Environmental Pollution 342 (2024) 123067

Contents lists available at ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

Metagenomic analysis of microbial communities and antibiotic resistant

genes in the Tijuana river, and potential sources☆
Shayla Shahar a, Karilyn E. Sant b, Nicholas Allsing a, Scott T. Kelley a, *
a
Department of Biology, San Diego State University, San Diego, CA 92182, USA
b
School of Public Health, San Diego State University, San Diego, CA 92182, USA

A R T I C L E I N F O A B S T R A C T

Keywords: The Tijuana River is a transborder river that flows northwest across the border from Baja California in Mexico
Metagenomics into Southern California before discharging into the Pacific Ocean. The river is frequently contaminated with raw
Tijuana river sewage due to inadequate sanitary infrastructure in Tijuana. To assess the type and degree of microbial
Surface water
contamination, water samples were collected monthly from a near-border and an estuarine site from August
SourceTracker
2020 until May 2021. A portion of each sample was used for epifluorescent microscopy and DNA was extracted
directly from the rest for shotgun metagenomic sequencing. After sequence quality checking and processing, we
used the rapid taxonomic identifier tool Kaiju to characterize the microbial diversity of the metagenomes and
matched the sequences against the Comprehensive Antibiotic Resistance Database (CARD) to examine antimi­
crobial resistance genes (ARGs). Bacterial and viral-like particle (VLP) abundance was consistently higher in the
near-border samples than in the estuarine samples, while alpha diversity (within sample biodiversity) was higher
in estuarine samples. Beta-diversity analysis found clear compositional separation between samples from the two
sites, and the near-border samples were more dissimilar to one another than were the estuarine sites. Near-border
samples were dominated by fecal-associated bacteria and bacteria associated with sewage sludge, while estuarine
sites were dominated by marine bacteria. ARGs were more abundant at the near-border site, but were also readily
detectable in the estuarine samples, and the most abundant ARGs had multi-resistance to beta-lactam antibiotics.
SourceTracker analysis identified human feces and sewage sludge to be the largest contributors to the near-
border samples, while marine waters dominated estuarine samples except for two sewage overflow dates with
high fecal contamination. Overall, our research determined human sewage microbes to be common in the
Tijuana River, and the prevalence of ARGs confirms the importance of planned infrastructure treatment upgrades
for environmental health.

1. Introduction Treating water is important because humans live, work, and play near
bodies of water, which if untreated can carry disease (Water Science
Pathogens in water quality can lead to health concerns ranging from School, 2018). Untreated water can also result in beach pollution and
gastrointestinal illnesses to neurological disorders (CDC, 2020). The contamination of wildlife in the ocean (Water Science School, 2018).
presence of bacteria, viruses, and other pathogens in bodies of water Antibiotic resistance is a major growing issue in public health (CDC,
have been found in the intestinal tracts of humans, indicating a corre­ 2022) and can be transmitted by bacteria, viruses, and even fungi (Singh
lation between these pathogens and disease (Cordy, 2014). Of the bac­ et al., 2022). Antibiotic resistance genes (ARGs) have been detected in
teria found in the human intestine, the majority are bacteria that shed in freshwater (Singh et al., 2022) and saltwater ecosystems (Jang et al.,
feces, live in symbiosis with the host, or are pathogenic to the host 2022), and even wastewater treatment plants (Jang et al., 2022; Singh
(Chahal et al., 2016). These pathogenic bacteria have been found to be et al., 2022). High levels of waterborne ARGs may put human health at
diverse and prevalent in sewage and wastewater, which is a major health risk. Zhang et al. found evidence of horizontal transfer of ARGs in
risk when it discharges into recreational waters (Chahal et al., 2016). aquatic environments (Zhang et al., 2009), and there is also evidence

☆
This paper has been recommended for acceptance by Dr. Sarah Harmon.
* Corresponding author. Department of Biology San Diego State University 5500 Campanile Drive San Diego, CA 92182-4614, USA.
E-mail address: skelley@sdsu.edu (S.T. Kelley).

https://doi.org/10.1016/j.envpol.2023.123067
Received 13 July 2023; Received in revised form 2 November 2023; Accepted 27 November 2023
Available online 1 December 2023
0269-7491/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Fig. 1. Map of the Tijuana River (blue) with both sites. The general location for the estuarine Boca Rio sample collections is indicated by the black circle. The near-
border Radio Club sample collection location is indicated with hollow red triangle. A scale bar is shown at the bottom left. (For interpretation of the references to
color in this figure legend, the reader is referred to the Web version of this article.)

that waterborne antibiotic resistant (ARG-containing) bacteria can 2. Materials and methods
transfer to humans (Larsson & Flach, 2022). Thus, characterization of
microbes and antibiotic resistance genes (ARGs) in contaminated water 2.1. Sample collection
will provide information on sewage contamination as the Tijuana River
discharges into recreational water. Thirteen samples were collected from two sites on August 28, 2020,
The Tijuana River lies across the California – Mexico border (Fig. 1) November 10, 2020, December 30, 2020, January 27, 2021, March 14,
with approximately 25% residing on the United States side (Sewage 2021, April 14, 2021, and May 5, 2021 (Fig. 1, Table 1). Boca Rio is an
Pollution within the Tijuana River Watershed | San Diego Regional Water estuary site located at the mouth of the river as it flows into the Pacific
Quality Control Board, n.d.). The northwest flowing river is primarily Ocean in Imperial Beach, CA. Radio Club is located near the U.S.-Mexico
surrounded by the City of Tijuana, Mexico (Sewage Pollution within the border in San Ysidro, CA, near the South Bay International Wastewater
Tijuana River Watershed | San Diego Regional Water Quality Control Board, Treatment Plant (SBIWTP). The SBIWTP is located within the bounds of
n.d.) and contains treated and untreated wastewater (US EPA, 2021). this watershed and can treat up to 20 million gallons per day of this
Due to the ongoing concern of Tijuana’s sewage problem, the South Bay untreated urban runoff and surface water discharge via an outfall off the
International Wastewater Treatment Plant (SBIWTP) was built in the U. coast in the Pacific Ocean. However, during storm events, this limit is
S. just north of Tijuana’s main wastewater pumping station to admin­ frequently exceeded, and excess water remains untreated and flows into
ister secondary treatment to the wastewater influx (South Bay IWTP, n. the estuary. Most of the land surrounding these sites is recreational,
d.). However, due to inadequate sewage infrastructure, sewage often open land managed by government agencies. However, there are several
collects in canyons and arroyos, which is then mobilized into the river residential homes within the bounds of the Tijuana River watershed in
during rain events. Therefore, the Tijuana River is well-known to have this area, and the northern and eastern bounds of this region are densely
high proportions of sewage, leading to frequent beach closures in the used for mostly residential and commercial purposes.
coastal town of Imperial Beach, CA. Water quality information is available in Rocha et al., since samples
In this study, we performed a metagenomic microbial and ARG were collected concurrently (Rocha et al., 2022). Five of the seven
contamination survey, and a microbial source tracking analysis, in samples collected in each location, indicated in Table 1, were collected
longitudinally collected water samples from cross-border sites and in the <72 h after rainfall ended. Each sample was collected using a telescopic
Tijuana River estuary. Our analysis found significant contributions of dipper sampler. The sampler was rinsed with river water 3 times prior to
fecal and wastewater contamination from cross-border flows and collecting the sample. Dipper samples were transferred to sterile
showed that the amount of fecal contamination in the estuary correlated Whirl-Pak® bags, transported to the lab in coolers with ice packs, and
with sewage overflows. In addition, we used microscopy to characterize immediately filtered through 0.22 μm Sterivex® filters until clogging
the microbial and viral loads at the two main sites and compared our prevented filtration. Between 50 and 100 mL of water was filtered per
metagenomic results to other fecal indicators E. coli (Criteria et al., 1985; sample.
Staley et al., 2018), and crAssphage (Park et al., 2020). The overarching
goal of this study was to characterize the abundance and types of mi­
crobes in the Tijuana River. By also assessing antibiotic resistance and 2.2. Microscopy (counts and biomass)
potential sources of contamination we can inform management pro­
tocols for waste management controls. Microscopy was performed to provide count estimates of microbes
and viral-like particles (VLPs). Samples were prepared by diluting 5 μL
(Radio Club) and 250 μL (Boca Rio) of each sample to 1 mL with mo­
lecular grade water into a 1.5 mL Eppendorf tube. Samples were inde­
pendently fixed for VLP and microbe counts and for biomass using
epifluorescent microscopy, as performed by Haas et al. (2014) and

