Biogeography of bacterioplankton in the tropical seawaters of Singapore


  • Stanley C.K. Lau,

    1. Division of Environmental Science and Engineering, National University of Singapore, Singapore
    Current affiliation:
    1. Division of Life Science, The Hong Kong University and Science and Technology, Clear Water Bay, USA
    Search for more papers by this author
  • Rui Zhang,

    1. Division of Environmental Science and Engineering, National University of Singapore, Singapore
    2. State Key Laboratory of Marine Environmental Science, Xiamen University, Xiamen, China
    Search for more papers by this author
  • Eoin L. Brodie,

    1. Division of Earth Science, Ecology Department, Lawrence Berkeley National Lab, Berkeley, CA, USA
    Search for more papers by this author
  • Yvette M. Piceno,

    1. Division of Earth Science, Ecology Department, Lawrence Berkeley National Lab, Berkeley, CA, USA
    Search for more papers by this author
  • Gary Andersen,

    1. Division of Earth Science, Ecology Department, Lawrence Berkeley National Lab, Berkeley, CA, USA
    Search for more papers by this author
  • Wen-Tso Liu

    Corresponding author
    1. Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA
    • Division of Environmental Science and Engineering, National University of Singapore, Singapore
    Search for more papers by this author

Correspondence: Wen-Tso Liu, Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA. Tel.: +1 217 3338442; fax: +1 217 3339464;e-mail:


Knowledge about the biogeography of marine bacterioplankton on the global scale in general and in Southeast Asia in particular has been scarce. This study investigated the biogeography of bacterioplankton community in Singapore seawaters. Twelve stations around Singapore island were sampled on different schedules over 1 year. Using PCR-DNA fingerprinting, DNA cloning and sequencing, and microarray hybridization of the 16S rRNA genes, we observed clear spatial variations of bacterioplankton diversity within the small area of the Singapore seas. Water samples collected from the Singapore Strait (south) throughout the year were dominated by DNA sequences affiliated with Cyanobacteria and Alphaproteobacteria that were believed to be associated with the influx of water from the open seas in Southeast Asia. On the contrary, water in the relatively polluted Johor Strait (north) were dominated by Betaproteobacteria, Gammaproteobacteria, and Bacteroidetes and that were presumably associated with river discharge and the relatively eutrophic conditions of the waterway. Bacterioplankton diversity was temporally stable, except for the episodic surge of Pseudoalteromonas, associated with algal blooms. Overall, these results provide valuable insights into the diversity of bacterioplankton communities in Singapore seas and the possible influences of hydrological conditions and anthropogenic activities on the dynamics of the communities.


There has been an increase in the number of studies on the diversity and distribution of bacterioplankton in the marine ecosystems; however, different and somehow contradicting observations have often been reported. Some studies indicated cosmopolitan distribution of marine bacterioplankton (Finlay, 2002; Ramette & Tiedje, 2007), whereas others revealed clear biogeography at different spatial scales ranging from a few hundred meters within a salt marsh to thousands of kilometers in the open ocean (Martiny et al., 2006). For macroorganisms, it is generally accepted that species richness increases from the poles to the equator, although there is a lack of consensus about an explanation. If the distribution of microbes follows the same gradient of species richness as observed for macroorganisms (Martiny et al., 2006), one can expect a higher microbial diversity in the tropical seas than in the colder regions.

A latitudinal gradient of marine bacterioplankton diversity was observed based on the PCR-DNA fingerprints of 103 samples collected from 57 locations around the globe (Fuhrman et al., 2008). Pommier et al. (2007) also observed a similar pattern of variation in diversity using DNA cloning of samples taken from nine different oceanic sites. However, an investigation along a Pacific Ocean transect spanning the area between the two polar regions did not show a clear trend for the species richness of bacterioplankton (Baldwin et al., 2005). At present, most studies have focused on the temperate and polar waters, leaving the tropical seas largely unexplored (Pommier et al., 2005).

In addition to natural environmental factors, human activities are other forces that directly and indirectly shape the diversity and community structure of bacterioplankton. For example, raw or treated wastewaters provide nutrients to coastal environments and subsequently affect the bottom-up control of bacterial assemblages. For the Ross Sea, the Antarctic, the influences of the human activities at the research station on bacterioplankton community were evidenced by the observation of DNA clones being affiliated with lineages that were typically associated with the effluent of wastewater treatment plants (Gentile et al., 2006). Long-term studies on the subtropical coastal waters of Hong Kong revealed spatial and temporal distribution of bacterioplankton diversity in relation to the discharge of treated sewage (Zhang et al., 2007, 2009). Similar results were obtained for the heavily polluted Guanabara Bay (Brazil) and the moderately exploited Mallorca Island (Spain) (Aguiló-Ferretjans et al., 2008; Vieira et al., 2008). However, in comparison with the well-documented human effects on animal and plant communities, detailed investigations of microbial response to anthropogenic pollution have been scarce, limiting our understanding of ecosystem responses to pollution impacts.

