Molecular Analysis of Bacterial Communities and Detection of Potential Pathogens in a Recirculating Aquaculture System for Scophthalmus maximus and Solea senegalensis

Abstract

The present study combined a DGGE and barcoded 16S rRNA pyrosequencing approach to assess bacterial composition in the water of a recirculating aquaculture system (RAS) with a shallow raceway system (SRS) for turbot (Scophthalmus maximus) and sole (Solea senegalensis). Barcoded pyrosequencing results were also used to determine the potential pathogen load in the RAS studied. Samples were collected from the water supply pipeline (Sup), fish production tanks (Pro), sedimentation filter (Sed), biofilter tank (Bio), and protein skimmer (Ozo; also used as an ozone reaction chamber) of twin RAS operating in parallel (one for each fish species). Our results revealed pronounced differences in bacterial community composition between turbot and sole RAS, suggesting that in the systems studied there is a strong species-specific effect on water bacterial communities. Proteobacteria was the most abundant phylum in the water supply and all RAS compartments. Other important taxonomic groups included the phylum Bacteriodetes. The saltwater supplied displayed a markedly lower richness and appeared to have very little influence on bacterial composition. The following potentially pathogenic species were detected: Photobacterium damselae in turbot (all compartments), Tenacibaculum discolor in turbot and sole (all compartments), Tenacibaculum soleae in turbot (all compartments) and sole (Pro, Sed and Bio), and Serratia marcescens in turbot (Sup, Sed, Bio and Ozo) and sole (only Sed) RAS. Despite the presence of these pathogens, no symptomatic fish were observed. Although we were able to identify potential pathogens, this approach should be employed with caution when monitoring aquaculture systems, as the required phylogenetic resolution for reliable identification of pathogens may not always be possible to achieve when employing 16S rRNA gene fragments.

About the Authors:

Patrícia Martins

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Daniel F. R. Cleary

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Ana C. C. Pires

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Ana Maria Rodrigues

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Victor Quintino

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Ricardo Calado

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Newton C. M. Gomes

* E-mail: [email protected]

Affiliation: Department of Biology and CESAM, University of Aveiro, Aveiro, Portugal

Introduction

The growing worldwide demand for fish has prompted research towards intensive aquaculture. Innovative technologies, such as recirculating aquaculture systems (RAS) and shallow raceway systems (SRS) [1], have been developed to improve the efficiency and sustainability of intensive aquaculture practices. RAS basically consists of the recycling and re-using of production water thus reducing adverse environmental impacts associated with water usage and release of nutrient-rich effluents [2-4]. SRS represent an improvement of common raceways, as they have an optimized hydrodynamic performance due to their low water level and plug-flow pattern thus enabling producers to employ higher fish stocking densities [1]. The organic matter resulting from unconsumed food and fish metabolites is recycled in RAS by a diverse array of microbes. Nitrogen-containing organic molecules are decomposed into ammonia by heterotrophic bacteria, with ammonia subsequently being converted into nitrite and nitrite into nitrate by nitrifying bacteria in biological filters. Given the key role played by these microorganisms in the operation of RAS, it is no surprise that most studies addressing bacterial communities in these production systems have mainly focused on biological filters [5,6]. However, there is a lack of information on the overall diversity and composition of bacterial communities in the different components of these intensive aquaculture systems. An in-depth analysis of these microbial communities will provide quantitative and qualitative outputs that should allow us to obtain a comprehensive definition of the 'standard' microbiome of a RAS. In turn this information can improve our ability to understand and control the microbial quality of production systems and reduce the risks associated with disease outbreak.

Traditionally, conventional microbiological techniques, such as culture-based approaches, serology and histology, have been used for pathogen detection in aquaculture. However, these techniques are often laborious and time-consuming. The development of molecular tools has allowed researchers to overcome these limitations [e.g. polymerase chain reaction (PCR) and real-time PCR (RT-PCR)] [7,8]. However, the application of these technologies depends on the selection of a range of pathogens to be targeted by the assay and, therefore, new or unexpected pathogens will not be detected. Alternatively, molecular fingerprint analyses of microbial communities [e.g. PCR - denaturing gradient gel electrophoresis (PCR-DGGE)] enable us to profile complex microbial communities in a range of environments [9-11]. Overall, these community fingerprint techniques are cost-effective, allow a fast characterization of different microbial assemblages in multiple samples and can easily be used to monitor microbial community structure in fish farms [12]. However, although this approach can provide important information on the structural diversity of microbes at different taxonomic levels (group specific PCR-DGGE for different kingdom, phylum, class, order, family, genus and species) [13], no information concerning the identity of microbial populations is provided.

In order to overcome these constraints, it is now possible to employ high throughput sequencing technologies (e.g. pyrosequencing) to achieve an in depth compositional analysis of complex microbial communities with an unprecedented level of resolution [14-16]. Additionally, these technologies can be especially interesting for monitoring fish disease in aquaculture systems. However, high throughput sequencing technologies demand specialized personnel and high-performance computing resources, and thus may not be readily available for most fish producers. Alternatively, DGGE can be used as a rapid proxy for the determination of compositional variation among different samples and/or experimental treatments. In this way, researchers can have a rapid overall characterization of the microbial communities being studied through DGGE and, at the same time, select the best strategy for sequencing analysis [15,17]. Here, for the first time, a DGGE - barcoded pyrosequencing approach was applied to explore the diversity of bacterial communities and detect potential fish pathogens in an intensive aquaculture operating RAS (with a SRS) for the production of Scophthalmus maximus (turbot) and Solea senegalensis (sole). The results obtained are critically discussed to highlight the advantages and limitations of this approach for the detection and characterization of bacterial pathogen assemblages in RAS.

Materials and Methods

Study site

The present study was carried out in a turbot (S. maximus) and sole (S. senegalensis) super-intensive fish farm located in northern Portugal, which employed RAS technology combined with SRS. Briefly, saltwater was pumped from a well and was strongly aerated and sand-filtered before entering each RAS; the water was recycled as it passed from the production tanks to the sedimentation tank where mechanical filtration was performed. The water then flowed to the biofilter tank for biological filtration and subsequently to the protein skimmer, which was also used as an ozone reaction chamber (Figure 1).

[Figure omitted. See PDF.]

Figure 1. Schematic representation of the sampling compartments in RAS.

Water circuit of the recirculating aquaculture system studied (not to scale).

https://doi.org/10.1371/journal.pone.0080847.g001

Sampling and DNA extraction

In RAS the external environmental parameters, such as temperature and natural photoperiod, have no influence on the system. Therefore, RAS display little to no seasonal variability in water parameters, allowing us to have a representative sample from the system by sampling at a single time point. Four sampling compartments for each species (S. maximus and S. senegalensis) were sampled: 1) shallow raceway tank (Pro) (containing fishes ~380 days old, weighing approximately 200 g to 300 g and densities with about 15 Kg/m²), 2) sedimentation tank (Sed), 3) biofilter tank (Bio) and 4) ozone tank (acting both as protein skimmer and ozone reaction chamber) (Ozo). We also sampled the water supply pipeline (Sup), which was the same for both RAS systems. The turbot and sole RAS were fully independent.

Water samples (three replicates) were collected at different parts of the tank in each RAS compartment studied. Ammonia (NH₃), nitrites (NO₂^-), nitrates (NO₃^-), bromine residuals (BR2), sulfates (SO₄^2-) and phosphates (PO₄^3–) were determined following the 8507, 8016 and 8155 methods described in the Hach Spectrophotometer (DR 2800) standard analytical procedures and according to EPA Method 300.1 and 351.2. Total organic carbon analysis (TOC) in the water was performed according to the European Norm 1484. Conventional physical-chemical parameters (temperature, pH, dissolved oxygen (DO), suspended particles and salinity) were also recorded.

For bacterial community analysis, water samples (250 ml) were filtered through 0.2 µm pore polycarbonate membranes (Poretics, Livermore, CA, USA) and total community (TC) DNA extraction was performed directly on the filter using an E.Z.N.A Soil DNA Extraction kit (Omega Bio-Tek, USA) following the manufacturer's instructions.

PCR-DGGE bacterial community fingerprinting

In this study DGGE fingerprinting was used as a rapid tool to determine compositional variation among bacterial communities in different samples prior to barcoded pyrosequencing [17]. After statistical analysis of DGGE profiles, we applied a more-in-depth compositional barcoded pyrosequencing analysis of composite samples. Amplified 16S rRNA gene fragments suitable for bacterial DGGE fingerprints of total microbial community DNA samples were obtained using a nested approach following Gomes et al. [18]. The V6-V8 regions of bacterial 16S rDNA fragments were amplified using the primers set 27F and 1494R [19,20] and 968GC - 1378R [21], for the first and second PCR, respectively. The PCRs were conducted in a Professional Thermocycler (Biometra). Positive and negative controls were run for each PCR.

