A Comparison of Microbial Water Quality and Diversity for Ballast and Tropical Harbor Waters
A Comparison of Microbial Water Quality and Diversity for Ballast and Tropical Harbor Waters
Charmaine Ng 0 1
Thai-Hoang Le 0 1
Shin Giek Goh 0 1
Liang Liang 0 1
Yiseul Kim 0 1
Joan B. Rose 0 1
Karina Gin Yew-Hoong 0 1
0 1 National University of Singapore, Department of Civil and Environmental Engineering, Singapore, Singapore, 2 Michigan State University, Department of Microbiology and Molecular Genetics , East Lansing, Michigan , United States of America, 3 National University of Singapore Environmental Research Institute (NERI), Singapore , Singapore
1 Editor: Adelaide Almeida, University of Aveiro , PORTUGAL
Indicator organisms and antibiotic resistance were used as a proxy to measure microbial water quality of ballast tanks of ships, and surface waters in a tropical harbor. The survival of marine bacteria in ballast tanks appeared to diminish over longer water retention time, with a reduction of cell viability observed after a week based on heterotrophic plate counts. Pyrosequencing of 16S rRNA genes showed distinct differences in microbial composition of ballast and harbor waters. The harbor waters had a higher abundance of operational taxonomic units (OTUs) assigned to Cyanobacteria (Synechococcus spp.) and α-proteobacteria (SAR11 members), while marine hydrocarbon degraders such as γ-proteobacteria (Ocenspirillaes spp., Thiotrchales spp.) and Bacteroidetes (Flavobacteriales spp.) dominated the ballast water samples. Screening of indicator organisms found Escherichia coli (E. coli), Enterococcus and Pseudomonas aeruginosa (P. aeruginosa) in two or more of the ballast and harbor water samples tested. Vibrio spp. and Salmonella spp. were detected exclusively in harbor water samples. Using quantitative PCR (qPCR), we screened for 13 antibiotic resistant gene (ARG) targets and found higher abundances of sul1 (4.13-3.44 x 102 copies/mL), dfrA (0.77-1.80 x10 copies/mL) and cfr (2.00-5.21 copies/mL) genes compared to the other ARG targets selected for this survey. These genes encode for resistance to sulfonamides, trimethoprim and chloramphenicol-florfenicol antibiotics, which are also known to persist in sediments of aquaculture farms and coastal environments. Among the ARGs screened, we found significant correlations (P<0.05) between ereA, ermG, cfr and tetO genes to one or more of the indicator organisms detected in this study, which may suggest that these members contribute to the environmental resistome. This study provides a baseline water quality survey, quantitatively assessing indicators of antibiotic resistance, potentially pathogenic organisms and a broad-brush description of difference in microbial composition and diversity between open oceans and tropical coastal environments through the use of next generation sequencing technology.
Data Availability Statement: All 16S rRNA gene
sequence data files are available from the NCBI SRA
the SRA archive with the accession number
Funding: Funding was provided by the Public
Utilities Board (PUB) adminstered under the
Environment & Water Industry Program Office
Incentive for Research & Innovation Scheme (EWI
IRIS) scheme (reference: 1102-IRIS-12-02), and
administrative support from the National University of
Singapore Environmental Institute (NERI).
Competing Interests: The authors have declared
that no competing interests exist.
The city-state of Singapore has one of the busiest ports in the world receiving an average arrival
of 140,000 vessels and 2.33 billon gross tonnes of ballast water per year . These commercial
ships use ballast water to provide stability and maneuverability, taking up waters when cargo is
being unloaded and discharging ballast waters at ports of call . This ballast water exchange
can contain a diverse community of organisms, including invasive species (e.g. toxic
dinoflagellates, Vibro spp.) and potentially impose an environmental and health threat to coastal regions,
where ballast waters are released [3–7]. Drake & Lodge (2004) identified regions in Southeast
Asia, which may function as hotspots for invasion due to high shipping activity, which in turn
facilitates microbial transport . Asexual reproduction and dormancy are strategies, which
aid the survival of non-indigenous invasive microbes under environmental stressed conditions.
These traits help preserve invasive microbes in the nutrient depleted, light scarce environments
of ballast water tanks, and upon discharge to coastal regions where conditions are more
optimal, enables successful colonization by rapid population growth .
The deballasting of waters from ships acts as a vehicle for the global distribution of
pathogens (and possibly antibiotic resistant forms) and waterborne diseases, which may have an
adverse impact on humans, marine animals and the aquatic ecosystem as a whole .
Thomson et al. (2003) also showed that horizontal gene transfer of antibiotic resistant genes (ARGs)
might occur due to the closed system and long water retention time within ballast tanks .
Cell death of antibiotic resistant bacteria (ARB) results in the release of intracellular ARGs,
which may eventually be transferred into other viable cells within the microbial community
imparting resistances in recipient cells. ARGs are increasingly being recognized as an emerging
contaminant hence there is an urgency towards early detection in an effort to prevent
widespread dissemination of antibiotic resistance .
In a study of bacteriological assessment of ballast waters of six ships docked at Singapore
harbors, Joachimsthal et al. (2004) used fluorescence in situ hybridization (FISH) coupled with
flow cytometer to enumerate total bacteria, and fluorescent tags to differentiate between
Enterobacteria, Vibrio spp., and E. coli. Using these techniques, their findings showed that
Enterobacteria (range: 0–1.45 x 105 cells/mL) and Escherichia coli (E. coli) (range: 0–7.87 x 105
cells/mL) cells were not always present in all samples, but Vibrio spp. (range: 1.45 x 104–1.07 x
105 cells/mL) cells were [
In this study, we investigated the microbial content of mid-ocean exchanged (MOE)
ballast waters obtained from grab samples from ballast tanks calling at the port of Singapore,
and harbor waters surrounding these ships to evaluate differences in microbial diversity
between sample types. We expected the ballast waters to represent the microbiological
content of the waters at the sites of ballast water uptake, taking into consideration the possibility
of the contribution of biofilms, which may develop on the surfaces of ballast water tanks
during the voyage. Collectively, we applied pyrosequencing of tagged 16S rRNA gene amplicons
as a semi-quantitative method to elucidate differences in microbial diversity, and a culturing
and quantitative PCR (qPCR) based approach to survey for the presence of indicator
organisms, ARB and associated resistance genes. Among known invasive species, we were
interested in the pathogens rather than phytoplankton. Specifically, we focused on monitoring
known opportunistic pathogens including E. coli, Enterococcus spp., Vibrio spp., and
Pseudomonas aeruginosa to determine if levels measured complied with the International Maritime
Organization (IMO) guidelines.
