Culturable bioaerosols along an urban waterfront are primarily associated with coarse particles

Sampling location and methods

Air sampling, at a height of 175 cm from the ground, was conducted during the months of July, August and September of 2015 along the waterfront of Flushing Bay, NY (Fig. 1), an embayment adjacent to the East River in western Long Island Sound. The waterfront is surrounded by industrial and urban residential developments, adjacent highways, LaGuardia International Airport and large recreational areas including Citi Field, the Billie Jean King National Tennis Center and Flushing Meadows Corona Park. A promenade stretches along the shoreline where sampling occurred, and is a site frequently used for recreational activities. Flushing Bay has many combined sewer overflow (CSO) discharges, including one of the largest CSO outfalls in the city, and water quality impairment has resulted in the need for a Long Term Control Plan for sewage pollution (http://www.nyc.gov/html/dep/html/cso_long_term_control_plan/flushing_bay_ltcp.shtml). Enteric bacteria concentrations in Flushing Bay are regularly monitored (http://riverkeeper.org) and previous studies have characterized the microbial composition of the water, including elevated antibiotic resistant bacteria concentrations that are linked to wet weather sewage discharge into the bay (Young, Juhl & O’Mullan, 2013).

Thirty-seven air sampling events were completed over fourteen distinct days, with 23 of the events occurring during onshore winds, on seven unique days, and the remainder completed during offshore wind conditions. Measurements for meteorological conditions, particle concentrations and culturable bioaerosols were conducted in parallel. The start time, duration and volume of air sampled for each event is available in supplemental material (Data S1), along with raw bioaerosol counts. Total aerosol particle concentrations and size distributions were measured at sixty second intervals using an Aerocet-531S Mass Particle Counter (http://www.metone.com). Local weather conditions were logged in sixty second intervals using a Kestrel 4500 weather meter (http://www.kestrelmeters.com). Reference data for local weather conditions were also obtained through the local LaGuardia Airport (KLGA) Weather Underground station (http://www.wunderground.com).

The phrase “culturable microbial aerosols” in this study refers to culturable bacteria and fungi, and they were sampled using an Andersen six-stage viable cascade impactor (Tisch 1800-x, Andersen Inc) with an air flow rate of 28.3 LPM. The cascade impactor fractionates particles into six size bins according to their aerodynamic diameter. The aerodynamic diameter bins are 0.65–1.1 µm (Stage 6), 1.1–2.1 µm (Stage 5), 2.1–3.3 µm (Stage 4), 3.3–4.7 µm (Stage 3), 4.7–7.0 µm (Stage 2), and >7.0 µm (Stage 1). Impacted particles are retained on the surface of agar plates housed in the interior of each chamber. The agar plates were prepared with 27 mL of R2A media, commonly used in aerosol and surface water studies, including other studies in the New York region and Flushing Bay (Young, Juhl & O’Mullan, 2013; Dueker et al., 2012b). Sampling occurred between late morning and late afternoon. At the end of each sampling event the agar plates were quickly covered and stored in a cool container away from sunlight. To assess potential contamination during field activities and the recovery of the plates, blank plates were transported to the field, exposed to air concurrently with the cascade plates for a period of two seconds and handled in parallel with the experimental plates. Because the size bins from the Aerocet Particle Counter and the Anderson Cascade Impactor were not identical, coarse particles were considered as >1 µm, while coarse bioaerosols were considered as >2.1 µm.

Microbial enumeration and molecular analysis

Following sampling, the agar plates were immediately transported to the lab, and incubated at a temperature of 27 °C for 72 h. Observable bacterial and fungal colonies were then visually enumerated. The raw colony counts were adjusted using the positive hole correction method (Andersen, 1958). Bacterial colonies were picked using sterile pipette tips into individual wells of 96-well plates, suspended in 30 µL of nuclease-free sterile water (Hyclone Laboratories, Inc. Logan, Utah, USA), lysed by boiling at 95 °C for 5 min, and frozen at −20 °C until subsequent molecular analysis. To determine taxonomic identity of bacterial colonies we utilized the approach of Dueker et al. (2012a); Dueker et al. (2012b) and Young, Juhl & O’Mullan (2013) including PCR amplification of lysed colony suspensions using 16s rRNA gene universal bacterial primers 8F (5′-AGRGTTTGATCCTGGCTCAG-3′), and 1492R (5′-CGGCTACCTTGTTACGACTT-3′) (Teske et al., 2002). Amplification reactions included 35 cycles of denaturation (45 s at 95 °C), annealing (45 s at 55 °C ) and elongation (60 s at 72 °C). PCR products were visualized by electrophoresis to confirm amplification of a single approximately 1,500 base pair product and were then processed for single pass Sanger sequencing, using the same 8F primer, by Eton Bioscience (Union, NJ, USA). Sequence quality control and editing was performed using the Geneious software package, and high quality sequences, including a minimum length of 236 base pairs, were uploaded to the Ribosomal Database Project (RDP) for alignment and taxonomic classification to the level of genera using the RDP naïve Bayesian rRNA classifier.

