Antimicrobial Activities of Sponge-Derived Microorganisms from Coastal Waters of Central Vietnam

Bacteria associated with invertebrates are considered as good sources for biologically active compounds. Sponge-derived bacteria were screened for antimicrobial activities, the presence of the cluster genes of polyketide synthases (PKSs) and non-ribosomal peptide synthetases (NRPSs), and through a colony picking method. Crude extracts of broth cultures were tested for microbial inhibition. Eleven out of 25 isolated strains showed inhibition of at least one of eight tested indicator microorganisms. Antimicrobial activities were observed in the strains coded HM5, HM6, and HM9 with the presence of NRPS and PKSII genes, whereas the isolate HM21 held both NRPS and PKSII and inhibited only the growth of Bacillus subtilis by the well diffusion method and only inhibited Serratia marcescens by the colony picking method. Two isolates, HM5 and HM6, belonged to the species of Bacillus. Interestingly, the isolate HM9 was nearest to Streptomyces mexicanus T NBRC100915 (GenBank accession number AB249966) with 94% sequence similarity. This potent strain HM9 could possibly be considered as a new species and a good source for bioactive compound discovery. Some isolates showed NRPS/PKS genes but did not exhibit antimicrobial activity. Thus, we suggested that both molecular and traditional methods should be conducted for the screening of antimicrobial producers.


Introduction
Reversing fossil evidence from about 635 to 750 million years ago, Porifera is one of the most ancient living organisms [1] and is considered as an excellent source of drug materials. It is evident that 99% of the total of 15,000 species of marine mammals are capable of serving as biologically active substance producers [2]. Polyketide synthases (PKSs) and non-ribosomal peptide synthetases (NRPSs) are a huge source of bioactive compounds, which have been found from Archaea and Eukarya, and are most commonly found in bacteria; they are abundant in the phyla of Actinobacteria, Firmicutes, Proteobacteria, and Cyanobacteria and in the fungal phylum of Ascomycota. Marine microorganisms have been reported as a source of peptides, such as those harvested from soil inhabiting Actinomycete and Bacillus [3]. Those gene clusters are coded for multifunctional enzymes involved in the synthesis of natural bioactive compounds, many of which were proved to become drugs for humans. The Bacillus genus is a good source of well-known antibiotics such as lipopeptides (LPs) and polyketides, which were dark polyethylene bag, stored in an ice-box, and transferred to the laboratory as soon as possible for further analysis.
Before collecting the sponge, the water environmental parameters were measured by the water quality profile of AAQ 1183 (ALEC, Japan). The temperature was 28.1-28.3 • C, and the oxygen dissolved concentration was 6.03-6.12 mg/L. Marine Living Resources, Institute of Oceanography, VAST. After collecting, the specimen was preserved in a sterile dark polyethylene bag, stored in an ice-box, and transferred to the laboratory as soon as possible for further analysis.
Before collecting the sponge, the water environmental parameters were measured by the water quality profile of AAQ 1183 (ALEC, Japan). The temperature was 28.1-28.3 °C, and the oxygen dissolved concentration was 6.03-6.12 mg/L.

