The seawater samples were collected from two sampling stations, the “port side” (36˚30'99"N and 140˚58'46"E) and “sea side” (36˚31'74"N and 140˚59'20"E), of the Oarai coastal area of the North Pacific Ocean, Ibaraki prefecture, Japan (Figure 1). Sampling stations are located closely, about 1 km far from one to another. One of them, the port side, is close to the Oarai port area which is semi-enclosed by a sea bank. The sea side sampling point, one the other hand, is thought to be subjected to the river inflow to some extent because, the Naka River, one of the class-one rivers of Japan, and the Hinuma River flows into the Pacific at the north end of the coastline.

In order to observe the similarities and dissimilarities in bacterial community structure and diversity at different

time scale, water samples were collected in March 2013 (early spring), October 2013 (autumn), February 2014 (winter), April 2014 (spring), and July 2014 (summer). At every sampling, about 5 liters of seawaters were collected in a previously sterilized screw-capped plastic bag and carried back to the laboratory in ice boxes within 2 hours after sampling.

About two litres of the seawater sample were filtered through 0.22 mm pore sized Sterivex-GP pressure filter unit (Millipore, Billerica, MA, USA) using a peristaltic pump to have the community compositions. The sterivex cartridge filters were immediately kept in sterile bags and stored at −80˚C until further processing. The filters in the sterivex units were cut aseptically and placed inside screw tubes just before DNA extraction.

The water temperature was measured at the time of sampling by using a mercurial thermometer. The seawater salinity was determined by a handy refractometer (IS/Mill-E, As One, ATAGO, Japan). The chlorophyll-a dataset with 4-km resolution was obtained from Level-3 MODerate resolution Imaging Spectroradiometer (MODIS) aqua standard−mapped image distributed by the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (oceancolor.gsfc.nasa.gov/cms). Averages of eight consecutive days used to prepare chlorophyll-a illustrated maps for each sampling time using Ocean Data View (ODV) software [12] .

DNA was extracted from the sterivex filters by combined use of ChargeSwitch Forensic DNA Purification Kits (Invitrogen^TM, Carlsbad, USA) and ZircoPrep Mini (FastGene^TM, Nippon Genetics Co. Ltd., Bunkyo-ku, Tokyo, Japan) beads beating with a slight modification of the manufacturer’s protocol. A MicroSmash (MS-100R, Tomy Medico., Ltd., Tokyo, Japan) was used for beads beating at 5000 rpm and 4˚C for 30 seconds for each filter under sterile conditions with a great care to avoid contamination. The extracted DNA was also cleaned using NucleoSpin (MACHEREY-NAGEL GmbH & Co. KG, Neumann-Neander-Str., Düren, Germany) gDNA clean- up kit according to the manufacturer protocol and stored −30˚C until amplification.

The V1-V3 hyper variable regions of 16S rDNA gene were amplified from the extracted DNA templates by polymerase chain reaction (PCR). The 27F with multiplex identifier (MID): 5’-CCATCTCATCCCTGCG- TGTCTCCGACTCAGXXXXXXXXXXAGAGTTTGATCMTGGCTCAG-3’, where X’s represents the sample-specific multiplex identifier-MID [13] was used as the forward primer and the 519R with adaptor: 5’-CCTATCCCCTGTGTG-CCTTGGCAGTCTCAG(GWATTACCGCGGCKGCTG)-3’ was used as the reverse primer. Each PCR reactions were carried out in a volume of 20 µL in triplicates while the mixture consisted of 2 µL DNA template, 13.1 µL molecular biological grade double distilled water, 0.6 µL (5 µM) each primer, 2 µL 10X TaKaRa Ex Taq Buffer, 1.6 µL TaKaRa dNTP mixture, and 0.1 µL TaKaRa Ex Taq HS Polymerase (TaKaRa, Japan). Thermal cycling was carried out for a total of 25 cycles as per the following conditions: initial denaturation at 94˚C for 4 mins, denaturation at 98˚C for 10 sec, annealing at 55˚C for 30 sec, elongation at 72˚C for 1 min and final elongation at 72˚C for 10 mins. After amplification, the desired length of the 16S rDNA gene was confirmed by agarose gel electrophoresis and any sort of contamination was carefully verified by observing the bands of the triplicates of the same samples. After confirming the desired length, amplified DNA products were purified and normalized using Agencourt AMPure XP (Beckman Coulter INC., USA) according to the guidance of the 454 Sequencing Amplicon Library Preparation Method Manual (GS Junior Titanium Series 2012, Roche, USA). The purified DNA amplicon was then quantified using Quant-iT Picogreen dsDNA Kit (Invitrogen, Carlsbad, USA). The bacterial 16S rDNA gene amplicons were then sequenced using the 454 GS Junior sequencer (Roche, USA) at Atmosphere and Ocean Research Institute (AORI), the University of Tokyo (Kashiwa, Chiba, Japan) according to the manufacturer’s protocol for 454 GS Junior Titanium Series.

