https://orcid.org/0000-0001-9820-3452
Abstract
This study explores the microbial diversity of sources which may influence boat microbial communities. We investigated the impact of dock, air and water microbial communities on the hull, transom and bilge microbial communities over the span of 11 days.
Using source tracking software, we investigated the extent to which each of our potential sources (air, water and dock) influenced the overall microbial community. This study concluded that the dock impacted 14–64% of the hull and transom microbial community. Micro‐organisms from the water were shown to impact 5·6% the bilge microbial community but had minimal impact on hull and transom microbial communities. Micro‐organisms from the air had minimal impact in all areas of the boat.
Our results demonstrate that micro‐organisms from sources other than water can influence the microbial community of a boat, suggesting that terrestrial micro‐organisms can impact the boat microbial community.
Outside of ballast tanks, microbial diversity on boats is largely unexplored. While ballast water is widely recognized as a route for dispersal of allochthonous micro‐organisms, comparatively little is known about the microbial diversity on other areas of the boat. If the organisms on a boat originate from sources other than water, there is potential that terrestrial micro‐organisms could be dispersed by shipping activity.
Introduction
For decades, boats have been moving large volumes of ballast water around the world to stabilize ships for safer and more efficient voyages. This movement of ballast water is recognized as a conduit for dispersal of invasive species (Carlton 1985; Vinogradov et al.1989; Carlton et al.1990; Cariton and Geller 1993; Mills et al.1993; Habrison and Volovik 1994; Ruiz et al.2000). Regulations have been put in place to manage the negative effects of moving large amounts of ballast water around the world to slow the spread of invasive organisms (IMO 2017a, 2018). Ships are now required to undergo ballast water exchanges and/or treat their ballast water to prevent the spread of organisms in ballast water (Kideys et al.2005; IMO 2017a, 2018). Most regulations have been designed to prevent the spread of macroscopic eukaryotic species. A few recent studies have suggested that ballast water exchanges are ineffective at removing micro‐organisms from ballast water (Seiden et al.2010; Emami et al.2012; Brinkmeyer 2016; Lymperopoulou and Dobbs 2017).
Although microbial life outside of the ballast tank of boats is less frequently discussed, it is a newly recognized problem, which impacts the health of coastal environments (Ruiz et al.2000; Brinkmeyer 2016; IMO 2017b). Bacteria have numerous survival strategies at their disposal (dormancy, asexual reproduction, biofilm formation, etc.; Madigan et al. 2012) which allow them to survive in the highly competitive environments, despite our best efforts to remove them. Microbial biofilms form on boat surfaces (on the hull and inside the ballast tank), which can be difficult to completely remove. These biofilms have been shown to lead to corrosion and fouling (Lappin‐Scott and Costerton 1998; Drake et al.2005; Lichter et al.2009; Hadfield 2010; Fitridge et al.2012; Watson et al.2015; Schaerer et al.2019). Additionally, several studies have demonstrated that bilge water sustains many micro‐organisms (Olivera et al.2003; Sivaraman et al.2011; Cappello et al.2012; Qun et al.2012; Out of the Bilge 2015; Schaerer et al.2019). Bilge water is water that collects at the bottom of the hull, which can be derived from rainwater, or outside seawater that makes its way into the boat through small leaks, typically in the propeller seal. Bilge water is discharged back into the environment and is not regulated for removal of biological organisms in the same way as ballast water (Federal Register 2012; IMO 2017a, 2018).
We previously characterized microbial communities of boats and shipping ports around the world (Schaerer et al.2019; Ghannam et al.2020). Using source tracking software, we identified that, on average, about 40% of the bilge microbial community and 52% of the boat surface microbial community were sourced from the water (Schaerer et al.2019). However, this percentage was highly variable between geographical locations and vessels, suggesting there are other factors which also influence the boat microbiome. For example, airborne micro‐organisms and micro‐organisms found on docks are potential sources for the boat microbial community. If these sources do contribute to the boat microbial community, there is potential that microbes from these sources could be carried with vessels when they travel.