The marine environment in Southeast Asia represents a vast area of diverse marine habitats in the equatorial Pacific. Thus far, their microbial diversity has not been systematically investigated. In particular, Singapore is encircled by two interconnected waterways with highly contrasting environmental conditions (Fig. 1). The Singapore Strait in the south is fast-flowing and predominantly influenced by the influx of oceanic waters. In contrast, the Johor Strait in the north is stagnant and influenced heavily by sewage and river discharges. Being one of the busiest seaports in the world, harbors in the Singapore Strait receive vast amount of ballast waters discharged from cargo ships and oil tankers. Ballast waters are an important vector for the transfer of invasive and pathogenic microorganisms (Joachimsthal et al., 2004). The highly contrasting marine environment within the small area of the Singapore seas provides a good opportunity to study the effects of human impacts on the community structure and population dynamics of marine microbes.

Figure 1.

Sampling stations in the Singapore Strait (S1–S4), the vicinity of the Tuas Bay (T1–T3), and the Johor Strait (J1–J5). Seawater samples (= 3) were taken from all 12 stations in February and March 2006, and were from four representative stations only (J1, J4, S4 and T2) in August and November 2006.

In this study, we conducted a 1-year investigation of the spatial and temporal variations of bacterioplankton community structure in Singapore seawaters. During the first and second expeditions around the Singapore island, water samples were taken from 12 locations that represented regions of contrasting water quality and hydrological conditions. Physicochemical parameters were collected concomitantly as proxies of water quality. DNA extracted from the water samples were compared for differences in bacterial community structure on the basis of PCR fingerprints of the 16S rRNA genes. Based on the variations observed, four representative stations were selected for further monitoring during the third and fourth expeditions. In addition to PCR fingerprinting, the 16S rRNA genes in the water samples collected during the first and second expeditions were also subjected to cloning and sequencing, and microarray hybridization, respectively, for the identification of dominant taxa. Ultimately, the spatial and temporal variations of bacterioplankton community structure were interpreted in light of the differences in water quality and hydrographic conditions that are known for different regions of the Singapore seas.

Materials and methods

Sampling stations

Singapore (1°09′–1°29′N, 103°38′–104°06′E) is enclosed by two interconnected waterways (Fig. 1). In the south, the Singapore Strait features relatively high-speed tidal currents up to 2 m s−1, and is dominated by the oceanic influx from the South China Sea (Pacific Ocean) and the Andaman Sea (Indian Ocean) (Chen et al., 2005). In the north, the Johor Strait is subject to river and wastewater discharges, and its water current is restricted by a causeway that divides the waterway into two semi-enclosed bodies (east and west) with limited exchange (Lim, 1984a, b). Water depth along the Singapore Strait (10–20 km wide) varies between tens of meters and > 130 m, whereas that of the Johor Strait (≤ 1 km wide) is between 10 and 15 m.

Twelve sampling stations were used in this study (Fig. 1). Stations J1 and J2 (western arm) and J3–J5 (eastern arm) were positioned along the Johor Strait. Stations S1–S4 were within the Singapore Strait. Stations T1–T3 were in the vicinity of the Tuas Bay. All sampling stations were located at least 500 m away from the shore. Water samples were collected from all stations in February and March 2006. Based on the spatial and temporal variability of bacterioplankton communities observed, four representative stations (J1, J4, S4 and T2) were selected for the expeditions in August and November 2006. Stations J1 and J4 were selected for being in the mid-reach of the western and eastern arms of the Johor Strait, respectively. S4 in the Singapore Strait was chosen being farthest away from the coastline while still remaining in Singapore waters. T2 in the Tuas Bay was chosen for being in the midst of the most intense shipyard operations in Singapore. All samples described hereafter are named according to the station and the sampling time. For example, samples collected from S4 in February 2006 are referred to as S4-Feb.