2
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Table 1
Sample collection data.a
Lab Sample Code Sample Matrix Sample Site Date Sampled Grab Time <72 h after rainfall

RC August 2020 Surface Water Tijuana River - Radio Club 8/28/20 12:20:00 p.m. No
RC November 2020 Surface Water Tijuana River - Radio Club 11/10/20 9:40:00 a.m. Yes
RC December 2020 Surface Water Tijuana River - Radio Club 12/30/20 10:16:00 a.m. Yes
RC January 2021 Surface Water Tijuana River - Radio Club 1/27/21 10:05:00 a.m. Yes
RC March 2021 Surface Water Tijuana River - Radio Club 3/14/21 10:20:00 a.m. Yes
RC May 2021 Surface Water Tijuana River - Radio Club 5/5/21 10:01:00 a.m. Yes
BR August 2020 Surface Water Tijuana River - Boca Rio 8/28/20 1:20:00 p.m. No
BR November 2020 Surface Water Tijuana River - Boca Rio 11/10/20 10:40:00 a.m. Yes
BR December 2020 Surface Water Tijuana River - Boca Rio 12/30/20 11:12:00 a.m. Yes
BR January 2021 Surface Water Tijuana River - Boca Rio 1/27/21 11:10:00 a.m. Yes
BR March 2021 Surface Water Tijuana River - Boca Rio 3/14/21 11:28:00 a.m. Yes
BR April 2021 Surface Water Tijuana River - Boca Rio 4/5/21 11:10:00 a.m. No
BR May 2021 Surface Water Tijuana River - Boca Rio 5/5/21 10:33:00 a.m. Yes
a
A Radio Club sampling for April is not present due to filtration issues with that particular sample.

Roach et al. (2020). SYBR Gold stain can be used to detect both bacteria in-depth set of protein sequences for the bacteria, archaea, eukaryotes,
and VLPs, while DAPI staining will only detect bacteria (Haas et al., and viruses. Read counts for sequences matching E. coli and crAssphage
2014; Porter and Feig, 1980). were determined from the Kaiju analysis and output tables since
Briefly, for viral and microbe counts, 16 μL of 32% para­ sequence data from both these organisms were available in the nr_euk
formaldehyde (PFA) were added and vortexed, for a final concentration database. Kaiju2table was used to convert output files into a summary
of 1% v/v. One μL of 10,000x SYBR Gold stain was added to each fixed table signified by the domain, kingdom, phylum, class, order, family,
sample. The entire 1 mL of PFA-fixed sample was resuspended in 5 mL of genus, and species with each read count. A python script converted
0.02 μm filtered Sigma water and then filtered with a 0.02 μm Anodisc kaiju2table′s tab separated values into one combined metagenomic
filter. Samples were mounted to a glass slide with 10–20 μL of Mountant. Operational Taxonomic Unit (OTU) Table with read counts for each
For microbial quantification, 20 μL of 25% glutaraldehyde were sample. Reads that did not match any sequences in the reference data­
added and vortexed, for a final concentration of 2% v/v. Samples were base were considered unclassified and removed from further analyses.
fixed in the dark for 15 min and 995 μL of each sample was added to an
Eppendorf tube containing 5 μL of 5 mg/mL DAPI. The entire 1 mL of
2.5. Antibiotic resistant genes
glut-fixed sample was resuspended in 5 mL of 0.2 μm filtered Sigma
water. The sample was then filtered with a 0.2 μm Anodisc filter. Sam­
Fastq sequences were converted to fasta sequences using seqtk (Li,
ples were mounted on a glass slide with 10–20 μL of Mountant.
2023/2023). Antibiotic resistance gene sequences (ARGs) were obtained
Images were taken using an epifluorescent microscope at 600X
from the Comprehensive Antibiotic Resistance Database (CARD) (The
magnification. Fluorescence parameters were set to ex./em. of
Comprehensive Antibiotic Resistance Database, n.d.) for antibiotic inacti­
325–375/537 and 358/461 nm for DAPI and SYBR Gold, respectively.
vation and antibiotic efflux mechanisms. Sequences were extracted from
Automated counts were performed using Image-Pro Plus. Within this
CARD using the filter terms “part_of”, “is_a”, “participates_in”, and
software, the counts of fluorescent features restricted to the relative size
“has_part”, to extract a broad diversity of ARG sequences. These se­
of VLPs and bacteria were automated, as described previously (Haas
quences were transformed into a blast database using blastdb and we
et al., 2014). Because samples were initially diluted, these numbers were
used blastn (McGinnis & Madden, 2004) with parameters set to 100%
then multiplied by the dilution factor to estimate counts per mL of water
identity and 100% query coverage to align metagenome reads to the
collected.
ARG database. The blastn output text files were converted into a comma
separated values table with the number of reads for each gene in each
2.3. DNA extraction, purification, and sequencing sample, and R was used to generate boxplots based on the ARG count
data.
DNA extraction, purification, and sequencing was performed simi­
larly to Allsing et al. (2023). Briefly, DNA was extracted and purified
2.6. Microbial source tracking
using the Qiagen DNeasy PowerWater Sterivex kit. Concentration and
purity were measured using a NanoDrop spectrophotometer. Aliquots of
The SourceTracker algorithm uses Bayes’ theorem and Gibbs sam­
DNA were shipped on dry ice to the Microbial Genome Sequencing
pling to estimate the proportion of sources in a sink (Knights et al., 2011;
Center (MiGS; Pittsburgh, PA, USA) who performed the library prepa­
McGhee et al., 2020). TJ samples were assigned as sinks and matched
ration and sequencing. Library preparation was done with the Nextera
against source metagenomes to perform a meta-SourceTracker analysis
XT DNA Library Preparation Kit and the resulting libraries were
in the manner of McGhee et al. (2020). We downloaded 164 source
sequenced on the NextSeq platform at a read depth of approximately 25
metagenomes from 6 different studies that represented environmental
M.
microbiomes that could potential contribute to the TJ samples: Fresh­
water (7 metagenomes), Gut (13) Ocean (50), Pacific Ocean (32), Soil
2.4. Taxonomic classification (49), and Wastewater (13) (Supplemental Table 1). To avoid “mixed”
sources, i.e., source samples that might contain microbes from more
All sequences were quality controlled using fastp (Chen et al., 2018), than one environment, we calculated the clr-transformed Euclidean
which filters reads where Phred quality is greater than for equal to 15. distances and used NMDS to determine similarity between samples.
Quality controlled sequences were run through Kaiju, a metagenomic Source metagenomes were filtered depending on “purity” based on
taxonomic profiling classification program, which uses an NMDS plot to: 7 Freshwater, 10 Gut, 10 Ocean, 10 Pacific Ocean, 10 Soil,
alignment-based method to align reads to a reference database (Menzel and 10 Wastewater samples. SourceTracker (and meta-SourceTracker)
et al., 2016). These reads were matched to NCBI BLAST database takes two input files: a biom formatted OTU table and a mapping file.
(nr_euk, February 2022) to output the highest accuracy as it includes an The OTU biom file combined the kaiju results from the TJ samples and

