The authors have declared that no competing interests exist.
Conceived and designed the experiments: KFS LAZ GER VS. Performed the experiments: GER VS. Analyzed the data: LAZ VS KSF GER. Contributed reagents/materials/analysis tools: KFS LAZ. Wrote the paper: KFS LAZ VS.
Ornamental fishes are among the most popular and fastest growing categories of pets in the United States (U.S.). The global scope and scale of the ornamental fish trade and growing popularity of pet fish in the U.S. are strong indicators of the myriad economic and social benefits the pet industry provides. Relatively little is known about the microbial communities associated with these ornamental fishes or the aquarium water in which they are transported and housed. Using conventional molecular approaches and next generation high-throughput amplicon sequencing of 16S ribosomal RNA gene hypervariable regions, we characterized the bacterial community of aquarium water containing common goldfish (
Ornamental fishes are the third most common group of pets in United States (U.S.) homes today. The 2011–2012 survey of the American Pet Products Manufacturers Association reported that 62% of U.S. households (73 million homes) own a pet. Of these, 17% own ornamental aquarium fishes, totaling 73 million homes with more than 151.1 million freshwater and 8.61 million saltwater fishes. During the past decade, fishes were one of the fastest growing categories of pets in the U.S., increasing in ownership by more than 20% over the previous decade
The pet industry provides many economic and social benefits and the global scope, scale and growing popularity of the ornamental fish trade are a testament to this. Unintended outcomes can occur, however, including the spread of potential pathogens that may cause disease in trade animals themselves or to other susceptible hosts encountered in supply chains, at pet shops, or end destination aquaria. In particular, carriage and aquarium tank water associated with ornamental fishes provide prime conditions for bacterial growth; most fishes in trade are tropical in origin
New molecular strategies introduced by an international effort to census marine life
Our sequencing of bacterial V3–V5 hypervariable regions of the 16S rRNA gene from two aquaria samples, each across three pet/aquarium stores, generated a total of 64,757 reads (mean 10,792 per sample, range 6,934–14,295). We sequenced the same six samples, plus one additional sample from a 4th store, using primers targeting
Sample Collection Information | 341F-926R (V3–V5) Amplicon Run |
518F-680R (V4) Amplicon Run |
|||||
Sample name |
Target species | Other species within tank | No. target/No. other | No. reads | Obs. OTUs | No. reads | Obs. OTUs |
|
|
19/3 | 9232 | 1312 | 6938 | 26 | |
|
|
|
14/42 | 11221 | 1664 | 7096 | 14 |
|
|||||||
|
|
None | 10/0 | 10268 | 580 | 7565 | 33 |
|
|
|
10/20 | 6934 | 380 | 7599 | 29 |
|
|
|
10/1 | 12807 | 917 | 7051 | 30 |
|
|
|
2/14 | 14295 | 1242 | 8464 | 61 |
|
|||||||
|
|
None | 10/0 | N/A | N/A | 8202 | 44 |
The 341F-926R primers were designed to capture overall bacterial diversity, whereas the 518F-680R primers were designed specifically to capture
Sample names consist of a store ID letter, followed by sample number taken at that store. Only a single sample exists from store B which was included only on the 518F-680R
Taxonomic analysis of the V3–V5 16S rRNA gene amplicon reads yielded a total of 30 phyla across all samples with the two most abundant, Proteobacteria (mean 51.8%) and Bacteriodetes (mean 17.6%), accounting for nearly 70% of all reads. Several phyla were extremely rare. Elusimicrobia, Deferribacteres and Tenericutes contained only a single read (<0.01%), and WS3, SR1 and TAO6 contained fewer than 10 reads each (<0.03%), despite being present in multiple samples (
Phylum-level bacterial diversity as revealed by pyrotag sequencing of the hypervariable V3–V5 region of 16S rRNA genes in our study samples (
Alpha diversity (within sample diversity) based on species richness estimation from our V3–V5 rRNA gene amplicon sequencing differed significantly and was higher between store A and both stores D and E, but not between stores D and E. Analyses using both phylogeny-based metrics (Phylogenetic Diversity (PD) Whole Tree) as implemented in Qiime v1.4.0
Alpha diversity within retail stores based on
Beta diversity metrics also showed strong groupings of samples taken from the same store but were not statistically significant (ANOSIM, p = 0.067). UNIFRAC distances
Beta diversity between samples based on
We tested for the direct presence of 12 known bacterial or eukaryotic potential pathogens in samples from all seven stores using specific primer sets. Five of the twelve genera (∼42%) were not detected in any of our samples:
It is important to note that a negative PCR result, even after multiple attempts, does not prove the absence of a given species. Despite the use of positive controls in PCR reactions, reasons for false negatives are numerous, including low target abundances, lack of optimized PCR reaction conditions for particular genomic DNA extractions, poor genomic DNA quality, or other methodological factors that may have prevented successful amplification of a given sample. Therefore we did not perform statistical analyses on inter-store differences.