The open-sourced MOTHUR program [14] was used for subsequent analysis, quality checking and arrangement of the obtained sequences following the guidelines available to the operation manual for the 454 [http://www.mothur.org/wiki/454_SOP]. Initially, the unique sequences were selected and then the similar sequences were clustered and aligned against the SILVA bacterial databases [15] . Then the pre-cluster method [16] was applied to reduce the sequencing errors by screening, filtering, and de-noising. The chimera. uchime command was used for checking and removing the chimeras. The sequences were subsequently classified against the ribosomal 160 database project (RDP) database and the inactive components such as chloroplast, mitochondria etc. organelles affiliated “former” bacterial sequences were removed from our dataset to improve the data quality. The qualified high-quality sequences were then used to generate distance matrix and clustered assigning to operational taxonomic units (OTUs) at 97% identity level [17] . A representative sequence from every OTU was used for classification by running the MOTHUR program based on the SILVA bacterial databases. To standardize the number of sequences between samples, they were randomly re-sampled to the sample with the fewest reads (2674 reads) using the MOTHUR program based on the OTU files clustered at 0.03 cut-off levels.

The species richness and diversity indices were considered to evaluate the biodiversity and analyse the rarefaction. For the species richness, the Chao1 index [18] , and for diversity, the inverse Simpson (Invsimpson) index [19] was calculated using the MOTHUR software at OTU definition at a distance of 0.03.

To check the correlations between bacterial communities and environmental factors, nonmetric multidimensional scaling fitting with the environmental features (metaMDS) was carried out based on the relative abundance data of each OTUs. The permutation test was used following the “MASS” [20] and “Vegan” package [21] from R software (R Development Core Team 2012). Redundancy analysis (RDA) was also carried out using the R software with “Vegan” package [21] based on the relative abundance data of each OTUs and environmental information. Toassess the similarities or dissimilarities between the bacterial groups of the two sampling stations, the clustering analysis (Bray-Curtis) test [22] was also performed using the R software with “Vegan” package [21] .

Changes in the water temperature, salinity, and chlorophyll-a values at the two sampling stations are shown in Table 1. Salinity showed similar fluctuation patterns at both stations, with the maximum values, around 35, were obtained in February 2014 and the minimum, around 24, in July 2014 after the rainy season for both the stations. Other features, such as water temperature and water depth were also similar at both the stations.

The chlorophyll-a values obtained from the satellite data were used to prepare chlorophyll-a illustrated maps (Figure 2) for each date of sampling using Ocean Data View (ODV) software. Then the chlorophyll-a values of the sampling locations were obtained from this illustrated maps considering the location (latitude and longitude) of the sampling stations. However, as the sampling stations are closely located, it was not possible to obtain data separately for the two stations (Table 1). The chlorophyll-a values varied from about 2.5 µg∙m⁻³ in February to about 5.0µg∙m⁻³ in March. No chlorophyll-a data was obtained during July due to cloudy weather (Table 1 and Figure 2).

After sequencing all the samples a total of 66,609 sequences were obtained which consists of 5249 different types of OTUs. The obtained sequences were analysed for community composition. The composition of the

^*PSU, practical salinity unit. ^**As the sampling stations are located closely, just about 1 km far from one another, it was not possible to get chlorophyll-a data separately for port side and sea side station rather a single value from the closest available location was noted here for both the stations.

major groups mostly at phylum/class level is presented to Figure 3.

The Phylum Proteobacteria was the most dominant one followed by the Bacteroidetes at almost all the sampling periods regardless of the sampling stations. Among different subgroups of Proteobacteria, the class Alphaproteobacteria was the most abundant one and the class Gammaproteobacteria was the second. In the case of the phylum Bacteroidetes, the class Flavobacteriia was mostly abundant. The relative abundance of the Flavobacteriia was higher in port side while the Gammaproteobacteria in the sea side at almost all the sampling periods. The class Alphaproteobacteria was almost equal or slightly higher in the port side station except April 2014. Among other groups the phylum Cyanobacteria, Verrucomicrobia and the class Betaproteobacteria were higher in abundance to the sea side as compared to the port side. Moreover, the unclassified members were also higher on the sea side. The class Flavobacteriia was most abundant in March, followed by the February at the port side, while the class Alphaproteobacteria in July at both the sampling points. The abundance of the phylum Cyano-

bacteria was also relatively higher in July and lower or almost absent in March and February; while the phylum Actinobacteria was more abundant in February. The phylum Deferribacteres was found abundant in October at sea side station (Figure 3).