2 / 22
Materials and Methods
Sampling of ballast and harbor waters
Ballast water samples were collected from three vessels with voyages within the Middle East
and Asian regions. Each ballast water sample was acquired in the morning, on-board the
docked vessel at the Port of Singapore. Seawater next to the hull of the docked ships was
collected on the same morning of ballast water sampling for comparative analysis. Ballast waters
1, 2 and 3, (abbreviated as BW1, BW2 and BW3) were taken from the South China Sea, Gulf of
Aden and to the west of South Korea in the East China Sea accordingly. The retention time of
BW1, 2 and 3 (from the time of collection from the source ports to docking at the Port of
Singapore) were 4, 107 and 11 days, respectively. Harbor waters corresponding to BW1, BW2 and
BW3 were abbreviated as HW1, HW2 and HW3. Permission for water sampling was granted
by the Maritime Port Authority of Singapore and did not involve endangered and protected
Environmental parameters including temperature, pH, salinity and turbidity were measured
on site using a YSI probe, model 63 (YSI incorporated, Yellow Springs, Ohio, USA). Using a
bucket, approximately 60 L of surface water was hauled up through a manhole leading to the
ballast tank. Likewise, 60 L of surface water surrounding the docked vessels was sampled. The
water samples were concentrated through hollow-fiber ultrafiltration (Hemodialyzer
RexeedTM-25S, Asahi Kasei Medical Co., Ltd) to a final volume of 500 mL according to the
method of Hill et al. 2005. A volume of 20 mL of concentrated water samples (equivalent to 2.4
L of original water sample), were used for total DNA extraction. Whole water samples were
taken for culturing work.
The bacterial DNA from the concentrated samples was extracted with the PowerWater DNA
Isolation kit (Mo Bio Laboratories, Inc, Carlsbad, CA). DNA yields and purity were measured
using a nanodrop spectrophotometer (NanoDrop 2000, Thermo Scientific, Wilmington, USA).
The extracted DNA was used for quantification of ARG and P. aeruginosa and Salmonella
Detection of ARG by PCR
A total of 14 primer pairs were used to detect the presence of ARGs, with one primer pair
targeting 16S rRNA gene sequences to detect the total bacterial population (Table 1). These
primers were also designed to quantify the target genes in real time or quantitative Polymerase
Chain Reactions (PCR). Among these, 10 primers were selected from published work,
reporting a good detection and quantification of relevant ARGs in qPCR, while 4 primers were
designed in this study to target and quantify the specific ARGs, i.e. ereA, ereB, sul1, and cfr
genes. These gene sequences were downloaded from the GeneBank Database (http://www.ncbi.
nlm.nih.gov/) and primers were designed using PRIMER 3 tool . PCR amplicon sizes were
specified in the range of 100–300 base pairs (bp) for qPCR analyses.
All PCR assays were performed on a Veriti 96-Well Fast Thermal Cycler (Applied
Biosystem, CA, USA). Each PCR reaction included 2 μL 10× Faststart PCR buffer (Roche, IN, USA),
0.25 mM dNTPs (Promega, Singapore), 0.0375 U/μL Taq Faststart, DNA polymerase (Roche,
IN, USA), 0.5 mM forward and reverse primers each (AITBiotech, Singapore), 20–200 ng
genomic DNA template, with the final volume adjusted to 20 μL using diethylpyrocarbonate
(DEPC)-treated distilled deionized water (PureLab Option, Elga, Singapore). The thermo
cycles for PCR amplifications were carried out under the following settings: 95°C for 5 min;
3 / 22
repeated 40 times with a cycle of denaturation at 95°C for 30 s, at annealing temperatures
specific for each primer set for 30 s, elongation at 72°C for 30 s, and finally 72°C for 10 min .
ARGs were chemically synthesized (AITBiotech, Singapore) and the gene products served as a
positive control. For a negative control, the genomic DNA of a susceptible E. coli strain (not
containing these ARGs) was used. PCR products were visualized using DNA electrophoresis
on a 2% agarose gel stained with GelRedTM (Biotium, Inc., Hayward, CA). The gel was
visualized under UV light with an E-Gel1 Imager System (Life Technologies, Carlsbad, CA).
Quantification of 16S rRNA genes and ARGs by Real-time qPCR
The calibration curves for qPCR quantification for 16S rRNA genes and each ARG were built
using the ten-fold dilution series of ARG standard DNA. Triplicate samples were used for each
dilution. Each qPCR reaction contained 10 μL SYBR Select Master Mix (Applied Biosystem,
CA, USA), 0.25 μL 10 mM forward primers and reverse primers each, 2 μL DNA template, and
7.6 μL DEPC-treated distilled deionized water. qPCR reactions were performed in a
StepOnePlus Real-Time PCR system (Applied Biosystem, CA, USA) with the following settings:
holding at 50°C for 2 min and 95°C for 2 min; 40 cycle amplification under conditions of 95°C for
15 s with annealing and elongating at 60°C for 1 min, and finally, a melting curve analysis at
95°C for 15 s, 60°C for 1 min, and 95°C for 15 s. The amplification profiles for each series of
DNA dilutions during 40 cycles were recorded. A threshold cycle (Ct) value was determined as
the cycle number where fluorescence data crossed the threshold line, which is manually defined
within the logarithmic increase phase. The gene copies of 16S rRNA genes and ARGs were
calculated as previously described . The molecular weight of a gene was determined by
multiplying the size of a gene in base pairs and the average molecular weight of a double-stranded
DNA base pair (649 Da). Absolute gene abundance was calculated as gene copies normalized
to the volume of ballast and harbor water samples, and relative gene abundance was calculated
as percentage of ARGs normalized to the abundance of 16S rRNA gene.
Determining the phenotypic concentrations of ARB
Marine agar (BD Diagnostics, Maryland, USA) was used to estimate the concentrations of viable
and culturable heterotrophic bacteria. Marine agar plates were supplemented with antibiotics to
determine the concentrations of ARB. The antibiotics (purchased from Sigma-Aldrich, USA)
of the following concentrations were used in the screening process; trimethoprim (4 μg/mL),
sulphanilamide (300 μg/mL), norfloxacin (10 μg/mL), lincomycin (100 μg/mL), tetracycline
5 / 22
(35 μg/mL), erythromycin (100 μg/mL), kanamycin (80 μg/mL), and linezolid (8 μg/mL).
Concentrations used for screening were higher than published minimum inhibitory concentration
(MIC) breakpoints in the 2013 Clinical and Laboratory Standards Institute (CLSI) performance
standards for Antimicrobial Susceptibility Testing report. Raw water samples were serially diluted
in M9 minimal media and 100 μL of samples were spread plated and incubated at 35°C for 24 h
and subsequently at room temperature for an additional 24 h. Plates containing colonies within
the range of 30–100 cells were counted.
Measuring indicator organisms using IDEXX kits and qPCR
ColilertTM and EnterolertTM (IDEXX Laboratories, Inc., Westbrook, Maine, USA) were used to
quantify the cell numbers of E. coli and Enterococcus, respectively. 100 mL of raw water and a
10 fold diluted raw water sample were tested to allow a wider range of detection due to high
fluctuations in bacterial concentration. The water sample was mixed with the ColilertTM or
EnterolertTM reagent in a sterile bag and poured into a multi-well tray (Quanti-Tray/2000).