We cloned and sequenced positive PCR amplicons from
From the same freshwater aquaria we also sequenced 93 PCR products targeting the near full-length 16S rRNA gene (
Our targeted
A
To the best of our knowledge this is the first survey to characterize the microbiome of water associated with freshwater ornamental aquarium fishes in the pet industry using high-throughput methods. Two earlier studies by Raja et al.
GAST Identified OTU Taxonomy | Number of Positive Samples | Primary Carrier Hosts & Environments |
Hosts that Acquire the Disease | Disease Manifestation | Primary Transmission Route |
|
|||||
|
2 | Isolated from aquatic environments, Domestic mammals, Birds |
Humans, Livestock, Other domestic mammals |
Q Fever (Humans), Respiratory disease and Abortion (Livestock) | Spore inhalation |
|
4 | Isolated from aquatic environments |
Freshwater fish |
Columnaris | Contaminated water |
1 | Freshwater amoebae |
Humans |
Legionaire's disease, Pontiac fever (Humans) | Spore inhalation |
|
1 | Unknown (presumed freshwater amoebae, soil) |
Rare in humans |
Pneumonia | Spore inhalation |
|
4 | Isolated from aquatic environments, Zooplankton |
Humans |
Cholera (Humans), Septicaemia (Fish) | Contaminated water and food |
|
1 | Isolated from aquatic environments |
Rare in humans |
Diarrhea | Contaminated water and food |
|
|
|||||
7 | See above | ||||
1 | Isolated from aquatic environments, Shellfish |
Humans |
Wound infections (Humans), Septicemia (Fish) | Contaminated water and food |
|
2 | Unknown | Human |
Intestinal infection (Humans) | Contaminated water and food |
|
|
|||||
|
N/A | Isolated from aquatic environments, Medical Leech symbiont |
Human |
Diarrhea (Humans), Epizootic ulcerative syndrome (Fish) | Contaminated water and food |
|
N/A | Isolated from aquatic environments |
Human |
Diarrhea (Human), Haemorrhagic septicaemia, Fin rot (Fish) | Contaminated water and food |
|
N/A | Isolated from terrestiral and aquatic environments, Fish, Reptiles, Birds, Mammals |
Humans |
Intestinal infection, diarrhea (Humans), Isolated from morbid individuals (Fish) | Contaminated water and food |
Many of these species have been isolated from environmental samples (eg. aquatic or terrestrial), this does not however imply that they are actively dividing outside of a host.
These species were also found using full-length 16S rRNA gene Sanger sequencing (see
Although these species were not found in either pyrosequencing run, the
It is difficult to make direct comparisons of our high-throughput results, but in typical freshwater environments (e.g. lakes, streams etc.) Actinobacteria dominate
Alpha diversity estimates were similar to those reported in the literature but these vary greatly depending on the freshwater environment, level of impact on the environment, and the contribution of landscape transfers such as soil to the overall community structure
Our survey identified 53 genera that contain potentially pathogenic species (
It is beyond the scope of this paper to discuss each of these potential pathogens in depth though some context is warranted. Many of the species in
Among the potential fish pathogens identified in this survey,
Vibrios are typically associated with marine and brackish environments but are occasionally detected in freshwater fishes and environments, as they were in this study.
