three sampling days. Subsamples were pooled and homogenized by sampling station and day, which gave a total of nine homogenized samples (three per sampling station) plus three field controls. Aerobiome samples. Passive sampling methods were used to collect low biomass aerobiome samples following established protocols (Mhuireach et al. 2016 (Mhuireach et al. , 2019 . Petri dishes (100 × 15 mm) were attached with decontaminated Velcro tabs on the sampling stations at four sampling heights: ground level, 0:5 m, 1 m, and 2 m. The total height of the sampling stations was 2 m from ground level (in a cohort study across Europe, North America, Australia, and East-Asia, 95% of adult human heights fell within 2 SD at 1:93 m for males and 1:78 m for females) (Jelenkovic et al. 2016 ). One meter is the average height of a 4-y-old child (RCPCH 2020)typically the maximum weaning age (Mutch 2004; Clayton et al. 2006) and the time when the gut microbiome is thought to become less plastic (Milani et al. 2017 )--and is the approximate height of a jogging stroller (Thule 2020). Fifty centimeters is the approximate height of an adult torso from the hip to the mouth (representing the height of an adult sitting on the floor) (Nikolova et al. 2017) , although this will vary depending on size and age. The ground surface is also considered to be an important sampling level, for example, representing the point of contact for a crawling child or an adult lying on the floor. The petri dish sampling plates were also decontaminated using the 5% Decon 90 solution prior to use. The petri dishes were secured to the sampling stations ( Figure 2 ) and left open for 6-8 h (Mhuireach et al. 2016) . At the end of the sampling period, we closed the petri dishes. A new set of gloves was worn for the handling of petri dishes at each vertical sampling point to reduce potential contamination. The petri dishes were then sealed using Parafilm, labeled, immediately placed on ice, and transported to the laboratory for storage at −80 C prior to DNA extraction (Mhuireach et al. 2019) . Unused petri dishes were left open for 60 s in the equipment box carried on site and then sealed at each site as field controls. Dishes were later swabbed during the DNA extraction process using nylon flocked swabs (FLOQSwabs Cat. No. 501CS01, Copan Diagnostics, Inc.) (Mhuireach et al. 2019; Bae et al. 2019; Liddicoat et al. 2020 ). We extracted DNA from samples at the Evolutionary Biology Unit (EBU), South Australian Museum. The order of processing samples was randomized using a digital number randomizer, including the soil samples (higher biomass), which were processed after the low biomass aerobiome samples to minimize cross-contamination. The petri dishes for each sampling station were swabbed with FLOQSwabs for 30 s (with consistent back and forth strokes) in a laminar flow cabinet type 1 (License No. 926207). The base and lid samples for each height, station, and date were then pooled, prior to extraction. The swabs were cut with decontaminated scissors directly into labeled 2-mL Eppendorf tubes. Extraction blank controls were used to demonstrate the absence of sample contamination during extraction and were the last samples in the extraction. Sterile water and reagents were used instead of a sample, and all DNA extraction steps were performed as if they were normal samples. We used Qiagen QIAamp DNA Blood Mini Kits to extract DNA from the swabs together with extraction blank controls, and Qiagen DNAeasy PowerLyzer Soil Kits to extract DNA from the soil samples (and extraction blank controls). We followed the manufacturer's instructions throughout the extraction process. Polymerase chain reaction (PCR) amplification was done in triplicate using the 341F/806R primer targeting the V3-V4 region of the 16S rRNA gene (5 0 -CCTAYGGGRBGCASCAG-3 0 /5 0 -GGACTACNNGGGTATCTAAT-3 0 ). The 300-bp paired end run was sequenced on an Illumina MiSeq platform at the Australian Genome Research Facility Ltd (AGRF) using two flow cells (ID 000000000-CW9V6 and 000000000-CVPGT). Image analysis was done in real time by the MiSeq Control Software (MCS) (version 2.6.2.1) and Real Time Analysis (RTA) (version 1.18.54). Then the Illumina bcl2fastq 2.20.0.422 pipeline was used to generate the sequence data. A minimum of 0:20 ng=uL of usable PCR product was required to generate sequencing output guarantee of 10,000 raw reads and to be included in the analysis. Paired-end reads were assembled by aligning the forward and reverse reads using PEAR (version 0.9.5). Primers were identified and trimmed. Trimmed reads were processed using Quantitative Insights into