Advances in next‐generation sequencing and culture‐independent methods have shown that the air is home to a diverse population of microbes (Lighthart 1997; Pace 1997; Brodie et al.2007). Air microbial communities vary significantly between different environments and change in response to several factors: season, climate, solar radiation and other environmental meteorological parameters (Lighthart 1997; Pace 1997; Brodie et al.2007; Bowers et al.2011). Studies have shown the microbial community of air is influenced by many different sources including soil, vegetation and faecal material (Lindemann et al.1982; Bowers et al.2011).
Aside from a study looking at iron‐oxidizing bacteria on docks made from steel sheet piling in the Duluth‐Superior harbour (Hicks 2007), we do not know of any studies looking at dock microbial communities. We expect that the dock will have similarities to dust microbial communities because dust may be carried onto the dock on the shoes of passengers or blown onto the dock by wind. Dust microbial communities are diverse and have been shown to vary between locations depending on climate (Barberán et al.2012). A study of the sources of microbes in household dust found that outdoor material tracked inside was a major contributing source to household dust (Rintala et al.2012). Because of this finding, dock microbial communities may be shaped by the microbial communities of soil. Soil has also been shown to be diversely populated by microbes including some human pathogens (Adams et al.2015; Fahimipour et al.2018). One study looking at patterns of soil microbial communities across North America, South America and Antarctica found soil samples to be dominated by Acidobacteria, Proteobacteria and Verrucomicrobia (Barberán et al.2012). Another study showed that soil samples from Oklahoma were dominated by Planctomycetes, Firmicutes and delta‐Proteobacteria (Adams and Errington 2009).
Here, in addition to the microbial community in water, we explore other sources which we expect to influence the boat microbial community. Our previous work primarily focused on water as a source of non‐native species to the bilge water and boat surface (Schaerer et al.2019). It is important to know the extent to which other boat surfaces are colonized by micro‐organisms and the environmental sources of these microbes. Here we look at other locations on a boat and evaluate the potential for these locations to serve as a vector for the spread of non‐native organisms. Additionally, the different physical conditions of the bilge compartment compared to surface settings may allow for differential seeding by different sources. For example, the bilge compartment is enclosed below the deck and is protected from weather, whereas the hull and transom are located on the outside surface of the boat and are exposed to weather and water splash. Looking at three locations on the boat (hull, transom and bilge), we attempt to quantify how much each of these sources contributes to the microbial community composition of the boat.
In particular, we investigated the influence of dock and air microbial communities on the boat microbiome, attempting to explain some of the variability in the boat microbial community identified in our previous study. Flow rate could also be an important factor affecting community growth and composition in different locations on the boat. Previous studies have shown that the thickest biofilms form on surfaces when the flow rate is very slow (Lau and Liu 1993). Biofilms can release cells in response to quorum sensing or in response to increased sheer stress related to flow rate (Donlan 2002). Additionally, we explore the variability in microbial communities on each of the different boat sites over time. Here we performed a pilot scale study which looked at only one boat over the course of 11 days.
We hypothesize that (i) dock and air microbial communities influence the microbial consortia present on the boat, further explaining the variable influence of water microbial communities. Because the flow of water against the hull of the boat is intermittent depending on the boat’s movement, we hypothesize (ii) that the microbial consortia on the hull and transom of the vessel will be more variable than the microbial consortia in the bilge water.
Materials and methods
Sample collection
Samples were taken from the Portage Canal in Houghton, MI during the summer of 2018 on 11 days between June 19 and July 3. We collected four sample types for this study: water, air, dock, bilge water, hull and transom. The number of each type of sample collected is shown in Table S1. Water samples were collected from the dock at the Great Lakes Research Center at Michigan Technological University (47·120571 N, −88·545425 W). For each day, we collected triplicate 1‐l water samples from the Portage Canal and triplicate 1‐l water samples near the dock. Each litre of water was filtered through a glass fibre pre‐filter and a 0·2 μm pore size polyether sulfone (PES) post‐filter using a peristaltic pump. After filtering, both filters were treated as independent samples, which is why the number of water samples is so much higher than the number of the other sample types. Boat samples were collected from the SV Osprey, which is a 24‐foot vessel with an inboard/outboard motor. Bilge and boat surface samples were collected using sterile swabs (puritan sterile polyester tipped swabs). Three swabs were collected and pooled into a single tube. Air samples were collected by attaching a 0·2‐μm filter to the end of a filter housing and using a vacuum pump to pull air through the filter for 10 min. Triplicate air samples were collected for each day. Figure S1 gives an overview of the sample collection process. Samples were frozen at −80°C until further analysis was performed. While our air and water samples can be considered independent of each other due to the constant movement of both air and water, our boat samples and dock samples may not be fully independent of each other due to collection of samples from the same boat at different times. However, for our statistical analysis, we considered each sample to be independent.