Our results, combined with evidence from the literature, suggest that ornamental fishes and aquarium tank water are an understudied system with highly diverse microbial communities and sources of potential pathogens of interest to the pet industry and public health. Many of the potentially pathogenic bacteria discovered in our survey cannot be eradicated as they are part of the normal microbial flora of myriad hosts and aquatic environments. And, as described above, they are not always harmful. Nevertheless, risks exist and so we encourage owners of ornamental fishes and the pet industry to take responsibility for the health of the animals in their care and the people caring for them. Risk reduction can benefit from additional science aimed at providing a deeper understanding of the microbial ecology of aquarium systems and especially the industry/consumer practices that influence microbial community diversity and facilitate opportunistic infections. Such knowledge can be distilled into specific consumer and industry outreach initiatives. Guidelines have been established to help prevent salmonellosis in reptile owners (see those from the Association of Reptilian and Amphibian Veterinarians and the Centers for Disease Control) and help industry eliminate pathogen-carrying ticks on reptiles imported to the U.S. for sale in the pet trade (PIJAC's National Reptile Improvement Plan). Similar agendas may be created for ornamental fishes, perhaps in line with the Marine Aquarium Council's certification program. Consumer education initiatives on the topic of healthy pets are already reaching more groups (i.e. PetWatch and CDC's Healthy Pets Healthy People), some of which include information on ornamental fishes. After a series of failed policy attempts to address disease in wildlife trade
Over a two-day period in November 2009, we purchased freshwater common goldfish (
We manually filtered water samples to concentrate microbial biomass immediately upon arrival at the lab. Sterile 60 mL syringes were used to transfer water directly from the plastic bags onto 0.2-µm Sterivex filter units (Millipore, Billerica, MA). We filtered a total of 600 mL of water per bag such that each filter corresponded to a single tank in a single store, yielding 14 filtered samples. Air pushed through each filter three times served to remove any residual water. After filtration was complete, we placed filter cartridges immediately on dry ice and stored them frozen at −80°C until transport on dry ice to the MBL at Woods Hole for further processing. Following sampling, one of the authors kept the fishes as personal pets.
DNA extraction followed Puregene (Qiagen, Valencia, CA) kit instructions with the following modifications. We removed the filter inside of the sterivex using a sterilized pvc pipe cutter. We then used a sterile razor blade to cut the filter into two halves and placed each half into a screw-cap tube containing Puregene lysis buffer. Cell lysis was accomplished via the addition of lytic enzyme and proteinase K incubation followed by bead beating with 0.1 mm zirconium beads (Biospec products #11079101z). We bead-beated the cells at 5000 rpm for 60 seconds using a Beatbeater 8 (Biospec Products, Bartlesville, OK). The remainder of the protocol followed the manufacturer's instructions. Water filtration and DNA extraction protocols are available for download at
We used diagnostic PCR primers to determine the presence or absence of 9 bacterial and 4 eukaryotic genera that contain common human pathogens across our 14 freshwater aquarium tank samples from the 7 surveyed pet stores. We based our primer selection on previously published reports or personal communication and targeted bacterial 16S rRNA gene, eukaryotic 18S rRNA gene, or protein-coding genes involved in pathogenicity (see
Amplifications employed the Phusion High-Fidelity PCR kit (Finnzymes, Espoo, Finland) at 98°C denaturation for 1 minute followed by 25 cycles at 98°C for 5 seconds, primer annealing temperature for 15 seconds, and 72°C for 30 seconds, followed by a final 5 minutes at 72°C. Annealing temperature varied depending on the melting temperature (Tm) of each primer set, but was generally 3°C above the lowest primer Tm. An amplification was labeled “negative” only after multiple failed amplifications, but we acknowledge that the lack of amplification is not conclusive proof of absence. For nested PCRs, outside amplifications ran under the same conditions but employed 5 fewer cycles.
We used the TOPO® cloning kit with Mach1™-T1R
Pyrosequencing methodologies for 16S rRNA gene amplicon sequencing have been described previously
We also performed separate pyrosequencing reactions using
We processed raw reads through the VAMPS pipeline
We calculated alpha diversity using both phylogenetic diversity (PD) and best-fit parametric based models using CatchAll
Our phylogenetic diversity estimates showed strong evidence for inter-store differences, however PD estimates are descriptive, sample based only and do not allow extrapolation to a population. To provide this additional context, we also calculated alpha diversity using CatchAll 3.0
Although the GAST strategy provides an efficient way to assign taxonomy to our OTUs, it is quite conservative. To further refine
(DOCX)
(DOCX)
(DOCX)
(XLSX)
We thank the various anonymous pet and aquarium shop owners who contributed to this work. We also thank Philippe Rocca-Serra and the ISA Tab team (