Microbial Ecology (QIIME 1.8.4), USEARCH (version 8.0.1623), and UPARSE software. Using USEARCH tools, reads were quality filtered; full-length duplicate reads were removed and sorted by abundance. Singletons or unique reads in the data set were discarded. Reads were clustered and chimeric reads were filtered using the "rdp_gold" database as a reference. To obtain the number of reads in each operational taxonomic unit (OTU), reads were mapped back to OTUs with a minimum identity of 97%. Taxonomy was assigned using QIIME. We used the phyloseq (version 1.24.0) package (McMurdie and Holmes 2013) in R to import and analyze the sequencing data, and decontam (version 1.1.2) ) to identify and exclude contaminants. Lower biomass samples (i.e., air, field blanks, and extraction blank controls) were analyzed using the isNotContaminant() function, where contaminants were identified by increased prevalence in negative controls. Higher biomass samples (i.e., soil, and corresponding extraction blanks) were analyzed using the isContaminant() function. Using isContaminant(), contaminants were identified by the frequency that varies inversely with sample DNA concentration, or by increased prevalence in negative controls. All taxa identified as contaminants were pooled and removed from further analysis. To estimate OTU alpha diversity, we derived Shannon Index values based on rarefied abundances (Liddicoat et al. 2020) in phyloseq. The lowest number of reads in a sample was used to rarefy the data sets (Liddicoat et al. 2020) . We generated box and violin plots with ggplot2 (version 3.0.0) to visualize the distribution of the alpha diversity scores for each sampling height. Microbial beta diversity was visualized using nonmetric multidimensional scaling (NMDS) ordination of Bray-Curtis distances based on rarefied OTU abundances. The ordinations plots show low-dimensional ordination space in which similar samples are plotted close together, and dissimilar samples are plotted far apart. We used permutational multivariate analysis of variance (PERMANOVA) to test for compositional differences between sampling heights. The Pearson's product-moment and Spearman's rank correlation tests were used to examine correlations between sampling height and alpha diversity scores. A Mann-Whitney-Wilcoxon test was used to examine differences in alpha diversity between merged air sampling heights (0:0-0:5 m and 1:0-2:0 m) and a Kruskal-Wallace chi-square test to explore differences in correlations between sites and dates. We also calculated OTU relative abundances using the phyloseq package in R to examine the distribution of taxa that have potential implications for public health. To compare presence and proportions of taxa, we used two-sample tests for equality of proportions with continuity corrections and created radial charts using pivot tables with comma separated value (csv) files. A data point was considered to be an outlier if it was more than 1:5 × above the third quartile or below the first quartile. We obtained 3,781,284 raw reads from air samples with an average length of 300 base pairs and 3,278,433 reads after quality control (QC). For soil, we obtained 1,830,395 raw reads and 1,287,303 reads after QC. The range of reads per samples after QC was 19,966-251,822. Reads were clustered into 10,563 OTUs. Overall, bacterial communities were diverse at each sampling height and bacterial phyla were dominated by: • Proteobacteria ( Ten bacterial phyla represented 100% of OTUs over 1% relative abundance, including: Proteobacteria, Actinobacteria, Bacteriodetes, Chloroflexi, Cyanobacteria, Firmicutes, Gemmatimondates, GN02, OD1, and TM7. Proteobacteria were dominant at upper sampling heights, and Actinobacteria were dominant at lower sampling heights. We observed a significant negative correlation between alpha diversity (air and soil for all sites/dates) and sampling height [r = − 0:58, degrees of freedom ðdfÞ = 38, p = < 0:01; Figure 3A ; Table 1 ]. Alpha diversity ranged from 1 to 6 and was highest at soil level followed by the lower air sampling levels (0:0 m-0:5 m) and the upper sampling levels (1:0 m-2:0 m), respectively. When the lower sampling heights and the upper sampling heights were merged (0.0 with 0:5 m; 1:0 m with 2:0 m), we observed a significant negative correlation between alpha diversity and sampling height (r = − 0:68, df = 38, p = < 0:01) ( Figure 3B ). Following an examination of alpha diversity scores for individual sites and dates, all variants showed negative correlations between alpha diversity and sampling height. Four out of six indicated strong and significant relationships (Day 