1. Introduction
In coastal areas, bioaerosols are produced from water surfaces and transported by onshore winds to the near-shore environment. The role of wind speed in modulating bioaerosol production and transport has not been fully examined, particularly in coastal urban settings where dense human populations are in close proximity to waterways suffering from microbial pollution due to sewage discharges. Whereas higher winds are known to dilute fine aerosols (and other air pollutants) in urban spaces, and therefore are correlated with improved air quality [
1], they are also likely to increase local production and transport of aerosols from adjacent urban aquatic environments [
2]. If water quality is poor due to sewage contamination, locally-produced aerosols may contain pollution-related microbes, thereby degrading air quality and affecting human health [
3,
4,
5,
6]. Previous studies have confirmed the transfer of viable bacterial cells from polluted surface waters to air in the urban waterfront, including bacteria from genera containing known human pathogens [
4,
7]. The public health implications for the movement of potentially pathogenic bacteria from aquatic sources to urban air remain largely unexplored, despite the stated need to assess and model sources, production, and transport of aerosolized pathogens in the outdoor environment [
8,
9].
The movement of microbes and microbial products from water to air through aerosol production has been demonstrated in a wide range of aquatic environments, including a superfund-designated waterway [
4], urban coastal areas [
10,
11,
12], coastal surf and public beaches [
7,
13], open ocean [
14,
15], riverine and estuarine environments [
16,
17,
18], and temperate lakes [
19,
20]. In urban settings, where surface waters are often sewage-contaminated, the delivery of surface water to air via aerosolization may represent a movement of sewage-associated bacteria [
5,
21,
22,
23,
24,
25,
26], endotoxins [
27,
28], and viruses [
29,
30,
31], into urban airspace. Despite the local connections between water quality and air quality suggested in these studies, urban bioaerosol studies have been primarily conducted in cities without extensive waterfronts [
32,
33,
34,
35,
36,
37] and have also focused on bacterial aerosols. Studies of viral aerosols have been conducted in sea spray aerosols [
18,
38,
39,
40] inland forest and industrial outdoor environments [
30,
31], but not in coastal urban environments. Nevertheless, these studies have confirmed the production of bacterial and viral aerosols from water surfaces, which is pertinent to public health management of water and air quality.
Aerosols are produced from water surfaces through the bursting of bubbles that release microscopic droplets into the air [
41]. Bubbles are introduced to the water column through wind-wave interactions, wave-shore interactions, recreational water use, or any other activity that disrupts water surfaces. The delivery of viable bacterial and viral particles from water to air and land requires onshore winds. In addition, physical wind-wave interactions at wind speeds above 4 m s
−1 are associated with the onset of whitecapping [
42], and, therefore, the production of new aerosol particles from the water’s surface [
43,
44]. Some of these new particles contain microbes [
39]. In coastal environments, aerosol particle number concentrations and inland transport distances increase with increasing onshore wind speeds [
45], resulting in marine particles traveling further before settling out [
46,
47]. Previous research has demonstrated that bacterial aerosols are predominantly detected in the coarse aerosol fraction [
10,
48], and often are attached to larger particles [
39]. These findings combined suggest that wind-facilitated production of coarse aerosols at wind speeds above 4 m s
−1 will result in the increased onshore transfer of large marine particles containing aquatic microbes to the coastal atmosphere.
We measured urban waterfront bacterial and viral aerosols under the influence of a range of onshore wind speeds in New York, NY, USA. We hypothesized that the production of new microbial aerosols (bacterial and viral) from water surfaces would result in increased transfer of culturable and total microbial aerosols to the near-shore environment when winds exceeded 4 m s−1. Specifically, we expected that elevated onshore wind speeds would (1) increase the concentrations of total and culturable microbial (viral and bacterial) aerosols in the near-shore environment, and (2) result in greater taxonomic similarity between water and air bacterial communities in the near-shore environment.