DNA extraction and sequencing library preparation
DNA was extracted from half of each filter and the other half was stored at −80°C as an archive. All three of the swabs were used for the extraction, due to the expected low biomass of these samples. We used the ZymoBIOMICS DNA Microprep kit (Zymo Research Corporation, Irvine, CA). The manufacturer’s protocol was followed with the following exceptions: samples were processed in a homogenizer for 200 s at 5 m s−1 and then centrifuged at 12 000 g for 1 min. At least one extraction blank was done for every 23 samples that were extracted. Extraction blanks were empty tubes from the extraction kit and did not include sterile swabs or filters. The swabs and filters were sterile, but there is some possibility for low level contamination introduced by the sampling instrument that was not controlled for here. 16S rRNA sequencing libraries were prepared according to a modified version of the Illumina 16S rRNA Metagenomic Sequencing Library Preparation protocol. Briefly, an initial PCR was performed using Thermo Scientific Phusion Flash PCR Master Mix to amplify the V4–V5 region of the 16S rRNA gene using the primers 515YF and 926R (Parada et al.2016). We used AxyPrep Mag PCR Clean‐Up beads to purify PCR products from the first round of amplification, according to the Illumina 16S rRNA protocol. A second 8‐cycle PCR was performed to add Illumina adaptors and indices for multiplexed sequencing. For each sample, distinct 12 base pair go‐lay barcodes were added to the amplicons. Samples were then pooled to result in roughly similar amounts of PCR product for each sample. The pool of 16S rRNA gene products was then diluted to 4 × 10−6 mol m−3 and sequenced using an Illumina MiSeq. Sequencing was done using a v3 600‐cycle Reagent Kit to produce 2x300 paired end run.
16S rRNA processing and statistical analysis
Raw 16S rRNA sequencing reads were demultiplexed by the Illumina MiSeq. We used the DADA2 (divisive amplicon denoising algorithm) package in R (Callahan et al.2017) to overlap paired‐end reads, merge, quality filter and remove internal standard (phiX). Amplicon sequence variants (ASVs) were inferred. Denoised reads were merged, bimeric reads were removed and ASVs were assigned using the Silva database v132. 16S rRNA sequence processing followed a similar process to the one described in Schaerer et al.(2019). The complete pipeline for our analysis can be found at https://github.com/lgschaer/boatbiome.
Analysis of sequences was performed in R (R Core Team 2013), ver. 3.6.0. The sequence table was subset to remove quality control blanks. Diversity analysis was performed with the phyloseq package (McMurdie and Holmes 2013). To determine the extent to which alternative sources contribute to the boat microbiome, we first needed to examine the microbial diversity of each sample type and determine if the microbial community in the samples were sufficiently different from one another so that we could differentiate between the sources. To begin, our data were rarefied using the ‘rarefy even depth’ function in phyloseq to a minimum sequencesample size of 1017 reads. Before further analysis, mitochondrial DNA and chloroplasts were removed from our dataset.
Alpha diversity was measured using the ‘estimate richness’ function in phyloseq; the metrics Shannon and Observed ASVs were used. To determine whether there was a significant difference between the sample types, a Kruskal–Wallis test was performed using the FSA (Ogle et al.2020) package in R. Due to our relatively small number of samples, Kruskal–Wallis was used because of its compatibility with non‐normally distributed data and imbalanced datasets. Base R (R Core Team 2013) was used to perform a Dunn test (post‐hoc) to determine between which sample types there was a significant difference. We also used phyloseq to examine differences in the microbial community composition between different sample types. A t‐SNE plot was used to visualize these differences using a Bray–Curtis dissimilarity matrix. A permutational multivariate analysis of variance (permanova) was performed using vegan (Dixon 2003) to look for statistically significant differences between pairwise comparisons of sample types.