1: r = − 0:76, p = 0:00; Day 3: r = − 0:64, p = 0:01; SC01: r = − 0:68, p = < 0:01; and, SC03: r = − 0:73, p = 0:01; Table 2 ). It is important to note that we omitted six samples from the lower heights due to failure to reach minimum DNA concentrations (as denoted by "-" in Table 1 ). With the merged sampling heights, all correlations increased in strength and were all statistically significant (Table 2) . A Mann-Whitney-Wilcoxson test for differences in alpha diversity between the merged air sampling heights (0:0 m-0:5 m and 1:0 m-2:0 m) showed a statistically significant difference (W = 188, p = < 0:01). A Kruskal-Wallace chi-squared test indicated no significant difference in correlations between sites or dates (p = 0:44). Using these same merged sampling heights, a two-sample test for equality of proportions with continuity correction showed a significant difference in proportions of taxa that occurred in lower air sampling heights (compared to upper sampling heights) that also occurred in the soil samples. The positive relationship between the proportion of taxa occurring in the air that also occurred in the soil decreased as vertical distance from the soil increased. For example, at the genus level, 84.4% of taxa in the lower air samples also occurred in the soil samples, whereas only 76.1% of the taxa in the upper air samples occurred in the soil. This difference was statistically significant (v 2 = 9:5376, df = 1, p = < 0:01; Figure 4 shows taxonomic breakdown). Sampling heights displayed distinct bacterial signatures ( Figure 5A ). Sampling height explained 22% of the variation in environmental microbiota when all air sampling heights and the soil level were included, and this was statistically significant (PERMANOVA df = 4, F = 2:50, R 2 = 0:22, p = < 0:01, permutations = 999). When analyzing air samples in isolation, sampling height explained 11% of the variation in environmental microbiota, however, this was not significant (df = 3, F = 1:18, R 2 = 0:11, p = 0:15, permutations = 999). When we merged within lower and upper sampling heights, sampling heights explained 6% of the variation and this was statistically significant (df = 1, F = 1:98, R 2 = 0:06, p = 0:01, permutations = 999) ( Figure 5B ). The dominant taxa in the soil and lower sampling heights were Actinobacteria (based on mean relative abundance >1%), and the dominant taxa in the upper sampling heights were Proteobacteria ( Figure 6 ; segments 1 and 9). A significantly greater proportion of Actinobacteria were present in lower air sampling heights (merged 0:0 m-0:5 m; 43.52% and 26.61%, respectively; x = 35:07%) compared to upper air sampling heights (merged 1:0 m-2:0 m; 17.52% and 19.67%, respectively; x = 18:59%) (v 2 = 6:1032, df = 1, p = 0:01). A significantly greater proportion of Proteobacteria was present in the upper air sampling heights (merged 1:0 m-2:0 m; 43.78% and 49.50% respectively; x = 46:64%) compared to the lower air sampling heights (merged 0:0 m-0:5 m; 27.11% and 28.14%, respectively; x = 27:63%) (v 2 = 6:9471, df = 1, p = < 0:01). A number of relatively abundant and notable taxa (contingent primarily on their implications for public health) were identified in the samples (Figure 7) . The relative abundance of these taxa differed across sampling heights, and all significantly correlated with sampling height, ranging from moderate to strong relationships (Table 3 ). The relative abundance of these taxa are as follows: Streptomyces (3.63% and 3.7% in soil and 0:0 m, respectively), Kingella (2% and 4.1% in 1:0 m and 2:0 m, respectively), Lactobacillus (5.9% and 3.8% in 1:0 m and 2:0 m, respectively), Flavobacterium (4.3% in 0:0 m, 7.5% in 0:5 m, 7.9% in 1:0 m, and 4.8% in 2:0 m), and Sphingomonas (4.3% in 0:0 m, 4.8% in 0:5 m, 6.5% 1:0 m, and 6.8% in 2:0 m). The potential implications of these taxa for public health are highlighted further in Table 4 in the Discussion. Here we show that vertical stratification of aerobiome alpha diversity occurred in an urban green space scrub habitat in Adelaide Parklands, South Australia. This transpired as a significant association in the reduction of bacterial alpha diversity as height increased (i.e., between the ground surface level and two vertical meters of the air column). When considering all sampling heights, alpha diversity reduced with greater height. This vertical stratification in alpha diversity was neither spatially (i.e., site specific) or temporally dependent. The strength of the negative relationship between alpha diversity and height increased when we merged lower sampling heights (0:0 m with 0:5 m) and the upper sampling heights (1:0 m with 2:0 m). This finding implies that the required spatial frequency to elucidate vertical stratification in alpha diversity--specifically, five sampling heights across a 2-m vertical transect--may have been overestimated. However, several omissions in the lower sampling heights due to failure to reach minimum DNA concentrations could have affected the strength of this association. The decay in observed alpha diversity as height increased could be the result of increasing distance from the primary source, i.e., potentially the soil. It is widely accepted that soil represents one of the most microbially diverse terrestrial habitats (Briones 2014; Bender et al. 2016; Dumbrell 2019; Zhu et al. 2019 ). Therefore, it seems reasonable to suggest that lower sampling heights may possess a higher level of microbial diversity because they are closer to a potentially greater concentration of microbiota. We observed that a greater proportion of bacteria taxa found in the lower sampling heights (in comparison with the upper sampling heights) were also present in the soil samples, Note: -, missing data (failed to reach minimum DNA concentrations: 0:20 ng=uL of usable PCR product was required to generate sequencing output of 10,000 raw reads); * A data point was considered to be an outlier if it was more than 1:5 × above the third quartile or below the first quartile. Scrub 1, 2, and 3 refer to samples collected from the scrub habitat study sites. SD, standard deviation. . Relative abundance of bacterial operational taxonomic units (OTUs) at the phylum taxonomic level (based on mean relative abundance >1% for each sampling height). Ring segments relate to phyla via the number key on the right; segment size corresponds to mean relative abundance across all heights; mini bar charts relate to relative abundance of taxa for individual sampling heights where applicable. Actinobacteria (1) dominate lower sampling heights (indicated by the darker colored bars), Proteobacteria (9) dominate upper sampling heights (indicated by the lighter colored bars). Table 4 for potential public health implications of notable taxa. Reference numbers within the relative abundance bars correspond to the number key and notable taxa displayed in the upper-right pane. both at genus and family levels. Together, these results suggest that soil does appear to play a key role in supplementing the local aerobiome, particularly at lower heights. The presence of vertical stratification of bacterial diversity in the aerobiome could have important implications for human health. Indeed, exposure to environmental microbes is thought to prime and "educate" the immune system (Belkaid and Hand 2014; Hanski 2014; Mincham et al. 2020 ) particularly in early life, and a recent mouse study suggests that exposure to environmental microbes such as the butyrate-producer Kineothrix alysoides could also have anxiolytic (anxiety-reducing) effects (Liddicoat et al. 2019) . The vertical stratification concept could also be important for exposome researchers, who investigate the types and methods of exposures to both endogenous and exogenous chemical composites (including microbes and their biological compounds across the life course) (Escher et al. 2017; Daiber et al. 2019; McCall et al. 2019) . The presence of vertical stratification implies that the potential for exposure to environmental microbial diversity may differ throughout the human life course due to age and gender differences in height, activity types, and methods of motion. However, our static experimental conditions fail to capture the dynamics of human movement and activity within and between environments. Further research is required to understand how vertical stratification may affect human colonization, with particular focus on the dynamic nature of human movement through environments. Additional research into aerobiome stratification could lead to improved design and management of three-dimensional urban structures and vegetation assemblages, which may influence aerobiome dynamics. In the future, this research could lead to ways of optimizing human-environmental microbe interactions. Humans are spending more time indoors (Ergan et al. 2019 ). Therefore, future aerobiome studies should also consider whether vertical stratification occurs indoors and consider the relative influence of the outdoor environment and the potential health implications of these dynamics. Understanding how patterns of human behavior influence exposure to airborne microbiota will also be important. For example, ongoing changes to commuting, recreation, and living environments may have