DGGE was performed on a DCode Universal Mutation Detection System (Bio-Rad, Hercules, CA, USA). Samples were loaded onto 8% (w/v) polyacrylamide gels in 1x Tris-acetate-EDTA (TAE) with the denaturing gradient ranging from 40% to 58% (100% denaturant contains 7 M urea and 40% formamide) and performed at 58 °C at 160 V during 16 hours. Gels were silver-stained according to Byun et al. [22], except for the stop solution that in our case was replaced by a Na₂CO₃ solution. The image was acquired using the Epson perfection V700 Photo Scanner.

Barcoded pyrosequencing

A barcoded pyrosequencing approach was used for in-depth compositional analyses of bacterial communities in turbot and sole RAS intensive cultures. Prior to pyrosequencing, TC-DNA of all three replicates of each sampling compartment was combined, forming one DNA library for each stage of the production system. This procedure was performed for the water supply and both fish species (S. maximus and S. senegalensis). The V3-V4 region of bacterial 16S rRNA gene amplicons were amplified using barcoded fusion primer V3 Forward (5´ -ACTCCTACGGGAGGCAG-3’) and V4 Reverse (5´ -TACNVRRGTHTCTAATYC-3’) with the Roche 454 titanium sequencing adapters. Pyrosequencing libraries were generated by the 454 Genome Sequencer FLX platform (Roche Diagnostics Ltd., West Sussex, United Kingdom) [23,24]. PicoGreen dsDNA quantitation kit (Invitrogen, Life Technologies, Carlsbad, California, USA) was used to quantify the PCR product, and was pooled at equimolar concentrations and sequenced in the A direction with GS 454 FLX Titanium chemistry, according to the manufacturer’s instructions (Roche, 454 Life Sciences, Brandford, CT, USA).

Data analyses

Analysis of the DGGE banding profile was performed with the software package BioNumerics v6.6 (Applied Maths, Belgium). Band standardization was carried out automatically by the program, but was always confirmed visually with changes made when necessary. Subsequently, the program constructed a matrix that incorporated the position and intensity of each band. Briefly, the matrix containing both band position and intensity were processed in a spreadsheet and transformed into relative abundance. The relative abundance matrix was uploaded to R, log₁₀ (x+1) transformed and a distance matrix constructed with the Bray Curtis similarity coefficient using the vegdist() function in the vegan package [25] in R (version 2.11.1; http://www.r-project.org/). Variation in composition was visualised using principal coordinates analysis (PCO) with the cmdscale() function in R. Differences in the bacterial composition of RAS and water supply were tested using the adonis() function in vegan.

Pyrosequencing libraries were first analysed using QIIME (Quantitative Insights Into Microbial Ecology) (http://qiime.sourceforge.net/). First, data was filtered using the split_libraries.py script, which removed forward primers, barcodes and reverse primers. Sequences shorter than 200 base pairs were also removed. Operational taxonomic units (OTUs) were selected (97% similarity) using the pick_otus.py script with the 'usearch_ref' method and the most recent Greengenes release (Greengenes 12_10; http://qiime.wordpress.com/2012/10/16/greengenes-12_10-is-released/). Chimera were identified and removed using de novo and reference based chimera checking based on a reference fasta file from the Greengenes 12_10 release. Representative sequences were selected using the pick_rep_set.py script with the 'most_abundant' method and taxonomic identity was determined using the assign_taxonomy.py script with the Ribosomal Database Project (RDP) method [26]. We used the make_out_table.py script in QIIME to produce an OTU by sample table containing the abundance and taxonomic assignment of all OTUs. Unique OTUs were identified by assigning them to arbitrary numbers. This table was uploaded to R and non-bacteria, chloroplasts and mitochondria were removed from the analysis. Rarefaction curves were made for each sampling compartment using a self-written function in R [14]. Variation in OTU composition was visualised using principal coordinates analysis following the same method used for DGGE band data. Variation in the relative abundance of the most abundant bacterial taxa (two phyla, eight classes and ten orders) was assessed using bar plot graphs. The relative abundance was calculated considering the total of reads for each taxonomic level. In addition to this, OTUs taxonomically classified into genera known to be fish pathogens [27-30] (Data S1) were selected and their phylogeny investigated. BLAST search (http://www.ncbi.nlm.nih.gov/) was used to obtain the closest relatives of selected OTUs (pathogens and abundant taxa, i.e., where the number of sequences > 400). These sequences were also aligned using ClustalW and a phylogenetic tree was constructed using the neighbour-joining method in Mega 5.1 (http://www.megasoftware.net/). The evolutionary distances were computed using the Maximum Composite Likelihood method with a gamma distribution (four categories) and 500 bootstraps.

The DNA sequences generated in this study were submitted to the NCBI SRA: Accession number SRP026529.

Results and Discussion

Bacterial Community Profiles

In this study, temperature, salinity and pH were similar in both RAS systems (Table 1). However, ammonia, nitrites and nitrates were present in lower concentrations in water samples collected from the RAS employed for sole production. The sole RAS also showed the highest values for solid particles and TOC. In line with these analyses, the water of different fish cultures also showed distinct microbial communities. The PCO ordination analysis of bacterial DGGE profiles (Figure 2) showed clear separation of three main groups in the ordination representing the water supply and the RAS compartments for turbot and sole production (Figure 2A). The differences in bacterial composition were highly significant (F_8,18 = 6.94, P < 0.001, R² = 0.755) among groups. The primary axis of variation of the PCO in Figure 2A revealed that bacterial communities from the water supplied to the RAS and the samples collected from the compartments of the RAS clearly differed. In order to gain a better insight on how the bacterial communities differed between the two fish RAS and among the different compartments of each RAS, a new PCO was performed (Figure 2B) excluding the samples from water supply. The second PCO (Figure 2B) confirmed the differences between the two fish RAS, as illustrated by the primary axis of variation, with a significant difference in composition among groups being recorded (F_7,16 = 7.48, P < 0.001, R² = 0.766). Previous studies have shown that gut fish microbes may colonize the biofilm of marine aquaculture systems and that their microbial composition will depend on the fish species being cultured [6,31]. In fact, fish faeces and unconsumed feed are important parameters controlled in RAS. Large amount of organic materials form suspended particles, which support the growth of heterotrophic bacteria. This adversely affects nitrifiers and increases concentrations of ammonia, nitrites and nitrates and may trigger the growth of pathogenic microorganisms [32-34]. However, in contrast to current knowledge [35,36], neither solid particulates nor TOC were associated with increased values of dissolved nitrogen in turbot RAS (Data S2). Sole RAS compartments showed the highest concentration of solid particulates and the lowest levels of dissolved nitrogen (Table 1). The PCO did not show pronounced separation between bacterial communities from different RAS compartments with the exception of the sedimentation compartment of the turbot RAS (Data S3).

Temperature pH DO Salinity Ammonia Nitrites Nitrates Bromine residuals S. Particules Sulfates Phosphates TOC

(C°) (mg /L) (mg /L) (mg /L) (mg /L) (mg /L) (mg /L) (mg /L) (mg /L) (mg /L)

Sup 17.60 ± 0.08 7.40 ± 0.02 6.87 ± 0.09 24 ± 0 1.47 ± 0.05 0.80 ± 0.14 <10 0.06 ± 0.01 4.80 ± 1.23 1782.00 ± 13.64 <10 2.10 ± 0.26

SolPro 18.40 ± 0.00 6.77 ± 0.08 13.37 ± 2.30 24 ± 0 1.03 ± 0.12 0.48 ± 0.06 32.67 ± 9.29 0.16 ± 0.01 7.60 ± 1.78 1766.33 ± 9.24 <10 8.00 ± 0.00

SolSed 18.60 ± 0.10 6.60 ± 0.07 10.93 ± 0.06 24 ± 0 1.40 ± 0.17 0.53 ± 0.04 36.00 ± 1.73 0.06 ± 0.01 185.27 ± 33.69 1787.33 ± 4.04 <10 6.67 ± 1.53

SolBio 18.43 ± 0.06 6.97 ± 0.01 6.73 ± 0.51 24 ± 0 0.73 ± 0.06 0.30 ± 0.00 36.00 ± 2.00 0.07 ± 0.01 20.80 ± 2.31 1745.67 ± 50.08 <10 9.67 ± 1.53

SolOzo 18.50 ± 0.00 7.18 ± 0.19 8.90 ± 0.00 24 ± 0 0.60 ± 0.17 0.27 ± 0.06 26.00 ± 10.82 0.13 ± 0.03 10.00 ± 3.94 1777.67 ± 12.66 <10 6.00 ± 0.00