Sample collection and handling

Seawater samples (= 3) were collected from 5 m below the water surface of each station. This sampling depth was initially chosen in approximation of the mid-depth of the shallower water of the Johor Strait (10–15 m); the same sampling depth was then used for all stations throughout this study. Each water sample was split into three portions: (1) 500 mL in autoclaved plastic bottles for DNA extraction, (2) 200 mL in autoclaved, acid-washed glass bottles for nutrient analysis, and (3) 15 mL in centrifuge tubes prefilled with 1.5 mL paraformaldehyde solution (4% w/v final concentration in the samples) for bacterial counts. All samples were stored in ice and transported to the lab within 3–4 h. Samples for DNA extraction were pre-filtered through hardened ashless cellulose filters (8-μm pore size, Grade 540; Whatman) to remove large particles and eukaryotes, and then through 0.2-μm polycarbonate membranes (ISOPORE™; Millipore) to collect bacterioplankton. Each membrane was cut into small pieces, submerged in 0.6 mL of DNA extraction buffer (100 mM Tris-HCl, 100 mM Na2-EDTA, 100 mM Na2HPO4, 1.5 M NaCl, 1% CTAB; pH 8) and kept at −80 °C until further processing.

Physical and chemical seawater parameters

Seawater temperature, salinity, pH and dissolved oxygen (DO) levels were measured in situ with a portable meter (Hach). The concentrations of total organic carbon and total nitrogen were determined in the laboratory using a total organic carbon analyzer (TOC-VCSH; Shimadzu, Japan) equipped with a total nitrogen measurement unit (TNM-1, Shimadzu) according to manufacturer's instructions. The values of these parameters were used as proxies of water quality when the conditions of different regions of Singapore seawaters were compared.

Enumeration of bacterioplankton

Aliquots of 2 mL of each paraformaldehyde-fixed sample were diluted to 20 mL with ultra-filtered water and slowly filtered onto a 0.2-μm (pore size) black polycarbonate membrane (20 mm diameter; Millipore) supported by a 1-μm (pore size) cellulose acetate membrane to obtain evenly distributed bacterial cells. After the filtration, the membranes were air dried and stained with 4′,6-diamidino-2-phenylindole (DAPI) for epifluorescence microscopy (BX51; Olympus, Japan) at 1000× magnification (Zhang et al., 2007). For each membrane, the number of bacterial cell-shaped, DAPI-stained particles (referred to as bacteria-like particles, BLP) in four randomly chosen fields of known sizes was manually counted. The mean value of the four counts represented the abundance of BLP captured on the membrane. The density of BLP (mL−1) of each water sample was determined based on the mean ± SD of three replicate membranes.

DNA extraction

Total DNA was extracted from the bacterioplankton captured on polycarbonate membranes using standard phenol-chloroform-isoamyl extraction procedures (Liu et al., 1997). Briefly, bacterial cells were lysed using a mixture of achromopeptidase, lysozyme, and proteinase K. DNA released was purified using phenol-chloroform-isoamyl extraction, and subsequently precipitated and washed twice in 70% ethanol. DNA pellets were air-dried, dissolved in 50 μL ultra-pure water, and purified using the DNeasy® Tissue Kit (Qiagen). Purified DNA was checked for molecular size and quantity using agarose gel electrophoresis.

DNA fingerprinting

The bacterial 16S rRNA genes in the DNA samples were amplified in PCR using the forward primer 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) labeled with the WellRed D4 dye (Beckman Coulter) and the reverse primer 927R (5′-ACCGCTTGTGCGGGCCC-3) without labeling. PCR amplicons were analyzed for terminal restriction fragment length polymorphism (T-RFLP) using the endonuclease RsaI as described previously (Liu et al., 1997; Chen et al., 2004). For each T-RFLP profile, the peak area of each individual terminal restriction fragment (T-RF) was transformed to the percent of the total area of all peaks. T-RFs with peak areas < 1% of the total were excluded (Liu, et al., 2010). A distance matrix was calculated based on Euclidean distances after square-root transformation of the percent peak areas, and visualized using non-metric multi-dimensional scaling (nMDS) ordinations. Samples of different stations were compared for differences in T-RFLP profiles using anosim (analysis of similarity). The T-RFLP profiles were also analyzed for possible correlation with water physicochemical parameters using canonical correspondence analysis (CCA) after a detrended correspondence analysis test, as described in detail elsewhere (Wu et al., 2006).

16S rRNA gene clone library construction

The bacterial 16S rRNA genes were PCR amplified using the primer pair 27F and 1492R (5′-GGTTACCTTGTTACGACTT-3′) as described previously (Zhang et al., 2007). PCR amplicons were cloned using a TOPO TA cloning kit (Invitrogen) according to the manufacturer's instructions. One hundred clones were randomly chosen from each library and classified into ribotypes on the basis of RFLP. The coverage of the clone libraries was calculated using the Good's coverage estimator (Good, 1953), where coverage = 1 − (the number of singleton ribotype/the total number of clones subjected to RFLP analysis). One clone was randomly selected from each ribotype and sequenced bidirectionally. The DNA sequences obtained were manually trimmed, edited, and assembled into contigs using the software sequencher (Gene Codes Corp., USA). Sequences that had passed the check for chimers using bellerophon on the greengenes server ( were subjected to blast analysis on the NCBI server for the identification of closest affiliates. A phylogenetic tree of the sequences was constructed using the mega software package (Tamura et al., 2007).