We used SourceTracker2 (Knights et al.2011) to quantify the mixing proportions for various sources of the boat microbial community. SourceTracker uses Latent Dirichlet Allocation and Gibbs sampling to deconvolute the mixing proportions of a sink sample into the source components. This approach has been used previously to understand the sources of microbes on surfaces in the built environment (Bowers et al.2011; Knights et al.2011; Shin et al.2015; Henry et al.2016). Our air, dock and water samples were used as sources and our boat (surfaces and bilge) samples were used as sinks. For our ASV table, we used the same table that we had already rarefied in Phyloseq so we did not rarefy our data in SourceTracker. Analysis was done with SourceTracker2 to look for additional sources of microbes which contribute to the boat microbial community (Fig. 5). Our model was trained using dock, air and water samples as sources. SourceTracker will classify any of sources it does not recognize into an ‘unknown’ category. Once the model was trained, we classified the proportion of these three sources in our sinks which were the different boat samples: bilge, transom and hull. SourceTracker was run five times, and the results are shown in Table S5. Any sequences which were not associated with one of these sources were put into the unknown category.
Results
Overall bacterial diversity
To compare richness and evenness of each sample type, we calculated alpha diversity of each of the different sample types using two measures: Shannon Diversity and Observed ASVs (Fig. 1). Overall, water had the widest range of diversity and the highest median values for both diversity metrics. Air had the lowest median diversity for both measures; however, air had the second highest range in diversity, suggesting that air samples are highly variable. Aside from water, bilge samples had the highest median diversity of any sample type for both alpha diversity measures (Fig. 1). Hull, transom and dock samples all had similar median diversity for both measures (5·04, 5·01 and 4·72, Shannon diversity, respectively).
Alpha diversity of each sample type for both observed ASVs and Shannon diversity. letters represent sample types which are significantly different from each other (Dunn post‐hoc test, NS = not significant).
A Kruskal–Wallis test showed a statistically significant difference between sample types for both metrics (Shannon: P value <0·001, chi‐squared 60·24, Degrees of Freedom 5. Observed: P value <0·001, chi‐squared 60·10, Degrees of Freedom 5, Table S2). A Dunn post‐hoc test was performed to determine the statistically significant difference between pairs of sample types. For Shannon diversity, the bilge versus air (P value = 0·02), water versus air (P value <0·01), water versus transom (P value <0·01) and water versus dock (P value <0·01) comparisons were significant (Fig. 1, Table S3). For Observed ASVs, the bilge versus air (P value = 0·03), water versus air (P value <0·01), water versus transom (P value <0·01) and water versus dock (P value <0·01) comparisons were significant (Fig. 1, Table S3).
A t‐Distributed Stochastic Neighbour Embedding plot (t‐SNE) was made to visually represent between sample diversity (Fig. 2). Close proximity of two points on the t‐SNE plot suggests that those samples have very similar microbial communities. Briefly, most of the water samples fall on the right side of the plot, with some division between the pre‐ and post‐filters, suggesting that there is some variability between the microbial community captured on two filter types. Most of the bilge samples form a tight cluster near the top left of the plot, suggesting that bilge water has a microbial community that is unique from the other sample types, although two bilge samples cluster more closely with the water samples which indicates some level of similarity between the water microbial community and the bilge microbial community. The air samples are scattered throughout the left side of the plot, demonstrating a high variability in air microbial community composition. The dock and transom form a tight cluster in the lower left corner of the plot suggesting that these sample types have more in common with each other in terms of microbial community composition than with the other sample types.
t‐SNE plot comparing the microbial community composition of samples to one another. Squares represent glass fibre filters, circles represent 0·2 micron filters, and diamonds represent swab samples. (
) Air, (
) Bilge, (
) Transom, (
) Hull, (
) Dock and (
) Water.
These observations were supported by the results from permanova analysis (Table S4). All comparisons were statistically significant (significance level of 0·05), indicating a statistically significant difference in sample composition between sample types. All P values were <0·001, except for the dock–transom comparison where the P value was equal to 0·03. This suggests, as observed in Fig. 2, that the dock and transom are more similar to each other than to water, bilge and air samples. With alpha diversity, we observed that the transom and dock were similar to each other in terms of the microbial community richness and evenness; here we observe that transom and dock samples are also similar in terms of microbial community composition.