To compare bioaerosol taxonomy with that of local water sources, bacterial 16S rRNA sequences previously obtained from Flushing Bay (Young, Juhl & O’Mullan, 2013), also during the summer season, were assembled into a surface water sequence library for comparison. After quality control filtering removed low quality sequences, a total of 266 16S rRNA sequences were obtained from cascade aerosol isolates and an additional 117 sequences from Flushing Bay surface water isolates. From these sequences three clone libraries were assembled for aerosol sequences, grouped by the wind direction (onshore or offshore) and the size fraction (coarse or fine) where these isolates were observed. Sequences were not obtained from offshore fine isolates. The libraries were designated as onshore fine (n = 56); onshore coarse (n = 126); offshore coarse (n = 84); and surface water (n = 117). DNA sequences from bioaerosol isolates have been submitted to the NCBI Genbank database using accession numbers KX959694 –KX959959.

Statistical analysis

Mann–Whitney nonparametric tests were performed to assess significant differences between the following groups: (1) total fine (<2.1 µm) and total coarse (>21 µm) microbial aerosols; (2) total coarse microbial aerosols under low (<4.0 m s⁻¹) and high (>4.0 m s⁻¹) wind conditions; and (3) total microbial aerosols (coarse + fine) over high and low wind conditions. To test for significant correlations between culturable microbial aerosols, wind speed, and total particle concentrations, Spearman’s coefficient was used. A Chi-squared test was performed on the distribution of microbial cells over the six cascade aerodynamic size fractions to determine if there was a significant change in size distribution over higher and lower wind speeds. These statistical tests were performed using R statistical software (R Core Team, 2015). For taxonomic analyses, significant differences between sequence libraries were determined using Ribosomal Database Project’s (RDP) Library Compare algorithm (Wang et al., 2007).

Aerosol particle size distributions

Total particle concentrations during offshore wind conditions (mean = 7.46 × 10⁶m⁻³ ± 2.17 × 10⁶) were approximately four times higher than during onshore wind conditions (mean =1.94 × 10⁶m⁻³ ± 4.52 × 10⁵), with the majority of aerosol particles in the fine fraction (88.6% for offshore, 90.4% for onshore) (Fig. 2). Despite increased concentrations in the offshore wind conditions, both onshore and offshore winds resulted in similar aerosol particle size distributions (Fig. 3).

Total culturable microbial aerosol counts were higher under offshore winds (mean = 778 CFU m⁻³ ± 66.7), with bacteria comprising the majority of colonies (58.5%), as compared to onshore winds (580 CFU m⁻³ ± 110) where fungi were dominant (91.1%) (Fig. 4). Bacterial concentrations were significantly greater during offshore winds (Mann–Whitney, p < 0.001), but fungal concentrations did not significantly differ with changing wind direction (Mann–Whitney, p = 0.56). The majority of bacterial colonies (90% during offshore, 68% during onshore) were found in the coarse fraction (>2.1 µm) (Fig. 5). Similarly, the majority of fungal colonies were in the coarse fraction (85% during offshore, 70% during onshore) but with a more distinct peak of fungal abundance in Stage 4 (aerodynamic sizes 2.1–3.3 µm).

Wind speed

Wind conditions were found to be important to the distribution and abundance of culturable microbial aerosols. Wind speed and total coarse bioaerosols measured during onshore winds were significantly correlated (Spearman, r = 0.73 and p < 0.001) (Fig. 6). Under high onshore wind conditions (>4 ms⁻¹) , the mean total (fine + coarse) microbial concentration in air was 900 CFU m⁻³ ± 180 (Fig. 6). It was significantly higher than microbial concentrations observed during lower wind conditions (mean = 180 CFU m⁻³ ± 27.6; Mann–Whitney, p < 0.001). The size distribution of culturable microbes significantly varied under low and high winds (Chi-squared, p < 0.05) for offshore bacteria, offshore fungi, and onshore fungi, with the percentage of fine aerosols increasing under high winds.

Taxonomic identification

Sequences within the three aerosol libraries were taxonomically identified and represent 49 genera, with 24 genera occurring as singletons (Table S1). The majority of genera belonged to the Actinobacteria, Proteobacteria, Firmicutes, and Bacteroidetes phyla (Fig. 7) with an additional three sequences in the onshore coarse and fine libraries belonging to Deinococcus–Thermus. Actinobacteria and Proteobacteria were found to be the two most abundant phyla present in all three aerosol libraries (Fig. 7). The most common genera shared by all three aerosol libraries were Streptomyces, Bacillus, Sphingomonas, Microbacterium, and Massilia, (Fig. 8).