The results were expressed as the most probable number (MPN) according to the presence of
fluorescence in individual wells when visualized under long-wave ultraviolet light (365 nm)
after 24 h of incubation at 37°C (for ColilertTM) and 41°C (for EnterolertTM).
Enumeration of pathogenic bacteria
Vibrio spp. were enumerated according to the method of Dickinson et al. (2013) with some
modifications. Briefly, ballast and harbor water samples at different dilutions (0.1mL, 1mL, 10mL and
100mL) were used to determine the most probable number (MPN) by enrichment in alkaline
peptone water (Sigma-Aldrich, St Louis, MO). The samples were incubated overnight at 37°C. A
volume of 1 mL of enriched sample was centrifuged at 10 000 g for 10 min to pellet the culture,
which was subsequently resuspended in 200 μL of nuclease-free water and heated to boiling for
10 min to release the DNA for the PCR analysis. The primers used for the detection of Vibrio
spp. are listed in Table 1. PCR was carried out in a thermal cycler (Eppendorf, Germany) and
performed in 25 μL volumes consisting 12.5 μL of 2x GoTaq1 master mix (Promega, Madison,
WI), 1.25 μL of each primer (10 μM), 7.5 μL of nuclease-free water and 2.5 μL of extracted DNA
from the Vibrio spp. cultures. PCR annealing temperatures and durations for different target
Vibrio spp. are shown in Table 1. The conditions for PCR amplification was an initial denaturing
cycle at 94°C for 2 min, followed by 30 cycles of denaturation at 94°C for 30 s, at an annealing
temperature as stated in Table 1 for 1 min and extension at 72°C for 1 min. A final extension step
at 72°C for 5 min was also included at the end of 30 cycles. The amplified products were
separated by electrophoresis on a 1.5% agarose gel stained with GelRedTM (Biotium, Inc., Hayward,
CA). The separated fragments on the agarose gel were visualized under UV light with the
E-Gel1 Imager System (Life Technologies, Carlsbad, CA).
P. aeruginosa and Salmonella spp. were measured with real-time qPCR method. Primer and
probe sets for both P. aeruginosa and Salmonella spp. are summarized in Table 1. A 20 μL
qPCR reaction was prepared which contained 10 μL of FastStart Universal Probe Master
(ROX) (Roche, Germany), 2 μL of forward and reverse primer each, 0.5 μL of probe, 0.5 μL of
nuclease-free reagent-grade water and 5 μL of DNA template. Each set of reactions included
both positive (recombinant plasmid DNA) and negative controls (double-deionized water).
qPCR amplification for both targets was performed on StepOnePlusTM Real-time PCR system
at 50°C for 2 min, 95°C for 10 min, and 45 cycles of 95°C for 15 s and 60°C for 1 min. The
threshold cycle values (Ct) obtained from qPCR analysis were converted to gene copies (GC)
based on the equations generated from standard curves [Salmonella spp.: Ct = -3.4651 log
(GC/μl) + 40.416; P. aeruginosa: Ct = -3.341 log (GC/μl) + 42.466].
6 / 22
16S rRNA tag encoded gene sequencing and data analysis
To determine microbial community composition, bacterial tag-encoded FLX amplicon
pyrosequencing (bTEFAP) using primers 515F (5’-CACGGTCGKCGGCGCCATT-3’) and 806R
(5’-GGACTACHVGGGTWTCTAAT-3’) targeting the V4 region of the 16S rRNA gene was
performed using the Roche 454 FLX Titanium platform , at the Molecular Research LP
(Texas, USA). Datasets obtained from the analyses were processed with using version 1.33.3
MOTHUR . Sequences were trimmed to remove barcodes and primers, quality checked to
have an average Phred quality score of 35 or above, with sequences exceeding a homopolymer
stretch of more than 8 bases being excluded from the dataset. Sequences were aligned by
MOTHUR against the SILVA bacterial reference database (14,956 sequences), screened,
filtered and de-noised. UCHIME was used to identify and remove chimeric sequences. Datasets
for each sample were normalized by subsampling each dataset to 3,230 reads . Operational
taxonomic units (OTUs) were assigned at a distance of 3% using an average
neighbor-clustering algorithm. Community diversity between samples was assessed using the observed number
of OTUs, Shannon and Simpson indices. Rarefaction curves were generated using MOTHUR
and taxonomy of OTUs were assigned using the SILVA database with abundances expressed as
percentage of sequences assigned to specific taxa at the phyla, class and family level over the
total number of sequenced reads for each sample.
Environmental and biological datasets were analyzed using the statistical software package
Primer version 6 [
]. A square-root transformation was applied to the OTU abundance, ARB,
ARG, indicator organisms, pathogenic bacteria and environmental datasets. Bray-Curtis
similarity values were calculated for the OTU dataset to determine the extent of similarity of
microbial community composition between samples and visualized as a dendrogram. A SIMPER
analysis was used to determine the taxa, which defined difference in the harbor and ballast
water samples. An ANOSIM test was used to verify the significance of observed differences in
the microbial community structure. A Principal component analysis (PCA) was used to
determine the relationship between the microbial community structure (based on bray-curtis
similarity indices), and other biological and environmental parameters. Only parameters that had a
Pearson correlation coefficient of >0.7, with a significance of <0.05 were represented in the
Bacterial cell and gene counts
To estimate the concentrations of viable bacterial cells in the two sample types, whole water
fractions were plated on marine agar. The average (n = 3) heterotrophic plate counts (HPC) in
ballast waters (4.31 x 103 CFU/mL) was approximately a magnitude lower than harbor waters
(1.06 x 104 CFU/mL, Fig 1). When determining concentrations by qPCR for total bacterial
populations including both the culturable and non-culturable bacteria, the average 16S rRNA
gene copies in ballast samples (1.40 x 105 CFU/mL) were higher compared to the harbor waters
(7.24 x 104 CFU/mL). Discrepancies in the counts of viable HPC and total bacterial populations
within ballast water samples could be due to long retention time and thus, nutrient depletion
and reduction of viable cells in the ballast tank over the voyage. This is evident in BW2 and
BW3 samples where the ballast waters had a longer retention time (107 and 11 days,
respectively), and showed higher 16S rRNA gene copy concentrations and lower HPC compared to
BW1 samples, which were held in the ballast water tanks for 4 days. Overall, two of the ballast
7 / 22
Fig 1. Concentrations of culturable bacteria and total 16S rRNA genes in ballast and harbor waters.
waters in our study had significantly lower concentrations of viable HPC than harbor waters
(with exceptions of BW2 and HW2).