TurPro 18.77 ± 0.06 6.76 ± 0.04 16.60 ± 2.43 24 ± 0 2.33 ± 0.12 1.67 ± 0.72 82.33 ± 2.08 0.06 ± 0.01 6.73 ± 1.36 1760.00 ± 39.89 <10 7.33 ± 0.58

TurSed 18.73 ± 0.06 6.69 ± 0.01 11.33 ± 0.15 24 ± 0 3.43 ± 0.49 1.23 ± 0.35 85.00 ± 15.62 0.13 ± 0.02 6.60 ± 1.60 1708.00 ± 129.32 <10 7.00 ± 1.00

TurBio 19.00 ± 0.00 6.88 ± 0.04 7.33 ± 0.06 24 ± 0 1.60 ± 0.17) 1.57 ± 0.15 86.00 ± 10.54 0.11 ± 0.02 10.20 ± 2.95 1720.33 ± 22.37 <10 8.00 ± 1.00

TurOzo 18.80 ± 0.00 6.98 ± 0.01 8.73 ± 0.06 24 ± 0 1.73 ± 0.21 2.00 ± 0.65 57.33 ± 11.72 0.15 ± 0.06 5.33 ± 2.14 1567.00 ± 208.32 <10 7.67 ± 0.58

Table 1. Mean values and standard deviation of temperature, pH, dissolved oxygen (DO), salinity, ammonia, nitrites, nitrates, bromine residuals, suspended particles, sulfates, phosphates and total organic carbon (TOC) in the aquaculture system [water supply (Sup), turbot production tank (TurPro), turbot sedimentation tank (TurSed), turbot biofilter tank (TurBio), turbot ozone tank (TurOzo), sole production tank (SolPro), sole sedimentation tank (SolSed), sole biofilter tank (SolBio), sole ozone tank (SolOzo)].

The sign < indicates values below detection limit.

CSV

Download CSV

[Figure omitted. See PDF.]

Figure 2. Ordination diagrams (PCO) of the bacterial community based on DGGE profile.

A) in all sampling compartments (water supply (Sup), turbot production tank (TurPro), turbot sedimentation tank (TurSed), turbot biofilter tank (TurBio), turbot ozone tank (TurOzo), sole production tank (SolPro), sole sedimentation tank (SolSed), sole biofilter tank (SolBio), sole ozone tank (SolOzo)); B) only in turbot and sole sampling compartments (without water supply).

https://doi.org/10.1371/journal.pone.0080847.g002

Overall Assessment of Bacterial Composition in RAS

Barcoded pyrosequencing analysis yielded 5553, 24214 and 22786 sequence reads for water supply, turbot and sole RAS compartments, respectively. In terms of bacterial community diversity, the water in the turbot RAS showed the highest bacterial richness (Figure 3). Controlling for sample size (n = 4300 individual sequences), OTU richness varied from 562.58 ± 7.58 OTUs in the shallow raceway tank to 527.65 ± 8.40 OTUs in the biofilter tank of turbot RAS. OTU richness of the sole RAS varied from 504.45 ± 7.70 OTUs in the biofilter tank to 445.03 ± 5.34 OTUs in the sedimentation tank. The water supply used in the RAS had the lowest richness, 33.24 ± 0.82 OTUs. Although the ozone compartment in sole RAS exhibited relatively low richness, the ozone compartment in turbot had similar values to those obtained in other compartments.

[Figure omitted. See PDF.]

Figure 3. Bacterial richness.

Rarefied OTU richness in all sampling compartments (water supply (Sup), turbot production tank (TurPro), turbot sedimentation tank (TurSed), turbot biofilter tank (TurBio), turbot ozone tank (TurOzo), sole production tank (SolPro), sole sedimentation tank (SolSed), sole biofilter tank (SolBio), sole ozone tank (SolOzo)).

https://doi.org/10.1371/journal.pone.0080847.g003

In line with the PCO of DGGE profiles (Figure 2), the ordination of barcoded-pyrosequencing data (OTU composition) showed that the water supply had the most distinct microbial community (Figure 4A). The PCO comprising only turbot and sole production systems also showed clear differences between the two RAS (Figure 4C) with a range of abundant OTUs (> 400 sequences) specific to each system (Figure 4C and D). Only few bacterial OTUs (~4) were found in the water supply and RAS. These results indicate that bacterial populations entering the system through the water supply are probably out-competed by those already established in turbot and sole RAS.

[Figure omitted. See PDF.]

Figure 4. Ordination diagrams (PCO) of the bacterial community based on barcoded pyrosequencing data.

(A) in all sampling compartments; (B) the most abundant OTU’s associated to all sampling points; (C) only in turbot and the sole sampling compartments (without water supply); (D) the most abundant OTUs associated to turbot and the sole sampling compartments.

https://doi.org/10.1371/journal.pone.0080847.g004

The overall taxonomic analyses grouped bacterial sequences into twenty-four phyla, forty-two classes and sixty-one orders. At the phylum level, about 8% of OTUs remained unclassified. Figure 5 shows the relative abundance of the most dominant bacterial groups (≥ 400 reads). In agreement with the PCO analysis, the water supply had the most distinct composition, showing the highest abundance values for Oscillatoriophycideae (20.7%), Rhizobiales (24.1%), Chroococcales (20.2%), Vibrionales (20.6%), Xanthomonadales (12.5%) and Sphingomonadales (10.6%) (Figure 5). Proteobacteria was the most abundant phylum in all fish sampling compartments and displayed a slight dominance in sole RAS. In general, its relative abundance ranged between 70% and 90%. This phylum is widely dispersed in marine environments and plays an important role in the processes of nutrient cycling and mineralization of organic compounds [37,38]. Bacteroidetes was the second most abundant phylum with a relative abundance ranging from 7% to 11% in sole RAS and 18% to 26% in turbot RAS compartments. The Bacteroidetes (previously Cytophaga-Flexibacter-Bacteroides) are dominant members of marine heterotrophic bacterioplankton and are frequently found colonizing macroscopic organic matter particles (marine snow) [39]. Further differences were also observed at lower taxonomic levels between turbot and sole RAS compartments (e.g., Gammaproteobacteria, Alphaproteobacteria, Deltaproteobacteria, Oceanospirillales and Verrucomicrobiae) (Figure 5).

[Figure omitted. See PDF.]

Figure 5. Dominant bacterial groups.

Relative abundance of the most dominant bacterial groups (2 phyla, 8 classes and 10 orders) in each sampling compartments (water supply (Sup), turbot production tank (TurPro), turbot sedimentation tank (TurSed), turbot biofilter tank (TurBio), turbot ozone tank (TurOzo), sole production tank (SolPro), sole sedimentation tank (SolSed), sole biofilter tank (SolBio), sole ozone tank (SolOzo)). Groups are present from the most abundant to the least abundant.

https://doi.org/10.1371/journal.pone.0080847.g005

Composition Analysis of Dominant OTUs

The dominance analysis revealed six OTUs found in the water supply: 40 (unknown Chroococcales), 142 (unknown Sphingobacteriales), 184 (Sphingomonas), 185 (Stenotrophomonas), 243 (Aliivibrio) and 321 (Phyllobacterium) (Figure 4 and Table 2). However, none of these OTUs were dominant in the RAS compartments.

OTU Reads Class Order Family Genus Acession Sq ident Source

Tur Sol Sup

40 0 0 1079 Oscillatoriophycideae Chroococcales Xenococcaceae Unclassified EU780251 99 disease affected Turbinaria mesenterina colony

54 10 681 0 Gammaproteobacteria Oceanospirillales Oleiphilaceae Unclassified JN092240 100 gut Nephrops norvegicus

88 281 361 0 Flavobacteriia Flavobacteriales Flavobacteriaceae Olleya JN175350 100 seawater particulates, water temperature 5°C

142 0 0 418 Sphingobacteriia Sphingobacteriales Unclassified Unclassified FJ178015 95 Austrocochlea concamerata

148 196 559 0 Gammaproteobacteria Alteromonadales Colwelliaceae Unclassified KC756863 100 Paralichthys olivaceus

177 466 9 0 Alphaproteobacteria Rhizobiales Hyphomicrobiaceae Unclassified FR647917 99 seawater

184 0 1 564 Alphaproteobacteria Sphingomonadales Sphingomonadaceae Sphingomonas JQ229600 100 Crater Cirque Lake

185 0 0 679 Gammaproteobacteria Xanthomonadales Xanthomonadaceae Stenotrophomonas EU438980 100 paper mill pulps containing recycled fibres

195 756 97 0 Flavobacteriia Flavobacteriales Flavobacteriaceae Kordia FJ015036 100 turbot larval rearing unit, tank surface

