Science of the Total Environment 472 (2014) 746–756

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Response of bacterial communities to environmental changes in a mesoscale subtropical watershed, Southeast

Anyi Hu a, Xiaoyong Yang a,b, Nengwang Chen c, Liyuan Hou a, Ying Ma d,Chang-PingYua,⁎ a Key Laboratory of Urban Environment and Health, Institute of Urban Environment, Chinese Academy of Sciences, 361021, China b College of Life Sciences, , Xiamen 361005, China c College of the Environment and Ecology, Xiamen University, Xiamen 361005, China d Fisheries College, , Xiamen 361021, China

HIGHLIGHTS

• Bacterial communities in were investigated by 454-pyrosequencing. • The Jiulong River harbored relatively low abundance of typical freshwater bacteria. • The Jiulong River could be contaminated by fecal pollution. • Water chemistry has effects on the diversity and structure of bacterial communities.

article info abstract

Article history: This study used 16S rRNA gene-based pyrosequencing (16S-pyrotag) to investigate both planktonic and benthic Received 3 June 2013 bacterial communities in two main tributaries (North River and West River) of the Jiulong River Watershed Received in revised form 27 September 2013 (JRW), a mesoscale subtropical watershed that has experienced intensive human perturbation in recent decades. Accepted 19 November 2013 The results of 16S-pyrotag showed that benthic bacterial communities were clearly more diverse and uniform Available online 10 December 2013 than surface bacterioplankton communities. The results of taxonomic assignments indicated that Betaproteobacteria, Actinobacteria and Firmicutes were significantly more abundant in planktonic than in benthic Keywords: fl Jiulong River watershed communities, whereas the relative abundances of Acidobacteria, Delta-, Gammaproteobacteria, Chloro exi and Environmental change Nitrospira were higher in sediment than in water samples. In particular, several sewer- and fecal-pollution bac- Bacterial community terial indicators were observed in water samples, implying that the water bodies of the JRW were contaminated Typical freshwater bacteria by fecal pollution. Using the typical freshwater bacteria (TFB) taxonomic framework, 57.6 ± 10%, 27.6 ± 10.9% 16S-pyrotag and 10.4 ± 6.9% of sequences recovered from planktonic communities could be assigned to lineages, clades and tribes of TFB, respectively. The relatively lower abundance of TFB implied that some unknown or unique autoch- thonous bacterioplankton populations occurred in the JRW. The principal coordinate analysis (PCoA) and one way analysis of similarity (ANOSIM) analysis demonstrated that planktonic bacterial community structures were significantly different between North River and West River, whereas benthic communities from these two tributaries were grouped together. Multivariate statistical analysis revealed that nutrient concentrations and stoichiometry were the key drivers of both α-andβ-diversity patterns of bacterioplankton communities. Overall, our results indicate that the diversity, composition and structure of planktonic bacterial communities are sensitive to water chemistry (e.g., nutrient concentrations and stoichiometry) in the JRW, and therefore can serve as a good sentinel of environmental changes in this watershed. © 2013 Elsevier B.V. All rights reserved.

1. Introduction as well as ecosystem functions (Shiklomanov, 1998). Freshwater ecolo- gists recently recognized that river ecosystems play a critical role in re- Rivers, an important component of the hydrological cycle, comprise gional and global biogeochemical cycles (Aufdenkampe et al., 2011), less than 0.01% of total freshwater reserves. However, river is widely and thus can serve as a good sentinel of environmental changes in ter- distributed across the landscape and provides major volume of water restrial and atmospheric processes (Crump et al., 2009; Williamson for maintaining human health, industrial and agricultural production et al., 2008). However, because of the increasing loading of nutrients, heavy metals and other chemical contaminants (e.g., pesticides, herbi- ⁎ Corresponding author. Tel./fax: +86 592 6190768. cides, hormones and antibiotics) resulting from human activities (pop- E-mail address: [email protected] (C.-P. Yu). ulation growth and rapid urbanization), the health of river ecosystems