3
S. Shahar et al. Environmental Pollution 342 (2024) 123067

the kaiju result from the chosen source metagenome samples. The 3. Results
mapping file indicates which samples are sinks and which samples are
sources. The meta-SourceTracker analyses were run with the entire 3.1. Microbial biomass consistently higher in near-border samples
metagenomes, then with bacteria, archaea, eukaryotes, and viruses
separately, and the outputs were converted to a bar plot using matplotlib Microscopy was used to estimate total counts of microbes (Fig. 2A)
in Python. and VLPs (Fig. 2B) per mL of water. The data were log transformed due
to high counts. The average microbe counts were 324,861,783 and
2.7. Data availability & statistical analysis 3,078,841 for Radio Club and Boca Rio, respectively. The average VLPs
were 315,219,367 and 7,345,869 for Radio Club and Boca Rio, respec­
Scripts for this analysis and sample input and output data can be tively. With the exception of the VLP count estimates on 04/2021, both
found at https://github.com/shaylashahar/Tijuana_River_2020_2021. microbial and VLP counts were 2 or greater orders of magnitude higher
Sequences used for SourceTracker metagenomes were all from the Eu­ on each of the collection days.
ropean Nucleotide Archive (ENA) and project accession numbers are
listed in Supplemental Table 1. Raw sequences from this study were 3.1.1. Near-border samples dominated by pathogens also found in estuarine
uploaded to ENA, Project number is PRJEB 63194. To assess differences samples
between sites across experiments, ANOVA tests were used, with a 95% Table 2 compares the mean relative abundances of the ten most
confidence level (α = 0.05). common species in Radio Club and Boca Rio metagenome samples as

Fig. 2. Results of microbial biomass estimation using epifluorescence microscopy. (A) Microbe counts and (B) Viral-like particles. Both microbial and viral particle
concentrations were consistently higher in the near-border samples than in the estuarine samples.

Table 2
Ten most common microbes based on read counts and percent relative abundances at (A) Radio Club and (B) Boca Rio.
A. Top 10 Microbe Average Relative Abundances at Radio Club

Species Radio Club Boca Rio

Counts (% Abundance) Counts (% Abundance)

Comamonas denitrificans 77,063 (2.94%) 6228 (0.17%)

Aliarcobacter cryaerophilus 73,212 (2.79%) 31,444 (0.84%)
Acinetobacter johnsonii 63,874 (2.44%) 1689 (0.04%)
Pseudoarcobacter acticola 46,040 (1.76%) 10,660 (0.29%)
Denitrificimonas caeni 45,867 (1.75%) 7624 (0.20%)
Acidovorax temperans 39,782 (1.52%) 5788 (0.15%)
Mesorhizobium sp. F7 39,507 (1.51%) 225 (0.000060%)
Chryseoglobus frigidaquae 35,779 (1.36%) 111 (0.000030%)
Comamonas aquatica 31,446 (1.20%) 4121 (0.11%)
uncultured Caudovirales phage 26,707 (1.02%) 31,294 (0.84%)
B. Top 10 Microbe Average Relative Abundances at Boca Rio
Species Boca Rio Radio Club
Counts (% Abundance) Counts (% Abundance)

Bacteroidetes bacterium 125,496 (3.36%) 7287 (0.28%)

Flavobacteriales bacterium 86,959 (2.32%) 779 (0.03%)
Rhodobacteraceae bacterium 65,833 (1.76%) 2395 (0.09%)
Planktomarina temperata 58,946 (1.58%) 31 (0.00001%)
Gammaproteobacteria bacterium 54,177 (1.45%) 5211 (0.20%)
Glaciecola sp. HTCC2999 49,698 (1.33%) 19 (0.000007%)
Rhodobacteraceae bacterium SB2 46,674 (1.25%) 38 (0.00001%)
SAR86 cluster bacterium 44,038 (1.18%) 76 (0.00003%)
Candidatus Pelagibacter ubique 34,943 (0.94%) 20 (0.000008%)
Rhodobacterales bacterium HTCC2255 33,293 (0.89%) 14 (0.000005%)

4
S. Shahar et al. Environmental Pollution 342 (2024) 123067

determined by Kaiju. With one exception (an uncultured Caudovirales level of taxonomic differentiation among the Radio Club samples is re­
phage), all the most common species were bacteria. Comamonas deni­ flected in the significantly higher mean beta-dispersion (i.e., the dis­
trificans was the most abundant in the Radio Club samples and Bacter­ tances of each sample to each site’s respective centroid) when compared
oidetes bacterium the most abundant among the Boca Rio samples. with the Boca Rio samples (Fig. 3D; p = 0.0082).