As the class Flavobacteriia and Alphaproteobacteria were 2 most dominant classes in both the stations throughout study periods, community composition within these classes at family or genus level was also evaluated. In the case of Flavobacteriia, the order Flavobacteriales was the only contributing group (Figure 4). Although, there were differences between the sampling periods, but in general, genera NS3a marine group, Polaribacter, and Winogradskyella of the family Flavobacteriaceae comprised most of the bacterial fractions at port side while the genera Flavobacterium, NS4 marine groups and NS5marine groups of the same family at the sea side station (Figure 4). Analysis of the members within Alphaproteobacteria also showed marked differences between the stations (Figure 5). The maximum contribution to the bacterial community was made by the members of the order Rhodobacterales. The genera Lentibacter, Nereida, Sulfitobacter and unclassified members of the family Rhodobacteraceae was relatively abundant in the port side station while the genus Roseobacter clade, order Rickettsiales and SAR11 contributed significantly in the sea side station. However, there were seasonal variations in their abundance. For example, the genus Sulfitobacter contributed mostly to the samples of March, February, and April while the order SAR11 to the samples of October, April and July (Figure 5).

Bacterial biodiversity was evaluated in terms of the species richness and richness-evenness considering the Chao index and inverse Simpson (invSimpson) index, respectively, as was shown in Figure 6. The Chao index values were higher at the sea side in March 2013, October 2013 and July 2014 while at the port side in February 2014 and April 2014. Compared to the port side in almost all the sampling periods except April 2014, Simpson index values were higher at the sea side. It seems reasonable, from these observations, to suppose that more diversified communities were made up in the sea side area. For both the indices, the highest values were observed in April and the lowest in October on the port side. Relatively higher Chao index values were observed in October and July on the sea side, while the lowest was shown in February (Figure 6). The rarefaction curves are showing the relationship between the numbers of obtained sequences and observed OTUs (Supplementary Figure 1A).

Non-metric Multidimensional Scaling fitting with the environmental features (metaMDS) based on the relative abundance data of the samples and seasonal environmental data was used to categorize bacterial community composition of the studied stations at different sampling periods. The samples were separated according to the sampling periods (r² = 0.63 and P = 0.1, based on 1000 permutations) as well as according to the stations (r² = 0.22 and P = 0.17, based on 1000 permutations); the community composition was similar between the stations at February (Figure 7).

The association between the bacterial community structure and environmental factors was examined by RDA. In Figure 8, the samples are plotted on a sample-to-sample basis, with respect to stations (r² = 0.3 and P = 0.02, based on 1000 permutations) and season (r² = 0.5 and P = 0.4, based on 1000 permutations). The water temperature (r² = 0.08 and P = 0.7, based on 1000 permutations) and salinity (r² = 0.1 and P = 0.6, based on 1000 permutations) had no significant influence in clustering (Figure 8).

We performed a Bray-Curtis clustering analysis based on the bacterial relative abundance data at phylogenetic level. Samples were aligned according to degrees of similarity in community composition on a sample-to- sample basis (Figure 9). However, the April sample of the port side station was aligned next to the sea side clade, whilst the February sample of the sea side was aligned next to the port side clade. This indicates that there were similarities in bacterial community composition among the samples of the February and April (Figure 9).

About 20.92% to 30.72% of the OTUs were common in the port side station and about 7.25% to 31.53% in the

sea side station between different sampling periods (Table 2). The percentages of common OTUs shared between March and October, October and February, February and April, and April and July were 6.95, 6.88, 10.20, and 5.79 respectively at the port side station while 7.63, 5.12, 8.25, and 7.90 respectively for the sea side station (Table 3). The overall observations showed that bacterial community composition was fluctuated highly between two consecutive sampling periods and the community composition was also dissimilar between the stations at most of the sampling periods (Table 2 and Table 3).

Only 14 OTUs out of 5249 (0.27%) was found common throughout the study regardless of the sampling periods and stations. However, their contribution to the total abundance was 10.7% in July to 27.36% in April on the port side station while 24.46% in October to 48.31% in March on the sea side station (Figure 10). Bacterial groups of these common OTUs were Candidatus Actinomarina (Acidimicrobiales) of phylum Actinobacteria; Fluviicola, NS4 marine group, NS5 marine group (2 OTUs), Owenweeksia and Polaribacter (2 OTUs) (Flavobacteriales) of phylum Bacteroidetes; Roseobacter clade (2 OTUs) (Rhodobacteriales) and unclassified SAR11 of Alphaproteobacteria; unclassified Alteromonadales and SAR86 (Oceanospirillales) of Gammaproteobacteria; and 12up (Rhodocyclales) of Betaproteobacteria (Figure 10).