0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.11.097 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 747 is seriously degraded, especially in developing countries (Chen and have led to water quality degradation, eutrophication and algal blooms Hong, 2012; Hilton et al., 2006; Vörösmarty et al., 2010). As a conse- in the JRW and the adjacent (Chen and Hong, 2012; Huang quence, deteriorating water quality reduces the availability of water re- et al., 2011; Li et al., 2011; Yan et al., 2012). Based on analysis of over 20- sources to general population and poses a human health risk. year monitoring dataset of the JRW, Chen et al. (2013) suggested that Furthermore, there is growing evidence that declining water quality the raising livestock breeding and overfertilization are the major may lead to the loss of biodiversity of aquatic communities including sources of nutrient pollution in NR and WR, respectively. In addition, a algae, invertebrates and fishes, which has a major effect on ecosystem variety of chemical pollutants including antibiotics, polycyclic aromatic functions and services (Villéger et al., 2011; Weijters et al., 2009). How- hydrocarbons, and endocrine disrupting compounds have been shown ever, it is still poorly known how river microbial community responds to contaminate the JRW (Maskaoui et al., 2002; Zhang et al., 2012; to human-caused environmental changes. Zheng et al., 2011). Current findings imply that anthropogenic forces Microbes represent by far the most diverse and numerically pre- pose potential threat to the health and sustainability of the JRW ecosys- dominant organisms in river ecosystems, and play the vital roles in me- tem as well as public health. In a previous dry season survey based on diating and regulating carbon and nutrient fluxes as well as the removal rDNA PCR-DGGE analysis, Liu et al. (2011) found that agricultural pollu- of contaminants (Ghai et al., 2011; Kirchman et al., 2004; Newton et al., tion is one of the main factors controlling planktonic eukaryotic and 2011). A large body of literature based on analysis of the 16S rRNA genes bacterial community composition in the JRW. However, DGGE method demonstrates that some phylogenetic groups or tribes of planktonic provides limited information about phylogenetic composition and Actinobacteria, Alpha-, Beta-, Gammaproteobacteria, Bacteroidetes, structure of JRW bacterial communities, and sediment bacterial assem- Cyanobacteria and Verrucomicrobia have a cosmopolitan distribution in blage in this watershed remains largely unknown. freshwater biome, and represent unique bacterial lineages found only Here, we investigated the diversity, composition and structure of in this habitat (i.e., so-called “typical freshwater bacteria”,(TFB)) planktonic and benthic bacterial communities in NR and WR using mul- (Newton et al., 2011; Zwart et al., 2002). However, the majority of tiple methods including physiochemical characterization, 16S rRNA those studies identifying TFB were based primarily on observations of gene-based pyrosequencing (16S-pyrotag) and multivariate statistical lentic environments, but relatively little is known about lotic bacterial analyses. The main goals of the study were: i) to compare the diversity communities (Logue and Lindström, 2008; Newton et al., 2011). Previ- and structure of both planktonic and benthic bacterial communities in ous studies have indicated that the assembly of freshwater bacterial two main tributaries (NR and WR) of the JRW; ii) to determine the dis- communities often tightly correlated with physicochemical and biolog- tribution of indigenous (typical freshwater bacteria) and exotic (fecal- ical factors including temperature, pH, dissolved oxygen (DO), nutri- or sewage-indicator bacteria) bacterial communities in the JRW; iii) to ents, carbon source, grazing and virus lysis (i.e., species-sorting) as identify the critical factor (local or regional factors) influencing the di- well as regional (dispersal) processes (Crump et al., 2012; Hu et al., versity and structure of planktonic and benthic bacterial communities 2010; Jones et al., 2009; Kirchman et al., 2004; Yannarell and Triplett, in NR and WR. 2005). In addition, some researchers noted that toxic contaminants could also cause a shift in the composition and function of river bacterial 2. Materials and methods communities (Vishnivetskaya et al., 2011; Yergeau et al., 2012). In view of this, anthropogenic water-quality changes would definitely lead to 2.1. Study area changes in indigenous bacterial communities, thereby affecting the functional characteristics of lotic ecosystems. However, only limited The JRW, situated in a subtropical monsoon zone, is one of the most studies have begun to explore the potential effects of human-induced developed areas in Province and consists of eight districts environmental changes on aquatic bacterial communities. (, , Zhangping, Hua'an, Changtai, Pinghe, Several recent studies demonstrated that bacterial community and Longhai) (Fig. 1). Annual discharges of NR and WR are 8.2 and 3.7 - might be a better choice as a biological indicator for assessing ecosystem billion m3, respectively. More than five million residents from south- health than conventional biological indicators such as macroinverte- eastern Fujian Province use the JR as the main source of water supply brate, fish and bird populations (Ager et al., 2010; Lear and Lewis, for domestic, agricultural and industrial activities (Huang et al., 2011). 2009). Lear and colleagues found that biofilm bacterial community Numerous studies have shown that the upstream area of NR structures in streams changed significantly along a catchment land- (Longyan district; from NR_1 to NR_2, Fig. 1) and the downstream use gradient (Lear et al., 2011; Lear and Lewis, 2009), while Wang area of WR (Zhangzhou district; from WR_7 downstream, Fig. 1)re- et al. (2011) suggested that changes in microbial community composi- ceived significantly higher nutrient loads from livestock breeding and tion in response to watershed urbanization (land-use change) may have domestic sewages, while cash crop production contributed great nutri- significant impact on ecosystem functioning. However, because of in- ents to the upper WR (Pinghe and Nangjing districts; from WR_1 to herent limitations of classical molecular techniques (e.g., genetic finger- WR_3, Fig. 1)(Chen et al., 2013, 2008; Li et al., 2011). Due to excessive printing, clone libraries and fluorescence in situ hybridization), those inorganic P loadings contributed by livestock breeding (waste dis- studies provided limited information about phylogenetic structure of charges) and cash crops production (fertilizers), the N:P ratio in the microbial communities and how they responded to anthropogenic dis- water body of JR declined in recent years, which have increased the turbances. With the emergence of next-generation massively parallel risk of eutrophication and algal bloom (Chen et al., 2013). sequencing technology, an improved sampling depth (103–106 se- quences per sample) is being introduced to allow the identification of 2.2. Field sampling and physiochemical analyses the dominant and rare populations within a community simultaneous- ly, thus it could reveal a significantly greater level of microbial diversity Sampling was carried out during 16–19th October, 2011 under nor- than conventional molecular tools (Binladen et al., 2007; Caporaso et al., mal hydrological condition (stream flow is in its normal condition). As 2011; Sogin et al., 2006). shown in Fig. 1, a total of 18 sampling sites were selected within the The Jiulong River Watershed (JRW) is a mesoscale watershed locat- JRW with nine sites for each tributary. Water temperature, pH, electron- ed in southeastern China, with a catchment area of 1.47 × 104 km2 ic conductivity (EC), DO (mg/L) and DO saturation (%) were determined (Fig. 1). The main stem of the Jiulong River (JR) consists of two major in situ using a WTW portable meter (WTW, Weilheim, Germany). tributaries (North River (NR) and West River (WR)) with a total length At each site, water samples were collected at the surface depth of approximately 258 km (Chen et al., 2008). Due to rapid economic de- (0.5 m). Separate water samples for nutrient analysis (NH4-N, NO2-N, velopment in this area in the past 30 years, various human activities NO3-N, total dissolved nitrogen (TDN), soluble reactive phosphorus such as agricultural practices, livestock breeding and urbanization (SRP), and total dissolved phosphorus (TDP)) were filtered through 748 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756

Fig. 1. Location of sampling stations in two main tributaries (NR and WR) of the JRW. This figure was modified from Fig. 1 of Chen et al. (2013).

0.45-μm pore-size cellulose filters (Millipore, Billerica, MA, USA) and (Lachat Instruments, Loveland, CO, USA). Sediment pH was determined frozen until analysis. Nutrients were determined by a colorimetric at a 1:2.5 sediment to water ratio (w/w). method using an AA3 Auto-Analyzer (Bran + Luebbe Co., Germany). Water samples for analysis of dissolved organic carbon (DOC) were fil- 2.3. DNA extraction, PCR amplification and 16S-pyrotag tered through pre-combusted GF/F filters (Whatman, Clifton, NJ, USA) and then acidified with HCl. The DOC concentrations were measured For molecular analysis, about 500 to 1000 mL water samples were using a Multi N/C 3100 TOC-TN analyzer (Analytik Jena, Germany). pre-filtered through 20 μm mesh (Millipore) and subsequently filtered Sediment samples were collected and transferred into sterile 50 mL onto 0.2-μm pore-size polycarbonate filters (47 mm, Millipore) and Falcon tubes, and stored at −80 °C until further analysis. All sediment stored at −80 °C until further analysis (the sample bottle containing samples were dried in a freeze-drier (Labconco, Kansas City, MO, NR_8 water subsample for molecular analysis was broken during trans- USA), and the organic matter contents (OrgC, OrgN and OrgS) were portation, and therefore, this sample was excluded from further molec- measured by a Vario Max CNS analyzer (Elementar, Hanau, Germany). ular analysis). DNA from water samples was extracted in duplicate Inorganic nitrogen was extracted from dried sediment with 2 M KCl using the modified protocol described by Urakawa et al. (2010).Briefly, and measured using Lachat QC8500 Flow Injection Autoanalyzer filters were cut into small pieces and placed in Lying matrix E tube A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 749