Isolation of Sponge-Associated Bacteria-SAB and Test Strain Vibrio sp.
In the laboratory, the natural seawater collected at the sampling site was filtered using Whatman Glass microfiber filters (GF/F 0.45 µm) to remove suspended sediment and particular organic matter; after that, the seawater was sterilized by a Hirayama HV-II autoclave at 121 °C and 1 atm for 20 min. The sterile natural seawater was used to treat sponge and coral specimens and for experimental applications.
In the laboratory, the sponge specimen was washed three times with the sterile natural seawater for 5 min to remove surface-attached microbes. The sponge specimen was sterile crushed with a pestle and mortar. A total of aerobic microorganisms in 1 g of fresh sponge was counted as 1.31 × 10 5 cfu/g, corresponding to 218 and 175 colonies at dilution concentrations of 10 −2 and 10 −3 with 80 and 103 colonies, respectively; at a dilution of 10 −4 , less than 30 colonies were recorded, and most of them were the same in shape and size. Then, it was diluted with the sterile natural seawater to a dilution of 10 −5 to have a homogenized sample. The 100 µL of the homogenized sample was spread onto the R2A agar plates; then, the streaking plate method was used to get pure isolates. The isolates were preserved in marine broth plus 50% glycerol at a room temperature of ca. 30 °C. A total of 25 distinct culture colonies in shapes, colors, and sizes was selected for antimicrobial screening.
A branch of bleaching coral, Acropora sp., at the same sampling site, was sampled for isolation of Vibrio sp. with selected media, namely TCBS (HiMedia-India) and ChromagarTM Vibrio (CHROMagar-France). The strain V1 was confirmed as Vibrio parahaemolyticus when it showed green color on TCBS and mauve color on Chromagar TM Vibrio and 99% sequence similarity to the 16s RNA gene. The strain V2 was found in yellow color and a large colony, and it changed the medium's color on TCBS and was colorless on CHROMagar TM . The partial sequence of 16s RNA showed that V2 could be Vibrio campbellii with 98% sequence identity. This occasional pathogenic bacteria V. parahaemolyticus and V. campbellii were used as test strains in this present study and were preserved in 20% glycerol at −80 °C.
The 16s RNA gene method to determinate Vibrio sp. is as follows: (1) Universal primer sets of 27F, 1500R were used for amplification of the 16S rRNA gene with the PuReTaq ™ Ready-To-Go ™ PCR Beads (Healthcare) with a total volume of 25 µL, including 5 µL DNA templates (50 ng), 10 pmol 27F, 10 pmol 1500R, and DNA-free H2O (Sigma). The PCR reaction conditions included initial denaturation (2 min at 94 °C), followed by 30 primer annealing cycles (40 s at 50 °C), and extended

Isolation of Sponge-Associated Bacteria-SAB and Test Strain Vibrio sp.
In the laboratory, the natural seawater collected at the sampling site was filtered using Whatman Glass microfiber filters (GF/F 0.45 µm) to remove suspended sediment and particular organic matter; after that, the seawater was sterilized by a Hirayama HV-II autoclave at 121 • C and 1 atm for 20 min. The sterile natural seawater was used to treat sponge and coral specimens and for experimental applications.
In the laboratory, the sponge specimen was washed three times with the sterile natural seawater for 5 min to remove surface-attached microbes. The sponge specimen was sterile crushed with a pestle and mortar. A total of aerobic microorganisms in 1 g of fresh sponge was counted as 1.31 × 10 5 cfu/g, corresponding to 218 and 175 colonies at dilution concentrations of 10 −2 and 10 −3 with 80 and 103 colonies, respectively; at a dilution of 10 −4 , less than 30 colonies were recorded, and most of them were the same in shape and size. Then, it was diluted with the sterile natural seawater to a dilution of 10 −5 to have a homogenized sample. The 100 µL of the homogenized sample was spread onto the R2A agar plates; then, the streaking plate method was used to get pure isolates. The isolates were preserved in marine broth plus 50% glycerol at a room temperature of ca. 30 • C. A total of 25 distinct culture colonies in shapes, colors, and sizes was selected for antimicrobial screening.
A branch of bleaching coral, Acropora sp., at the same sampling site, was sampled for isolation of Vibrio sp. with selected media, namely TCBS (HiMedia-India) and ChromagarTM Vibrio (CHROMagar-France). The strain V1 was confirmed as Vibrio parahaemolyticus when it showed green color on TCBS and mauve color on Chromagar TM Vibrio and 99% sequence similarity to the 16s RNA gene. The strain V2 was found in yellow color and a large colony, and it changed the medium's color on TCBS and was colorless on CHROMagar TM . The partial sequence of 16s RNA showed that V2 could be Vibrio campbellii with 98% sequence identity. This occasional pathogenic bacteria V. parahaemolyticus and V. campbellii were used as test strains in this present study and were preserved in 20% glycerol at −80 • C.
All the PCR products of PKS and NRPS were visible under ultraviolet fluorescence after Gold View staining with a molecular size marker 2000 bp (DL 2000 DNA Marker, TaKaRa, Japan). The PCR conditions for PKSII (IPF6, IPR6) were 95 • C for 1 min for initial denaturation followed by 30 cycles of 94 • C for 1 min and 58 • C for 40 s and 72 • C for 2 min with a final extension of 10 min at 72 • C. The PCR conditions for NRPS were 95 • C for 1 min followed by 35 cycles of 94 • C for 40 s and 59 • C for 1 min and 72 • C for 1.5 min with a final extension of 10 min at 72 • C. For PKSI (K1F-M6R) the reaction was done at 95 • C for 1 min for initial denaturation followed by 30 cycles of 94 • C for 40 s and 55 • C for 1 min and 72 • C for 1.5 min with a final extension of 10 min at 72 • C. All the PCR were performed in 25 µL tubes each containing 12.5 µL of 2 × EasyTaq ® PCR SuperMix (TransGen Biotech, Beijing, China), 1 µL of forwarding primer, 1 µL of reverse primer (each primer of 10 pmol), 1 µL of 100 ng/µL DNA, 1% (v/v) of DMSO, and DNA-free water to a final volume of 25 µL. Universal primer sets of 27F, 1500R were used for amplifying the 16s rRNA gene [23]. All PCR amplicons were carried out in TaKaRa PCR Thermal Cycler Dice Touch at the key lab in the school of Science Sun Yat-Sen University, Guangzhou, China.