3.1.2. Microbial biodiversity greater in estuarine than near-border samples 3.1.4. Diverse ARGs detected in near-border and estuarine samples
Measurements of species richness and species evenness showed sig­ The five most abundantly detected ARGs with efflux antibiotic
nificant differences between both sites, with Boca Rio samples being resistance mechanisms were: (1) tetracycline efflux pumps found in
both richer and more even in species diversity than Radio Club (Fig. 3A gram-negative bacteria tet(39) (CARD Accession:3000566) and (2) tet(c)
and B). Richness metrics showed significant differences between Radio (CARD Accession: 3000167), (3) a plasmid or transposon-encoded
Club and Boca Rio (p = 0.0022). Evenness metrics showed significant chloramphenicol exporter found in Escherichia coli (CARD Accession:
differences between Radio Club and Boca Rio (p = 0.0012). 3002695), (4) a resistance gene that confers to antiseptics QacEdelta1
(CARD Accession: 3005010), and (5) a plasmid or transposon-encoded
3.1.3. Greater taxonomic compositional similarity and lower dispersion in chloramphenicol exporter found in Pseudomonas aeruginosa and Klebsi­
estuarine samples ella pneumoniae (CARD Accession: 3002693). The five most abundantly
The data were centered-log ratio transformed, and pairwise detected ARGs with inactivation antibiotic resistance mechanisms were
Euclidean distances reduced to 2-dimentions using Non-Metric Multi­ (1) a resistance gene found on plasmid pRSB105 mphE (CARD Accession:
dimensional Scaling (NMDS) (Fig. 3C). There was a clear separation of 3003741), (2) a beta-lactamase found in the Enterobacteriaceae family
the Radio Club and Boca Rio samples, with the Radio Club samples much (CARD Accession: 3001397), (3) an aminoglycoside nucleotidyl­
more dissimilar to each other than the Boca Rio samples. The greater transferase identified in Acinetobacter ANT(3”)-IIa (CARD Accession:

Fig. 3. Alpha and beta diversity analyses showing that estuarine samples had higher species richness and evenness and had greater compositional similarity over
time. (A) Species richness, (B) Species evenness, (C) Non-metric multidimensional scaling plot observing beta diversity between all samples from both sites, and (D)
Levels of beta dispersion among samples at each of the sampling locations (p = 0.0082).

5
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Fig. 4. Results of metagenome sequence matching to CARD database. Antibiotic resistance genes abundance at each site showing the (A) Mean total read counts of
all ARGs detected, and mean counts of ARGS with (B) Inactivation and (C) Efflux mechanisms, respectively. The mean combined read counts matching both ARG
types was higher in the Radio Club samples, though the results were not significant. The Efflux ARG read counts were significantly higher in the Boca Rio samples.

3004089), (4) an aminoglycoside nucleotidyltransferase identified in Ocean though the November 2020 and April 2021 had high proportions
Enterobacteriaceae, Acinetobacter baumannii, P. aeruginosa, and Vibrio of Wastewater. The Eukaryote source proportions in Radio Club samples
cholerae (CARD Accession: 3002601), and (5) a beta-lactamase that were more evenly split between Wastewater, Soil and Freshwater, while
confers resistance to ampicillin and cephalothin OXA-540 (CARD Ocean and Freshwater source dominated as Boca Rio sources. The
Accession: 3005314). Despite trends, when relative abundances were Bacteria and Virus source proportions were most like each other and to
summed, no significant differences were found between Radio Club and the combined results in Fig. 6. However, the Wastewater and Freshwater
Boca Rio for efflux (p = 0.051) (Fig. 4A) or inactivation (p = 0.073) proportions tended to be greater for the Viruses in both Radio Club and
(Fig. 4B) mechanisms. Boca Rio samples. Both Figs. 6 and 7 present the data after removing the
estimated fraction that was unknown, which was especially high in the
3.1.5. Microbial source tracking determined fecal sources dominated near- bacterial results and obscured the interpretation. The full meta-
border and marine sources dominated estuarine except after sewage SourceTracker results are shown in Supplemental Figs. 4–5.
overflows
Fig. 5 shows the results of an NMDS analysis using clr-transformed 4. Discussion
Euclidean distances between metagenomics samples from putative
source environments (Supplemental Table 1). This was done to allow Our longitudinal comparative analysis of near-border (Radio Club)
selection of samples from all the environments with no visible overlap (i. and downstream estuarine (Boca Rio) water samples from the geopo­
e., “pure” environment source samples). Fig. 5A found strong visual litically important Tijuana River estuary provided substantial insights
separation of many of the Soil, Ocean and the Pacific Ocean meta­ into levels and diversity of microbial contamination, ARG diversity,
genome samples, though numerous samples from these environments patterns of biodiversity and composition, and insights into microbial
clustered with the Soil, Wastewater, and Gut samples. After selecting a source origins. Radio Club samples were highly contaminated with fecal
“pure” subset (no overlap with any other environment) of these meta­ and sewage sludge bacteria that included several known pathogens and
genomes (10 per source). Fig. 5B shows the NMDS analysis with the we detected numerous ARGs in the metagenomes. While the estuarine
remaining Freshwater, Gut, and Wastewater with the circles indicating samples were dominated by marine microbes, the same pathogens,
the 10 “pure” samples used in downstream meta-SourceTracker sludge bacteria, and ARGs found in Radio Club were also found in Boca
analyses. Rio. Boca Rio samples were significantly more biodiverse and more
Fig. 6 shows the results of the meta-SourceTracker analysis with the taxonomically consistent over time than the Radio Club samples, which
entire metagenome. Radio Club samples were dominated by Gut and were highly dissimilar to one another. The near-border Radio Club
Wastewater microbial communities, while Boca Rio samples were samples were dominated by fecal and sewage sludge communities, while
dominated by ocean and Pacific Ocean microbial communities (Fig. 6). Boca Rio samples were dominated by marine-associated microorganisms
Fig. 7 shows the results of meta-SourceTracker for each individual do­ except for two collections days showing high proportions of fecal
mains. The highest source proportion of the Archaea in Radio Club contamination after incidences of cross-border sewage overflow.
samples was from Wastewater, while the Boca Rio Archaea came from

6
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Fig. 5. Results of NMDS analysis of pairwise Euclidean distances between putative source samples based on the clr-transformed kaiju taxonomic count tables. Each
dot represents a particular sample from a given source environment indicated by the colors. The closer the dots are to one another, the more similar the taxonomic
composition of those samples. (A) Samples from all putative source metagenomes; (B) Samples from a subset of metagenomes after removing Ocean, Pacific Ocean,
and Soil samples. The circles indicate the “pure” samples used for downstream meta-SourceTracker analyses.