important implications for aerobiome characteristics and exposure potential. We also showed vertical stratification of aerobiome beta diversity, where sampling height explained 22% of the variation in environmental microbiota when all sampling heights were included. This variation was corroborated by the analysis of equality of taxonomic proportions between the air and the soil samples. As mentioned, the proportion of bacterial taxa from the air samples that were also present in the soil decreased as altitude increased. This finding provides preliminary evidence that soil has a stronger influence on aerobiome composition at lower heights, and allochthonous sources make a key contribution to the aerobiome higher up. It is likely that distance to source makes a key contribution to aerobiome vertical stratification. However, there may be other important biophysical driving factors. For example, the size range of bacterial cells can vary by eight orders of magnitude (from 0:013 lm to 750 lm) (Levin and Angert 2015) . However, many bacteria are thought to occur in the 0:3-5 lm range (Schaechter 2016) . Bacteria can also nucleate and exist as "clumps" or adhere to larger suspended particles (Tham and Zuraimi 2005; Haas et al. 2013; Gong et al. 2020) , thus altering their net particle size that would influence their fluid dynamics. Airborne bacterial concentrations can be influenced by several factors, including ambient temperature, humidity, wind dynamics, and particulate matter concentrations (Gong et al. 2020) , and these factors could also play important roles in vertical stratification and warrant further research. There also appeared to be some mixing of aerobiome signals within fine vertical resolution strata, whereas more sensible patterns emerged in larger vertical strata. These findings are consistent with the phenomenon of turbulent mixed (nonlaminar) flow, and we might expect some level of vertical mixing in the aerobiome where turbulent flow occurs over and around obstacles and over rough surfaces. Vertical stratification in bacterial beta diversity could also have important implications for public health. For example, our results point to intriguing questions, such as: a) Are there where n is the number of bivariate observations and rs is Spearman's rank correlation coefficient. Reference number refers to the number key and corresponding notable taxa in Figure 7 and Table 4 . * 0.05. ** 0.01. *** <0:01. Streptomyces spp. These Actinobacteria are relatively more abundant at lower (vertically) sampling levels. They are soil-associated but also considered to be "old friends" with potential beneficial implications for human health (Bolourian and Mojtahedi 2018). 2 Kingella spp. Higher relative abundance at upper (vertical) levels. The Gram-negative K. kingae is considered to be pathogenic to humans, causing osteomyelitis and septic arthritis, particularly in children (Kiang et al. 2005; Nguyen et al. 2018 ). 3 Lactobacillus spp. Gram positive Firmicutes, relatively more abundant at upper levels. Some species are widely considered to be beneficial "old friends" and probiotics in humans and other ecosystems (Rook et al. 2014 ) (e.g., L. acidophilus; L. plantarum; L. rhamnosus). 4 Flavobacterium spp. Soil and water-dwelling Bacteroidetes bacteria. These are present in all levels but with highest relative abundance at upper levels. Generally not considered to be pathogenic to humans. Spatial distribution suggests potential allochthonous deposition. 5 Sphingomonas spp. These are Proteobacteria, found in a variety of environments. Relatively abundant in all sampling heights but less so in the soil level. These organisms are not considered to be pathogenic to humans and can in fact be highly beneficial via their ability to break down polycyclic aromatic hydrocarbons, which are deleterious to human health (Macchi et al. 2017; Asaf et al. 2020 ). Note: The taxa in this table may have important public health implications as highlighted in the third column. Reference number refers to the number key and corresponding notable taxa in Figure 7 . significant and consistent differences in potentially beneficial and pathogenic bacterial assemblages at different altitudes in the aerobiome? b) Does this affect exposure and colonization in humans across the life course? and c) What are the downstream health implications of this, if any? We provide a preliminary contribution towards answering the first question, as discussed in the section, "Relative Abundances and Notable Taxa." Future research could also consider the potential influence of physicochemical (e.g., antimicrobials, pesticide