Primary Screening of Antimicrobial Activities by Colony Picking Method
The primary screening for antimicrobial activities was performed by the colony picking method according to Hettiarachchi et al. [20] with slight modification (Figure 2). In brief, the test strains were prepared into 4 mL of marine broth (5.0 g/L peptone from soya bean, 1.0 g/L yeast extract, 0.3 g/L dipotassium hydrogen phosphate, 1.0 g/L glucose, 5.0 g/L sodium chloride), shaking at 120 rpm, and incubation at room temperature (approximately 30 • C due to the in-situ temperature of 28.1-28.3 • C when collecting samples) for 18-24 h until the concentration of bacteria was a measure of the 0.5 McFarland standard [24], approximately 10 8 cfu/mL. Then, the test trains were diluted 200 times in the culture medium to have the test strains of 5 × 10 5 cfu/mL. A 100 µL culture of 5 × 10 5 cfu/mL for each test strain was spread onto Mueller Hinton agar (MHA, India) plates, and then the SAB strains were picked up from a 3 day cultivation colony on R2A agar medium and introduced to make a lawn on agar plates; one sponge-associated bacterium was applied for one test strain only. The plates were incubated with test strains B. subtilis, S. aureus, S. typhimurium, and E. coli at 37 • C, V. campbellii, V. parahaemolyticus, and S. marcescens at 25 • C, and the yeast C. albicans at 30 • C for up to 3, 5, or 7 days.  The SAB strains were streaked in marine agar plates for 3-5 days and then inoculated into 300 mL Erlenmeyer flasks containing 100 mL BM medium (yeast extract: 1 g/L, beef extract: 1 g/L, tryptone: 2 g/L, glucose: 10 g/L, and 1000 mL filtered sea water). After 72 h of incubation at 30 °C with shaking at 120 rpm, the bacterial cells and the supernatants were homogenized using an ultrasonic processor for 30 s to break the cells. The homogenized broth was extracted with ethyl acetate 1/1 (v/v). Crude extract was dried and re-suspended in 1 mL of methanol (Merck). The methanolic extract was applied for an antimicrobial activity test.

Antimicrobial Activity Testing by Well Diffusion Method
The antimicrobial activity was conducted by the well diffusion method [25]. Each 100 µL of cultured broth of overnight tested strains was spread on the MHA plates; the wells were punched by 6 mm diameter with a sterile Corning long disposable glass Pasteur pipette. A 30 µL methanolic extract of each SAB was pipetted into four available wells on MHA containing tested bacteria. The same amount of methanol without extract was used as a negative control. A disc of 30 µg tetracycline (BioRad France) was used as a positive control. All the plates tested with B. subtilis, S. typhimurium, and E. coli were incubated at 37 °C for 24 h. The plates tested with S. marcescens were incubated at 25 °C for 24 h. The zone of inhibition was measured, and the value was expressed as a mean of measured wells excluding the well diameter ( Figure 3). The SAB strains were streaked in marine agar plates for 3-5 days and then inoculated into 300 mL Erlenmeyer flasks containing 100 mL BM medium (yeast extract: 1 g/L, beef extract: 1 g/L, tryptone: 2 g/L, glucose: 10 g/L, and 1000 mL filtered sea water). After 72 h of incubation at 30 • C with shaking at 120 rpm, the bacterial cells and the supernatants were homogenized using an ultrasonic processor for 30 s to break the cells. The homogenized broth was extracted with ethyl acetate 1/1 (v/v). Crude extract was dried and re-suspended in 1 mL of methanol (Merck). The methanolic extract was applied for an antimicrobial activity test.