Sources and sinks
Analysis with SourceTracker showed that overall, air was not a significant source for the boat microbial community (Fig. 3, Table S5). Air only contributed a small percentage (<1 ± 0·006%) to bilge, hull and transom. Water seemed to influence a small portion of the bilge microbial community (5·63 ± 0·164%), but water was shown to contribute <1 ± 0·014% to hull and transom. The dock was shown to be a substantial source of the microbial community to hull and transom (14·40 ± 0·343% and 64·48 ± 0·200%, respectively).
Summary of proportion of each boat sample (sink) attributed to each source by analysis with SourceTracker. (
) Air, (
) Dock, (
) Water and (
) Unknown.
Stability of microbial communities
To further explore differences in the microbial community composition between our sample types, we looked at the proportion of each phyla present in each sample type. Across all sample types, there was a relatively high proportion of Bacteroidetes and Proteobacteria making up a substantial portion of the microbial community. On the hull, there was a considerable portion of Deinococcus‐Thermus which was higher than any other sample type. Verrucomicrobia are present in higher proportions in air, bilge, transom and water than in the dock samples. Acidobacteria are present in the air, transom and hull samples at higher proportions than the other sample types. Both Verrucomicrobia and Acidobacteria are in relatively low proportions overall. Cyanobacteria and Planctomycetes are present in relatively low proportions across all sample types.
To observe the change in taxa across the time of sampling, we separated the taxa plot by sampling date (Fig. 4). Overall, the microbial community composition did not change considerably over the course of sampling with a few exceptions. The hull samples had higher abundances of Deinococcus‐Thermus from June 26 to June 28. Towards the end of sample collection, the samples became overall less diverse and were dominated by Actinobacteria, Bacteroidetes and Proteobacteria in addition to Deinococcus‐Thermus.
Taxa present in each sample type presented as an average percentage over the entire study. Proteobacteria are shown at class level, all other taxa are represented at phylum level. (
) Acidobacteria, (
) Actinobacteria, (
) Alphaproteobacteria, (
) Armatimonadetes, (
) Bacteroidetes, (
) Chloroflexi, (
) Cyanobacteria, (
) Deinococcus‐Thermus, (
) Deltaproteobacteria, (
) Firmicutes, (
) Fusobacteria, (
) Gammaproteobacteria, (
) Gammatimonadetes, (
) Planctomycetes, (
) Spirochaetes, (
) Verrucomicrobia and (
) Taxa <1% relative abundance.
Overall, air samples had the most variation in phyla present throughout the study. Cyanobacteria were intermittently present in air samples between 1 and 4% of total reads. Planctomycetes and Chloroflexi were noted on June 28th at higher abundances (2 and 1%, respectively) than any other day. Actinobacteria were present in air in highly variable amounts ranging from 44 to 4%. Verrucomicrobia were noted in the air on June 21 (4%) but were not present in any other days at abundances >1% (Fig. 5). The bilge samples in this present study were dominated by Actinobacteria, Bacteroidetes and Proteobacteria (1–12%, 3–39% and 49–65%, respectively). Due to limited replication in our dataset, these observations were not confirmed with a statistical test.
Taxa present in each sample type presented as an average percentage for each day during the study to illustrate the changes in microbial community composition over time Proteobacteria are shown at class level, all other taxa are represented at phylum level. Missing samples were removed during rarefying due to lack of sequencing depth. Proteobacteria are shown at class level, all other taxa are represented at phylum level. () Acidobacteria, () Actinobacteria, () Alphaproteobacteria, () Armatimonadetes, () Bacteroidetes, () Chloroflexi, () Cyanobacteria, () Deinococcus‐Thermus, () Deltaproteobacteria, () Firmicutes, () Fusobacteria, () Gammaproteobacteria, () Gammatimonadetes, () Planctomycetes, () Spirochaetes, () Verrucomicrobia and () taxa <1% relative abundance.
While the water and air are constantly in motion, it is likely that the individual samples are independent of each other. However, it is possible that the hull, transom and dock samples are not fully independent of each other since the same microbial community was re‐sampled on multiple time points throughout the study. This limits our ability to measure variation in the microbial community throughout the study.