(Qbiogene-MP Biomedicals, Irvine, CA, USA), followed by addition of TFB were identified using 85%, 95% and 97% identity cutoffs, 0.7 mL lysis buffer (100 mM Tris–HCl [pH 8.0], 40 mM EDTA [pH 8.0], respectively.). 200 mM NaCl, 2% SDS). After bead-beating (5.5 m/s, 90s), 0.5 mg/mL of proteinase K was added to each tube and incubated at 55 °C for 4 h. 2.5. Statistical analyses The tubes were then centrifuged at 14,000 g for 5 min. The supernatant was transferred to a clean tube, purified with phenol/chloroform/ We used phylogenetic diversity index (Faith's PD), Shannon diversi- isoamyl alcohol (25:24:1; pH 8.0) and chloroform/isoamyl alcohol ty index (H′), Gini evenness coefficient, and an abundance-based cover- (24:1). DNA was precipitated with 20 μg linear polyacrylamide age estimators Chao1 as measures of α-diversity. A multiple ordinary (Sigma-Aldrich, St Louis, MO, U.S.A.) and 2 volumes of PEG solution least squares (OLS) regression was performed to examine the relation- (30% PEG 6000, 1.6 M NaCl), redissolved in 50 μL TE buffer. For sedi- ship between environmental variables and community α-diversity pat- ment samples, environmental DNA was extracted from ~0.3 g of each tern using SAM v4.0 (Rangel et al., 2010). Since spatial autocorrelation sample in duplicate using the FastDNA SPIN Kit for Soil (Qbiogene-MP may cause Type I errors, an eigenvector-based spatial filtering was in- Biomedicals) according to the manufacturer's instructions. The DNA of corporated into multiple OLS regression analysis (Diniz and Bini, 2005). subsamples were pooled and stored at −80 °C until further molecular The pattern of the bacterial community structures was analyzed by analysis. principal coordinate analysis (PCoA) based on unweighted or weighted The V1–V3 region of bacterial 16S rRNA genes was amplified using UniFrac distance matrices using QIIME v1.4.0. One way analysis of sim- the primer pair: 454 adaptor A-barcode-27 F (5′-AGA GTT TGA TYM ilarity (ANOSIM) was performed to verify the significance of bacterial TGG CTC AG-3′) and 454 adaptor B-534R (5′-ATT ACC GCG GCT GCT community compositions from different environmental categories of GG-3′). Each sample is tagged with a unique 10 bp barcode. The PCR samples. BIOENV was used to select the first three environmental pa- mixture (50 μL) consisted of 25 μL Failsafe Premix E (Epicentre Biotech- rameters that best correlate with bacterial community structures, and nologies, Madison, WI, U.S.A.), 0.4 μM of each primer, 2.5 U of BVSTEP was used to identify a combination of environmental variables AccuPrime™ Taq DNA polymerase High Fidelity (Invitrogen, Carlsbad, that best explain the community pattern (Fortunato and Crump, CA, U.S.A.) and ~20 ng DNA template. PCR were performed under the 2011). Prior to performing BIOENV and BVSTEP analyses, highly following conditions: 94 °C for 5 min, followed by 25 cycles of 94 °C autocorrelated environmental variables (Spearman's correlation: for 30 s, 55 °C for 30 s, and 72 °C for 90 s, and a cycle of 72 °C for r N 0.80, P b 0.05) were removed to ensure more interpretable results. 7 min. Each sample was amplified in triplicate, pooled and purified The OTU data and environmental variables were standardized by using the QIAquick PCR purification kit (Qiagen, Valencia, CA, U.S.A). Hellinger and z-score transformation, respectively. A partial Mantel The concentration of amplicon was measured in duplicate using test was used to determine the effects of local environmental variables Quant-iT dsDNA BR assay kit (Molecular Probes, Sunnyvale, CA, U.S.A) on bacterial community composition by controlling geographical dis- on the Qubit fluorometer (Molecular Probes). Then all amplicons were tance between sampling sites. ANOSIM, BIOENV and BVSTEP were per- mixed in equimolar concentration and sent to the Genomics In- formed with PRIMER software (Clarke and Ainsworth, 1993), and the stitute (, China) for pyrosequencing on a Roche 454 GS-FLX Ti- Mantel and partial Mantel tests were performed using R v2.14 with tanium platform. The raw sequence data were deposited into the NCBI the packages vegan (Oksanen et al., 2009). The correlation and differ- short reads archive database (accession number: SRX193676). ence between two independent groups were assessed with Spearman's correlation coefficient and Mann–Whitney test respectively, using SPSS v13.0 (SPSS, Inc., Chicago, IL). 2.4. Sequence analyses 3. Results Pyrosequencing data were processed using QIIME v1.4.0 (Caporaso et al., 2010) by following the procedure described in our previous 3.1. Environmental characteristics study (Yusoff et al., 2013). In brief, raw sequences were processed to re- move poor-quality reads using the following thresholds: i) minimum The main physicochemical characteristics of JR surface water are read length of 150 bp; ii) minimum average quality score of 25; iii) no shown in Fig. S1 and in Table S1 in the Supplemental Materials. Gener- ambiguous bases (N) and maximum homopolymers of 6 bp in the en- ally, the physical properties (temperature, pH, EC, DO and DO%) of sur- tire sequence; and iv) sequences containing correct barcode or primer face water were comparable between two tributaries except that water sequences. The pyrosequencing errors were further corrected using temperature was significantly higher in WR than in NR (Mann–Whitney Acacia v1.52 (Bragg et al., 2012). Operational taxonomic units (OTUs) test, P b 0.01) (Table S1). For chemical properties, there was a pro- were identified using 97% identity cutoff with USEARCH (Edgar, nounced decreasing trend in dissolved inorganic nitrogen (DIN) from 2010), and α- (species diversity within a single sample) and β diversity upstream to downstream in both tributaries although the DIN increased