200 1405 1060 0 Gammaproteobacteria Thiotrichales Thiotrichaceae Leucothrix GU451651 100 macroalgal surface

209 347 398 0 Alphaproteobacteria Rhodobacterales Rhodobacteraceae Phaeobacter NR_043888 100 Phaeobacter arcticus DSM 23566

243 7 8 998 Gammaproteobacteria Vibrionales Vibrionaceae Aliivibrio AY292946 100 Aliivibrio fischeri

298 314 1481 0 Gammaproteobacteria Oceanospirillales Oleiphilaceae Unclassified EF215752 99 inert artificial surfaces submerged in marine water

321 7 0 1003 Alphaproteobacteria Rhizobiales Phyllobacteriaceae Phyllobacterium JQ316262 100 soil from Fazenda Nova Vida

340 1749 28 0 Deltaproteobacteria PB19 Unclassified Unclassified EU283402 95 Phaeobacter arcticus DSM 23566

377 1060 4681 0 Gammaproteobacteria Alteromonadales Pseudoalteromonadaceae Pseudoalteromonas HQ401050 100 biofilm from surface of coral reef

460 1526 1926 0 Gammaproteobacteria Alteromonadales Colwelliaceae Thalassomonas HM237288 98 Thalassomonas sp. M-M1

862 528 87 0 Flavobacteriia Flavobacteriales Flavobacteriaceae Polaribacter EU586892 100 aquaculture seawater

887 362 421 0 Gammaproteobacteria Alteromonadales Colwelliaceae Thalassomonas HM237288 100 Thalassomonas sp. M-M1

904 64 418 0 Gammaproteobacteria Alteromonadales Pseudoalteromonadaceae Pseudoalteromonas FJ154991 99 ocean water

1164 143 687 0 Gammaproteobacteria Alteromonadales Pseudoalteromonadaceae Pseudoalteromonas AB257337 100 Pseudoalteromonas mariniglutinosa

1544 297 378 0 Gammaproteobacteria Alteromonadales Colwelliaceae Thalassomonas HM237288 98 Thalassomonas sp. M-M1

Table 2. Taxonomic affiliation of the most abundant OTUs (>400 reads) in water supply, turbot and sole RAS-SRS, and their closest relatives (using Blast) with the respective accession number, sequence identity (Sq ident) and source.

Reads indicates the number of sequences obtained for each OTU in water supply (Sup), turbot (Tur) and sole (Sol).

CSV

Download CSV

The most abundant OTUs in turbot RAS compartments were assigned to Kordia (OTU 195), Polaribacter (OTU 862), unknown Hyphomicrobiaceae (OTU 177) and unknown PB19 (OTU 340) group (Figure 4 and Table 2). Previous studies on Kordia spp. have shown that members of this genus can exhibit algicidal activity and produce extracellular proteases responsible for the cell lysis of diatom species [40]. Kordia spp. was also previously detected in biofilter media employed in RAS for the culture of goldfish Carassius auratus [41]. Goméz-Consarnau et al. [42] showed that some Polaribacter species have the ability to produce specific bacterial compounds (namely rhodopsins) that induce growth when associated with light (photoheterotrophy). OTU 177 was classified as an unknown member of the family Hyphomicrobiaceae and was closely related with a partial sequence of a 16S rRNA gene retrieved from marine bacterioplankton communities after environmental disturbance [43] (Figure 6). Species in this family have been recognized as important methylotrophic denitrifiers in fluidized bed reactors and activated sludge [44,45]. The most dominant OTU in the turbot RAS was OTU 340 (1749 reads). This OTU was classified as unknown PB19, and according to Blast (http://www.ncbi.nlm.nih.gov/) was related to an uncultured bacterium [GenBank accession number (acc.) EU283402] isolated from activated sludge produced by an aerated submerged membrane bioreactor for domestic wastewater treatment [46] (Figure 6).

[Figure omitted. See PDF.]

Figure 6. Neighbour-joining phylogenetic tree (16S rRNA gene sequences) of the most dominant OTUs in turbot and sole RAS and their closest relatives (accession number-species).

Bootstrap values generated from 500 replicates. Values lower than 50% were omitted.

https://doi.org/10.1371/journal.pone.0080847.g006

OTUs 54, 377, 904 and 1164 were the most abundant in the sole RAS compartments, and were classified as Pseudoalteromonas spp. (Alteromonadales), except for OTU 54 that was identified as an unclassified Oceanospirillales (Figure 4 and Table 2). According to Wawrik et al. [47] in a study of assimilatory nitrate utilization by bacteria on the West Florida Shelf, Alteromonadales and Oceanospirillales orders were identified as relevant marine heterotrophic bacteria involved in the uptake of dissolved inorganic nitrogen (DIN). Therefore, their higher abundance in sole RAS compartments may be in part responsible for the lower DIN values observed in this system when compared to the turbot RAS (Table 1). Members of the order Oceanospirillales are also often involved in symbiotic interactions with marine animals; however, their putative functional role in fish is yet to be determined [48]. In line with the higher-taxon analysis (Figure 5), the most abundant OTUs (OTUs 1164, 377 and 904) detected in sole RAS were related to Pseudoalteromonas mariniglutinosa (acc. AB257337) (Figure 6). A member of the Alteromonadales order previously isolated from an organically enriched sediment below fish farms [49]. Recently, Aranda et al. [50] showed that Pseudoalteromonas sp. strains (related to P. mariniglutinosa) could be used as Vibrio-biocontrol agents, as they were able to produce a putatively novel class of bacteriostatic compounds.

The dominant OTUs 88 (Olleya), 148 (unclassified Alteromonadales), 200 (Leucothrix), 209 (Phaeobacter), 298 (unclassified Oceanospirillales), 460 (Thalassomonas), 887 (Thalassomonas) and 1544 (Thalassomonas) were abundant in both RAS (Figure 4D). OTU 88 was assigned to the genus Olleya and was closely related to Olleya marilimosa (Figure 6) isolated from a Danish turbot farm [51]. This bacterium produces an exopolysaccharide in liquid media which may contribute to the capture, sinking and mineralization of organic substances in natural marine environments [38,52]. OTUs 148 and 298 were closely related with Colwellia aestuarii (acc. KC756863) and an unknown member of the Oceanospirillales order (acc. EF215752), respectively (Figure 6). These taxa comprise marine bacterial guilds often associated with nitrate reduction [47,53]. OTU 200 was classified as Leucothrix and was closely related to an epibiont bacteria (acc. GU451651). The epibiont Leucothrix mucor is the most studied member of this genus and is known to cause fouling diseases in prawns, although it is not considered to be a true pathogen [54].

Interestingly, our analysis detected OTUs phylogenetically related to taxa comprising bacterial species with potential activity against several aquaculture pathogens. OTUs 460, 887 and 1544 were closely related with Thalassomonas agariperforans previously isolated from marine sand [55]. Recently, Torres et al [56] showed that a member of this genus (Thalassomonas sp. PP2-459) isolated from a bivalve hatchery was capable of degrading N-acylhomoserine lactone (AHL) signal molecules (quorum quenching). The degradation of this molecule may affect the quorum sensing system of potential pathogens. Quorum sensing is a mechanism that allows bacteria to coordinate the expression of certain genes (including virulence genes) in response to the presence of small signal molecules such as AHL [57]. Recent studies showed that the use of bacterial strains with quorum quenching activity might be a useful strategy to biocontrol aquaculture pathogens [58,59]. We also detected an abundant OTU (209) closely related to Phaeobacter arcticus. Recent studies showed that members of the genus Phaeobacter can also have antagonistic activity against known aquaculture pathogens [60-62].

Phylogenetic analyses of potential fish pathogens

In order to understand the distribution of potential fish pathogens in the aquaculture system studied, a list of the most frequent bacterial pathogens responsible for fish diseases in aquacultures located in Europe was compiled [27-30] (Data S1). The 16S rRNA gene sequences of these bacteria were then phylogenetically compared with related OTUs (same genera) detected in this study (potential fish pathogens) and their closest GenBank relatives (blastn tool - http://blast.ncbi.nlm.nih.gov/) (Figure 7). In addition to this, the spatial distribution and the relative abundance of potential pathogens in RAS (turbot and sole) were investigated (Table 3). OTUs 257 and 467 were closely related to an unknown Photobacterium sp. and P. damselae, respectively. P. damselae subsp. piscicida is one of the most common pathogens associated with sole aquaculture [63,64]. However, in our study, P. damselae (OTU 467) was only detected in the turbot RAS compartments, and was more abundant in production tanks and the sedimentation filter. In contrast, we detected OTUs closely related to T. discolor (OTU 107) and T. soleae (OUT 350) in nearly all turbot and sole RAS compartments. T. soleae and T. discolor are known fish pathogens responsible for tenacibaculosis disease and were first isolated from diseased sole (S. senegalensis and S. solea) and turbot (S. maximus) in Spain [65,66]. Eight OTUs were classified (OTUs 12, 219, 428, 503, 847, 865, 1127 and 1135) as members of the genus Vibrio (data not shown). OTUs 12, 1135, 1127 and 428 were closely related with 16S rRNA gene sequences of V. ichthyoenteri, V. parahaemolyticus, V. gallaecicus and V. xuii, respectively (Figure 7). All these Vibrio species were isolated from marine aquaculture environments. However, only V. ichthyoenteri and V. parahaemolyticus are potentially pathogenic, as they are often associated with diseased animals [67,68]. These species were detected in nearly all RAS compartments. Our analysis also detected an OTU with strong homology to S. marcescens (OTU 17). This species is known to be an opportunistic pathogen previously detected in brackish and freshwaters and is a causative disease agent in natural population of white perch [69]. This potential pathogen was also isolated from diseased tilapia fish (Oreochromis niloticus) in Malaysia [70].