Microarray hybridization

PCR amplicons (obtained as described above) were hybridized to the PhyloChip (G2 version), which contained 297 851 probes for 8935 operational taxonomic units (OTUs) that represented all 121 delimited prokaryotic orders (Brodie et al., 2006). Each OTU, representing a group of 16S rRNA gene sequences sharing ≥ 97% similarity, was targeted by a probe set that contained at least 11 perfect match–mismatch probe pairs. The detection of an OTU was regarded as positive when 92% or more of the relevant probe pairs gave positive signals (i.e. positive fraction ≥ 0.92) (Brodie et al., 2007). The procedures for PCR product preparation and hybridization were performed as previously described (Brodie et al., 2006, 2007).

Accession numbers

DNA sequences obtained in this study are available from GenBank under the accession numbers EU010127EU010235.


Physical and chemical seawater parameters

Clear spatial variations of physicochemical parameters were observed for the 12 sampling stations during the course of this study (Supporting Information, Fig. S1). In general, as observed over the four sampling events, seawater in the Singapore Strait (S1–S4) and the Tuas Bay (T1–T3) was more saline and less polluted than that in the Johor Strait (J1–J5) using higher levels of pH and DO, and lower levels of dissolved organic carbon and total nitrogen as proxies. Within the Johor Strait, seawater quality deteriorated along the east (J3, J4 and J5) and west (J1 and J2) channels towards the Causeway. Temporally, seawater physicochemical parameters within the Singapore Strait and the Tuas Bay were relatively stable in comparison with those measured for the Johor Strait (Fig. S1). Overall, the biplot of CCA indicated that the values of salinity, DO, and pH were negatively correlated with those of temperature, dissolved organic carbon, and total nitrogen (Fig. 2), conforming to the differences in water quality known for the Singapore Strait and the Johor Strait (Lim, 1984a, b).

Figure 2.

Biplot of CCA of the relationships between T-RFLP profiles and the environmental variables of the Singapore seawaters measured in this study. The numbers indicate T-RFs whose size could be matched with DNA clone sequences (Table S2).

Bacterioplankton density

The density of BLP observed over the course of this study ranged between 9.6 × 105 and 1.3 × 107 mL−1 (Fig. S2). The Singapore Strait (S1–S4) and the Tuas Bay (T1–T3) had a lower BLP density than the Johor Strait (J1–J5) throughout the course of this study. For each sampling event, the highest density was always associated with the samples taken from the east Johor Strait (J3–J5).

T-RFLP profiles

Four distinct types of T-RFLP profiles were observed for the samples collected from the 12 stations in February and March 2006 (Fig. 3). All samples of the east Johor Strait (J3–J5) in February contained a unique 563-bp T-RF that represented 53.8 ± 9.3% to 87.1 ± 6.3% (= 3) of the total signal intensity (Fig. 3a). This profile is referred to as type A hereafter. All other samples of the Johor Strait (February and March) produced another profile (type B) that contained the largest number of discernible T-RFs (> 20), with those of 300–320-bp and 840-bp in size being dominant (Fig. 3b). Type C profile was observed for all samples of the Singapore Strait (S1–S4) that contained a predominant 421-bp T-RF, representing 46.7 ± 5.4% to 81.5 ± 2.6% (= 3) of the total intensity (Fig. 3c). This T-RF was also observed in some samples of the other three types of profiles but in much lower abundance. The samples of the Tuas Bay (T1–T3) resulted in the type D profile that had features of the type B (840-bp T-RF) and C (421-bp T-RF) profiles (Fig. 3d).

Figure 3.

Four types of T-RFLP profiles (a–d) were observed for the seawater samples collected over the course of this study. The table shows that the association of the profiles with samples collected from different stations. The profiles shown exemplify the samples collected in February 2006. Data presented for each profile are the fragment size (bp) and the signal intensity of each TRF relative to the total of the sample. The phylogenetic affiliations of the TRFs were estimated using in silico digestion of cloned 16S rRNA gene sequences.