Culture-independent microscopy analysis detected 105 times more environments, can colonize and cause disease in humans, such as bac­
microbes and 43 times more VLPs at Radio Club than at Boca Rio teraemia (Feng et al., 2016), and Comamonas denitrificans is a biomarker
(Fig. 2A and B; Supplemental Fig. 1). The much higher level of microbial for denitrification in activated sludge (Gumaelius et al., 2001). Se­
and viral contamination at the Radio Club site undoubtedly reflected the quences counts of the commonly used fecal indicator virus crAssphage
significant fecal contamination at these sites. Metagenomic analysis of from the metagenomes were significantly higher at Radio Club than
samples from the transborder and estuary sites provided insight into the Boca Rio (p = 0.014) (Supplemental Fig. 2). E. coli were also higher at
degree and types of cross-border contamination and its effect on the Radio Club compared to Boca Rio, though the difference was not sig­
downstream water quality. The 10 most abundant microbes detected nificant (p = 0.078) (Supplemental Fig. 3). Collectively, these data
collectively in the Radio Club samples made up 18.29% of the total, confirm the high levels of cross-border fecal contamination at Radio
while those same 10 microbes are found in 2.64% of the Boca Rio Club and show that this contamination impacts the downstream water
samples (Table 2A). Many of the most commonly detected species in the quality at Boca Rio.
Radio Club samples were fecal contaminants and disease-causing mi­ The most abundant microbes found at Boca Rio made up 16.06% of
crobes. For example, Aliarcobacter cryaerophilus, the seventh most the total average microbes found at that site, and only 0.60% of the
common organism isolated from human feces, is a potential threat to microbes in the Radio Club samples (Table 2B). The Boca Rio samples
human health as it is an emerging foodborne and zoonotic pathogen were much more diverse and were dominated by marine bacteria, many
(Müller et al., 2020). The fact that the fecal-associated Aliarcobacter of them poorly characterized, which clearly reflected the estuarine na­
cryaerophilus comprised almost 1% of the Boca Rio metagenome se­ ture of this site. The high diversity and low degree of species-level
quences indicated that this organism likely came from cross-border identification in the Boca Rio metagenomes also reflects the generally
contamination. Acinetobacter johnsonii, commonly found in aquatic high diversity of marine ecosystems, most of which is still

7
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Fig. 6. Result of the meta-SourceTracker results based on the full metagenome analysis excluding estimated proportions from unknown environments. The colors
indicate the relative contribution of each source environment to the given sink (water) sample. Gut (feces) and wastewater (sewage sludge) sources dominate near-
border samples, while marine sources dominate estuarine samples except for two day (BR November 2020 and BR 4/2021) that had high proportions of from
gut sources.

uncharacterized. As a result, most of the reads were aggregated at a between the two sites. The similarity of the Boca Rio samples in terms of
higher taxonomic level. For example, the three most abundant matches microbial composition again probably reflects the similarity of the ma­
in Boca Rio metagenomes were not to individual species but rather to rine community over time. The Radio Club samples, on the other hand,
taxonomic groups consolidated at the phylum, order, and family level, probably have highly variable inputs particularly of sewage but also
namely the Bacteroidetes, Flavobacteriales, and Rhodobacteraceae, freshwater from rainfall. This would lead to the lower overall similarity
respectively (Table 2B). Bacteroidetes, Flavobacteriales, and Rhodo­ of the Radio Club sites which had significantly greater beta dispersion
bacteraceae are commonly found in marine environments (Rhodo­ compared with the Boca Rio samples (p = 0.0082) (Fig. 3D).
bacteraceae - an Overview | ScienceDirect Topics, n.d.; Thomas et al., 2011; ARGs are an important indicator of potential drug resistance within
Yang et al., 2017). It is likely that within these samples there are microbial communities. Two primary mechanisms, efflux and inhibi­
numerous uncharacterized species of Bacteroidetes, Flavobacteriales, and tion, can be used to characterize the mode of resistance conferral.
Rhodobacteraceae. Bacteroidetes bacterium colonize numerous diverse Elevated efflux activity facilitates the expedited and efficient removal of
environments and play a beneficial role in degrading organic matter, toxicants (such as antibiotics) from microbes, leading to a reduction in
though they can also act as an opportunistic pathogens (Thomas et al., microbial concentrations of these drugs (Levy, 2002; Nikaido, 1998;
2011). The most common single identifiable species at Boca Rio was Webber and Piddock, 2003). For this reason, mutations in these genes
Planktomarina temperate (family Rhodobacteraceae), a planktonic bacte­ can improve the export of these toxic compounds, and therefore can
rium found in marine environments (Giebel et al., 2013). However, increase survival of microbes. Inhibitory enzymes, such as
while the most common microorganisms at Boca Rio were marine in beta-lactamases, can also confer resistance through the slowing or in­
origin, there were also high relative levels of fecal and sludge associated hibition of key enzymes that metabolize drugs and xenobiotics (Hooban
organisms also detected in Radio Club samples, including Aliacrobacter et al., 2020; Majiduddin et al., 2002). Increased activity of these en­
cryaerophilus and Comamonas denitrificans (Table 2A). zymes or variation of these genes can improve the efficiency of xeno­
Comparisons of alpha diversity between Radio Club and Boca Rio biotic metabolism in microbes, also increasing survival. Because the
metagenomic samples analyses showed high species richness and prevalence of these genes in metagenomic samples may indicate overall
evenness at Boca Rio compared with Radio Club (Fig. 3A and B). This drug resistance in the population, we quantified ARGs in order to un­
makes sense given that Boca Rio is an estuary site with high volumes of derstand potential implications for public health.
ocean water that mixes with the river water. Ocean waters are extremely As antibiotic resistance increases (CDC, 2022), it is important to
diverse compared to freshwater systems (Lee & Eom, 2016), and the mix understand the types of ARGs present in environments where people
of two different ecosystems should lead to an even higher level of species interact. The total ARGs, and namely the inactivation ARGs, were more
richness. The fact that Boca Rio has a higher level of evenness probably concentrated at Radio Club, which is closer to the human population
indicates a consistently high degree of biodiversity without many center and receives more of the runoff and wastewater associated with
dominant taxa, which would be typical of ocean systems. The Radio this urbanization. In terms of ARGs, beta-lactamases comprised six of the
Club samples, on the other hand, were dominated by a smaller set of ten most abundant ARGs with inactivation mechanism (Supplemental
mainly fecal taxa, which would result in much lower levels of evenness. Table 2A). The most common ARGs with efflux mechanisms were
Beta diversity analysis (Fig. 3C) showed strong clustering of the Boca Rio tetracycline pumps and plasmid or transposon-encoded chloramphen­
sites, high dispersion among Radio Club sites, and distinct separation icol (Supplemental Table 2B). Tet 39, the most common ARG identified