[Figure omitted. See PDF.]

Figure 7. Neighbour-joining phylogenetic tree (16S rRNA gene sequences) of OTUs related potential fish pathogens and their closest relatives (with accession number).

Black circles indicate the most frequent pathogens described in literature. Bootstrap values generated from 500 replicates. Values lower than 50% were omitted.

https://doi.org/10.1371/journal.pone.0080847.g007

Potential fish pathogen OTU Sup SolPro SolSed SolBio SolOzo TurPro TurSed TurBio TurOzo

Chryseobacterium scophthalmum 1271 ND ND ND 0.016 ND ND ND ND ND

Mycobacterium conceptionense 599 ND ND ND 0.016 ND ND 0.023 ND 0.015

Photobacterium damselae 467 ND ND ND ND ND 0.064 0.069 0.014 0.030

Pseudomonas putida 127 0.018 0.079 0.039 0.065 0.114 0.081 0.092 0.141 0.227

Serratia marcescens 17 0.990 ND 0.039 ND ND ND 0.046 0.042 0.030

Streptococcus infantarius 811 ND ND ND ND ND ND 0.023 ND ND

Tenacibaculum discolor 107 ND 0.189 0.138 0.114 0.152 0.307 0.253 0.212 0.242

Tenacibaculum soleae 350 ND 0.031 0.039 0.033 ND 0.081 0.069 0.014 0.015

Vibrio ichthyoenteri 12 ND 0.252 0.454 0.147 0.133 0.275 1.034 0.198 0.167

Vibrio parahaemolyticus 1135 ND 0.031 0.020 0.016 0.038 0.032 0.184 0.014 ND

Table 3. Values of relative abundance (%) of potential fish pathogens detected in water supply (Sup), turbot production tank (TurPro), turbot sedimentation tank (TurSed), turbot biofilter tank (TurBio), turbot ozone tank (TurOzo), sole production tank (SolPro), sole sedimentation tank (SolSed), sole biofilter tank (SolBio), sole ozone tank (SolOzo).

ND – Not detected

CSV

Download CSV

Some selected OTUs in this study were closely related to potential new fish pathogens. For example, OTU 127 was close related to P. putida. This species is known to be an opportunistic human pathogen, although a previous report suggests that P. putida may also be a disease causative agent in rainbow trout aquaculture [71]. OTUs 599 and 881 were closely related to Mycobacterium conceptionense and Streptococcus infantarius, respectively. These species can be found colonizing human and environmental samples, but there is no previous report on their occurrence in aquaculture systems. However, it is important to note that the genera Mycobacterium and Streptococcus comprise several members able to cause mycobacteriosis and streptococcosis diseases among both wild and captive fishes worldwide [72,73].

Despite the presence of pathogens, no diseased fish were registered during the study period. The composition and relative abundance analysis of the potential bacterial pathogens in turbot and sole RAS (Table 3) indicated that OTUs closely related to fish pathogens were present at a low abundance (low infective concentration) in both RAS. Fish density and infectious dose have been considered as key factors to control fish mortality [74,75]. Possibly, the high abundance of naturally occurring antagonistic strains detected in this study (e.g., bacterial populations closely related to known antagonistic strains belonging to the genus Thalassomonas and Phaeobacter; see above) may have contributed to suppress the development of potential fish pathogens.

Critical evaluation of barcoded pyrosequencing of 16S rRNA gene fragments for fish disease detection

The results obtained in this study were critically evaluated to underline the advantages and disadvantages of using barcoded pyrosequencing of 16S rRNA gene fragments for monitoring potential fish pathogens in aquaculture systems. The 16S rRNA gene is widely used in phylogenetic studies and is an important marker for molecular diagnostics and molecular ecology studies. However, this gene may fail to provide a sufficient phylogenetic resolution or a correct classification of some bacterial pathogens. For example, the 16S rRNA gene fragments used in this study, are unable to resolve differences between two different P. damselae subspecies, namely, damselae and piscicida [76]. These subspecies may, however, cause very different infections in a variety of fish species [28,63]. In addition to this, P. damselae subsp. piscicida is more infectious. This bacterium is the causative agent of pasteurellosis, one of the most common diseases in marine fish farms in the Iberian Peninsula [64,77]. The same problem may be observed at the species level for other bacteria. The 16S rRNA gene phylogeny provides an accurate classification of Vibrio species at family and genus level, but due to the high similarity between distinct species, it fails to provide an accurate identification at species level [78]. This was the case for V. ichthyoenteri (OTU 12), V. parahaemolyticus (OTU 847 and 1135), V. gallaecicus (OTU 1127) and V. xuii (OTU 428) detected in this study. Thompson et al. [79], showed that V. ichthyoenteri and V. scophthalmi and V. nereis and V. xuii had 99% 16S rDNA sequence similarity but only shared 90% recA gene sequence similarity. According to Osorio and Klose [76], V. parahaemolyticus and V. alginolyticus sequences were 99.8% similar using the 16S rRNA marker gene while using partial toxR gene showed only 61.7% similarity. Beaz-Hidalgo et al. [80] also showed that the recA gene is a better genetic marker to discriminate V. gallaecicus than the 16S rRNA gene. Several other studies reported that the degree of resolution obtained with the 16S rRNA gene is not sufficiently robust for phylogenetic analysis of some known bacterial fish pathogens [81-86].

Another additional problem is that none of the next generation sequencing technologies developed until now can provide long sequence reads of gene fragments (pyrosequencing provides the longest fragment size ≤ 600 bp). Very often long stretch nucleotide sequences or complete nucleotide sequences of the 16S rRNA gene (~1,500 bp) are necessary for clear differentiation of closely related species. Other phylogenetic markers can be used, however, while the 16S rRNA gene sequence database (The Ribosomal Database Project) has currently over 2,765,278 sequences, other genetic databases are substantially smaller and incomplete (lacking representative sequences for all taxa and limited numbers of sampled ecotypes) (http://rdp.cme.msu.edu/; Accessed 30 July 2013). Therefore, this problem will hamper a fast identification of a range of bacterial species or the establishment of relationships between the sequence retrieved and their closest relative in the GenBank database. Currently, this approach allows us to associate 16S rRNA gene sequences with other sequences in the database previously detected in a specific environment or case of study (e.g., ecotypes and case studies about emergent causative agents of disease outbreaks).

Conclusion

In this study we applied a combined DGGE and pyrosequencing approach to assess the structural variation and composition of bacterial communities in RAS for turbot and sole production. The DGGE approach revealed significant structural differences between bacterial communities from turbot and sole RAS. This result suggests a strong fish species-specific effect on bacterial communities of both RAS studied. The pyrosequencing approach provided fundamental information about the bacterial composition and pathogen load in different RAS compartments. However, despite detecting potential fish pathogens in sole and turbot RAS, no symptomatic fish were observed during this study. Our goal in the future should be to identify the triggering mechanism(s) causing fish infection and disease progress in aquaculture facilities. The use of a high throughput sequencing approach using the 16S rRNA gene allowed us an unprecedented means to detect pathogens in the aquaculture systems studied. However, while the use of the 16S rRNA gene is commonly recognized to be a suitable option when studying microbial composition, it may not be optimal for detecting some bacterial groups, including potential fish pathogens. Therefore, data obtained needs to be carefully examined and critically evaluated in terms of the level of resolution provided by the 16S RNA gene. Other conventional molecular tools may be used in combination with this technology to ensure the correct identification of some specific bacteria [e.g. Real Time PCR, Fluorescence In Situ Hybridization (FISH) and Isothermal DNA amplification], as well as the use of a different phylogenetic marker.

High-throughput sequencing technologies are now widely used in scientific research and, given the rapid reduction in their operating costs, it is likely that they will soon be routinely used to screen for pathogens and compare bacterial communities in aquaculture systems.