MDS ordination placed the samples collected during February and March in three separate clusters with a statistically significant stress value of < 0.2 (Fig. 4a). One cluster was for samples displaying the type A profile; another cluster for samples showing the type C profile. Samples of type B and D profiles were grouped into a third cluster. In support of the MDS ordination, the results of anosim indicated that differences between clusters were significant (R ≥ 0.918, = 0.001) while differences within cluster were not (R ≤ 0.165, > 0.084).

Figure 4.

MDS ordinations of the T-RFLP profiles for the samples collected (a) from all 12 stations in the February and March 2006, and (b) from four representative stations in all four sampling rounds in 2006. An MDS ordination is considered statistically significant when its stress value is < 0.2.

Since samples from within the same region of the Singapore sea resulted in highly similar T-RFLP profiles, we decided to reduce the number of stations to four during the sampling events in August and November 2006. Samples collected during those two expeditions resulted in the same four types of T-RFLP profiles as observed for February and March (Fig. 3). MDS ordination of the samples collected from J1, J4, S4 and T2 over the four sampling times resulted in three distinct clusters (stress value = 0.050) (Fig. 4b). Samples of J1 and J4 exhibited temporal variations between type A and B profiles (Fig. 4b). Based on our observation in the field, the occurrence of the type A profile was believed to be associated with algal blooms in the water. Although all samples of T2 were regarded as having the type D profile based on visual examination (Fig. 3d), samples collected in November were clustered with the samples of S4 (type C), and those collected in three other months were clustered with the majority of samples from J1 and J4 (type B) (Fig. 4b).

CCA conducted for all samples of the four sampling events revealed that T-RFLP profiles derived from samples of the Johor Strait were associated with warmer and more eutrophic conditions, whereas those from the Singapore Strait and the Tuas Bay were associated with waters that had higher levels of salinity, DO, and pH (Fig. 2). The two axes of the biplot together explained 67.68% of the variation in the T-RFLP profiles.

16S rRNA gene clone libraries

The samples of J1, J4, and S4 collected in February, as representatives of the three distinct clusters in the MDS ordination (Fig. 4a), were subjected to the cloning and sequencing of 16S rRNA genes. After screening for RFLP ribotypes and chimeric sequences, the number of sequences obtained for J1-Feb, J4-Feb, and S4-Feb were 34, 15, and 54, respectively. The coverage of the three clone libraries, on the basis of the RFLP profiles, were 66.7% for J1-Feb, 70.8% for J4-Feb and 44.8% for S4-Feb. The relatively low coverage of the S4-Feb clone library may be related to a high micro-diversity of the 16S rRNA gene sequences therein, as 63% of the DNA sequences retrieved shared over 99% similarity with at least one and up to five other sequences in the library.

At the phylum level, the three clone libraries were dominated by sequences affiliated with Proteobacteria followed by Bacteroidetes. The remaining sequences were affiliated with Actinobacteria, Cyanobacteria, Firmicutes, Verrucomicrobia and unclassified taxa (Table 1). The majority of the sequences of sample S4-Feb were affiliated with Alphaproteobacteria (46.3% of the total), whereas those of sample J1-Feb were predominately Gammaproteobacteria (35.3%) (Fig. S3). The vast majority of sequences of the two samples were affiliated with taxa with no cultured representative (Table S1). Conversely, 13 of the 15 sequences obtained for sample J4-Feb were affiliated with cultured members of Gammaproteobacteria, including Pseudoalteromonas spp., Vibrio spp., and Shewanella spp. (Table S1).

Table 1. Phylogenetic affiliation (phylum level) of PCR amplicons detected by DNA cloning and PhyloChip, respectively
Month (2006)FebruaryMarchFebruaryMarchFebruaryMarchFebruaryMarch
  1. Data shown are the number of DNA clone sequences and PhyloChip OTUs that were affiliated with each phylum.

  2. Numbers in parentheses indicate percent of the total of each sample.

  3. ND, not detected.

Acidobacteria ND39 (2.6%)ND57 (3.0%)ND34 (2.5%)49 (2.6%)
Actinobacteria 1 (2.8%)125 (8.3%)ND177 (9.4%)4 (6.9%)109 (8.1%)160 (8.5%)
Bacteroidetes 6 (16.7%)139 (9.2%)1 (6.7%)149 (7.9%)10 (17.2%)102 (7.6%)141 (7.5%)
Cyanobacteria 2 (5.6%)56 (3.7%)ND57 (3.0%)7 (12.1%)57 (4.2%)54 (2.9%)
Firmicutes ND152 (10.1%)ND221 (11.8%)ND158 (11.7%)273 (14.5%)
Proteobacteria24 (66.7%)816 (54.1%)14 (93.3%)981 (52.2%)34 (58.6%)709 (52.6%)963 (51.3%)
Others3 (8.3%)180 (11.9%)ND237 (12.6%)3 (5.2%)180 (13.3%)238 (12.7%)