use) and social (e.g., crowd gathering, isolation, or distancing) practices on microbial vertical stratification. For example, efforts to reduce infectious agents such as COVID-19 may disrupt out relationship with environmental microbiomes. Therefore, understanding whether and how these changes affect human-environmental microbial interactions will be essential in the future. Following the analyses of relative abundances, the dominant taxa in the soil and lower sampling heights were found to be Actinobacteria, and the dominant taxa in the upper sampling heights were Proteobacteria. This finding is not surprising, given that a large proportion of terrestrial Actinobacteria are soildwelling organisms (Barka et al. 2016; Zhang et al. 2019) , and both phyla are among the largest in the bacterial domain (Verma et al. 2013; Polkade et al. 2016; Rizzatti et al. 2017) . Other studies have shown similar dominant roles for these phyla in the aerobiome (Arfken et al. 2015; Maki et al. 2017; Li et al. 2018 ), but vertical stratification has not, to our knowledge, been explored. We identified a number of notable dominant taxa at the genus level, including Streptomyces, Kingella, Lactobacillus, Flavobacterium, and Sphingomonas. With the exception of Flavobacterium, species in these genera are considered to have potentially beneficial or pathogenic impacts on human health. For example, the Actinobacteria Streptomyces spp., is considered to be a microbial "old friend" and potentially beneficial to human health via production and regulation of antiproliferative, antiinflammatory and antibiotic compounds (Bolourian and Mojtahedi 2018; Nguyen et al. 2020 ). This genus had higher relative abundance at lower sampling heights. On the other hand, members of the Kingella genus such as K. kingae are considered to be pathogenic to humans, for example--causing debilitating conditions such as osteomyelitis and septic arthritis, particularly in children (Kiang et al. 2005; Nguyen et al. 2018; Ingersoll et al. 2019 ). These findings warrant further research because, if consistent across time and space, the spatial and compositional differences in microbiota have the potential to be important considerations for public health through the modulation of exposure. As a proof of concept study, we have demonstrated, for the first time, the presence of vertical stratification of microbial alpha and beta diversity at lower levels of the biosphere (ground level to 2:0 m high). However, data from a larger number of replicates from different environments and geographical areas will be required to establish the generalizability of our findings; i.e., will our results be consistent outside of the Adelaide Parklands environment? We also used OTU picking methods at the bioinformatics stage. We recognize that although this approach has value for short-read platforms and many studies still use this approach (Dei-Cas et al. 2020; Derilus et al. 2020; Sato et al. 2020) , Amplicon Sequence Variant (ASV) analysis would have provided a more detailed taxonomic picture of vertical stratification. Further, following the DNA extraction process, three samples (each at SC03 0:0 m) failed to reach sufficient DNA concentrations to enable PCR and sequencing, which may have affected the vertical stratification relationship; we can only speculate that the relationship would have been stronger with their inclusion. There are many sensitive variables involved with processing low biomass samples (Eisenhofer et al. 2019; McArdle and Kaforou 2020) , and perhaps even more stringent workflows are required for passive sampling. We provide support for the presence of aerobiome vertical stratification in bacterial diversity (alpha and beta), and demonstrate that significant spatial differences in potentially pathogenic and beneficial bacterial taxa may occur. Although the need to promote healthy ecosystems and understand environmental microbial exposures has always been important, in light of the COVID-19 pandemic, it is now justifiably at the forefront of many public health agendas worldwide. As discussed, there is growing evidence to suggest that exposure to the microbiome in biodiverse green spaces contributes towards educating the immune system (Rook et al. 2003 (Rook et al. , 2013 Tasnim et al. 2017; Liddicoat et al. 2020) . Furthermore, the microbiome is thought to support the immune system's defensive role against pathogens and prevent hyperinflammatory responses and metabolic dysregulation, which are risk factors for severe COVID-19 (Torres et al. 2019; Guo et al. 2020) . Gaining a greater understanding of the transmission routes and physical