Microbial diversity of ballast and harbor waters
A total of 57,576 sequences assigned to 1,395 unique OTUs were identified at the genus level
(97% similarity cut off) from all samples. Sequences were deposited into the NCBI SRA archive
website (http://www.ncbi.nlm.nih.gov/) and available under SRA accession number
SRP065174. There were 136 OTUs shared by microbial communities, with 619 and 640 OTUs
exclusive to the ballast and harbor communities, respectively (Fig 2). Within the ballast and
harbor samples, the harbor waters were more similar in microbial composition (73 common
OTUs) in comparison to the ballast waters (20 common OTUs).
Rarefaction curves indicated that a sufficient number of sequences were sampled to cover
the majority of species in the microbial communities (S1 Fig). Shannon and Simpson indices
showed that BW1 had highest species diversity (Shannon– 4.315, Simpson– 0.9501) and
richness (441 OTUs) compared to the two ballast water samples which had values lower than those
of the HW samples (Table 2). In general, the HW samples appeared to be more diverse than
the BW samples with the exception of BW1, which was taken up at from the South China Sea.
Dominant taxa representing an average of >1% of total assigned sequences across all sample
types included Proteobacteria (49.36 ± 22.54), Cyanobacteria (25.34 ± 27.84), Bacteroidetes
(11.78 ± 6.25), Planctomyces (4.43 ± 2.93), Chloroflexi (3.39 ± 7.78), Actinobacteria
8 / 22
Fig 2. Total OTUs identified in ballast (BW) and harbor waters (HW) from community sequencing analysis.
(1.80 ± 0.88) and Aquificae (1.59 ± 2.32) (Fig 3). A SIMPER analysis showed an average
dissimilarity of 52% between the ballast and harbor water sample types. The harbor waters were
enriched in Cyanobacteria sequences (~30 fold) while ballast waters had more Proteobacteria
(~2 fold), Bacteroidetes (~2 fold) and Chloroflexi (~18 fold) sequence. A SIMPER analysis
showed that these taxa accounted for 90% of the total dissimilarity between the microbial
communities of the two sample types.
At the family level, among the dominant phyla, Sphingomonadales, Oceanospirillales,
Thiotrichales and Flavobacteriales were more represented in the ballast waters (Table 3). The higher
relative abundance of Chloroflexi differentiated BW1 among the other water samples, and
9 / 22
deeper inspection of the sequence reads showed that Caldilineales was the main contributor to
the abundance. One of the striking differences between the ballast and harbor water sample
types was the prevalence of SAR11 and Synechococcus spp. in the harbor waters.
Quantification of indicator organisms
Enterococcus and E.coli counts were higher in harbor water samples, with the exception of BW3.
Viable indicator organisms were detected in the harbor water samples and together with BW3
(from the Gulf of Aden) exceeded the acceptable levels of (100 CFU/100 mL) of intestinal
Fig 3. Relative abundance of total sequences assigned at the phyla level.
10 / 22
Enterococci according to IMO guidelines. E. coli was detected in BW1 but at levels within the
acceptable range. Indicator organisms were present in all the harbor water samples at levels
higher than those in the ballast samples. Vibrio species (including toxin-producing species) were
below the detection limit in all ballast water samples, but were detected in two of the harbor
water samples (HW1 and HW2, Table 4). V. parahaemolyticus was absent in all samples,
however V. cholerae and V. vulnificus were detected in two of the harbor water samples. MPN-PCR
results showed that BW1 and BW2 were positive for P. aerunginosa with a relatively low
concentration of 19–20 MPN/100 mL in comparison to two harbor samples, HW1 and HW2, which
had much higher concentrations (1,322–19,393 MPN/100 mL, Table 4). Higher levels of P.
aeruginosa gene copies were detected in two of the harbor water samples (HW1 and HW2) with
lower levels (100–1,000 fold less) in the ballast waters (BW1 and BW2, Table 4).
Detection of ARBs and ARGs
There was no colony growth on norfloxacin-supplemented plates across all samples. BW1 and
HW1 samples showed resistance to the remaining 7 antibiotics while BW2 and BW3 as well as
HW2 and HW3 showed resistance to 5 antibiotics (Fig 4). In most cases, the ballast water
samples had significantly lower counts of ARB compared to the harbor water samples (Fig 4).
Among the harbor water samples, HW1 had the highest resistant counts to trimethoprim,
sulfanilamide and lincomycin (1.8–7.0 x 105 CFU/mL).
Fig 4. The concentrations of heterotrophic bacteria and their resistance to 8 antibiotics in ballast and harbor waters (A) BW1 and HW1, (B) BW2
and HW2, (C) BW3 and HW3. TRI: Trimethoprim; SUL: Sulfanilamide; NOR: Norfloxacin; LIN: Lincomycin; TET: Tetracycline; ERY: Erythromycin; KAN:
Kanamycin; LINE: Linezolid. Two-tailed T-test, *P<0.05, **P<0.01.
Genes linked to resistance of sulfanamide (sul1), trimethoprim (dfrA) and
chloramphenicol-florfenicol (cfr) were most abundant across all samples with levels ranging from 4.13–
3.44 x 102, 0.77–1.80 x 101, 2.00–5.21 gene copies/mL respectively (Fig 5). Sul1 genes were at
12 / 22
Fig 5. The abundance of the ARGs in ballast and harbor waters. (A) BW and HW1, (B) BW2 and HW2, (C) BW3 and HW3. Two-tailed T-test, *P<0.05,
**P<0.01. ARGs with non-detects in both BW and HW samples were excluded.
higher concentrations in harbor waters, while sul2 and dfrA concentrations were higher in the
harbor waters with the exception of HW2. In general, harbor water samples had higher
measurements of ARGs in comparison to the ballast waters.
13 / 22
Correlations between microbial community composition, environmental variables, indicator organism loads and antibiotic resistance
The pH of water samples ranged between 5.98–7.87, with HW2 having lowest value measured
(pH 5.98, Table 5). HW1 and HW2 samples also showed slightly higher temperatures and
turbidity measurements than the rest of the samples. HW1 had the lowest salinity levels, which
could be attributed to dilution due to rainfall prior to sampling. A Pearson correlation of
environmental variables to ARG, ARB and indicator organisms showed negative correlations
between pH and crf gene copies, linezolid resistant counts and P. aeruginosa gene copy
numbers (P<0.05, Table 6). Temperature and turbidity measurements were positively correlated to
V. vulnificus (P<0.05, Table 6). Significant correlations were found between the ARGs eraA,
ermG, cfr and tetO and indicator organisms (P<0.05, Table 7).
A clustering analysis showed that the microbial community in the harbor waters was more
similar in microbial composition than ballast waters at a similarity cut off of 75% (Fig 6). At
the same level of similarity, BW2 and BW3 showed more resemblance compared to BW1, and
this clustering pattern observed on a Principal Component Analysis (PCA) plot was
statistically verified using an ANOSIM analysis (P<0.05).