Antimicrobial Activity Testing by Well Diffusion Method
The antimicrobial activity was conducted by the well diffusion method [25]. Each 100 µL of cultured broth of overnight tested strains was spread on the MHA plates; the wells were punched by 6 mm diameter with a sterile Corning long disposable glass Pasteur pipette. A 30 µL methanolic extract of each SAB was pipetted into four available wells on MHA containing tested bacteria. The same amount of methanol without extract was used as a negative control. A disc of 30 µg tetracycline (BioRad France) was used as a positive control. All the plates tested with B. subtilis, S. typhimurium, and E. coli were incubated at 37 • C for 24 h. The plates tested with S. marcescens were incubated at 25 • C for 24 h. The zone of inhibition was measured, and the value was expressed as a mean of measured wells excluding the well diameter ( Figure 3).

Identification of Antimicrobial Producing Strains
Antibacterial producing strains were identified by the traditional method with the addition of 16s rRNA gene analysis [26].

Data Analysis
All data were processed using Microsoft Excel 2016. All experiments were repeated three times. The results are stated as the mean ± standard deviation.

Identification of Antimicrobial Producing Strains
Antibacterial producing strains were identified by the traditional method with the addition of 16s rRNA gene analysis [26].

Data Analysis
All data were processed using Microsoft Excel 2016. All experiments were repeated three times. The results are stated as the mean ± standard deviation.

Sponge Associated Bacteria with PKS-and NRPS-Holders
The results of holder genes strains (Table 1 and

Identification of Antimicrobial Producing Strains
Antibacterial producing strains were identified by the traditional method with the addition of 16s rRNA gene analysis [26].

Data Analysis
All data were processed using Microsoft Excel 2016. All experiments were repeated three times. The results are stated as the mean ± standard deviation.

Antimicrobial Tests
The results of the primary screening of the holder cluster genes of PKS/NRPS strains for antimicrobial producers are exhibited in Table 2 and Figure 5. When four of seven holder bioactive cluster genes strains showed antagonistic activities, other SAB strains did not present any antimicrobial activities to all tested strains by the colony picking method ( Table 2). Table 2. Primary antibiotic producer strains by colony picking method (Notation: (+) inhibition zone observed; (−) no inhibition zone observed).

Test Strains
Holder PKS/NRPS Strains HM2 HM5 HM6 HM9 HM12 HM21 HM22  Antimicrobial activity of the SAB ( Figure 6 and Table 3) indicated that among 25 isolates tested, eleven strains (44%) exhibited antibacterial activities. The three strains, HM2, HM21, and HM23, showed a pattern of B. subtilis only. Two other strains, HM5 and HM8, possessed patterns of B. subtilis and E. coli, whereas the strain HM12 showed patterns of B. subtilis and S. marcescens. The isolates HM19 possessed patterns of E. coli, S. marcescens, and S. typhimurium. Particularly, the strains HM6 and HM20 were shown to be very potent antimicrobial producers, with patterns of B. subtilis, E. coli, S. marcescens, and S. typhimurium. The strain HM22 showed activity against only coral pathogenic S. Antimicrobial activity of the SAB (Figure 6 and Table 3) indicated that among 25 isolates tested, eleven strains (44%) exhibited antibacterial activities. The three strains, HM2, HM21, and HM23, showed a pattern of B. subtilis only. Two other strains, HM5 and HM8, possessed patterns of B. subtilis and E. coli, whereas the strain HM12 showed patterns of B. subtilis and S. marcescens. The isolates HM19 possessed patterns of E. coli, S. marcescens, and S. typhimurium. Particularly, the strains HM6 and HM20 were shown to be very potent antimicrobial producers, with patterns of B. subtilis, E. coli, S. marcescens, and S. typhimurium. The strain HM22 showed activity against only coral pathogenic S. marcescens. Interestingly, it exhibited the highest level of inhibition compared to other strains which were inhibiting the growth of this tested bacterium. Genetic screening revealed that the strain HM20 had neither PKS nor NRPS. Additionally, the colony picking method showed that this strain was not active against any tested bacteria. However, the crude extract from this isolate inhibited against four tested strains.