Supporting Information

[Figure omitted. See PDF.]

Data S1.

Most frequent bacterial pathogens in aquaculture systems of turbot and sole production described in literature and the pathogen species identified in this study.

https://doi.org/10.1371/journal.pone.0080847.s001

(DOCX)

Data S2.

Principal Coordinates Analysis of the RAS bacterial communities. Water parameters (ammonia, nitrites, nitrates, bromine residuals, sulfates, total organic carbon, temperature, pH, dissolved oxygen, suspended particles and salinity) are represented by vectors.

https://doi.org/10.1371/journal.pone.0080847.s002

(DOCX)

Data S3.

ANOSIM, pairwise test comparing among sampling compartments (global R=0.62).

https://doi.org/10.1371/journal.pone.0080847.s003

(DOCX)

Author Contributions

Conceived and designed the experiments: NG. Performed the experiments: PM ACCP. Analyzed the data: PM NG DFRC. Contributed reagents/materials/analysis tools: NG RC. Wrote the manuscript: PM DFRC ACCP AMR VQ RC NG.

Citation: Martins P, Cleary DFR, Pires ACC, Rodrigues AM, Quintino V, Calado R, et al. (2013) Molecular Analysis of Bacterial Communities and Detection of Potential Pathogens in a Recirculating Aquaculture System for Scophthalmus maximus and Solea senegalensis. PLoS ONE 8(11): e80847. https://doi.org/10.1371/journal.pone.0080847

References

1. Øiestad V (1999) Shallow raceways as a compact, resource-maximizing farming procedure for marine fish species. Aquac Res 30: 831-840. doi:10.1046/j.1365-2109.1999.00408.x.

2. Lekang O -I (2013) Removal of Particles: Traditional Methods: In: Wiley, J. & Sons, editors, Aquaculture Engineering: Oxford. pp. 50-65.

3. Blancheton JP (2000) Developments in recirculation systems for Mediterranean fish species. Aquacult Eng 22: 17-31. doi:10.1016/S0144-8609(00)00030-3.

4. Borges M-T, Sousa A, De Marco P, Matos A, Hönigová P et al. (2008) Aerobic and anoxic growth and nitrate removal capacity of a marine denitrifying bacterium isolated from a recirculation aquaculture system. Microb Ecol 55: 107-118. doi:10.1007/s00248-007-9256-8. PubMed: 17674087.

5. Schreier HJ, Mirzoyan N, Saito K (2010) Microbial diversity of biological filters in recirculating aquaculture systems. Curr Opin Biotechnol 21: 318-325. doi:10.1016/j.copbio.2010.03.011. PubMed: 20371171.

6. Sugita H, Nakamura H, Shimada T (2005) Microbial communities associated with filter materials in recirculating aquaculture systems of freshwater fish. Aquaculture 243: 403-409. doi:10.1016/j.aquaculture.2004.09.028.

7. García-González P, García-Lamas N, Fuentes Edfuf C, Santos Y (2011) Development of a PCR method for the specific identification of the marine fish pathogen Tenacibaculum soleae. Aquaculture 319: 1-4. doi:10.1016/j.aquaculture.2011.06.013.

8. Saulnier D, De Decker S, Haffner P (2009) Real-time PCR assay for rapid detection and quantification of Vibrio aestuarianus in oyster and seawater: A useful tool for epidemiologic studies. J Microbiol Methods 77: 191-197. doi:10.1016/j.mimet.2009.01.021. PubMed: 19318049.

9. Acinas SG, Rodrı ́guez-Valera F, Pedrós-Alió C (1997) Spatial and temporal variation in marine bacterioplankton diversity as shown by RFLP fingerprinting of PCR amplified 16S rDNA. FEMS Microbiol Ecol 24: 27-40. doi:10.1111/j.1574-6941.1997.tb00420.x.

10. Muyzer G, de Waal EC, Uitterlinden AG (1993) Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl Environ Microbiol 59: 695-700. PubMed: 7683183.

11. Schwieger F, Tebbe CC (1998) A new approach to utilize PCR–single-strand-conformation polymorphism for 16S rRNA gene-based microbial community analysis. Appl and Environ Microbiol 64: 4870-4876.

12. De Schryver P, Crab R, Defoirdt T, Boon N, Verstraete W (2008) The basics of bio-flocs technology: The added value for aquaculture. Aquaculture 277: 125-137. doi:10.1016/j.aquaculture.2008.02.019.

13. Gomes NCM, Landi L, Smalla K, Nannipieri P, Brookes PC et al. (2010) Effects of Cd- and Zn-enriched sewage sludge on soil bacterial and fungal communities. Ecotoxicol Environ Saf 73: 1255-1263. doi:10.1016/j.ecoenv.2010.07.027. PubMed: 20688396.

14. Gomes NCM, Cleary DFR, Pinto FN, Egas C, Almeida A et al. (2010) Taking root: enduring effect of rhizosphere bacterial colonization in mangroves. PLOS ONE 5: e14065. doi:10.1371/journal.pone.0014065. PubMed: 21124923.

15. Pires ACC, Cleary DFR, Almeida A, Cunha Â, Dealtry S et al. (2012) Denaturing gradient gel electrophoresis and barcoded pyrosequencing reveal unprecedented Archaeal diversity in mangrove sediment and rhizosphere samples. Appl Environl Microbiol 78: 5520-5528. doi:10.1128/AEM.00386-12. PubMed: 22660713.

16. Roesch LF, Fulthorpe RR, Riva A, Casella G, Hadwin AK et al. (2007) Pyrosequencing enumerates and contrasts soil microbial diversity. ISME J 1: 283-290. PubMed: 18043639.

17. Cleary DFR, Smalla K, Mendonça-Hagler LCS, Gomes NCM (2012) Assessment of variation in bacterial composition among microhabitats in a mangrove environment using DGGE fingerprints and barcoded pyrosequencing. PLOS ONE 7: e29380. doi:10.1371/journal.pone.0029380. PubMed: 22247774.

18. Gomes NCM, Borges LR, Paranhos R, Pinto FN, Mendonça-Hagler LCS et al. (2008) Exploring the diversity of bacterial communities in sediments of urban mangrove forests. FEMS Microbiol Ecol 66: 96-109. doi:10.1111/j.1574-6941.2008.00519.x. PubMed: 18537833.

19. Gomes NCM, Heuer H, Schönfeld J, Costa RS, Mendonça-Hagler LCS et al. (2001) Bacterial diversity of the rhizosphere of maize (Zea mays) grown in tropical soil studied by temperature gradient gel electrophoresis. Plant Soil 232: 167-180. doi:10.1023/A:1010350406708.

20. Weisburg WG, Barns SM, Pelletier DA, Lane DJ (1991) 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol 173: 697-703. PubMed: 1987160.

21. Nübel U, Engelen B, Felske A, Snaidr J, Wieshuber A et al. (1996) Sequence heterogeneities of genes encoding 16S rRNAs in Paenibacillus polymyxa detected by temperature gradient gel electrophoresis. J Bacteriol 178: 5636-5643. PubMed: 8824607.

22. Byun SO, Fang Q, Zhou H, Hickford JGH (2009) An effective method for silver-staining DNA in large numbers of polyacrylamide gels. Anal Biochem 385: 174-175. doi:10.1016/j.ab.2008.10.024. PubMed: 19013423.

23. Wang Y, Qian P-Y (2009) Conservative fragments in bacterial 16S rRNA genes and primer design for 16S Ribosomal DNA amplicons in metagenomic studies. PLOS ONE 4: e7401. doi:10.1371/journal.pone.0007401. PubMed: 19816594.

24. Cleary DFR, Becking LE, de Voogd NJ, Pires ACC, Polónia ARM et al. (2013) Habitat and host related variation in sponge bacterial symbiont communities in Indonesian waters. FEMS Microbiol Ecol 85: 465–482. doi:10.1111/1574-6941.12135. PubMed: 23607753.

25. Oksanen J (2011) Vegan: ecological diversity. Available: . http://cran.r-project.org/web/packages/vegan/vignettes/diversity-vegan.pdf. Accessed 7 February 2013.

26. Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) Naïve Bayesian Classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol 73: 5261-5267. doi:10.1128/AEM.00062-07. PubMed: 17586664.

27. Imsland AK, Foss A, Conceição LEC, Dinis MT, Delbare D et al. (2003) A review of the culture potential of Solea solea and S. senegalensis. Rev Fish Biol Fisher 13: 379-408. doi:10.1007/s11160-004-1632-6.

28. Toranzo AE, Magariños B, Romalde JL (2005) A review of the main bacterial fish diseases in mariculture systems. Aquaculture 246: 37-61. doi:10.1016/j.aquaculture.2005.01.002.