2. Experiments
2.1. Sampling Sites and Meteorological Context
Sampling for this study was conducted at two New York, NY, USA coastal sites: Louis Valentino Pier (LVP) on the NY Harbor in Brooklyn, NY, USA (40.67838° N, 74.01966° W), and Flushing Bay (FB), Queens (40.761179° N, 73.846655° W) during onshore wind conditions (
Figure S1). Both of these sites are highly-trafficked public parks with waterfront access. The first field campaign was conducted at LVP from 6 April to 8 June 2011, and the second field campaign was conducted at LVP and FB from 23 April to 18 June 2014 (
Table S1). One-minute wind speed, wind direction, relative humidity, and temperature were collected using a Vantage Pro2 Plus Weather Station (Davis Instruments, Hayward, CA, USA). At both locations, the weather station was deployed at 1.8 m above ground level (this translated to 2–5 m above water-level, depending on tidal height). Wind speed was used to evaluate environmental controls on the transport and production of aerosol particles. Wind direction was used to assess local origin of sampled aerosols (onshore or offshore). At both LVP and FB, fetch over harbor waters, approximated using imagery from Google Earth (
www.google.com/earth), ranged from 3–8+ km depending on wind direction. Relative humidity (RH) and temperature were measured as parameters known to affect aerosol particle size and thought to influence the viability of aerosolized bacteria [
49,
50,
51].
During the first field campaign, aerosol particle concentrations were measured using a stationary Met One 9012 Ambient Aerosol Particulate Profiler (Met One Instruments, Grants Pass, OR, USA). The profiler was placed at the same height as the weather station (2.0 m from the pier decking). One-minute measurements of aerosol particle number concentrations were continuously logged in particle cut-off bins of 0.3, 0.7, 1, 2, 3, 5, 7 and 10 µm diameter (D
p), with a particle size cut-off of approximately D
p = 30 μm. This range of particle sizes covers both the fine (D
p = 0.3–2 µm) and coarse (D
p = 2–30 µm) aerosol particle modes [
1]. Due to the high variability of RH at LVP during the first field campaign, and the strong dependence of aerosol particle size on RH [
1], coarse and fine aerosol particle size concentrations were normalized to 80% RH according to Fitzgerald et al. [
52] to allow for analyses pooled across sampling events. To assess wind speed influence on coarse aerosol number concentrations and the presence of culturable bacterial aerosols, results were binned by wind speed.
2.2. Depositing Culturable Bacterial Aerosols and Sub-Surface Water Bacteria (First Field Campaign)
The presence of depositing culturable bacterial aerosols was measured by simultaneously exposing both R2A (BD Difco, Fisher Scientific, Hampton, NH, USA) and LB (LB Miller, BD Difco, Fisher Scientific) agar media plates (each in triplicate) to ambient aerosols from a wind-rotated platform 2.0 m above the pier decking. Replicate plates were horizontally aligned, perpendicular to wind direction, and exposure times varied from 15–50 min (depending on aerosol loading and wind conditions). Each exposure event consisted of exposing a total of six media plates, and each sampling day consisted of 4–6 exposure events. Sampling during the first field campaign (at LVP) occurred on nine separate days at LVP, from early morning to mid-afternoon in each case, resulting in 42 bacterial aerosol exposure events (a total of 252 plates exposed).
It is important to note that this method does not measure total depositing bacterial aerosols, since not all bacterial aerosols are settling and not all bacterial aerosols are capable of growth on the media provided. This first field campaign was instead focused on making consistent relative measurements of depositing culturable bacterial aerosols (e.g., [
46,
53]) in relation to onshore wind speed. In contrast to culture-independent methods, this approach allows for confidence that the bacteria that were counted and molecularly characterized were viable, intact cells at the time of sampling. Prior publications provide strong support for the use of culture-based methods to assess environmental controls on viable bacterial aerosols [
4,
45,
46]. Measurements were conducted using two different media types to assess and reduce bias in media selection: R2A, a relatively low-nutrient media typically used for heterotrophic bacterial plate counts; and LB, a high-nutrient media known to discourage fungal overgrowth [
53]. In previous studies, both LB and R2A media have been used to grow diverse microbial assemblages from aerosols and water from freshwater and brackish environments [
46,
53,
54].