Using the PERMANOVA analysis in the Primer v6 software, microbial community patterns
in the harbor samples were positively correlated (using a multivariate analysis where Pearson
correlation coefficient of >0.7) to measurements of E. coli, V. cholera, P. aeruginosa, V.
vulnificus, temperature, tetO and ermF1 genes (Fig 7). The PCO1 and PCO2 explained 60.3% and
26.4% of the variation. The ARG ermC was positively correlated to microbial composition of
BW1, while BW2 and 3 correlated to 16S rRNA gene sequence (Fig 7).
P-values are represented in parenthesis with
Mid ocean exchange (MOE) of ballast waters has been implemented by IMO to prevent the
dissemination of invasive aquatic species. This method requires ships to exchange ballast water at
least 200 nautical miles away from shore at an ocean depth of at least 200 m. It is proposed that
replacement of ballast water with mid ocean waters (containing oligotrophic marine bacteria)
flushes or dilutes any potential coastal dwelling invasive organisms, which may have been
picked up during the voyage of ships. In a few other studies, this method of water replacement
has shown no significant decrease in bacterial counts (pre- versus post-exchange). Rather, a
decline in bacterial abundance in ballast waters was mainly attributed to longer water retention
time in ballast tanks [
8, 35, 36
]. Overall, two of the ballast waters in our study had a significant
lower abundance of viable HPC than harbor waters (with exceptions of BW2 and HW2). In an
earlier study using flow cytometry, Joachimsthal et al. (2003) reported bacterial concentrations
of ballast waters in Singapore ranging from 2.35 × 106–5.87 × 107 cells/mL, with differences
between harbor and ballast waters attributed to differences in location and season of ballast
water uptake and open ocean water samplings [
In our study, the levels of indicator organisms measured in ballast waters were within limits
administered by IMO, with the exception of one ballast water sample where Enterococcus
concentrations exceeded guideline values. In comparison to another study using FISH techniques,
levels of Vibrio spp. and E. coli in our study was generally much lower than previously reported
]. We found a correlation between higher temperatures, turbidity and V. vulnificus
abundance in harbor waters (P<0.05). Studies have shown that marine pathogens such as Vibrio
spp. may preferentially thrive in coastal areas with increased water temperatures and reduced
salinity (NaCl <25 ppb, [
]) due to terrestrial runoff.
Pseudomonas spp. and Salmonella spp. are not listed as ballast water indicator organisms of
concern according to IMO guidelines. However, we selected these bacteria for testing based on
detection in other ballast water studies. For example, Burkholder et al. (2007) reported the
presence of other pathogens including P. aeruginosa in 48% of ballast waters surveyed from
vessels in the east and west coast of the United States [
]. In another study using chromogenic
agar, Altug et al. (2012) were able to detect viable Salmonella spp. (300–500 CFU/100 mL) in
15 / 22
Fig 6. Clustering analysis of microbial composition in ballast and harbor waters based on Bray-Curtis similarity index.
ballast waters coming from different regions in the world but not in the Sea of Marmara,
Turkey where the ships docked [
]. For this study, although Pseudomonas spp. in ballast waters
were detected at levels lower (0–20 GC/mL) relative to the other indicator organisms, their
ability to form biofilms on the surfaces of tanks and resistance to environmental stress make them
important targets to monitor [
Compared to soil, antibiotic resistance of bacteria in marine environments has not been
well documented. The ocean is a natural habitat of antibiotic-producing bacteria with marine
aquaculture introducing more antibiotics to treat infections and boost aquacuture production.
Plasmids are mobile vectors for the transfer of resistance genes that confer an advantage to
their host’s adaptation to different ecological niches [
]. In this study, samples were analyzed
for the presence of ARB and associated ARGs, which are increasingly being recognized as
xenobiotics, and emerging contaminants in the aquaculture industry. As a whole, the
culturable marine microbial communities in ballast and harbor waters were most resistant to
trimethoprim, sulfanilamide, lincomycin and tetracycline. Shah et al. (2014) reported the same
patterns of ARB (by proxy of pheno- and genotypic screening) in Chilean marine sediments of
16 / 22
Fig 7. Principal component analysis (PCA) of the microbial community structure of ballast and harbor waters in relationship with pathogen, ARB
and ARG data. Only variables with a Pearson correlation coefficient of >0.7 were included in the PCA plot. EC–E. coli, VC–V. cholera, P. aeruginosa, VV–V.
salmon fish farms, which suggests that aquaculture activity likely contributes to antibiotic
resistance patterns in the marine environment [
]. Interestingly, BW1 had the lowest levels of
phenotypic resistance to all antibiotics tested, which could be attributed to its more complex and
diverse microbial community structure resulting in its distinct characteristics compared to the
other samples. The ARB population in BW1 samples could be more fastidious in nature,
17 / 22
incapable of growing on marine media supplemented by antibiotics. This reasoning is
supported by comparing the pheno- and genotypic data which shows the identification of a wide
array of ARG targets in BW1 compared to the other samples. This highlights the importance of
using both culturing and screening of antibiotic resistance gene markers for a more thorough
assessment of antibiotic resistance given that pheno- and genotypic results may contradict
each other due to confounding effects of cell viability and other growth impediments.
In most of the samples for which ARGs were detected, higher concentrations were found in
the harbor water samples than a non-impacted freshwater body in Singapore, where ARG
concentrations were at least half or those reported in the harbor samples (non-impacted site: sul1–
1.11 x 102 gene copies/mL, cfr–below detection limit). The persistence of sul1 and dfrA genes
together with the relatively high levels of viable ARB counts (for most but not all of the
samples), suggests successive pressure exerted by sulfamethoxazole, trimethoprim or
co-trimoxazole, a combination of the former two antibiotics. Sulfonamide and tetracycline are commonly
used to treat bacterial and protozoan infections in aquaculture leading to the prevalent problem
of resistant bacteria in these environments [
]. Even without selective pressure,
sulfonamide-resistant bacteria can remain stable in the aqueous environment for 5 to 10 years [
Other genes including tetM, crf, ermB1, C, F1 and G were also identified at lower
concentrations in all samples except for BW3 and HW3. Most marine environmental surveys, particularly
in fish farms, have identified tetracycline, chloramphenicol and macrolide resistant gene markers
]. Strong positive correlations were shown between cfr and P. aeruginosa gene copies
(in HW2), and E. coli and ereA gene copies (in HW1) while ARGs ermG and tetO correlated with
a combination of at least four of the indicator organisms tested in this study. These correlation
patterns suggest that these indicator organisms may be contributing to the resistome of coastal
environments. Plasmids carrying the cfr gene mediating resistance to linezolid and
chloramphenical have only been decribed in Staphylococcus and Enterococcus [
] but not in other taxa of
pathogens. Likewise, there has been no reported association of resistance to macrolides in Vibrio
spp. encoded by the ermG gene. In aquacuture sources, it has been observed that genera of
Pseudomonas carry florfenicol resistant genes [
]. It would be worthwhile to taxonomically identify
a variety of antibiotic resistant phenotypes and test for the presence of these ARGs to verify this
correlation. Determining this would enable us to estimate the risk of dissemination of ARGs
between the environmental reservoir and human pathogens.