Identification of T-RFs

In silico digestion of the DNA sequences resulted in T-RF sizes (Table S2) that matched with many of the actual T-RFs observed (Fig. 3). An arbitrary range of ±3 bp was used for the matching so as to compensate for errors that might have occurred during DNA sequencing and for ambiguities that might be associated with actual T-RF size determination. The 421-bp T-RF that dominated the type C profile matched with three Alphaproteobacteria sequences and all Cyanobacteria sequences obtained for sample S4-Feb. This T-RF associated strongly with conditions of high salinity, pH, and DO as revealed using the CCA biplot (Fig. 2). For the type A profile (Fig. 3a), the unique 563-bp T-RF and its neighboring minor T-RFs were associated with the Pseudoalteromonas-like sequences of sample J4-Feb (Table S2). A cluster of T-RFs with sizes between 312 and 318 bp that occurred primarily in type B profile were identified as being affiliated with Bacteriodetes. These T-RFs in the type A and B profiles were associated with warmer and more eutrophic conditions of the waters (Fig. 2). Several T-RFs (i.e. 840, 883, and 893 bp) matched with a number of Alpha and Gammaproteobacteria sequences. Although these T-RFs were ubiquitously present in all profile types, they had a stronger association with warmer and more eutrophic conditions than waters of higher salinity, pH, and DO (Fig. 2).

Microarray hybridization

T-RFLP analysis is limited to the detection of dominant taxa. To further differentiate bacterioplankton diversity in different regions of the Singapore seas (particularly for Tusa Bay for its intense shipyard operation), microarray hybridizations of the PCR amplicons of the 16S rRNA genes were used. Samples of the four representative stations (J1, J4, S4, and T2) were selected for this analysis. Instead of using samples collected in February (same as for DNA cloning and sequencing), microarray hybridization was conducted with samples collected in March. This was to avoid the episodic algal bloom observed in February masking the differences/similarity in bacterioplankton diversity that may otherwise normally exist between the east and the west Johor Strait.

The hybridization of PCR amplicons to the PhyloChip detected 1349–1879 OTUs for samples J1-Mar, J4-Mar, T2-Mar, and S4-Mar (Table 1). The OTUs belonged to 42 different phyla with Proteobacteria (51.3–54.1%), Firmicutes (10.1–14.5%), Actinobacteria (8.1–9.4%), Bacteroidetes (7.5–9.2%), and Cyanobacteria (2.9–4.2%) dominating every sample (Table 1). Within Proteobacteria, the Gamma-subclass was most dominant (39.3–42.4%), followed by the Alpha-subclass (28.6–31.4%) and the Delta-subclass (10.3–12.6%) (Table S3). Betaproteobacteria, being cosmopolitan members of freshwater habitats, had a lower number of OTUs (68) present in the sample of the Singapore Strait (S4-Mar) compared with those of other stations (101–145 OTUs).

The four samples shared 1138 OTUs in common (Table 2). T2-Mar and J4-Mar had higher numbers (158 and 143, respectively) of unique OTUs compared with the other two samples (34 and 35, respectively). Furthermore, cluster analysis based on the intensity of hybridization signal indicated that the samples of T2 and J4 shared a higher level of similarity than with the two other samples (Fig. 5). Although the samples of the Tuas Bay had T-RFLP profiles that appeared to be a mixture of the Singapore Strait and the west Johor Strait, sample T2-Mar had the highest number of PhyoChip OTUs (158) that were unique to itself (Table 2). This was particularly the case for the OTUs of Firmicutes; among the 123 OTUs detected for all samples, 68 of them were unique to T2-Mar. In contrast, most of the Cyanobacteria OTUs were found in all four samples.

Table 2. Number of PhyloChip OTUs that were common to all four samples analyzed, and unique to each sample, respectively
  1. Numbers in parentheses indicate the percent of OTUs of the total detected for all samples.

  2. ND, not detected.

Acidobacteria 29 (49.2%)ND8ND1
Actinobacteria 86 (42.0%)52239
Bacteroidetes 87 (49.4%)71029
Cyanobacteria 48 (76.2%)1111
Firmicutes 123 (39.3%)417668
Proteobacteria 620 (54.9%)22761254
Total1138 (51.3%)4114332155
Figure 5.

Heatmap of the signal intensity of individual OTUs detected by the PhyloChip for samples J1-Mar, J4-Mar, T2-Mar, and S4-Mar. C: OTUs common for all samples; U followed by a station name: OTUs unique to a given station. Phylogenetic composition of C and U are shown in Table 2.