After exposure, plates were incubated for five days in the dark at 25 °C and then colony forming units (CFU) were counted. During protocol optimization, five days was optimal in terms of balancing CFU detection with overgrowth and plate desiccation. A “CFU accumulation rate” (CFU m−2 s−1) was calculated using the surface area of the exposed petri dishes (0.0079 m2) and the duration of exposure. To assess culturable sub-surface water bacteria at the site, near-shore surface water (<1 m depth) was collected near the aerosol sampling site during each exposure. Surface water subsamples were plated, incubated and enumerated under the same conditions described for aerosol exposures. For both media types, control plates (unexposed) were regularly incubated under the same conditions to ensure that poured media was sterile. All statistical analyses were performed using R statistical software (R Development Project 2008).
2.3. Total Microbial Aerosols (Second Field Campaign)
Because culture-based methods are only able to measure bacterial cells capable of growth on the specific media used, and our method sampled only depositing bacterial aerosols, we conducted a second field campaign using culture independent methods. Samples were taken over the course of 10 days during the second field campaign (2 days at LVP, 8 days at FB). Samples were gathered using a Coriolis cyclonic air sampler (Bertin, Inc., Montigny-le-Bretonneux, France). In brief, a measured volume of air was directed by the sampler into a 15 mL solution of sterile, endotoxin-free water with 0.001% final concentration of surfactant TritonX (Dow Chemical) to reduce surface tension of collection liquid. Immediately after sampling, 6 ml of the solution were fixed for approximately 30 min at 4 °C with electron microscopy-grade glutaraldehyde (0.5% final concentration, Sigma-Aldrich, St. Louis, MO, USA) for enumeration of bacteria and viruses by flow cytometry. The fixed samples were then flash-frozen in liquid-N2 and stored at −80 °C until further analysis.
Total bacteria and viruses were measured using a BD FACScan flow cytometer (Beckton Dickinson, Franklin Lakes, NJ, USA), equipped with an air-cooled laser providing 50 mW at 488 nm with standard filter set-up, as previously described [
55,
56]. Briefly, samples were stained with the fluorescent nucleic acid dye SYBR
® Green I (Invitrogen Co., Carlsbad, CA, USA) and viruses and bacteria were discriminated based on their side scatter and acquired green fluorescence (DNA) signals.
2.4. Bacterial Taxonomic Identification (First Field Campaign)
Bacterial colonies grown on R2A media agar plates from aerosols and sub-surface water samples collected at the LVP site were identified through sequencing of the 16S ribosomal gene. Specifically, after CFU enumeration, media plates were stored at 4 °C until colonies were individually picked and suspended in 50 μL of HyClone sterile water (ThermoScientific, Logan, UT, USA) in 96-well microplates and then boiled for 5 min to lyse the cells. This lysed-cell suspension was immediately frozen at −20 °C until PCR was performed. 16S rRNA genes were amplified from lysed-cell suspensions using TopTaq DNA Polymerase (Qiagen, Valencia, CA, USA) and universal bacterial primers 8F (5’-AGRGTTTGATCCTGGCTCAG-3’) and 1492R (5’-CGGCTACCTTGTTACGACTT-3’) [
57] with 35 polymerase chain reaction (PCR) cycles of 45 s of denaturation at 94 °C, 45 s of annealing at 55 °C, and 1 min elongation at 72 °C.
Single-read sequencing using the 8F primer was performed on amplified PCR products by SeqWright Laboratories (Houston, TX, USA). Sequences were quality-checked and edited using Geneious software [
58]. Edited sequences yielding less than 200 base pairs of high quality sequence were removed from further analyses. Remaining sequences (GenBank accession #’s MG270584–MG271746) were aligned using the Ribosomal Database Project (RDP) [
59] and taxonomically classified using RDP’s naive Bayesian rRNA classifier at an 80% confidence level [
60].