29. Toranzo AE, Novoa B, Romalde JL, Núñez S, Devesa S et al. (1993) Microflora associated with healthy and diseased turbot (Scophthalmus maximus) from three farms in northwest Spain. Aquaculture 114: 189-202. doi:10.1016/0044-8486(93)90295-A.

30. Vigneulle M, Laurencin FB (1995) Serratia liquefaciens: a case report in turbot (Scophthalmus maximus) cultured in floating cages in France. Aquaculture 132: 121-124. doi:10.1016/0044-8486(94)00375-X.

31. Lahav O, Massada IB, Yackoubov D, Zelikson R, Mozes N et al. (2009) Quantification of anammox activity in a denitrification reactor for a recirculating aquaculture system. Aquaculture 288: 76-82. doi:10.1016/j.aquaculture.2008.11.020.

32. Hess-Erga O, Attramadal KK, Vadstein O (2008) Biotic and abiotic particles protect marine heterotrophic bacteria during UV and ozone disinfection. Aquat Biol 4: 147-154. doi:10.3354/ab00105.

33. Liltved H, Cripps SJ (1999) Removal of particle-associated bacteria by prefiltration and ultraviolet irradiation. Aquacult Res 30: 445-450. doi:10.1046/j.1365-2109.1999.00349.x.

34. Michaud L, Blancheton JP, Bruni V, Piedrahita R (2006) Effect of particulate organic carbon on heterotrophic bacterial populations and nitrification efficiency in biological filters. Aquacult Eng 34: 224-233. doi:10.1016/j.aquaeng.2005.07.005.

35. Zhu S, Chen S (2001) Effects of organic carbon on nitrification rate in fixed film biofilters. Aquacult Eng 25: 1-11. doi:10.1016/S0144-8609(01)00071-1.

36. Welch EB, Lindell T (1992) Ecological effects of wastewater. Appl Limnol Poll Eff. Chapman and Hall, London. p. 343p. PubMed: 1507591.

37. Kersters K, De Vos P, Gillis M, Swings J, Vandamme P et al. (2006) Introduction to the Proteobacteria. In: M. DworkinS. FalkowE. RosenbergK-H SchleiferE. Stackebrandt. The Prokaryotes: a handbook on the biology of bacteria. Springer. pp 3–37.

38. Kirchman DL (2002) The ecology of Cytophaga–Flavobacteria in aquatic environments. FEMS Microbiol Ecol 39: 91-100. doi:10.1111/j.1574-6941.2002.tb00910.x. PubMed: 19709188.

39. Woebken D, Fuchs BA, Kuypers MAA, Amann R (2007) Potential interactions of particle-associated anammox bacteria with bacterial and archaeal partners in the Namibian upwelling system. Appl Environ Microbiol 73: 4648–4657. doi:10.1128/AEM.02774-06. PubMed: 17526789.

40. Paul C, Pohnert G (2011) Interactions of the algicidal bacterium Kordia algicida with diatoms: regulated protease excretion for specific algal lysis. PLOS ONE 6: e21032. doi:10.1371/journal.pone.0021032. PubMed: 21695044.

41. Itoi S, Niki A, Sugita H (2006) Changes in microbial communities associated with the conditioning of filter material in recirculating aquaculture systems of the pufferfish Takifugu rubripes. Aquaculture 256: 287-295. doi:10.1016/j.aquaculture.2006.02.037.

42. Gómez-Consarnau L, González JM, Coll-Lladó M, Gourdon P, Pascher T et al. (2007) Light stimulates growth of proteorhodopsin-containing marine Flavobacteria. Nature 445: 210-213. doi:10.1038/nature05381. PubMed: 17215843.

43. Sjöstedt J, Koch-Schmidt P, Pontarp M, Canbäck B, Tunlid A et al. (2012) Recruitment of members from the rare biosphere of marine bacterioplankton communities after an environmental disturbance. Appl Environ Microbiol 78: 1361-1369. doi:10.1128/AEM.05542-11. PubMed: 22194288.

44. Osaka T, Yoshie S, Tsuneda S, Hirata A, Iwami N et al. (2006) Identification of acetate- or methanol-assimilating bacteria under nitrate-reducing conditions by stable-isotope probing. Microb Ecol 52: 253-266. doi:10.1007/s00248-006-9071-7. PubMed: 16897304.

45. Liessens J, Germonpre R, Kersters I, Beernaert S, Verstraete W (1993) Removing nitrate with a methylotrophic fluidized bed: microbiological water quality. J Am Water Works Assoc 85: 155–161.

46. Du C, Wu Z, Xiao E, Zhou Q, Cheng S et al. (2008) Bacterial diversity in activated sludge from a consecutively aerated submerged membrane bioreactor treating domestic wastewater. J Environ Sci (China) 20: 1210-1217. doi:10.1016/S1001-0742(08)62211-1. PubMed: 19143345.

47. Wawrik B, Boling WB, Van Nostrand JD, Xie J, Zhou J et al. (2012) Assimilatory nitrate utilization by bacteria on the west Florida shelf as determined by stable isotope probing and functional microarray analysis. FEMS Microbiol Ecol 79: 400-411. doi:10.1111/j.1574-6941.2011.01226.x. PubMed: 22092701.

48. Jensen S, Duperron S, Birkeland N-K, Hovland M (2010) Intracellular Oceanospirillales bacteria inhabit gills of Acesta bivalves. FEMS Microbiol Ecol 74: 523-533. doi:10.1111/j.1574-6941.2010.00981.x. PubMed: 21044098.

49. Wada M, Zhang D, Do H-K, Nishimura M, Tsutsumi H et al. (2008) Co-inoculation of Capitella sp. I with its synergistic bacteria enhances degradation of organic matter in organically enriched sediment below fish farms. Mar Pollut Bull 57: 86-93. doi:10.1016/j.marpolbul.2007.10.004. PubMed: 18037450.

50. Aranda CP, Valenzuela C, Barrientos J, Paredes J, Leal P et al. (2012) Bacteriostatic anti-Vibrio parahaemolyticus activity of Pseudoalteromonas sp. strains DIT09, DIT44 and DIT46 isolated from Southern Chilean intertidal Perumytilus purpuratus. World J Microbiol Biotechnol 28: 2365-2374. doi:10.1007/s11274-012-1044-z. PubMed: 22806110.

51. Porsby CH, Nielsen KF, Gram L (2008) Phaeobacter and Ruegeria species of the Roseobacter clade colonize separate niches in a danish turbot (Scophthalmus maximus) - rearing farm and antagonize Vibrio anguillarum under different growth conditions. Appl Environ Microbiol 74: 7356-7364. doi:10.1128/AEM.01738-08. PubMed: 18952864.

52. Nichols CM, Bowman JP, Guezennec J (2005) Olleya marilimosa gen. nov., sp. nov., an exopolysaccharide-producing marine bacterium from the family Flavobacteriaceae, isolated from the Southern Ocean. Int J Syst Evol Microbiol 55: 1557-1561. doi:10.1099/ijs.0.63642-0. PubMed: 16014481.

53. Jung S-Y, Oh T-K, Yoon J-H (2006) Colwellia aestuarii sp. nov., isolated from a tidal flat sediment in Korea. Int Jof Syst Evol Microbiol 56: 33-37. doi:10.1099/ijs.0.63920-0. PubMed: 16403863.

54. Gutiérrez-Salazar GJ, Molina-Garza ZJ, Hernández-Acosta M, García-Salas JA, Mercado-Hernández R et al. (2011) Pathogens in Pacific white shrimp (Litopenaeus vannamei Boone, 1931) and their relationship with physicochemical parameters in three different culture systems in Tamaulipas, Mexico. Aquaculture 321: 34-40. doi:10.1016/j.aquaculture.2011.08.032.

55. Park S, Choi W-C, Oh T-K, Yoon J-H (2011) Thalassomonas agariperforans sp. nov., an agarolytic bacterium isolated from marine sand. Int J Syst Evol Microbiol 61: 2573-2576. doi:10.1099/ijs.0.027821-0. PubMed: 21131503.

56. Torres M, Romero M, Prado S, Dubert J, Tahrioui A et al. (2013) N-acylhomoserine lactone-degrading bacteria isolated from hatchery bivalve larval cultures. Microbiol Res 168: 547–554. doi:10.1016/j.micres.2013.04.011. PubMed: 23743010.

57. Waters CM, Bassler BL (2005) Quorum Sensing: cell-to-cell communication in bacteria. Annu Rev Cell Dev Biol 21: 319-346. doi:10.1146/annurev.cellbio.21.012704.131001. PubMed: 16212498.