To investigate our second hypothesis that the transom and hull were the most variable sites on the boat, we measured the CV to summarize the changes in within‐sample diversity. A high CV indicates a large variability in diversity within a group of samples, whereas a low CV suggests that samples in a given group have very little variability in diversity. We did this calculation for both sources and sinks. According to CV analysis, the air was the most variable source followed by the dock and the water (Fig. 6). Of the boat samples, the transom had the highest average CV (0·166) closely followed by the hull (0·160). The bilge had the lowest CV of any boat site (0·148).
Coefficient of variation showing the variability in the mean alpha diversity compared to the standard deviation of alpha diversity in each sample type over the course of the study.
Discussion
Microbial diversity of boat sites
We previously characterized boat microbial communities from vessels around the world, showing a diverse microbial community in the bilge compartment and on the boat surface (Schaerer et al.2019). In this present study, we looked at the role of the air and dock microbial communities in influencing the microbial community of boats and explored how dock and airborne sources can influence the boat microbial community. The dominant taxa in our boat samples (hull and transom) were Proteobacteria, Bacteroidetes and Actinobacteria. Hull samples also had higher proportions of Deinococcus‐Thermus.
The taxa present in the hull and transom samples varied throughout the course of the study. This is supported by previous studies which found diverse microbial communities in air (Lighthart 1997; Brodie et al.2007; Bowers et al.2011; Leung et al. 2014; Lymperopoulou et al.2016), soil (Feinstein et al.2009; Youssef and Elshahed 2009; Barberán et al.2012) and dust (Barberán et al.2015; Fahimipour et al.2018) samples. Our air samples were primarily made up of Alphaproteobacteria, Gammaproteobacteria, Actinobacteria and Bacteroidetes; in a survey of outdoor air microbial communities in the Midwest, Bowers et al. found the same phyla to be dominant (Bowers et al.2011). Our dock samples were dominated primarily by Alphaproteobacteria throughout the study, which is supported by a survey of outdoor dust microbiomes in which samples were also dominated by Alphaproteobacteria (Barberán et al.2015).
Taxa present in the hull, transom and bilge varied considerably between the different sample types over the course of the study. The transom had the highest concentration of Cyanobacteria compared to the other boat sample types. This makes sense as the boat was docked with the rear facing east during the course of the study, allowing for lots of morning sunlight. The boat would have been in the shade during the remainder of the day which is a possible explanation as to why the hull did not have the same large population of Cyanobacteria. The bilge is a mostly enclosed environment, so it makes sense that the microbial community composition remained relatively consistent throughout the course of this study and was not as strongly influenced by the environment. This experiment was carried out on a pilot scale and the results of this study should be confirmed with additional studies using more than one boat in additional locations for more robust results.
Sources and sinks
SourceTracker has been used to previously characterize many different types of samples including air (Lymperopoulou et al.2016), boat (Schaerer et al.2019) and dust (Fahimipour et al.2018) samples. It has also been successfully used to detect contamination of coastal areas with wastewater (Henry et al.2016) and identify microbial sources in several environments including ICU wards, classrooms and office environments (Adams et al.2015). The SourceTracker results from this study suggest the water microbial community influences a small portion of the bilge microbial community. In contrast, the water microbial community was not shown to be a significant source to the microbial community of the hull or transom. This contradicts what we found previously in our global study where we found that on average 52% of the hull microbial community was derived from the water (Schaerer et al.2019).
We hypothesize this discrepancy is due to the fact that the boat in this study was docked for the entire period of the study, decreasing the opportunity for micro‐organisms from the water to colonize the boat due to reduced water splash on the sides of the boat. Additionally, since the boat was docked, this increased the opportunity for micro‐organisms from the dock to colonize the boat. We hypothesize that the dock was an important source to the boat microbial community because it was influenced by the same air and dust microbial communities as the dock. In addition to being initially influenced by similar sources, the boat and the dock had similar environments for the duration of the study so it is likely that similar forces such as heat, moisture and sunlight would select for similar micro‐organisms on both objects.