(variation in species composition among samples) were further ana- slightly again in downstream of WR (Fig. S1). However, NH4-N was the lyzed using corresponding Python scripts in QIIME v1.4.0. Finally, the main form of DIN in the first two upstream sites (Longyan district) of high-quality sequences were assigned to taxonomic rank using RDP NR, whereas more than 90% of DIN in the upstream regions of WR classi fier with a 50% bootstrap confidence threshold. Indicator taxa was in the form of NO3-N (Fig. S1). NO2-N was detected in all water were identified within each environmental category using LEfSe soft- samples but its concentrations were generally lower in WR than in NR ware (Segata et al., 2011). (Mann–Whitney test, P b 0.01). SRP and DOC concentrations ranged Newton et al. (2011) have collected N11,500 16S rRNA gene se- from 0.04 to 0.38 mg/L and 1.04 to 2.53 mg/L respectively, and their quences of lake bacterioplankton from previous freshwater surveys, variations in these two tributaries were closely related to NH4-N and set up a TFB taxonomy (phylum/lineage/clade/tribe) based on phy- (Spearman's correlation: r N 0.602, P b 0.001) (Fig. S1). The DIN/SRP ra- logenetic analysis as well as sequence identity. In order to identify na- tios were always higher than 50 in JR, but a contrasting pattern of DIN/ tive freshwater bacterial populations in the JRW, the pyrosequencing SRP ratio distribution along the hydrological flow path was observed in sequences recovered from this study were also assigned into TFB taxon- NR and WR (Fig. S1). Dissolved organic nitrogen (DON = TDN−DIN) omy framework using CREST v1.0 software (Lanzen et al., 2012)with and dissolved organic phosphorus (DOP = TDP−SRP) concentrations default parameters (minimum bit-score = 155, LCA range = 2%) in were calculated, and the results showed that DON and DOP only combination with a modified 16S rRNA gene database of TFB (Newton accounted for, on average, 7% and 10.5% of TDN and TDP, respectively. et al., 2011). The minimum similarity filter was included to ensure the The physicochemical characteristics of sediment in the JRW are also accuracy of taxonomic assignments (the lineages, clades and tribes of shown in Table S1. Sediment pH varied from 5.87 and 7.8, and was 750 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 closely related to water EC (Spearman's correlation, r = 0.613, water communities respectively, whereas α-diversity of benthic com- P b 0.01). OrgC and OrgN in JRW sediment samples ranged from 0.14% munities was best explained by sediment OrgN and water pH (Table 1). to 2.51% and 0.02% to 0.2%, respectively. Sediment OrgS and OrgC/ OrgN were significantly higher in NR than in WR (Mann–Whitney 3.3. Distribution pattern of bacterial taxa in the JRW test, P b 0.05). Sediment NH4-N ranged from 4.6 to 33.93 mg/kg, and was the main form of total inorganic nitrogen. Based on the RDP Naïve Bayesian Classifier, a total of 35 and 33 phyla were determined in the JRW planktonic and benthic bacterial commu- nities respectively, with significantly higher unclassified sequences re- 3.2. α-Diversity covered from sediment samples (Fig. 3). Among all of the phyla, Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes, Acidobacteria, In order to compare the diversity, composition and structure of Cyanobacteria, Chloroflexi, Verrucomicrobia, Planctomycetes and planktonic and benthic bacterial communities between two main tribu- Gemmatimonadetes were the 10 most abundant phyla, and comprised taries (NR and WR) of the JRW, 16S-pyrotag data were produced from 88.7% of the total sequences (Fig. 3). all above samples. Overall, a total of 204,216 high-quality reads were As expected, most of bacterial taxa did not evenly distribute between obtained after the quality filtering, ranging from 3785 to 10,703 reads river water and sediments, although the relative abundances of Alpha-, of each sample with an average of 5835 ± 1797 reads (Table S2). Be- Epsilonproteobacteria, Bacteroidetes, Cyanobacteria, Verrucomicrobia, cause the uneven sampling efforts may have an effect on diversity anal- Planctomycetes and OP10 were similar between both planktonic and ysis, 3,500 reads were randomly selected from each sample for benthic communities (Fig. 3). Betaproteobacteria, Actinobacteria and subsequent α-andβ-diversity analyses. Generally, a total of 36,785 Firmicutes were significantly more abundant in water than in sediment OTUs were identified at the 97% identity at the equal sampling depth, samples, whereas the relative abundances of Acidobacteria, Delta-, ranged from 1086 to 2613. The majority (67%, 24,616 OTUs) of these Gammaproteobacteria, Chloroflexi and Nitrospira were obviously higher OTUs were singletons, and accounted for 20.1% of the total subsampling in benthic than in planktonic communities (Mann–Whitney test, sequences. The sediment communities harbored an average of P b 0.01) (Fig. 3). Moreover, the total relative abundance of minor 2346 ± 217 OTUs per sample, which was significantly higher than phyla including TM7, WS3, Deinococcus–Thermus and Spirochaetes, those in the water samples (1460 ± 218 OTUs per sample) (Mann– was also higher in benthic communities than in planktonic communi- Whitney test, P b 0.001) (Table S2). Likewise, phylogenetic diversity ties (Mann–Whitney test, P b 0.001) (Fig. 3). The indicator taxa specific Faith's PD, Shannon diversity (H′) and Chao1 richness indices were sig- to water and sediment communities from two tributaries (four catego- nificantly higher in sediment communities than those of water commu- ries: NR_Water, NR_Sediment, WR_Water and WR_Sediment) were nities (Fig. 2). However, the planktonic bacterial communities were less identified using LEfSe tool (Segata et al., 2011), and the detailed infor- uniform than the benthic bacterial communities as indicated by Gini mation were shown in the Supplemental Materials (Fig. S2). evenness index (Mann–Whitney test, P b 0.05) (Fig. 2). In other words, the JWR sediments contained a greater bacterial diversity and higher evenness. In addition, a significant difference was found in the 3.4. Do typical freshwater bacterial taxa widely occur in the JRW? α-diversity of NR and WR planktonic communities, but it seemed that benthic communities possessed comparable α-diversity between To explore the distribution of TFB in the JRW, we used the CREST these two tributaries of the JRW (Fig. 2). software (Lanzen et al., 2012), which applied the lowest common There was no apparent pattern of bacterial α-diversity along the hy- ancestor (LCA) to assign our 16S-pyrotag data to TFB taxonomic drological flow path in NR or WR (Table S2). The multiple OLS regres- framework based on a TFB 16S rRNA gene database (Newton et al., sion analysis demonstrated that environmental variables could 2011). The results indicated that, on average, 57.6 ± 10%, provide relatively high explanatory power for JRW bacterial richness 27.6 ± 10.9% and 10.4 ± 6.9% of planktonic communities could be and diversity even though spatial autocorrelation was considered assigned to lineages, clades and tribes respectively, which were sig- (r2 N 0.56) except for Faith's PD and Gini indices of planktonic commu- nificantly higher than those of benthic communities (about 37.1%, nities (Table 1). For water community, NO2-N and NH4-N in the water 5.7% and 2.2% of benthic communities were assigned to lineages, were the most important factors correlated with Chao1 and H′ of clades and tribes, respectively) (Mann–Whitney test, P b 0.001)

Fig. 2. Comparison of α-diversity of planktonic and benthic bacterial communities from two main tributaries (NR and WR) of the JRW. Significant differences were tested at P b 0.05, letters indicate significant differences between different environmental groups. Bars denote standard deviation of the mean. A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 751

Table 1 Results of multiple OLS regression using Akaike's information criterion, correlating α-diversity of bacterial communities with environmental factors.