58. Defoirdt T, Boon N, Bossier P, Verstraete W (2004) Disruption of bacterial quorum sensing: an unexplored strategy to fight infections in aquaculture. Aquaculture 240: 69-88. doi:10.1016/j.aquaculture.2004.06.031.

59. Defoirdt T, Sorgeloos P, Bossier P (2011) Alternatives to antibiotics for the control of bacterial disease in aquaculture. Curr Opin Microbiol 14: 251-258. doi:10.1016/j.mib.2011.03.004. PubMed: 21489864.

60. D’Alvise PW, Lillebø S, Prol-Garcia MJ, Wergeland HI, Nielsen KF et al. (2012) Phaeobacter gallaeciensis reduces Vibrio anguillarum in cultures of microalgae and rotifers, and prevents vibriosis in Cod Larvae. PLOS ONE 7: e43996. doi:10.1371/journal.pone.0043996. PubMed: 22928051.

61. Prol MJ, Bruhn JB, Pintado J, Gram L (2009) Real-time PCR detection and quantification of fish probiotic Phaeobacter strain 27-4 and fish pathogenic Vibrio in microalgae, rotifer, Artemia and first feeding turbot (Psetta maxima) larvae. J Appl Microbiol 106: 1292-1303. doi:10.1111/j.1365-2672.2008.04096.x. PubMed: 19187159.

62. Prado S, Montes J, Romalde JL, Barja JL (2009) Inhibitory activity of Phaeobacter strains against aquaculture pathogenic bacteria. Int Microbiol 12: 107–114. PubMed: 19784930.

63. Romalde JL (2002) Photobacterium damselae subsp. piscicida: an integrated view of a bacterial fish pathogen. Int Microbiol 5: 3-9. doi:10.1007/s10123-002-0051-6. PubMed: 12102234.

64. Zorrilla I, Balebona MC, Moriñigo MA, Sarasquete C, Borrego JJ (1999) Isolation and characterization of the causative agent of pasteurellosis, Photobacterium damsela ssp. piscicida, from sole, Solea senegalensis (Kaup). J Fish Dis 22: 167-172. doi:10.1046/j.1365-2761.1999.00157.x.

65. Piñeiro-Vidal M, Riaza A, Santos Y (2008) Tenacibaculum discolor sp. nov. and Tenacibaculum gallaicum sp. nov., isolated from sole (Solea senegalensis) and turbot (Psetta maxima) culture systems. Int J Syst Evol Microbiol 58: 21-25. doi:10.1099/ijs.0.65397-0. PubMed: 18175676.

66. Piñeiro-Vidal M, Carballas CG, Gómez-Barreiro O, Riaza A, Santos Y (2008) Tenacibaculum soleae sp. nov., isolated from diseased sole (Solea senegalensis Kaup). Int J Syst Evol Microbiol 58: 881-885. doi:10.1099/ijs.0.65539-0. PubMed: 18398187.

67. Gauger E, Smolowitz R, Uhlinger K, Casey J, Gómez-Chiarri M (2006) Vibrio harveyi and other bacterial pathogens in cultured summer flounder, Paralichthys dentatus. Aquaculture 260: 10-20. doi:10.1016/j.aquaculture.2006.06.012.

68. Zorrilla I, Arijo S, Chabrillon M, Diaz P, Martinez-Manzanares E et al. (2003) Vibrio species isolated from diseased farmed sole, Solea senegalensis (Kaup), and evaluation of the potential virulence role of their extracellular products. J Fish Dis 26: 103-108. doi:10.1046/j.1365-2761.2003.00437.x. PubMed: 12962218.

69. Baya AM, Toranzo AE, Lupiani B, Santos Y, Hetrick FM (1992) Serratia marcescens: a potential pathogen for fish. J Fish Dis 15: 15-26. doi:10.1111/j.1365-2761.1992.tb00632.x.

70. Chang C-Y, Koh C-L, Sam C-K, Chan X-Y, Yin WF et al. (2012) Unusual long-chain N-Acyl homoserine lactone production by and presence of quorum quenching activity in bacterial isolates from diseased tilapia fish. PLOS ONE 7: e44034. doi:10.1371/journal.pone.0044034. PubMed: 22952864.

71. Altinok I, Kayis S, Capkin E (2006) Pseudomonas putida infection in rainbow trout. Aquaculture 261: 850-855. doi:10.1016/j.aquaculture.2006.09.009.

72. Chinabut S (1999) Mycobacteriosis and nocardiosis. In: PTK WooDW Bruno. Fish Diseases and Disorders. CAB International. pp. 319–340.

73. Salati F (2011) Enterococcus seriolicida and Streptococcus spp. (S. iniae, S. agalactiae and S. dysgalactiae). In: PTK WooDW Bruno. Fish Diseases and Disorders. CAB International. pp 375-396.

74. Agnew W, Barnes AC (2007) Streptococcus iniae: An aquatic pathogen of global veterinary significance and a challenging candidate for reliable vaccination. Vet Microbiol 122: 1-15. doi:10.1016/j.vetmic.2007.03.002. PubMed: 17418985.

75. Holm JC, Refstie T, Bø S (1990) The effect of fish density and feeding regimes on individual growth rate and mortality in rainbow trout (Oncorhynchus mykiss). Aquaculture 89: 225-232. doi:10.1016/0044-8486(90)90128-A.

76. Osorio CR, Klose KE (2000) A Region of the transmembrane regulatory protein ToxR that tethers the transcriptional activation domain to the cytoplasmic membrane displays wide divergence among Vibrio species. J Bacteriol 182: 526-528. doi:10.1128/JB.182.2.526-528.2000. PubMed: 10629204.

77. Fouz B, Larsen JL, Nielsen B, Barja JL, Toranzo AE (1992) Characterization of Vibrio damsela strains isolated from turbot Scophthalmus maximus in Spain. Dis Aquat Org 12: 155-166. doi:10.3354/dao012155.

78. Thompson FL, Gevers D, Thompson CC, Dawyndt P, Naser S et al. (2005) Phylogeny and molecular identification of Vibrios on the basis of multilocus sequence analysis. Appl Environ Microbiol 71: 5107-5115. doi:10.1128/AEM.71.9.5107-5115.2005. PubMed: 16151093.

79. Thompson CC, Thompson FL, Vandemeulebroecke K, Hoste B, Dawyndt P et al. (2004) Use of recA as an alternative phylogenetic marker in the family Vibrionaceae. Int J Syst Evol Microbiol 54: 919-924. doi:10.1099/ijs.0.02963-0. PubMed: 15143042.

80. Beaz-Hidalgo R, Romalde JL, Prado S (2012) Identificación de bacterias del género Vibrio asociadas al cultivo de la almeja. Caracterización y patogénesis. AquaTIC 36: 1-2. Available: . http://www.revistaaquatic.com/aquatic/pdf/37_8.pdf. Accessed 14 March 2012.

81. Dauga C (2002) Evolution of the gyrB gene and the molecular phylogeny of Enterobacteriaceae: a model molecule for molecular systematic studies. Int J Syst Evol Microbiol 52: 531-547. PubMed: 11931166.

82. Mignard S, Flandrois JP (2006) 16S rRNA sequencing in routine bacterial identification: A 30-month experiment. J Microbiol Methods 67: 574-581. doi:10.1016/j.mimet.2006.05.009. PubMed: 16859787.

83. Ruimy R, Breittmayer V, Elbaze P, Lafay B, Boussemart O et al. (1994) Phylogenetic analysis and assessment of the genera Vibrio, Photobacterium, Aeromonas, and Plesiomonas deduced from small-subunit rRNA sequences. Int J Syst Bacteriol 44: 416-426. doi:10.1099/00207713-44-3-416. PubMed: 7520733.

84. Suzuki M, Nakagawa Y, Harayama S, Yamamoto S (2001) Phylogenetic analysis and taxonomic study of marine Cytophaga-like bacteria: proposal for Tenacibaculum gen. nov. with Tenacibaculum maritimum comb. nov. and Tenacibaculum ovolyticum comb. nov., and description of Tenacibaculum mesophilum sp. nov. and Tenacibaculum amylolyticum sp. nov. Int J Syst Evol Microbiol 51: 1639-1652. doi:10.1099/00207713-51-5-1639. PubMed: 11594591.

85. Venkateswaran K, Dohmoto N, Harayama S (1998) Cloning and nucleotide sequence of the gyrB gene of Vibrio parahaemolyticus and its application in detection of this pathogen in shrimp. Appl Environ Microbiol 64: 681-687. PubMed: 9464408.

86. Yamamoto S, Harayama S (1995) PCR amplification and direct sequencing of gyrB genes with universal primers and their application to the detection and taxonomic analysis of Pseudomonas putida strains. Appl Environ Microbiol 61: 1104–1109. PubMed: 7793912.