The PCA analysis showed a tighter clustering of the microbial composition of harbor waters
indicating higher similarities in structure as compared to the ballast waters which are more
dispersed. Although there are some distinguishing bacterial populations among the ballast water
samples, it is apparent that diversity is largely dependent on the site of water uptake. The
harbor waters were more similar in composition and showed significant positive correlations to
indicator organisms (E. coli), and pathogens (V. cholera, P. aeruginosa, V. vulnificus),
temperature and ARGs (tetO and ermF1 genes).
By comparing microbial communities of ballast and harbor waters, we found low levels of
similarity at the species level, which is expected given that the harbor samples are more
terrestrially impacted than off-shore oligotrophic seawater samples. At the phyla level,
Proteobacteria dominated across all samples, with SAR11 showing notably higher levels of abundance in
harbor waters. Members of the SAR11 clade are ubiquitous, and highly adapted to nutrient
poor oligotrophic environments such as surface waters of open oceans [
]. Open ocean
SAR11 populations, unlike their coastal counterparts, exhibit diel oscillations at the transcript
level and may regulate light-dark cycles of transcriptional activity by tracking surrounding
environmental conditions [
]. Steindler et al. (2011) found that in light-deprived, nutrient
starved conditions, Candidatus Pelgibacter ubique, a member of the SAR11 clade, are able to
utilize carbon reserves, shrinking to a volume of ~0.014 μm3 [
]. The marine cyanobacteria,
18 / 22
Synechococcus, were also detected at higher levels in harbor waters. The dominance of
phototrophs in the harbor waters versus ballast waters (where illumination is absent in the tanks) is
influenced by sampling in the late morning at the surface, within the photic zone of the water
column. In contrast, nutrient and light-stress in ballast tanks have been shown to take a toll on
autotrophic nanoflagellates and mixotrophic populations, with the loss of photosynthetic
pigments and dramatic decreases in counts over time [
]. Together, these findings offer a rational
explanation for observed low levels of phototrophic cyanobacteria in the ballast water samples
and also may suggest a selection of open ocean SAR11 phenotypes.
Taxa associated with marine hydrocarbon degradation (Oceanspirillales, Thiotrichales,
Flavobacteria) were enriched in the ballast water samples. Members of these taxa are capable of
hydrocarbon oxidation, and have been identified in sediments of marine hydrocarbon seeps
from 16S rRNA sequence clone libraries of environmental samples [
]. BW1 had the highest
microbial diversity with high abundances of Caldilineales and Sphingobacteriales, which are
commonly found in activated sludge in wastewater treatment plants [
]. Over summer
periods, Zhang et al. (2014) detected the prevalence of Caldilineales in the subtropical Pearl
River Estuary sediments which receives large volumes of inflow from the South China Sea and
anthropogenic inputs from surrounding industries . Interestingly, the last point of ballast
water uptake for BW1 was in the South China Sea which possibly suggests that a combination
of an upwelling event, sediment dispersal and water circulating patterns facilitated the
hitchhiking of these microbes in ballast water from this point of origin to our local ports.
The habor waters showed more similarity in microbial community structure while the
ballast waters had more variability. This is expected given that ballast water of the ships surveyed
were situated at diferent locations during uptake. Incorporating methods of assessing microbial
community composition by next-generation sequencing (NGS) techniques, in tandem with
monitoring indictor organisms and nutrient inputs with the effect on microbial growth and
potential harmful algal provides a comprehensive approach of understanding ecological impact
of shipping activity and ballast water dispersal.
In conclusion, we observed differences in the bacterial community composition, indicator
organisms (E. coli, Enterococcus, P. aeruginosa, Salmonella spp., Vibrio spp.), and the concentrations
of ARB and ARGs in ballast water of 3 ships and harbor waters at their port of call. Harbor waters
exhibited a general trend of higher concentrations of indicator organisms, however we detected the
presence of Enterococcus in all ballast water samples, with one sample, in particular, exceeding
levels recommended by IMO. We found correlations between the concentrations of indicator
organisms (E. coli, V. cholera, V. vulnificus, P. aeruginosa) and ARGs, which suggests that these species in
ballast and harbor waters may be carriers of ARGs. Selective culturing, isolation and antibiotic
susceptibilty testing of these marine species will be required to confirm if these potential pathogens are
indeed resistant and carriers of ARGs. We applied NGS, different molecular assays and
conventional microbiological methods to provide detailed microbial measurements and other predictive
parameters which may dictate the threat to microbiological safety standards in ballast waters.
S1 Fig. Rarefaction curves of species richness (97% cut off) in ballast and harbor waters.
We would like to thank the Public Utilities Board (PUB) for funding support adminstered
under and the Environment & Water Industry Program Office Incentive for Research &
19 / 22
Innovation Scheme (EWI IRIS) scheme (reference: 1102-IRIS-12-02), and administrative
support from the National University of Singapore Environmental Institute (NERI). We also
thank Aurore Trottet, Guillaume Drillet and Rui Shan Ker from the Danish Hydraulic Institute
(DHI), and Genevieve Gabrielle Rose Valdes Vergara, Fang Haoming and Laurence Haller for
assistance with sampling and engaging in scientific discussion.
Conceived and designed the experiments: CN THL SGG LL YK JBR KGYH. Performed the
experiments: CN THL SGG LL YK. Analyzed the data: CN THL SGG LL. Contributed
reagents/materials/analysis tools: JBR KGYH. Wrote the paper: CN THL JBR KGYH.
20 / 22
21 / 22
Singapore MPA . Annual report . 2013 .
Ruiz GM , Rawlings TK , Dobbs FC , Drake LA , T. M, A. H , et al. Global spread of microorganisms by ships . Nature . 2000 ; 408 : 2 .
Hallegraeff GM , Steffensen DA , Wetherbee R . Three estuarine Australian dinoflagellates that can produce paralytic shellfish toxins . Journal of plankton research . 1988 ; 10 : 9 .
McCarthy SA , Khambaty FM . International diseemination of epidemic Vibrio cholerae by cargo ship ballast and other non potable waters . Applied and Environmental Microbiology . 1994 ; 60 : 5 .
Marine Ecology Progress Series . 1997 ; 161 : 8 .
Knight I , Wells CS , Wiggins B , Russell H , Reynolds KA . Dection and enumeration of fecal indicators and pathogens in the ballast water of transoceanic vessels entering the Great Lakes . General Meeting of American Society for Microbiology American Society for Microbiology. 1999 .
Mimura H , Katakura R , Ishida H . Changes of microbial populations in a ship's ballast water and sediments on a voyage from Japan to Qatar . Marine Pollution Bulletin. 2005 ; 50 : 7 .
Drake JM , Lodge DM . Glocal hot spots of biological invasions: evaluating option for ballast-water management . The royal society of publishing . 2004 ; 271 : 6 .