Groups Diversity index r2 P n Explanatory variables (beta weights)a,b ⁎⁎⁎ ⁎ Water community Chao1 0.565 0.007 17 NO2-N (−1.08) ,PO4-P (0.702) ⁎ Faith's PD 0.299 0.023 17 EC (−0.547) ⁎⁎ ⁎⁎ H′ 0.843 b0.00117NH4-N (−0.475) , Temp (0.347) , ⁎ NO2-N (−0.353) ⁎⁎⁎ Gini 0.251 b0.00117NO2-N (0.608) ⁎⁎ ⁎ Water TFB community Chao1 0.497 0.015 17 NO2-N (−1.014) ,PO4-P (0.668) ⁎⁎⁎ ⁎ Faith's PD 0.551 0.005 17 EC (−1.097) ,PO4-P (0.655) ⁎⁎ ⁎ H′ 0.642 0.003 17 NH4-N (−1.398) ,PO4-P (0.800) ⁎⁎⁎ ⁎ Gini 0.582 0.008 17 NO2-N (1.082) ,PO4-P (−0.632) ⁎⁎⁎ ⁎⁎⁎ Sediment community Chao1 0.792 b0.001 18 OrgN (0.67) ,pH(−0.619) , ⁎⁎ PO4-P (−0.549) ⁎⁎⁎ ⁎⁎⁎ Faith's PD 0.948 b0.001 18 OrgN (0.62) ,pH(−0.689) , ⁎⁎⁎ ⁎⁎⁎ PO4-P (−1.16) , DOC (0.693) , ⁎⁎⁎ DON (0.54) ,NO2-N (0.443) ⁎⁎⁎ H′ 0.817 b0.00118pH(−0.904) ⁎⁎⁎ ⁎ Gini 0.808 b0.00118pH(0.781),spH(0.255)

a The spatial autocorrelation in the model residuals was considered. b spH, sediment pH. ⁎⁎⁎ P b 0.001. ⁎⁎ P b 0.01. ⁎ P b 0.05.

(Fig. 4). However, a lower proportion of total sequences from the community were significantly positively correlated with those of the first two upstream sites of NR were assigned into TFB at all taxonom- total bacterioplankton community (Spearman's correlation, r N 0.716, ic resolution levels (Fig. 4a). P b 0.01) (Table S3). The multiple OLS regression analysis indicated

To assess the diversity pattern of JRW indigenous bacterial commu- that water NH4-N and NO2-N were the most important factors associat- nity, 1200 TFB sequences were randomly selected from each sample to ed with diversity (H′) and richness (Chao1) of water TFB communities, calculate α-diversity indices. In general, the α-diversity indices of TFB respectively (Table 1).

Fig. 3. The relative abundance of dominant bacterial groups in water and sediment samples obtained from NR and WR. Minor phyla accounting for b0.5% of total sequences are summa- rized in the group ‘other’. Bars denote standard error of the mean. Asterisks in the superscripts on the taxonomic name indicate that there were statistically significant differences in spe- cific bacterial abundances between water and sediment communities (***P b 0.001, **P b 0.01, *P b 0.05). 752 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756

Fig. 4. The relative abundance of TFB (lineages, clades and tribes) within JRW planktonic (a, b) and benthic communities (c, d).

3.5. Comparison of bacterial community structures among subwatersheds

In general, ANOSIM analysis supported the result of taxonomic classification that the planktonic and benthic communities in the JRW were significantly different (R N 0.87, P b 0.01). Furthermore, a large spatial difference were found between NR and WR bacterioplankton communities (R = 0.8, P b 0.001) but not for the benthic communities from two tributaries (R = 0.14, P N 0.1). After removing the singletons (in order to highlight the shared OTUs as described by Stearns et al. (2011)), Venn analysis demonstrated that, on average, 21.6 ± 6.9%, 17.3 ± 4.2%, 12.3 ± 3.3%, and 12.9 ± 2.9% of non-singleton OTUs of each group were shared between any two samples from NR water, WR water, NR sediment and WR sediment communities, respectively (Fig. S3). In the surface water of two tributaries, the nearer neighbor sites harbored more similar bacterioplankton communities, but the sed- iment communities did not show a similar pattern (Fig. S3). This result was confirmed by linear regression analyses which showed a significant positive correlation between phylogenetic distance of water communi- ties and channel distance within each tributary (R N 0.5, P b 0.01), but not for the JRW benthic communities (P N 0.1) (Fig. S4). Consistent with the results of ANOSIM analysis, both unweighted and weighted PCoA analyses clearly distinguished the JRW water com- munities from communities in JRW sediments, and the whole water communities were further separated into two main groups correspond- Fig. 5. PCoA analysis of JRW bacterial community structures using the unweighted (circle) ing to their origin by PC2 axis (Fig. 5). These results were consistent re- and weighted (triangle) UniFrac distances. Procrustes analysis indicated higher concor- – 2 gardless of which ordination method is used (e.g. Bray Curtis or Jaccard dance between unweighted and weighted UniFrac distance matrices (M =0.215, similarity matrices) (Mantel test: r N 0.84, P b 0.001). P b 0.001, 1000 Monte Carlo iterations). A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 753

3.6. Relationship between bacterial community structure and environmen- long history of human disturbances using 16S-pyrotag, which can pro- tal variables duce tens to hundreds times of 16S rRNA gene sequences than tradi- tional molecular methods (Binladen et al., 2007; Sogin et al., 2006; The relationship between bacterial community structure and envi- Yusoff et al., 2013). Our study indicated that JRW sediment bacterial ronmental variables was determined using BVSTEP and BIOENV analy- communities had higher richness and evenness than those of surface ses. At the whole watershed scale, in general, BVSTEP and BIOENV bacterioplankton communities (Table S2), which is expected since sed- analyses indicated that nutrient parameters such as SRP and DIN/SRP iments are generally more heterogeneous than the water bodies were the factors strongly correlated with surface bacterioplankton com- (Lozupone and Knight, 2007). Multiple OLS regression analysis demon- munity structures (r = 0.643, n = 17), while the sediment bacterial strated that water NH4-N and NO2-N were of pivotal importance in de- community structures were significantly correlated with several water termining the biodiversity pattern of the whole bacterioplankton and and sediment environmental variables, including DIN/SRP, DON, NO3- water TFB communities, while sediment OrgN and water pH played a N, sediment pH and OrgC, and less variability could be explained certain role in controlling the benthic community diversity (Table 1), (r=0.388,n=18)(Table 2). Within each tributary group, local envi- supporting previous observations that resource availability is a major ronmental variables provided more explanatory power as shown by the driver of bacterial diversity either for planktonic (Logue et al., 2011) Spearman's correlation coefficient except for NR sediment community or benthic communities (Bienhold et al., 2012). However, in contrast (Table 2). SRP and DIN/SRP were once again identified as the most im- to the negative relationship between bacterial diversity (H′) and rich- portant factors in determining community variability in NR water ness (Chao1) and water NH4-N and NO2-N observed in the present (r = 0.932, n = 8) and WR water (r = 0.660, n = 9) groups, respec- study (Table 1), Logue et al. (2011) found that the supply of available tively. In addition, water temperature, a physical environment factor, nutrients (TOC, TN and TP) positively influence bacterioplankton rich- emerged as the second key variable correlated with bacterial communi- ness in 14 oligotrophic lakes from Sweden. Nevertheless, our findings ty structures for both NR water (r = 0.750, n = 8) and WR water are in accordance with the results of Van Horn et al. (2011), who report- (r = 0.627, n = 9) groups in the BIOENV analysis. For the sediment ed that bacterial community diversity in stream biofilms decreases in groups, a combination of several water (pH, SRP, DOC, DIN/SRP, DON nutrient enrichment. This is reasonable since not only total resource and DOP) and one sediment (OrgN) environmental variables explained available but also the balance of resources play critical roles in main- a large fraction of the variability in structure of WR sediment communi- taining community's diversity (Cardinale et al., 2009; Kassen et al., ty (r = 0.728, n = 9), whereas sediment OrgC was responsible for NR 2000; Van Horn et al., 2011). In the JRW, the intensive agricultural activ- sediment community assembly (r = 0.285, n = 9). ities (livestock breeding and cash crops production) resulted in exces- Because the distance-decay relationship was observed in NR and WR sive inorganic P loading, which leaded to resource imbalance in the water communities, a partial Mantel test was used to distinguish be- water body of Jiulong River (Chen et al., 2013). tween the effects of local environmental variables (species-sorting) RDP classifications indicated that Beta- (22.6%), Alphaproteobacteria and geographical distance (dispersal) on the bacterial communities. (17.5%), Actinobacteria (18.6%), Bacteroidetes (8%) and Cyanobacteria The results indicated that significant correlation between correspond- (3.2%) dominated the surface bacterioplankton communities of the ing environmental variables identified by BIOENV and BVSTEP analyses JRW (Fig. 3). This result is consistent with previous studies that investi- and planktonic bacterial community structures still existed when con- gated planktonic bacterial community in lotic environments (Crump trolling by channel distances (r N 0.716, P b 0.01). et al., 2012; Crump et al., 2009; Ghai et al., 2011; Sekiguchi et al., 2002; Winter et al., 2007), but it is quite unanticipated that the Firmicutes 4. Discussion was the fourth most abundant phylum in planktonic communities and accounted for about 14.1% of the total bacterioplankton sequences In the current study we investigated both planktonic and benthic (Fig. 3). Similar results were also observed by a prior field survey of mi- bacterial communities in a mesoscale subtropical watershed with a crobial community in surface JRW, in which 4 out of 19 (~21%) 16S rRNA