8
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Fig. 7. Result of the meta-SourceTracker results based on domain-specific analysis excluding estimated proportions from unknown environments. The colors indicate
the relative contribution of each source environment to the given sink (water) sample. Domain-specific meta-SourceTracker results for (A) Archaea, (B) Bacteria, (C)
Eukaryotes, and (D) Viruses. These results show clear domain specific patterns for each of the taxonomic groupings (e.g., a much higher proportion of the archaea are
sourced from wastewater sludge).

in our analysis is common in Acinetobacter spp. and can be found in samples overlap significantly with Gut, Wastewater, and Freshwater
aquatic environments (Agersø & Guardabassi, 2005). As tet(39) is samples. This makes sense particularly with coastal marine samples
resistant to tetracycline antibiotics, which are commonly used to treat which can be mixed with freshwater from rivers and contaminated by
pneumonia and other respiratory tract infections and other bacterial sewage and wastewater (Avakian, 2021). Interestingly, when these
infections (Tetracycline, n.d.), the appearance of such resistance genes in samples were removed, the NMDS plot was able determine clear sepa­
these samples is concerning considering there is a high likelihood of rations between Freshwater, Gut and Wastewater samples. This strategy
these bacteria surviving due to resistance. The plasmid or of plotting and suspected contaminated sample removal allowed us to
transposon-encoded chloramphenicol ARGs are responsible for the identify and select “pure” source samples which is critical for accurate
bacterial resistance to the antibiotic chloramphenicol, commonly used source-tracking. This analysis also allowed us to observe a very clear
to treat bacterial eye infections (Leslie et al., 1988; Oong & Tadi, 2023). differentiation between the marine sample types we termed Ocean and
The dominance of these plasmids in the sewage and estuary is worrisome Pacific Ocean, which came from two different studies.
to public health, as horizontal gene transfer can fuel antibiotic resistance To determine the contamination flow from Radio Club to Boca Rio, we
even further (Arnold et al., 2022). performed a meta-SourceTracker analysis. Using a balanced set of source
Prior to performing the meta- SourceTracker analysis, we examined metagenomes (10 per source environment) the meta-SourceTracker
six source metagenomes and selected only “pure” source samples. The analysis of the full metagenomes (complete taxonomic counts from the
meta-SourceTracker analysis of metagenomes from putative source en­ Kaiju analysis) found that Gut (feces) and Wastewater (sewages sludge)
vironments identified many samples that appeared to be contaminated, environments dominated as sources of the Radio Club environments,
particularly the Ocean, Pacific Ocean, and Soil metagenome source while Boca Rio samples sources were dominated by Ocean and Pacific
samples. The NMDS plot in Fig. 5A shows many of these environmental (marine) microbial communities (Fig. 6). However, the relative

9
S. Shahar et al. Environmental Pollution 342 (2024) 123067

proportion of sources, particularly Gut, varied significantly across time. Declaration of competing interest
Two Boca Rio sampling dates had particularly noteworthy proportions of
Gut source contamination, November 2020 and April 2021. This higher The authors declare that they have no known competing financial
proportion of Gut can be explained by the fact that these samples were interests or personal relationships that could have appeared to influence
taken days after an influx of sewage laden Tijuana River water into the the work reported in this paper.
estuary (International Boundary and Water Commission Spill Reports | San
Diego Regional Water Quality Control Board, n.d.). Data availability
Similar to the results of the original meta-SourceTracker paper
(McGhee et al., 2020), a domain-specific microbial source-tracking Code shared via github link in manuscript and data has been
analysis identified taxon-dependent patterns of environmental sources. uploaded on the ENA. The ENA project number is in the maunscript.
The Radio Club samples show that archaea came primarily from the
Wastewater (Fig. 7A), bacteria from Gut (Fig. 7B), eukaryotes from Acknowledgements
Freshwater and Soil (Fig. 7C), and viruses from Freshwater, Soil, and
Wastewater (Fig. 7D). For the Boca Rio samples, all the taxonomic do­ Research reported in this manuscript was supported by a pilot project
mains generally traced back to marine sources, though Freshwater was a to KES and SK through the San Diego State University HealthLINK
significant source for both eukaryotes and viruses, particularly on April Center for Transdisciplinary Health Disparities Research, a center sup­
2021 which showed a pronounced influence of both Wastewater and ported by the National Institute on Minority Health and Health Dis­
Freshwater for all the domains. Interestingly, while the numerically parities (U54 MD012397). Support for KES was provided by the
dominant bacteria (Fig. 7B) were more like the full metagenome results National Institute of Environmental Health Sciences (K01 ES031640).
(Fig. 6), the Gut signal in the Boca Rio (and some Radio Club) samples We would like to thank Alma Rocha for assistance collecting field
was much stronger with the full metagenome, suggesting a synergistic samples, and Dr. Jeff Crooks and the Tijuana River National Estuarine
effect of all the taxonomic data. Research Reserve for providing crucial guidance.