Thomson FK , Heinemann SA , Dobbs FC . Patterns of antibiotic resistance in cholera bacteria isolated from ships' ballast water . Proceedings of the Third International Conference on Marine Bioinvasion; La Jolla , California: Old Dominion University; 2003 .
Pruden A , Pei R , Storteboom H , Carlson K. Antibiotic Resistance Genes as Emerging Contaminants: Studies in Northern Colorado . Environmental Science & Technology . 2006 ;( 40 : ): 6 .
11. Joachimsthal EL , Ivanov V , Tay ST , Tay JH . Bacteriological examination of ballast water in Singapore Harbour by flow cytometry with FISH . Mar Pollut Bull . 2004 ; 49 ( 4 ): 334 - 43 . doi: 10 .1016/j.marpolbul. 2004 . 02 .036 PMID: 15341828 .
12. Joachimsthal EL , Ivanov V , Tay ST -L, Tay J-H . Bacteriological examination of ballast water in Singapore Habour by flow cytometry with FISH . Marine Pollution Bulletin 2004 ; 49 : 10 .
Untergasser A , Cutcutache I , Koressaar T , Ye J , Faircloth BC , Remm M , et al. Primer3 -new capabilities and interfaces . Nucleic Acids Research . 2012 ; 240 ( 15 ).
Le TH , Sivachidambaram V , Yi X , Li X , Zhou Z. Quantification of polyketide synthase genes in tropical urban soils using real-time PCR . J Microbiol Methods . 2014 ; 106 : 135 - 42 . doi: 10 .1016/j.mimet. 2014 .
08.010 PMID: 25181693 .
Ausubel FM , Brent R , Kingston RE , Moore DD , Seidman JG , Smith JA , et al. Short protocols in molecular biology . Trends in Biotechnology . 1992 ; 10 ( 0 ): 456 . doi: http://dx.doi.org/10.1016/ 0167 - 7799 ( 92 ) 90304 -E.
Dowd SE , Callaway TR , Wolcott RD , Sun Y , McKeehan T , Hagevoort RG , et al. Evaluation of the bacterial diversity in the feces of cattle using 16S rDNA bacterial tag-encoded FLX amplicon pyrosequencing (bTEFAP) . BMC Microbiology . 2008 ; 8 .
Schloss PD , Westcott SL , Ryabin T , Hall JR , Hartmann M , Hollister EB , et al. Introducing mothur: opensource, platform-independent, community-supported software for describing and comparing microbial communities . Applied and environmental microbiology . 2009 ; 75 : 5 .
Edgar RC , Haas BJ , Clemente JC , Quince C , Knight R. UCHIME improves sensitivity and speed of chimera detection . Bioinformatics . 2011 ; 27 ( 16 ): 7 .
19. Clarke KR , Warwick RM . Change in marine communities: an approach to statistical analysis and interpretation . 2nd ed2001.
20. Amann RI , Binder BJ , Olson RJ , Chisholm SW , Devereux R , Stahl DA . Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations . Applied and Environmental Microbiology . 1990 ; 56 ( 6 ): 7 .
21. Muyzer G , de Waal EC , Uitterlinden AG . Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA . Applied and Environmental Microbiology . 1993 ; 59 ( 3 ): 6 .
22. Koike S , Aminov RI , Yannarell AC , Gans HD , Krapac IG , Chee-Sanford JC , et al. Molecular ecology of macrolide-lincosamide-streptogramin B methylases in waste lagoons and subsurface waters associated with swine production . Microbial Ecology . 2010 ; 59 ( 3 ): 12 .
23. Knapp CW , Zhang W , Sturm BS , Graham DW . Differential fate of erythromycin and beta-lactam resistance genes from swine lagoon waste under different aquatic conditions . Environ Pollut . 2010 ; 158 ( 5 ): 1506 - 12 . doi: 10 .1016/j.envpol. 2009 . 12 .020 PMID: 20053492 .
24. Zhu YG , Johnson TA , Su JQ , Qiao M , Guo GX , Stedtfeld RD , et al. Diverse and abundant antibiotic resistance genes in Chinese swine farms . Proceedings of the National Academy of Sciences . 2013 ; 110 ( 9 ): 3435 - 40 . doi: 10 .1073/pnas.1222743110
25. Dahmen S , Mansour W , Boujaafar N , Arlet G , Boualle`gue O. Distribution of Cotrimoxazole Resistance Genes Associated with Class 1 Integrons in Clinical Isolates of Enterobacteriaceae in a University Hospital in Tunisia . MICROBIAL DRUG RESISTANCE . 2010 ; 16 : 5 .
26. Lee JC , Oh JY , Cho JW , Park JC , Kim JM , Seo SY , et al. The prevalence of trimethoprim-resistanceconferring dihydrofolate reductase genes in urinary isolates of Escherichi coli in Korea . Journal of Antimicrobial Chemotherapy . 2001 ; 47 : 6 .
27. Rapid Detection of Bacterial Antibiotic Resistance: Preliminary Evaluation of PCR Assays Targeting Tetracycline Resistance Genes [Internet] . smithAustralia: Human Protection and Performance Division DSTO Defence Science and Technology Organisation; 2007
28. Smith MS , Yang RK , Knapp CW , Niu Y , Peak N , Hanfelt MM , et al. Quantification of Tetracycline Resistance Genes in Feedlot Lagoons by Real-Time PCR . Applied and Environmental Microbiology . 2004 ; 70 ( 12 ): 7372 - 7 . doi: 10 .1128/aem.70.12. 7372 - 7377 . 2004 PMID: 15574938
29. Dickinson G , Lim K -y, Jiang SC. Quantitative Microbial Risk Assessment of Pathogenic Vibrios in Marine Recreational Waters of Southern California. Applied and Environmental Microbiology . 2013 ; 79 : 9 .
30. Panicker G , Call DR , Krug MJ , Bej AK . Detection of Pathogenic Vibrio spp. in Shellfish by Using Multiplex PCR and DNA Microarrays . Applied and Environmental Microbiology . 2004 ; 70 : 9 .
31. Brasher CW , DePaola A , Jones DD , Bej AK . Detection of Microbial Pathogens in Shellfish with Multiplex PCR . Curr Microbiol . 1998 ; 37 : 7 .
32. Chun J , Huq A , Colwell RR . Analysis of 16S-23S rRNA Intergenic Spacer Regions of Vibrio cholerae and Vibrio mimicus. Applied and Environmental Microbiology . 1999 ; 65 : 7 .
33. Vezzulli L , Pezzati E , Moreno M , Fabiano M , Pane L , Pruzzo C , et al. Benthic ecology of Vibrio spp. and pathogenic Vibrio species in a coastal Mediterranean environment (La Spezia Gulf , Italy). Microbial Ecology . 2009 ; 58 : 11 .
34. Lee D-Y , Shannon K , Beaudette LA . Detection of bacterial pathogens in municipal wastewater using an oligonucleotide microarray and real-time quantitative PCR . Journal of Microbiological Methods . 2006 ; 65 : 15 .