Stability of communities
The dock samples were dominated by Proteobacteria (94–65% throughout the sampling period) and a considerably smaller population of Bacteroidetes (15–8% throughout the sampling period). We hypothesize that dock microbial communities will be similar to soil microbial communities because soil can be tracked onto docks and dust can be blown onto docks by wind. Our results showed that the dock microbial community varied over time in terms of which phyla were present. This is consistent with previous studies which have shown soil microbial communities to be diverse and highly variable between sites (Tringe et al. 2005; Roesch et al.2007). Few microbes are able to grow in dust due to lack of moisture; therefore, most of the microbes in dust originate from other sources such as outdoor air and soil, humans and animals (Rintala et al.2012). Very little work has been done previously to characterize the microbial communities of hull, transom and bilge water (Schaerer et al.2019). Because the hull and transom are exposed to the environment (wind, rain, etc.), we hypothesize that this allows for many different variables to impact the boat microbiome.
A study of the outdoor air microbiome of eight different Midwest cities showed that the air contained a diverse bacterial community with seven bacterial phyla represented with a considerable proportion of the community made up of Proteobacteria, Actinobacteria, Bacteroidetes and Firmicutes (Bowers et al.2011). The phyla represented in the Midwest study are consistent with the phyla found in our samples. In our air samples, 14 bacterial phyla were represented at abundances >1%; Proteobacteria, Bacteroidetes and Actinobacteria dominated our air samples, which agrees with the Midwest study. The high variability in the diversity of our air samples agrees with previous research showing links between weather changes and variability in air microbial communities (Lighthart 1997; Pace 1997; Brodie et al.2007; Bowers et al.2011). Our samples were collected over the span of 13 days (no samples were collected on June 23, June 24, June 30 and July 1) in varying weather conditions (Table S6).
Coefficient of variation analysis showed that air was the most variable of our proposed sources to the boat microbiome. This agrees with previous studies which found air to have a highly variable microbiome (Lighthart 1997; Bowers et al.2011; Lymperopoulou et al.2016). This high variability in air could also be due to the limited sample collection time used in this study. We collected triplicate air samples, filtering for 10 min per sample. A longer time of filtration may have resulted in a more representative air sample. Dock was the next most variable of our proposed sources. There is very little research on dock microbiomes; however, soil microbial communities have been found to be highly variable (Bowers et al.2011; Barberán et al.2012). We show the transom and hull microbial communities to be more variable than the bilge microbial communities. This agrees with our previous work on boat microbiomes which showed that port water was a bigger source to boat surfaces than to bilge water (Schaerer et al.2019). We hypothesize that this is because the bilge is mostly closed off and somewhat sheltered from the environment. This could have implications on whether or not microbes are able to persist in the bilge as a boat travels.
In conclusion, we evaluated the influence of dock and air microbial communities on the boat microbial community. We found that the air contributed relatively little to the overall boat microbiome. However, we showed that the dock microbial community contributes substantially to the hull and transom microbial communities (between 14 and 64%). This suggests that terrestrial sources such as dust may be an underappreciated source of microbes to boats and other vessels which could serve as a vector to move terrestrial micro‐organisms around. This should be confirmed with other studies looking at more boats in more locations. Additionally, studies should investigate how long terrestrial and aquatic microbial signatures persist on a boat when it travels. This is important to know before making conclusions about boats as vectors of transport for micro‐organisms. Our second hypothesis demonstrated that the transom (closely followed by the hull) was the most variable part of the boat, supporting our hypothesis that the transom would be the most variable part of the boat. This shows that the outside surfaces of the boat are highly variable in terms of the microbial community and change in response to travel and other changes in conditions.
This pilot scale study was done on a small scale to explore microbial diversity on various sites of a boat. Since only one boat was sampled during this study, the conclusions we make here should be confirmed by further studies with additional boats over longer periods of time. Additionally, since this study was performed over 11 days, there may be larger trends, such as seasonal changes in how the various sources impact the boat, which are not described here.
Acknowledgements
This work was funded by DARPA Young Faculty Award D16AP00146. We would like to thank the Michigan Technological University Great Lakes Research Center for help with sample collection from the SV Osprey.
Conflict of Interest
The authors declare no conflict of interests.
Author contributions
L.G.S. processed the samples, analysed the data and wrote the paper. P.N.W. helped design the experiment, collected samples and processed samples. A.C. helped to design the experiment and collected samples. W.C.C. collected the samples and helped to write the paper. S.M.T. helped design the experiment, oversaw the data analysis and helped to write the paper.