Table 2 BVSTEP and BIOENV analyses showing the correlations between bacterial community structures and environmental variables.

Groups BVSTEP BIOENVa

Corr.b Env. factors Corr. Env. factors ⁎⁎⁎ All water community 0.643 SRP, DIN/SRP 0.542 DIN/SRP ⁎⁎ 0.411 SRP ⁎⁎ 0.351 Temp ⁎⁎ NR water community 0.932 SRP 0.932 SRP ⁎ 0.865 Temp, NO3-N, SRP, DOP 0.750 Temp ⁎ 0.664 NO2-N ⁎ WR water community 0.755 Temp, pH, EC, DIN/SRP 0.660 DIN/SRP ⁎ 0.627 Temp

0.482 NO2-N All sediment community 0.388 DIN/SRP, DON, spH, OrgC 0.216 OrgC

0.362 NO3-N, DIN/SRP, spH, OrgC 0.176 NO3-N

0.168 NO2-N NR sediment community 0.285 OrgC 0.285 OrgC

0.175 NO3-N, DIN/SRP, OrgC, OrgC/OrgN 0.139 EC 0.088 OrgC/OrgN ⁎ WR sediment community 0.728 pH, SRP, DOC, DIN/SRP, DON, DOP, OrgN 0.425 DOC 0.721 Temp, pH, EC, DOC, DIN/SRP, DON, DOP, spH 0.325 SRP 0.322 Temp

a For BIOENV, Spearman rank coefficients for the top three environmental variables are shown. b Corr.: Correlation; Env. factors: Environmental factors. ⁎⁎⁎ P b 0.001. ⁎⁎ P b 0.01. ⁎ P b 0.05. 754 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756

Fig. 6. The relative abundance of sewer- and fecal-associated bacterial indicators identified by Newton et al. (2013) within JRW planktonic communities.

gene sequences from DGGE bands were closely related to phylum because benthic bacterial 16S rRNA gene sequences were not considered Firmicutes (Liu et al., 2011). A high relative abundance of Firmicutes ob- in this TFB database (Newton et al., 2011), supporting the results of pre- served here is surprising because this phylum is one of the most domi- vious works that sediment harbor different bacteria groups or lineages. nant members of gastrointestinal ecosystem of both animals and Nonetheless, the classification percentages at each level of TFB taxonom- humans (Ley et al., 2008; Stearns et al., 2011), but generally has a patchy ic framework in the present study were lower than the results of a previ- distribution and low abundance (b1%) in surface freshwater (Newton ous study investigating the bacterioplankton community of Lake Erken, et al., 2011). Moreover, LEfSe analysis indicated that a family of the phy- Sweden (Eiler et al., 2012). This may be attributed to two reasons. First, lum Firmicutes, Clostridiaceae, which frequently occur in the mammalian the partial-length sequences (average length of 360 bp in this study) intestinal tracts (Newton et al., 2013), and likely prefer to live in the an- generated by 454 pyrosequencing might not provide enough informa- aerobic niches (Wiegel et al., 2006), is an indicator taxon for WR water tion for accurate taxonomic classification especially for high resolution communities (Figs. S2 and 6), raising the possibility that the water bodies taxonomic level (e.g. tribe) (Fig. 4)(Lanzen et al., 2012). Second, the of the JRW are contaminated by fecal pollution, especially for WR. TFB database only included bacterial 16S rRNA gene sequences recov- In a more recent study, Newton et al. (2013) performed a meta- ered from freshwater lakes, and there is relatively limited information analysis of two categories of bacterial 16S-pyrotag data which were de- available on bacterioplankton community composition in flowing waters rived from 40 sewage influent and 48 human fecal samples, respective- (Logue and Lindström, 2008; Newton et al., 2011). Therefore, it is possi- ly. The authors found that three bacterial genera, Acinetobacter ble that some unknown species may occur in lotic habitats (Lemke et al., (Gammoproteobacteria), Arcobacter (Epsilonproteobacteria)and 2009) or some unique autochthonous bacterioplankton populations are Trichococcus (Firmicutes) were prevalent in sewer samples, whereas present in JR water. two Bacteroidetes families, Bacteroidaceae and Porphyromonadaceae, It has been suggested that most microorganisms had a limited distri- and three Firmicutes families, Clostridiaceae, Lachnospiraceae and bution among different habitat types (freshwater, saline water, soils and Ruminococcaceae were consistently more abundant in human fecal sediments, etc.) due to the influence of environmental preferences, as samples, and thus could serve as indicators of sewer- and fecal pollu- observed for macroorganisms (Ley et al., 2008; Lozupone and Knight, tion, respectively (Newton et al., 2013). Further, these bacterial taxa 2007; Newton et al., 2011). Consistent with those observations, distinct were successfully used to track sewer- and fecal contamination in bacterial community compositions were found in the JRW water and Lake Michigan (Newton et al., 2013). Our results indicated that the sediments (Figs. 4 and 5). The PCoA and ANOSIM analysis further con- magnitude of these fecal signatures was generally higher in surface firmed this difference, indicating that JRW planktonic and benthic com- water of the JRW (N2%) than those from Lake Michigan (Newton munity structures were significantly different from each other et al., 2013) except for NR downstream, while the sewer signatures (ANOSIM, P b 0.001). Moreover, NR and WR planktonic bacterial com- had significantly lower abundance (b1%) than fecal signatures munities formed two distinct groups, whereas there was no significant (Mann–Whitney test, P b 0.001) (Fig. 6a), suggesting again that JRW difference between NR and WR benthic community structures water was likely contaminated with human and animal wastes which (ANOSIM, P N 0.1) (Fig. 5). Mounting evidence suggests that contempo- could be originated from non-point sources (Chen et al., 2013). Howev- rary environmental factors (light, temperate, pH, nutrients, and phyto- er, it is dif ficult to exclude the possibility that some members of these plankton composition and biomass) as well as the features of the bacterial taxa might be truly native to the water of the JRW since so surrounding environments (such as land use) govern the assembly of far many species or subspecies within these taxa do not have pure cul- river bacterial communities (Crump et al., 2012; Fierer et al., 2007; tures, and further multi-method surveys are needed to shed light on the Portillo et al., 2012; Wang et al., 2011). Similarly, our results strongly in- occurrence and sources of fecal pollution in the JRW. dicated that several water physical–chemical variables including SRP, Using a taxonomic framework of TFB, it is reasonable that significant- DIN/SRP and water temperature were the main determinants of JRW ly more sequences from JRW planktonic communities were assigned planktonic bacterial community structures, whereas factors affecting into lineages, clades and tribes than those obtained from sediments JRW sediment community structures remained unresolved except for A. Hu et al. / Science of the Total Environment 472 (2014) 746–756 755