factors (such as the vertical differential) affecting our exposure to environmental microbiomes--including potentially beneficial and pathogenic species--is likely to play an increasingly important role in the health sciences. Strategies to explicitly consider the microbiome as part of health-promoting urban green spaces have recently been proposed, such as Microbiome-Inspired Green Infrastructure (MIGI) (Robinson et al. 2018; Watkins et al. 2020) . Further exploration of aerobiome vertical stratification could make an important contribution to this approach. For example, there could be value in determining whether different habitats and vegetation management regimes affect vertical stratification in urban green spaces and in elucidating the downstream health effects on urban dwellers. Building on our findings--that vertical stratification did occur in an urban green space aerobiome--has the potential to inform future exposome research, urban biodiversity management, and disease prevention strategies. Fungal biodiversity in the air of Turin Vertical variation in Urticaceae airborne pollen concentration Comparison of airborne bacterial communities from a hog farm and spray field Sphingomonas: from diversity and genomics to functional role in environmental remediation and plant growth Genetic effects on the commensal microbiota in inflammatory bowel disease patients A culture-dependent and metagenomic approach of household drinking water from the source to point of use in a developing country Taxonomy, physiology, and natural products of Actinobacteria Role of the microbiota in immunity and inflammation An underground revolution: biodiversity and soil ecological engineering for agricultural sustainability Streptomyces, shared microbiome member of soil and gut, as 'old friends' against colon cancer Soil fauna and soil functions: a jigsaw puzzle Culturing of 'unculturable' human microbiota reveals novel taxa and extensive sporulation Diversity of nasal microbiota and its interaction with surface microbiota among residents in healthcare institutes Weaning age among foragers at Matjes River Rock Shelter, South Africa, from stable nitrogen and carbon isotope analyses Isolation of pathogenic yeasts in the air from hospital environments in the city of Fortaleza, northeast Brazil The "exposome" concept-how environmental risk factors influence cardiovascular health Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data Metagenomic analysis of gut microbiota in non-treated plaque psoriasis patients stratified by disease severity: development of a new Psoriasis-Microbiome Index An in-depth survey of the microbial landscape of the walls of a neonatal operating room Atmospheric Microbial Community Sampling System for Varying Altitude Collection Size matters in regulating the biodiversity of tropical forest soils Contamination in low microbial biomass microbiome studies: issues and recommendations Quantifying human experience in architectural spaces with integrated virtual reality and body sensor networks From the exposome to mechanistic understanding of chemical-induced adverse effects Urbanisation reduces the abundance and diversity of airborne microbes-but what does that mean for our health? A systematic review Modeling particle dispersion and deposition in indoor environments Asthma and allergic symptoms in relation to house dust endotoxin: Phase Two of the International Study on Asthma and Allergies in Childhood (ISAAC II) Concentration, viability and size distribution of bacteria in atmospheric bioaerosols under different types of pollution Diabetes is a risk factor for the progression and prognosis of COVID-19 The concentrations of culturable microorganisms in relation to particulate matter in urban air Biodiversity, microbes and human well-being The gut microbiome in myalgic encephalomyelitis Inflammatory markers limitations in the diagnosis of pediatric calcaneal osteomyelitis Genetic and environmental influences on height from infancy to early adulthood: an individual-based pooled analysis of 45 twin cohorts Bioaerosol generation by raindrops on soil Outbreak of osteomyelitis/septic arthritis caused by Kingella kingae among child care center attendees Brain-gut-microbiota axis in Alzheimer's disease Influence of pig farming on the human nasal microbiota: key role of airborne microbial communities Impact of environmental microbiota on human microbiota of workers in academic mouse research facilities: an observational study Small but mighty: cell size and bacteria Concentration and community of airborne bacteria in response to cyclical haze events during the fall and midwinter