35. Tomaru A , Kawachi M , Demura M , Fukuyo Y. Changes in Microbial Communities, Including both Uncultured and Culturable Bacteria, with Mid-Ocean Ballast-Water Exchange during a Voyage from Japan to Australia . PLoS One 2014 ; 9 ( 5 ).
36. J.M. S , C.J. W , R.B. R. Microbial hitchhikers: dynamics of bacterial populations in ballast water during a trans-Pacific voyage of a bulk carrier . Aquatic Invasions 2010 ; 5 : 10 .
37. Joachimsthal EL , Tay STL , Ivanov V , Tay JH . Monitoring the bacteriological quality of seawater by flow cytometry: development of a rapid method for the control of ballast water . Proceedings of the 2nd International Conference on Marine Waste Water Discharges; Istanbul, Turkey2002.
38. Joachimsthal EL , Ivanov V , Tay STL , Tay JH . Quantification of whole-cell in situ hybridisation with oligonucleotide probes by flow cytometry of Escherichia coli cells . World Journal of Microbiology and Biotechnology . 2003 ; 19 : 7 .
39. Baker-Austin C , Trinanes JA , Taylor NGH , Hartnell R , Siitonen A , Martinez-Urtaza J. Emerging Vibrio risj at high latitudes in response to ocean warming . Nature Climate Change . 2013 ; 3 : 5 .
40. Burkholder JM , Hallegraef GM , Melia G , Cohen A , Bowers HA , Oldach DW , et al. Phytoplankton and bacterial assemblages in ballast water of U.S. Military ships as a function of port origin, voyage time, and ocean exchange practices . Harmful Algae . 2007 ; 6 : 33 .
41. Altug G , Gurun S , Cardak M , Ciftci PS , Kalkan S. The occurance of pathogenic bacteria in some ships' ballast water incoming from various marine regions to the Sea of Marmara, Turkey . Marine Environmental Research . 2012 ; 81 : 8 .
42. Landini P. Cross-talk mechanisms in biofilm formation and responses to environmental and physiological stress in Escherichia coli . Research in Microbiology. 2009 ; 160 : 8 .
43. Brown Kav A , Sasson G , Jami E , Doron-Faigenboim A , Benhar I , Mizrahi I. Insights into the bovine rumen plasmidome . Proceedings of the National Academy of Sciences . 2012 ; 109 : 6 .
44. Shah SQ , Cabello FC , L' abée-Lund TM , Tomova A , Godfrey HP , Buschmann AH , et al. Antimicrobial resistance and antimicrobial resistance genes in marine bacteria from salmon aquaculture and nonaquaculture sites . Environmental Microbiology . 2014 ; 16 ( 5 ): 20 .
45. Tamminen M , Karkman A , Lohmus A , Muziasari WI , Takasu H , Wada S , et al. Tetracycline resistance genes persist at aquaculture farms in the absence of selection pressure . Environmental Science & Technology . 2011 ; 45 : 6 .
46. P.T.P. H , L. N , P.H. V, S. S. Detection of the sul1, sul2, and sul3 genes in sulfonamide-resistant bacteria from wastewater and shrimp ponds of north Vietnam . Science of the Total Environment . 2008 ; 405 : 8 .
47. Kim SR , Nonaka L , Suzuki S. Occurrence of tetracycline resistance genes tet(M) and tet(S) in bacteria from marine aquaculture sites . FEMS Microbiology Letters . 2004 ; 237 : 10 .
48. Gao P , Mao D , Luo Y , Wang L , Xu B , Xu L . Occurance of sulfonamide and tetracycline-resistant bacteria and resistance genes in aquaculture environment . Water Research . 2012 ; 46 : 10 .
49. Akinbowale OL , eng H , Barton MD . Antimicrobial resistance in bacteria isolated from aquaculture sources in Australia . J Appl Microbiol . 2006 ; 100 : 11 .
50. Akinbowale OL , Peng H , Barton MD . Diversity of tetracycline resistance genes in bacteria from aquaculture sources in Australia . Journal of Applied Microbiology . 2007 ; 103 : 10 .
51. Di Cesare A , Vignaroli C , Luna GM , Pasquaroli S , Biavasco F. Antibiotic-resistant enterococci in seawater and sediments from a coastal fish farm . Microbial Drug Resistance . 2012 .
52. Buschmann AH , Tomova A , Lopez A , Maldonado MA , Henriquez LA , Ivanova L . Salmon aquaculture and antimicrobial resistance in the marine environment . PLoS ONE . 2012 ; 7:e42724 . doi: 10.1371/ journal.pone.0042724 PMID: 22905164
53. L. D, P. K, R.E. M, D. P, K.V. S , C. A. A. Transferable plasmid-mediated resistance to linezolid due to cfr in a human clinical isolate of Enterococcus faecalis . Antimicrobial Agents and Chemotherapy . 2012 ; 56 ( 7 ).
54. Tripp HJ . The Unique Metabolism of SAR11 Aquatic Bacteria . Journal of Microbiolology . 2013 ; 51 : 5 .
55. Ottesen EA , Young CR , Gifford SM , Eppley JM , Marin R , Schuster SC , et al. Multispecies Diel Transcriptional Oscillations in Open Ocean Heterotrophic Bacterial Assemblages. Science 345 , 207 - 212 . Science. 2014 ; 245 :6. doi: 10 .1126/science.1252476 PMID: 25013074
56. Steindler L , Schwalbach MS , Smith DP , Chan F , Giovannoni SJ . Energy Starved Candidatus Pelagibacter Ubique Substitutes Light-Mediated ATP Production for Endogenous Carbon Respiration . PLoS One . 2011 ; 6:e19725 . doi: 10.1371/journal.pone.0019725 PMID: 21573025
57. Redmond MC , Valentine DL , Sessions AL . Identification of Novel Methane- , Ethane- , and Propane-Oxidizing Bacteria at marine Hydrocarbon Seeps by Stable Isotope Probing . Applied and Environmental Microbiology . 2010 ; 76 : 11 .
58. Sekiguchi Y , Yamada T , Hanada S , Ohashi A , Harada H . Anaerolinea thermophila gen . nov., sp. nov. and Caldilinea aerophila gen . nov., sp. nov., novel filamentous thermophiles that represent a previously uncultured lineage of the domain Bacteria at the subphylum level . International Journal of Systematic and Evolutionary Microbiology . 2003 ; I53: 8 .
59. Ye L , Zhang T . Bacterial communities in different sections of a municipal wastewater treatment plant revealed by 16S rDNA 454 pyrosequencing . Applied Microbiology and Biotechnology . 2012 ; 97 : 9 .
60. Zhang W , Bougouffa S , Wang Y , Lee OO , Yang JY , Chan C , et al. Towards Understanding the Dynamics of Microbial Communities in an Estuarine System . PLoS One . 2014 ; 9 ( 4 ).