3. Results
In the first field campaign, sampling conditions covered a range of temperatures (8.5–24.5 °C), daily onshore wind speeds (0.9–7.7 m s
−1) and RH (29.4–85%) (
Supplemental Table S1). Onshore SW winds measured in this study were <4 m s
−1 (low wind), while NW winds were >4 m s
−1 (high wind); W winds were either low or high winds on different sampling dates (
Supplemental Table S1). During the second field campaign, temperatures ranged from 9.1–27.8 °C, onshore wind speeds ranged from 1.5–10.8 m s
−1, and RH ranged from 32.7–86.0% (
Supplemental Table S1). In this second campaign, the LVP site had the lowest temperatures (9.1 °C) and the highest wind speeds (10.8 m s
−1).
In the first field campaign, regardless of wind direction, the geometric mean of humidity-corrected aerosol number concentrations (binned by wind speed) decreased with wind speed (
Figure 1). Relationships of aerosol number concentrations to wind speed were non-linear, and similar to those found using non-humidity-corrected concentrations. The declines of both fine (0.3–2 µm) and coarse aerosols were more marked for speeds up to 4 m s
−1 (
Figure 1). Although both fine and coarse particle counts declined with elevated wind speed, fine particle counts declined to a greater extent, indicating a shift in the particle size spectrum towards higher relative abundance of coarse particles with higher wind speeds.
The CFU accumulation rate (measured during the first field campaign) remained relatively low with wind speeds <4 m s
−1, with a geometric mean of 0.86 ± 0.11 CFU m
−2 s
−1 for R2A media and 0.32 ± 0.05 CFU m
−2 s
−1 for LB media. At wind speeds above 4 m s
−1, both LB and R2A CFU accumulation rates increased significantly above low wind speed values (
p < 0.01) with a geometric mean of 1.95 ± 0.20 CFU m
−2 s
−1 for R2A and 0.70 ± 0.09 CFU m
−2 s
−1 for LB. Regression analysis of CFU accumulation rate for both types of media revealed linear relationships with increasing wind speed (
Figure 2) (R2A media:
R2 = 0.39,
p < 0.01; LB media:
R2 = 0.28,
p < 0.01).
In the second field campaign, mean viral aerosol loading was 5.3 ± 1.2 × 10
4 m
−3 and mean bacterial aerosol loading was 1.6 ± 0.4 × 10
4 m
−3, yielding a virus to bacteria ratio (VBR) of 3.5 ± 0.7. Abundances of both bacterial and viral aerosols increased linearly with onshore wind speed (bacteria:
R2 = 0.49,
p < 0.05; viruses:
R2 = 0.54,
p < 0.01) (
Figure 3). At wind speeds below 4 m s
−1, the VBR was 1.1 ± 0.3, and at wind speeds above 4 m s
−1 the VBR was significantly higher at 4.57 ± 0.8 (
p < 0.01).
Sub-surface water spreads on R2A media yielded bacterial concentrations almost twice as high as concentrations from LB media (R2A mean: 891 CFU mL
−1, LB mean: 476 CFU mL
−1) (significant difference,
p < 0.05). Daily CFU accumulation rates and surface water bacterial concentrations were log-linearly related, regardless of media type (
R2 = 0.47,
p < 0.01) (
Figure 4). Near-shore culturable bacterial aerosol deposition was sensitive to hourly shifts in onshore wind speed and coarse aerosol number concentrations, as detailed on two consecutive sampling days in
Supplemental Figure S2.
Sequencing of bacterial isolates on R2A plates from aerosols (615 isolates) and surface waters (548 isolates) revealed a diverse bacterial assemblage, representing bacteria from Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes, and Deinococcus-Thermus (
Figure 5). 131 bacterial genera were represented, 39 of which were shared between surface water and aerosol sequence libraries, representing greater than 75% of both libraries (
Table 1). At least 25 of these shared genera are known to contain potential human pathogens. There were more bacterial genera shared between water surfaces and aerosols under wind conditions above 4 m s
−1 (37 shared genera) than under low-wind conditions (33 shared genera).