5. Conclusion Appendix A. Supplementary data

Overall, these data heavily suggest the presence of fecal contami­ Supplementary data to this article can be found online at https://doi.
nants at Radio Club and the flow from Radio Club to Boca Rio. These org/10.1016/j.envpol.2023.123067.
results, much like Allsing et al. (2023), indicate the potential for
application of metagenomic analysis to water quality monitoring. We References
showed that metagenomic analysis could simultaneously determine the
presence of pathogens and enumerate commonly used fecal indicator Agersø, Y., Guardabassi, L., 2005. Identification of Tet 39, a novel class of tetracycline
taxa, detect ARGs, track the source origins of environments of microbial resistance determinant in Acinetobacter spp. of environmental and clinical origin.
J. Antimicrob. Chemother. 55 (4), 566–569. https://doi.org/10.1093/jac/dki051.
communities, and provide insights into the general ecological diversity Allsing, N., Kelley, S.T., Fox, A.N., Sant, K.E., 2023. Metagenomic analysis of microbial
of ecosystems over time. As it becomes cheaper and more available, contamination in the U.S. Portion of the Tijuana River Watershed. Int. J. Environ.
metagenomic analysis should be considered as an important tool for Res. Publ. Health 20 (1), 1. https://doi.org/10.3390/ijerph20010600.
Arnold, B.J., Huang, I.-T., Hanage, W.P., 2022. Horizontal gene transfer and adaptive
understanding the ecology and water quality of contaminated water evolution in bacteria. Nat. Rev. Microbiol. 20 (4), 4 https://doi.org/10.1038/
systems. s41579-021-00650-4.
ARGs found in both the transborder and estuary site raise major Avakian, M., 2021. New study finds ocean pollution a threat to human health. Global
Environmental Health Newsletter. https://www.niehs.nih.gov/research/programs/
concern to the increase in antibiotic resistance. Often tides pull this geh/geh_newsletter/2021/2/articles/new_study_finds_ocean_pollution_a_threat_to_h
water south to Mexico seasonally, making this a binational issue uman_health.cfm.
requiring coordinated surveillance. As expected, the samples were CDC, 2020. Importance of Water Quality and Testing | Public Water Systems | Drinking
Water | Healthy Water | CDC. https://www.cdc.gov/healthywater/drinking/
contaminated with fecal microbes, but the meta-SourceTracker analysis
public/water_quality.html.
indicated influences for other potential sources. This method is useful in CDC, 2022. National infection & death estimates for AR. Centers for Disease Control and
understanding how the downstream contamination from the Tijuana Prevention. https://www.cdc.gov/drugresistance/national-estimates.html.
River flows into the Pacific Ocean at Imperial Beach. These findings are a Chahal, C., van den Akker, B., Young, F., Franco, C., Blackbeard, J., Monis, P., 2016.
Pathogen and particle associations in wastewater. Adv. Appl. Microbiol. 97, 63–119.
significant concern to public health as people reside in and near these https://doi.org/10.1016/bs.aambs.2016.08.001.
waters. Future work could improve to show the specificity of the path­ Chen, S., Zhou, Y., Chen, Y., Gu, J., 2018. fastp: an ultra-fast all-in-one FASTQ
ogens and environmental source identification. Recently, 300 million preprocessor. Bioinformatics 34 (17), i884–i890. https://doi.org/10.1093/
bioinformatics/bty560.
dollar commitments were made to expand the international water Cordy, G., 2014. FS-027-01—A Primer on Water Quality. https://pubs.usgs.gov/fs/f
treatment plants in San Ysidro, CA, build a new water treatment plant in s-027-01/.
Tijuana, and to upgrade the sewer infrastructure in Tijuana—which Criteria, N.R.C., Us, S, on, M., 1985. Selection of indicator organisms and agents as
components of microbiological criteria. In: An Evaluation of the Role of
could substantially reduce the contamination of the Tijuana River Microbiological Criteria for Foods and Food Ingredients. National Academies Press
watershed (US EPA, 2023) The data from this current study may serve as (US). https://www.ncbi.nlm.nih.gov/books/NBK216669/.
a baseline to contextualize improvements to the watershed after infra­ Feng, Y., Yang, P., Wang, X., Zong, Z., 2016. Characterization of Acinetobacter johnsonii
isolate XBB1 carrying nine plasmids and encoding NDM-1, OXA-58 and PER-1 by
structure implementation, and can provide crucial information for genome sequencing. J. Antimicrob. Chemother. 71 (1), 71–75. https://doi.org/
which microbial targets and sources could be priorities for monitoring in 10.1093/jac/dkv324.
water and community health. Giebel, H.-A., Kalhoefer, D., Gahl-Janssen, R., Choo, Y.-J., Lee, K., Cho, J.-C., Tindall, B.
J., Rhiel, E., Beardsley, C., Aydogmus, Ö.O., Voget, S., Daniel, R., Simon, M.,
Brinkhoff, T., 2013. Planktomarina temperata gen. Nov., sp. Nov., belonging to the
CRediT authorship contribution statement globally distributed RCA cluster of the marine Roseobacter clade, isolated from the
German Wadden Sea. Int. J. Syst. Evol. Microbiol. 63 (Pt_11), 4207–4217. https://
doi.org/10.1099/ijs.0.053249-0.
Shayla Shahar: Investigation, Writing - original draft, Software,
Gumaelius, L., Magnusson, G., Pettersson, B., Dalhammar, G., 2001. Comamonas
Formal analysis, Visualization. Karilyn E. Sant: Conceptualization, denitrificans sp. Nov., an efficient denitrifying bacterium isolated from activated
Methodology, Resources, Writing - review & editing, Supervision, sludge. Int. J. Syst. Evol. Microbiol. 51 (Pt 3), 999–1006. https://doi.org/10.1099/
Funding acquisition. Nicholas Allsing: Software, Methodology. Scott 00207713-51-3-999.
Haas, A.F., Knowles, B., Lim, Y.W., Somera, T.M., Kelly, L.W., Hatay, M., Rohwer, F.,
T. Kelley: Conceptualization, Methodology, Investigation, Resources, 2014. Unraveling the unseen players in the Ocean - a field guide to water chemistry
Writing - review & editing, Supervision, Project administration. and marine microbiology. JoVE, e52131. https://doi.org/10.3791/52131.