The tropical seas of Southeast Asia have been and are still one of the regions least investigated for marine microbial diversity. However, using 97% similarity in the 16S rRNA gene sequences as a delimiter, only 5% of the clone sequences obtained in this study could be regarded as belonging to species not discovered before. Indeed, many of the sequences obtained were closely affiliated with those that have been recovered previously from seawater in Hawaii, the Atlantic, and the polar regions (> 98% similarity).

The Singapore Strait lies where currents cross from the South China Sea on the east and the Andaman Sea on the west, as driven by the interplay of seasonal monsoons and equatorial currents. There is a net drift of water mass from the South China Sea through the Singapore Strait into the Andaman Sea during the northeast monsoon (the winter in the northern hemisphere) (Chen et al., 2005). The reverse would occur during the southwest monsoon (winter in the southern hemisphere). In this study, the expeditions during February and August were during the NE and SW monsoons, respectively; the expeditions during March and November are regarded as in the inter-monsoon periods.

Throughout the course of this study, the samples of the Singapore Strait were dominated by a 421-bp T-RF (Fig. 3), which was affiliated with the sequences of Cyanobacteria and Alphaproteobacteria (Table S2). These two taxa dominate the open waters globally, including the South China Sea (Ma et al., 2004; Zhang et al., 2011) and the Indian Ocean (Burkill et al., 1993). Indeed, among the 10 clone library sequences (sample S4-Feb) that were associated with the 421-bp T-RF, seven were > 99.6% similar to the sequences of Cyanobacteria and Alphaproteobacteria previously retrieved from the South China Sea (Zhang et al., 2011) (Tables S1 and S2). These observations, together with the circulation pattern described above, offer a plausible explanation for the year-round dominance of the 421-bp T-RF in the samples of the Singapore Strait. Nonetheless, the 421-bp T-RF was also found in samples collected from other regions of Singapore waters. Being strongly associated with the conditions of high DO, salinity, and pH (Singapore Strait) (Fig. 2), the 421-bp T-RF was found in much lower relative abundance in the less saline and more eutrophic water of the Johor Strait (Fig. 3).

The Johor Strait receives freshwater discharged from several rivers on the Malaysian side (most rivers on the Singaporean side have been dammed up to create reservoirs). The river discharge might have led to a higher diversity of Betaproteobacteria (cosmopolitan freshwater bacteria) found in samples of the Johor Strait than the Singapore Strait (Table S1). The number of Betaproteobacteria OTUs detected by the PhyloChip for the east and west Johor Strait were respectively 2.1 and 1.5 times the number for the Singapore Strait. Nonetheless, the number of Betaproteobacteria OTUs detected for each sample varied between 68 and 145, representing only 5–6% of the total of each sample. This low relative abundance may be a reason that the sequences of Betaproteobacteria were not found in the clone libraries (Fig. S3).

In addition to the freshwater discharge, the Johor Strait is also under the influence of treated and raw sewage, leading to levels of nitrogen and dissolved organic carbon being several fold higher than those in the Singapore Strait; concomitantly, the Johor Strait also had lower levels of DO and pH (Fig. S1). The eutrophic conditions of the Johor Strait supported the highest BLP densities observed within the Singapore seawaters (Fig. S2), especially towards the causeway that divides the waterway into east and west, with limited water exchange (Fig. 1). A number of T-RFs that were affiliated with the sequences of Bacteroidetes (132 and 312–318 bp) and Gammaproteobacteria (563–567 bp) were found to be strongly associated with the eutrophic conditions of the Johor Strait (Figs 2 and 3, Table S2). This is supported by the results of PhyloChip hybridization, which detected a higher diversity of the two taxa in the Johor Strait (Table S3). Whereas Bacteroidetes are important consumers of high molecular weight dissolved organic matter (Cottrell & Kirchman, 2000), many members of Gammaproteobacteria are opportunistic heterotrophs that specialize in the colonization of organic aggregates or other nutrient-rich microniches (Pernthaler & Amann, 2005).