Identification of Potential Strains.
Results of analysis of 16s RNA genes of active strains (Table 3) showed that most strains were identified as Bacillus sp., except strain HM9, which was Streptomyces sp. The strain HM9 was the nearest to the type strain Streptomyces mexicanus T NBRC100915 (AB249966: 94%). This actinomyces was thought to be a well-known source of antibiotics so far, for instance, gougerotin, a water-soluble antibiotic with non-antifungal activity [27].
Additionally, members of Bacillus species showed effective inhibitory activity to all four test microorganisms, and they were reported as potential strains for biologically active substance  Table 3. Classification of antibiotic producing strains (Notation: ND, not detected; NRPS, non-ribosomal peptide synthetases; PKSII, polyketide synthases II; the letters A, B, C, D, E, and F denote types of antimicrobial patterns against Bacillus subtilis, Escherichia coli, Serratia marcescens, Salmonella typhimurium, Vibrio campbellii, and Vibrio parahaemolyticus, respectively).

Identification of Potential Strains.
Results of analysis of 16s RNA genes of active strains (Table 3) showed that most strains were identified as Bacillus sp., except strain HM9, which was Streptomyces sp. The strain HM9 was the nearest to the type strain Streptomyces mexicanus T NBRC100915 (AB249966: 94%). This actinomyces was thought to be a well-known source of antibiotics so far, for instance, gougerotin, a water-soluble antibiotic with non-antifungal activity [27].
Additionally, members of Bacillus species showed effective inhibitory activity to all four test microorganisms, and they were reported as potential strains for biologically active substance producers. In addition, pure substances obtained from Bacillus sp. were capable of inhibiting some cancer cells and some important enzymes [28]. Most recently, Bacillus sp. strain BC028 isolated from the blue mussel Mytilus edulis also was reported as a new hexapeptide producer; this strain produced a natural biological active compound considered as a new antibiotic, and it showed inhibition of the growth of opportunistic pathogenic Enterococcus faecalis and pathogenic S. aureus with rather low minimum inhibition concentration values of 8 and 12 µM, respectively [29].
Although the strains HM8, HM19, HM20, and HM23 showed antimicrobial activity in well diffusion with broth culture extract, they did not show antimicrobial activities in the colony picking method, and all three types of potent bioactive coding genes were not found ( Table 3).
The bioactive coding genes were found in some strains including HM6, HM9, and HM12, and these strains also showed antimicrobial ability in both the colony picking method and well diffusion of crude extracts. The result of the presence of PKS genes in HM21 and HM22 strains are somewhat uncertain (Figure 2), but the strain HM21 showed inhibition against S. marcescens with the colony picking method, and its crude extract inhibited against B. subtilis, while the strain HM22 did not have any inhibition activity against indicator strains but crude extract of this strain inhibited the coral pathogen S. marcescens.