There were significant differences in surface water and aerosol taxonomy. 10% of the surface water isolates were classified as Flavobacterium, a genera that was not detected in the aerosol library. Pseudomonas comprised 18% of the surface water isolates, yet only comprised 2 to 6 % of the isolates in each of the aerosol libraries. Pairwise comparison of the aerosol and water libraries using the thetayc dissimilarity index (Yue & Clayton, 2005) showed that the three aerosol libraries were more similar to each other than to surface water. The onshore coarse and offshore coarse libraries showed the greatest similarity (thetayc index = 0.375) (Fig. S1).

In the onshore fine aerosol library, Streptomyces (21.4%) and Bacillus (16.1%) were abundant but were not found in surface waters. In the onshore coarse aerosol library, Bacillus (12.0%), Sphingomonas (11.1%), and Acinetobacter (8.7%) were the most abundant genera and Sphingomonas, Streptomyces, and Bacillus were significantly more abundant in the onshore coarse library than in surface waters (p < 0.05). For the offshore coarse library, Streptomyces (14.3%) and Microbacterium (13.1%) the most common genera, which were significantly more abundant than in surface water (p < 0.05).

Wind speed

Wind speed was correlated with total numbers of airborne bacteria and fungi. Regardless of particle size, large winds may transport bacterial and fungal spores kilometers from their source, affecting air quality on a regional or even global scale. Dust from Saharan Desert storms often penetrate Europe, and even cross the Atlantic to deposit in the Caribbean Ocean (Gyan et al., 2005). This has important health and environmental connotations. Strong wind conditions can transport airborne pathogens through large, populated areas. Such was the case during a foot and mouth disease epidemic in England (Hugh-Jones & Wright, 1970). Prior studies involving culturable microbes and meteorological factors have established a positive correlation between wind speed and airborne bacteria concentrations (Lighthart, 1997); however, other studies did not find a correlation between wind speed and culturable bacteria (Li et al., 2011). The association of wind speed and coarse aerosols in this study reinforces the importance of local production sources for this urban waterfront.

Taxonomy

Actinobacteria and Proteobacteria were the predominant bacterial phyla present in aerosol sequences. Proteobacteria is one of the most diverse phyla among the bacteria and has been found to be common in aerosols across a wide variety of regions and environments (Dueker et al., 2012a). Actinobacteria have been found in cloud water, and urban and rural aerosols (Amato et al., 2005; Maron et al., 2005; Lee et al., 2010; Dueker et al., 2012b), but were not dominant in Flushing Bay surface water samples, where Proteobacteria and Bacteroidetes were the most abundant. Some genera were detected in aerosol samples but were absent in surface water libraries, such as Streptomyces and Bacillus which are generally associated with soil, and are also capable of forming spores (Nicholson, 2002; Vetsigian, Jajoo & Kishony, 2011). This would enable them to remain viable in the atmosphere, and they may be transported long distances, providing a regional signal disconnected from local water sources.

Many of the genera observed in Flushing Bay air have also been observed in prior aerosol studies. For example, using a similar sampling approach, Wang et al. (2012) found that some of the dominant genera inside cave air included Micrococcus, Sphingomonas, Pseudomonas, and Bacillus; which were also dominant in our aerosol samples. Urbano et al. (2011) found a dominance of Bacillus in marine aerosol sampled in California. Using culture independent approaches, Amato et al. (2005) found a dominance of Streptomyces, Microbacterium, Micrococcus, Pseudomonas, and Bacillus in cloud water. The wide spread presence of these bacteria among diverse environments, including urban and rural environments, suggests the importance of cellular adaptations to aerosol formation or atmospheric persistence.

Aerosol particle size may control microbial exposure to environmental conditions. Small particle diameters imply a higher degree of exposure to the harsh conditions of the atmosphere, while bacteria that clump to each other or adhere to larger particles may be better sheltered from these stresses. The observed difference in coarse and fine onshore taxonomy may therefore reflect adaptation to much harsher conditions in the atmosphere. For example, isolates identified as Deinococcus–Thermus, a phyla adapted to tolerate high levels of radiation and desiccation (Battista, Earl & Park, 1999), were found in our aerosol samples.

Although transport models (e.g., Fig. S2) often indicate a range for particles of several hundred kilometers, this may be an overestimate for most bacteria, as the models generally do not account for viability or culturability. There are likely a subset of taxa (e.g., Deinococcus–Thermus) that may remain viable longer and their transport may be more constrained by physical, rather than biological constraints. However, biological response to stress is likely to be a major constraint for most taxa. Viability and culturability are expected to decrease with transport distance, making the actual range of viable bacterial transport much shorter than most models predict. This is consistent with the observed low number of culturable microbes in the fine fraction, which would be expected to travel the longest distances.