During the course of this study, the bacterioplankton community structure revealed by T-RFLP analysis was largely temporally and spatially stable within the entire Johor Strait (type B profile), with the exception of the samples collected from the west (J3–J5) in February and from the east (J1) in November, which featured the type A profile (Fig. 3). This type of profile was dominated by a cluster of T-RFs (563–567 bp) that were associated with sequences of Pseudoalteromonas. This is supported by the clone library of sample J4-Feb being dominated by sequences of Gammaproteobacteria (13 of 15 sequences), with nearly 70% being associated with Pseudoalteromonas (Fig. S3 and Table S1). We attribute the dominance of Pseudoalteromonas to the algal blooms observed at the monitoring stations during sampling. Many Pseudoalteromonas strains found in the marine environments are known to be associated with algal blooms and can produce algicidal extracellular metabolites (Skerratt et al., 2002). The algicidal activities are believed to give Pseudoalteromonas competitive advantages in the acquisition of dissolved organic matter (from lysed algal cells) and the colonization of debris (dead algal cells) (Holmström & Kjelleberg, 1999). It is interesting to note that 12 of the 13 Gammaproteobacteria sequences retrieved from J4-Feb were affiliated with cultivated strains. Many members of Gammaproteobacteria are opportunistic heterotrophs that grow well and outcompete other taxa on nutrient-rich media in the laboratory. Perhaps the particulate and dissolved organics associated with algal blooms had stimulated the growth of the culturable taxa of Gammaproteobacteria and led to the dominance of their sequences in the clone library.

Hydrologically, the west Johor Strait can be divided into the inner reach and the outer reach (Lim, 1984a, b). In this study, stations J1 and J2 fell within the inner reach, and station T3 in the outer reach (Fig. 1). Whereas the water of the inner reach is stagnant and heavily influenced by river discharge, the outer reach is turbulent and subject to exchange with the Singapore Strait due to the combined effects of monsoon winds and tidal currents (Lim, 1984a, b). Water samples collected from station T3 resulted in the type D T-RFLP profile that contained T-RFs related to the type B profile of the Johor Strait and the type A profile of the Singapore Strait (Fig. 3). This observation is similar to the effects of water masses mixing on bacterioplankton community structure reported for the Columbia River estuarine, where river and ocean waters met and led to the formation of a community that contained taxa from both habitats (Fortunato et al., 2011).

The type D T-RFLP profile was also observed for the samples of stations T1 and T2, suggesting that the effects of mixing of water masses on bacterioplankton communities were not limited to the lower reach of the Johor Strait but extended to the nearby Tuas Bay. Although the mixing of water currents is suggested to be an important factor determining the dominant taxa observed for the Tuas Bay, the hybridization of the PCR amplicon of sample T2-Mar to the PhyloChip revealed a large number of Firmicutes OTUs that were unique to the sample (Table 2). The vicinity of T2 is different from other areas of the Singapore Seas in having very intense shipyard operations. However, it is premature to associate the detection of unique Firmicutes OTUs with shipyard operations.

To conclude, the results in this study indicate a clear biogeography of bacterioplankton diversity within the small area of Singapore seawaters that could be largely explained by the interplay of oceanic influx, freshwater discharge, and sewage pollution already known for the marine environment of Singapore. The differences in physiochemical parameters observed between different regions of the Singapore seas support these known conditions. Using T-RFLP together with DNA cloning and sequencing, we identified taxa (421-bp T-RF associated with Cyanobacteria and Alphaproteobacteria) pertaining to the relatively oceanic conditions of the Singapore Strait. Water entering the Johor Strait (from the Singapore Strait) as a result of tidal cycles (Chen et al., 2005) is under the influence of river discharge and sewage disposal (Lim, 1984a, b). As such, the 421-bp T-RF gave way to a high diversity of Bacteroidetes, Betaproteobacteria, and Gammaproteobacteria that are presumably associated with freshwater discharge and the eutrophic conditions of the Johor Strait as depicted by the CCA biplot of T-RFLP profiles and also the result of microarray hybridization. In the Tusa Bay, the interplay of water in the Singapore Strait and the Johor Strait led to bacterioplankton communities that contained taxa pertaining to the two water bodies. In addition, microarray hybridization revealed a large number of Firmicutes OTUs that were unique to the water sample collected from Tusa Bay in March. It is not currently known what conditions enriched those OTUs and whether they were persistent in Tusa Bay. In contrast to the clear seasonality observed for the marine and freshwater habitats in the temperate regions (Fuhrman et al., 2006; Pinhassi et al., 2006), bacterioplankton community structure within the tropical seawater of Singapore was temporally stable, except for the episodic surge in the population size of opportunistic heterotrophs that were presumably associated with algal blooms (Figs 3 and 4). Overall, the results of this study provide valuable insights into the bacterioplankton communities in tropical seawaters and the possible influences of hydrological conditions and anthropogenic activities on the dynamics of microbial populations within the Singapore seas.


This work was supported by a research grant of National University of Singapore to Wen-Tso Liu. Part of this work was performed at Lawrence Berkeley National Laboratory, supported by the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Rui Zhang was partially supported by Program for New Century Excellent Talents in Xiamen University (NCET 09-0683).