Discussion
Research on microorganisms from sponges in Son Cha revealed the number of microorganisms isolated from six different species of sponges ranging from 150-820 cfu/g. However, the microorganisms in that study were isolated from samples stored at −20 • C [30]. The fresh hard coral-associated bacteria reported 1.32 ± 0.98 × 10 6 cfu/g wet sample by the culture based method, whereas the total number of heterotrophic bacteria counted directly from the fluorescence dye method was 2.07 ± 0.58 × 10 8 cells/g [31]. In addition, bacteria also were isolated from marine sediment, and the heterotrophic bacteria were recorded at 1.5 × 10 6 cfu/g sediment in the waters of Mot Island [32]. Microorganisms that live with the hosts, such as sponges, urchins, corals, mollusks, or other invertebrates were reported to have very high specificity and strongly depended on their hosts. Moreover, they also have inherent characteristics of microorganisms such as nutrient requirements, culture conditions, and so on, and of course, initial host states had crucial effectiveness for the isolation of microorganisms. Those important impact factors have a huge influence on the number of isolated microorganisms and types of isolates as well. From the large number of microorganisms counted on the various dilution levels mentioned above, this study isolated 25 pure strains with distinct shapes, color characteristics, and different cell sizes. The community of microbials associated with the Chondrilla nucular in the Mediterranean Sea was very diverse, as expressed by shape and size under scanning electron microscopy and by analysis of the 16s rRNA gene and the visual genes with 22 different phylotypes of associated bacteria [33]. Microorganisms consisted of 40 to 60% of the total biomass of their hosts and reached more than 10 9 cells/cm 3 , which was three to four times higher than those of marine bacteria with 10 6 mL/L [34]. Imhoff et al. [10] reported that actinobacteria groups and Firmicutes were found to be a source of holders of those bioactive coding genes, and in particular, the PKSII clusters were more successfully detected than the others.
Using degenerate primers MDPQQRf and HGTGTr for all the heterotrophic bacteria (non-actinomyces), amplification of the PKSI genes was efficiently done, and PKSI were detected with abundant bacterial strains such as Ruegeria arenilitoris, Pseudomonas aeruginosa, Bacillus aquimaris, Marinobacter vinifirmus, and Pseudovibrio sp. Similar to the detection of PKSI for non-actinomyces strains, primers MTf and MTr were used to amplify NRPS, and thus the study found antimicrobial activity strains such as Pseudomonas azotoformans, and Ruegeria arenilitoris strain G-M8. The well-known set primers of A3F and A7R were used to amplify NRPS in actinomyces successfully, with Kocuria rhizophilia strain TA68 possessing NRPS and showing activity against K. pneumoniae, E. coli, and Candida albicans. Other actinomyces, the Sanguibacter inulinus strain ST50, showed inhibition of the fungi A. niger (ATCC 16888) and the human pathogenic yeast C. albicans (ATCC 10231). Actually, the set of primers K1, M6R and A3F, A7R targeting for PKSI and NRPS, respectively, used in this study have been successfully applied for actinomyces rather than heterotrophic bacteria. That may be a reason why non-PKSI were detected in this present study when using primers K1, M6R. Other actinomycetes isolated from the sponge Kytococcus sedentarius strain DSM 20547, K. pneumoniae (ATCC BAA-1705), A. niger (ATCC 16888), or Bacillus safensis strain NBRC 100,820 possessed antimicrobial activity that inhibited E. coli, while none of both bioactive genes were detected. Previously, a cross streak method was used for antimicrobial producer screening; all the associated bacteria isolated from the hard coral Acropora muricata exhibited non-antibiotic producers [13]. However, other papers reported that the cross-streak method was a suitable method for screening of the actinomyces group. A total of 21 actinomycetes isolated from marine environments were tested for antimicrobial activities against eight pathogenic bacteria, and the results revealed that all tested strains showed inhibition of at least one of eight tested bacteria [35]. The actinomycetes isolated from the soil also showed as a good source of antibiotic producers after screening by the cross streak method [36,37]. Hettiarachchi et al. [20] introduced the colony picking method for screening of antibiotic producers; that method seemed to be fast, easy, and efficient for other groups of bacteria, such as Pseudoalteromonas, rather than actinomyces Recently, our other study reported that associated bacteria with coral species in central Vietnam fluctuated by following time [31,38]. In particular, the strain Bacillus sp. was the most common found among isolates from the coral Acropora hyacinthus and inhibition against coral pathogenic bacterium S. marcescens strain PDL100 but no inhibition of other indicators such as B. subtilis, S. typhimurium, and E. coli 0157 by the agar well diffusion method [31]. So far, Nguyen et al. [39] reported that 7 out of 11 bacteria isolated from the hard corals were active strains and inhibited the growth of B. subtilis, S. typhimurium, and E. coli, whereas the other four strains inhibited activity against S. typhimurium. Noteworthily, the strain S. marcescens, known to cause a severe disease named "white spot" on reef-building coral A. palmata in the Florida Keys, United States, was used as a test strain [40].
This presented study reported that bacterial isolates showed antimicrobial activity, but both cluster genes were not detected. On the other hand, some bacteria strains possessed PKS/NRPS and no inhibition of all tested microorganisms. We suggested screening of secondary metabolites from microbials by firstly using a molecular approach besides traditional methods that may be efficient for choosing marine strains for natural compound discovery. In fact, PKS and NRPS were thought to be easily found in active strains. However, this results depended mostly on the primers. The efficient primers were tested for actinomyces in contrast to the poor investigation of primers for marine heterotrophic bacteria. The sponge-associated bacteria were screened for bioactive compound producers with well-known primers for actinomyces. This approach will be more convincing when more specific primers for marine bacteria are used.