in Beijing Naturally-diverse airborne environmental microbial exposures modulate the gut microbiome and may provide anxiolytic benefits in mice Can bacterial indicators of a grassy woodland restoration inform ecosystem assessment and microbiota-mediated human health? Effect of air pollution on the total bacteria and pathogenic bacteria in different sizes of particulate matter Insights into the genome and proteome of Sphingomonas paucimobilis strain 20006FA involved in the regulation of polycyclic aromatic hydrocarbon degradation Variations in the structure of airborne bacterial communities in Tsogt-Ovoo of Gobi desert area during dust events Bioaerosols play a major role in the nasopharyngeal microbiota content in agricultural environment Sensitivity of shotgun metagenomics to host DNA: abundance estimates depend on bioinformatic tools and contamination is the main issue Characterization of the workplace chemical exposome using untargeted LC-MS/MS: a case study phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data Spatiotemporal controls on the urban aerobiome The first microbial colonizers of the human gut: composition, activities, and health implications of the infant gut microbiota Vertical stratification of ammonia in a broiler house Centrality of the XBP1-ERN1 axis in transplacental innate immune training via a microbial-derived immunomodulator Weaning from the breast Streptomyces sp. VN1, a producer of diverse metabolites including nonnatural furan-type anticancer compound Imaging of Kingella kingae musculoskeletal infections in children: a series of 5 cases Computer and mathematical modelling of the female human body: determination of mass-inertial characteristics in basic body positions Quorum sensing: an underexplored phenomenon in the phylum Actinobacteria The skin microbiome: impact of modern environments on skin ecology, barrier integrity, and systemic immune programming RCPCH (Royal College of Paediatrics and Child Health). 2020. UK-WHO growth charts -2-18 years Proteobacteria: a common factor in human diseases Green prescriptions and their co-benefits: integrative strategies for public and environmental health Let nature be thy medicine: a socioecological exploration of green prescribing in the UK Walking ecosystems in microbiomeinspired green infrastructure: an ecological perspective on enhancing personal and planetary health Regulation of the immune system by biodiversity from the natural environment: an ecosystem service essential to health Innate immune responses to mycobacteria and the downregulation of atopic responses In: Microbial Endocrinology: The Microbiota-Gut-Brain Axis in Health and Disease Environment dominates over host genetics in shaping human gut microbiota Aeolian dispersal of bacteria in southwest Greenland: their sources, abundance, diversity and physiological states Effects of the oral adsorbent AST-120 on fecal p-cresol and indole levels and on the gut microbiota composition the Company of Microbes: Ten Years of Small Things Considered Transfer of environmental microbes to the skin and respiratory tract of humans after urban green space exposure Nature-based interventions for improving health and wellbeing: the purpose, the people and the outcomes The nasal microbiota of dairy farmers is more complex than oral microbiota, reflects occupational exposure, and provides competition for staphylococci Microbiome and asthma Innate immunity and asthma risk in Amish and Hutterite farm children Linking the gut microbial ecosystem with the environment: does gut health depend on where we live? Size relationship between airborne viable bacteria and particles in a controlled indoor environment study Global airborne microbial communities controlled by surrounding landscapes and wind conditions Exposure to a healthy gut microbiome protects against reproductive and metabolic dysregulation in a PCOS mouse model Phylogenetic analyses of phylum Actinobacteria based on whole genome sequences Microbiomeinspired green infrastructure: a toolkit for multidisciplinary landscape design The soil microbiome influences grapevine-associated microbiota Variation in actinobacterial community composition and potential function in different soil ecosystems belonging to the arid Heihe river basin of Northwest China Soil biota, antimicrobial resistance and planetary health The authors acknowledge and pay their respects to the Kaurna people, the traditional custodians of their ancestral lands on which we conducted the research for this study. The authors also acknowledge the City of Adelaide, Australia, which helped facilitate this study through time and resources.