10
S. Shahar et al. Environmental Pollution 342 (2024) 123067

Hooban, B., Joyce, A., Fitzhenry, K., Chique, C., Morris, D., 2020. The role of the natural Park, G.W., Ng, T.F.F., Freeland, A.L., Marconi, V.C., Boom, J.A., Staat, M.A.,
aquatic environment in the dissemination of extended spectrum beta-lactamase and Montmayeur, A.M., Browne, H., Narayanan, J., Payne, D.C., Cardemil, C.V.,
carbapenemase encoding genes: a scoping review. Water Res. 180, 115880 https:// Treffiletti, A., Vinjé, J., 2020. CrAssphage as a novel tool to detect human fecal
doi.org/10.1016/j.watres.2020.115880. contamination on environmental surfaces and hands. Emerg. Infect. Dis. 26 (8),
Jang, J., Park, J., Hwang, C.Y., Choi, J., Shin, J., Kim, Y.M., Cho, K.H., Kim, J.-H., Lee, Y. 1731–1739. https://doi.org/10.3201/eid2608.200346.
M., Lee, B.Y., 2022. Abundance and diversity of antibiotic resistance genes and Porter, K.G., Feig, Y.S., 1980. The use of DAPI for identifying and counting aquatic
bacterial communities in the western Pacific and Southern Oceans. Sci. Total microflora 1. Limnol. Oceanogr. 25, 943–948. https://doi.org/10.4319/
Environ. 822, 153360 https://doi.org/10.1016/j.scitotenv.2022.153360. lo.1980.25.5.0943.
Knights, D., Kuczynski, J., Charlson, E.S., Zaneveld, J., Mozer, M.C., Collman, R.G., Roach, T.N.F., Little, M., Arts, M.G.I., Huckeba, J., Haas, A.F., George, E.E., Quinn, R.A.,
Bushman, F.D., Knight, R., Kelley, S.T., 2011. Bayesian community-wide culture- Cobián-Güemes, A.G., Naliboff, D.S., Silveira, C.B., Vermeij, M.J.A., Kelly, L.W.,
independent microbial source tracking. Nat. Methods 8 (9), 9. https://doi.org/ Dorrestein, P.C., Rohwer, F., 2020. A multiomic analysis of in situ coral–turf algal
10.1038/nmeth.1650. interactions. Proc. Natl. Acad. Sci. USA 117, 13588–13595. https://doi.org/
Larsson, D.G.J., Flach, C.-F., 2022. Antibiotic resistance in the environment. Nat. Rev. 10.1073/pnas.1915455117.
Microbiol. 20 (5), 5 https://doi.org/10.1038/s41579-021-00649-x. Rocha, A.Y., Verbyla, M.E., Sant, K.E., Mladenov, N., 2022. Detection, quantification,
Lee, S.-Y., Eom, Y.-B., 2016. Analysis of microbial composition associated with and simplified wastewater surveillance model of SARS-CoV-2 RNA in the Tijuana
freshwater and seawater. Biomedical Science Letters 22 (4), 150–159. https://doi. River. ACS ES&T Water 2 (11), 2134–2143. https://doi.org/10.1021/
org/10.15616/BSL.2016.22.4.150. acsestwater.2c00062.
Leslie, A.G., Moody, P.C., Shaw, W.V., 1988. Structure of chloramphenicol Singh, A.K., Kaur, R., Verma, S., Singh, S., 2022. Antimicrobials and antibiotic resistance
acetyltransferase at 1.75-A resolution. Proc. Natl. Acad. Sci. U.S.A. 85 (12), genes in water bodies: pollution, risk, and control. Front. Environ. Sci. 10. htt
4133–4137. ps://www.frontiersin.org/articles/10.3389/fenvs.2022.830861.
Levy, S.B., 2002. Active efflux, a common mechanism for biocide and antibiotic Staley, C., Kaiser, T., Lobos, A., Ahmed, W., Harwood, V.J., Brown, C.M., Sadowsky, M.
resistance. J. Appl. Microbiol. 92, 65S–71S. https://doi.org/10.1046/j.1365- J., 2018. Application of SourceTracker for accurate identification of fecal pollution
2672.92.5s1.4.x. in recreational freshwater: a double-blinded study. Environ. Sci. Technol. 52 (7),
Li, H., 2023. Lh3/seqtk [C]. https://github.com/lh3/seqtk (Original work published 4207–4217. https://doi.org/10.1021/acs.est.7b05401.
2012). Thomas, F., Hehemann, J.-H., Rebuffet, E., Czjzek, M., Michel, G., 2011. Environmental
Majiduddin, F.K., Materon, I.C., Palzkill, T.G., 2002. Molecular analysis of beta- and gut Bacteroidetes: the food connection. Front. Microbiol. 2. https://www.fronti
lactamase structure and function. International Journal of Medical Microbiology ersin.org/articles/10.3389/fmicb.2011.00093.
292, 127–137. https://doi.org/10.1078/1438-4221-00198. [Overviews and Factsheets] US EPA, O., 2021. USMCA Tijuana River Watershed. https
McGhee, J.J., Rawson, N., Bailey, B.A., Fernandez-Guerra, A., Sisk-Hackworth, L., ://www.epa.gov/sustainable-water-infrastructure/usmca-tijuana-river-watershed.
Kelley, S.T., 2020. Meta-SourceTracker: application of Bayesian source tracking to US EPA, O., 2023. Projects to mitigate transborder water pollution can now proceed, as
shotgun metagenomics. PeerJ 8, e8783. https://doi.org/10.7717/peerj.8783. EPA and USIBWC sign Record of decision (California). In: https://www.epa.gov/newsr
McGinnis, S., Madden, T.L., 2004. BLAST: at the core of a powerful and diverse set of eleases/projects-mitigate-transborder-water-pollution-can-now-proceed-epa-and-us
sequence analysis tools. Nucleic Acids Res. 32, W20–W25. https://doi.org/10.1093/ ibwc-sign.
nar/gkh435. Web Server issue). Water Science School, 2018. Wastewater Treatment Water Use. U.S. Geological Survey.
Menzel, P., Ng, K.L., Krogh, A., 2016. Fast and sensitive taxonomic classification for https://www.usgs.gov/special-topics/water-science-school/science/wastewater-tre
metagenomics with Kaiju. Nat. Commun. 7 (1), 1 https://doi.org/10.1038/ atment-water-use.
ncomms11257. Webber, M.A., Piddock, L.J.V., 2003. The importance of efflux pumps in bacterial
Müller, E., Hotzel, H., Ahlers, C., Hänel, I., Tomaso, H., Abdel-Glil, M.Y., 2020. Genomic antibiotic resistance. J. Antimicrob. Chemother. 51, 9–11. https://doi.org/10.1093/
analysis and antimicrobial resistance of aliarcobacter cryaerophilus strains from jac/dkg050.
German water poultry. Front. Microbiol. 11. https://www.frontiersin.org/arti Yang, J.-A., Kwon, K.K., Oh, H.-M., 2017. Complete genome sequence of Flavobacteriales
cles/10.3389/fmicb.2020.01549. bacterium strain UJ101 isolated from a xanthid crab. Genome Announc. 5 (5),
Nikaido, H., 1998. Multiple antibiotic resistance and efflux. Curr. Opin. Microbiol. 1, e01551 https://doi.org/10.1128/genomeA.01551-16, 16.
516–523. https://doi.org/10.1016/S1369-5274(98)80083-0. Zhang, X.-X., Zhang, T., Fang, H.H.P., 2009. Antibiotic resistance genes in water
Oong, G.C., Tadi, P., 2023. Chloramphenicol. In: StatPearls. StatPearls Publishing. http: environment. Appl. Microbiol. Biotechnol. 82 (3), 397–414. https://doi.org/
//www.ncbi.nlm.nih.gov/books/NBK555966/. 10.1007/s00253-008-1829-z.

11