WR sediment community (Table 2). Also, it is worth noting that other Cardinale BJ, Hillebrand H, Harpole W, Gross K, Ptacnik R. Separating the influence of re- source ‘availability’from resource ‘imbalance’on productivity–diversity relationships. environmental factors, such as the quality of dissolved organic matter Ecol Lett 2009;12:475–87. (DOM) which was different between NR and WR (Yang et al., 2012), Chen NW, Hong HS. Integrated management of nutrients from the watershed to coast in might play a certain role in influencing JRW bacterial community struc- the subtropical region. Curr Opin Environ Sustain 2012;4:233–42. Chen NW, Hong HS, Zhang LP, Cao WZ. Nitrogen sources and exports in an agricultural tures (Jones et al., 2009; Kirchman et al., 2004), although surface water watershed in Southeast China. Biogeochemistry 2008;87:169–79. DOC concentrations were comparable in both tributaries (Fig. S1). Fur- Chen N, Peng B, Hong H, Turyaheebwa N, Cui S, Mo X. Nutrient enrichment and N: P ratio ther studies are needed to examine the effect of DOM quality, chemical decline in a coastal bay-river system in southeast China: the need for a dual nutrient – contaminants and other factors which were not considered here on the (N and P) management strategy. Ocean Coast Manag 2013;81:7 13. Clarke K, Ainsworth M. A method of linking multivariate community structure to environ- diversity, composition and structures of JRW microbial communities. mental variables. Mar Ecol Prog Ser 1993;92:205. In summary, our study provides a comprehensive view of the phylo- Crump BC, Peterson BJ, Raymond PA, Amon RMW, Rinehart A, McClelland JW, et al. Cir- genetic makeup and structure of both planktonic and benthic bacterial cumpolar synchrony in big river bacterioplankton. Proc Natl Acad Sci U S A 2009;106:21208–12. communities in the JRW which has been intensively affected by human Crump BC, Amaral-Zettler LA, Kling GW. Microbial diversity in arctic freshwaters is struc- activities. Our results demonstrated that water and sediment of the tured by inoculation of microbes from soils. ISME J 2012;6:1629–39. JRW harbored different bacterial lineages, and water chemistry (e.g., nu- Diniz JAF, Bini LM. Modelling geographical patterns in species richness using eigenvector-based spatial filters. Glob Ecol Biogeogr 2005;14:177–85. trient concentrations and stoichiometry) had a great impact on JRW bac- Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics terial communities especially for the planktonic communities. The 2010;26:2460–1. results of our study therefore expand our knowledge about the diversity Eiler A, Heinrich F, Bertilsson S. Coherent dynamics and association networks among lake bacterioplankton taxa. ISME J 2012;6:330–. 42 and structure of lotic bacterial communities in a watershed of intensive Fierer N, Morse JL, Berthrong ST, Bernhardt ES, Jackson RB. Environmental controls on the agricultural use, and their responses to environmental changes. landscape-scale biogeography of stream bacterial communities. Ecology 2007;88: 2162–73. Fortunato CS, Crump BC. Bacterioplankton community variation across river to ocean en- Conflict of interest vironmental gradients. Microb Ecol 2011;62:374–82. Ghai R, Rodriguez-Valera F, McMahon KD, Toyama D, Rinke R, de Oliveira TCS, et al. Metagenomics of the water column in the pristine upper course of the Amazon All authors including Anyi Hu, Xiaoyong Yang, Nengwang Chen, River. PLoS ONE 2011;6:e23785. Liyuan Hou, Ying Ma, Chang-Ping Yu, declare that there are no conflicts Hilton J, O'Hare M, Bowes MJ, Jones JI. How green is my river? A new paradigm of eutro- – of interest to this work. phication in rivers. Sci Total Environ 2006;365:66 83. Hu A, Yao T, Jiao N, Liu Y, Yang Z, Liu X. Community structures of ammonia-oxidising ar- chaea and bacteria in high-altitude lakes on the Tibetan Plateau. Freshw Biol 2010;55:2375–90. Acknowledgments Huang J, Li Q, Pontius RG, Klemas V, Hong H. Detecting the dynamic linkage between landscape characteristics and water quality in a subtropical coastal watershed, south- – We thank Jiajun Hong and Jiangwei Li, and Jiezhong Wu for their as- east China. Environ Manag 2011:1 13. Jones SE, Newton RJ, McMahon KD. Evidence for structuring of bacterial community compo- sistance in collecting the samples and lab analysis, Dr. Anders Lanzén for sition by organic carbon source in temperate lakes. Environ Microbiol 2009;11: the assistance in using the CREST software, and Dr. Ryan J. Newton for 2463–72. providing the 16S rRNA gene database of freshwater bacteria. We also Kassen R, Buckling A, Bell G, Rainey PB. Diversity peaks at intermediate productivity in a laboratory microcosm. Nature 2000;406:508–12. thank three anonymous referees for their constructive comments Kirchman DL, Dittel AI, Findlay SEG, Fischer D. Changes in bacterial activity and commu- which improved this manuscript. nity structure in response to dissolved organic matter in the Hudson River, New York. This work was supported by the NSFC project (41106096), the Aquat Microb Ecol 2004;35:243–57. Lanzen A, Jorgensen SL, Huson DH, Gorfer M, Grindhaug SH, Jonassen I, et al. CREST Knowledge Innovation Program of the Chinese Academy of Sciences — Classification resources for environmental sequence tags. PLoS ONE 2012;7: (IUEQN201307), Technology Foundation for Selected Overseas Chinese e49334. Scholar of MOHRSS, Science and Technology Planning Project of Xiamen, Lear G, Lewis GD. Impact of catchment land use on bacterial communities within stream fi – China (3502Z20102017), the CAS/SAFEA International Partnership bio lms. Ecol Indic 2009;9:848 55. Lear G, Dopheide A, Ancion P, Lewis GD. A comparison of bacterial, ciliate and macroin- Program for Creative Research Teams (KZCX2-YW-T08), and the vertebrate indicators of stream ecological health. Aquat Ecol 2011;45:517–27. Open Fund of Key Laboratory of Urban Environment and Health, Institute Lemke MJ, Lienau EK, Rothe J, Pagioro TA, Rosenfeld J, DeSalle R. Description of freshwater of Urban Environment, CAS (KLUEH201105). bacterial assemblages from the Upper Parana River Floodpulse System, Brazil. Microb Ecol 2009;57:94–103. Ley RE, Lozupone CA, Hamady M, Knight R, Gordon JI. Worlds within worlds: evolution of the vertebrate gut microbiota. Nat Rev Microbiol 2008;6:776–88. Appendix A. Supplementary data Li Y, Cao WZ, Su CX, Hong HS. Nutrient sources and composition of recent algal blooms and eutrophication in the northern Jiulong River, Southeast China. Mar Pollut Bull – Supplementary data to this article can be found online at http://dx. 2011;63:249 54. Liu LM, Yang J, Zhang YY. Genetic diversity patterns of microbial communities in a subtrop- doi.org/10.1016/j.scitotenv.2013.11.097. ical riverine ecosystem (Jiulong River, southeast China). Hydrobiologia 2011;678: 113–25. Logue JB, Lindström ES. Biogeography of bacterioplankton in inland waters. Freshw Rev References 2008;1:99–114. Logue JB, Langenheder S, Andersson AF, Bertilsson S, Drakare S, Lanzén A, et al. Freshwa- Ager D, Evans S, Li H, Lilley AK, van der Gast CJ. Anthropogenic disturbance affects the ter bacterioplankton richness in oligotrophic lakes depends on nutrient availability structure of bacterial communities. Environ Microbiol 2010;12:670–8. rather than on species–area relationships. ISME J 2011;6:1127–36. Aufdenkampe AK, Mayorga E, Raymond PA, Melack JM, Doney SC, Alin SR, et al. Riverine Lozupone C, Knight R. Global patterns in bacterial diversity. Proc Natl Acad Sci U S A coupling of biogeochemical cycles between land, oceans, and atmosphere. Front Ecol 2007;104:11436–40. Environ 2011;9:53–60. Maskaoui K, Zhou JL, Hong HS, Zhang ZL. Contamination by polycyclic aromatic hydrocar- Bienhold C, Boetius A, Ramette A. The energy–diversity relationship of complex bacterial bons in the Jiulong River Estuary and Western Xiamen Sea, China. Environ Pollut communities in Arctic deep-sea sediments. ISME J 2012;6:724–32. 2002;118:109–22. Binladen J, Gilbert MTP, Bollback JP, Panitz F, Bendixen C, Nielsen R, et al. The use of coded Newton RJ, Jones SE, Eiler A, McMahon KD, Bertilsson S. A guide to the natural history of PCR primers enables high-throughput sequencing of multiple homolog amplification freshwater lake bacteria. Microbiol Mol Biol Rev 2011;75:14–49. products by 454 parallel sequencing. PLoS ONE 2007;2:e197. Newton RJ, Bootsma MJ, Morrison HG, Sogin ML, McLellan SL. A microbial signature ap- Bragg L, Stone G, Imelfort M, Hugenholtz P, Tyson GW. Fast, accurate error-correction of proach to identify fecal pollution in the waters off an urbanized coast of Lake Michigan. amplicon pyrosequences using Acacia. Nat Methods 2012;9:425–6. Microb Ecol 2013;65:1011–23. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME Oksanen J, Kindt R, Legendre P, O'Hara B, Simpson G, Solymos P, et al. vegan: community allows analysis of high-throughput community sequencing data. Nat Methods ecology package. R package version 1.15-2; 2009. 2010;7:335–6. Portillo MC, Anderson SP, Fierer N. Temporal variability in the diversity and composition Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Lozupone CA, Turnbaugh PJ, et al. of stream bacterioplankton communities. Environ Microbiol 2012;14:2417–28. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Rangel TF, Diniz JAF, Bini LM. SAM: a comprehensive application for Spatial Analysis in Proc Natl Acad Sci U S A 2011;108:4516. Macroecology. Ecography 2010;33:46–50. 756 A. Hu et al. / Science of the Total Environment 472 (2014) 746–756

Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett WS, et al. Metagenomic bio- Wiegel J, Tanner R, Rainey FA. An introduction to the family Clostridiaceae. Proc Natl Acad marker discovery and explanation. Genome Biol 2011;12:R60. Sci U S A 2006;4:654–78. Sekiguchi H, Watanabe M, Nakahara T, Xu BH, Uchiyama H. Succession of bacterial Williamson CE, Dodds W, Kratz TK, Palmer MA. Lakes and streams as sentinels of environ- community structure along the Changjiang River determined by denaturing gra- mental change in terrestrial and atmospheric processes. Front Ecol Environ 2008;6: dient gel electrophoresis and clone library analysis. Appl Environ Microbiol 247–54. 2002;68:5142–50. Winter C, Hein T, Kavka G, Mach RL, Farnleitner AH. Longitudinal changes in the bacterial Shiklomanov IA. World water resources: a new appraisal and assessment for the 21st community composition of the Danube River: a whole-river approach. Appl Environ century. Paris: UNESCO Press; 1998. Microbiol 2007;73:421–31. Sogin ML, Morrison HG, Huber JA, Welch DM, Huse SM, Neal PR, et al. Microbial diversity Yan XL, Zhai WD, Hong HS, Li Y, Guo WD, Huang X. Distribution, fluxes and decadal in the deep sea and the underexplored “rare biosphere”. Proc Natl Acad Sci U S A changes of nutrients in the Jiulong River Estuary, Southwest . Chin Sci 2006;103:12115–20. Bull 2012;57:2307–18. Stearns JC, Lynch MDJ, Senadheera DB, Tenenbaum HC, Goldberg MB, Cvitkovitch DG, Yang LY, Hong HS, Guo WD, Huang JL, Li QS, Yu XX. Effects of changing land use on dis- et al. Bacterial biogeography of the human digestive tract. Sci Rep 2011;1:170. solved organic matter in a subtropical river watershed, southeast China. Reg Environ Urakawa H, Martens-Habbena W, Stahl DA. High abundance of ammonia-oxidizing ar- Chang 2012;12:145–51. chaea in coastal waters, determined using a modified DNA extraction method. Appl Yannarell AC, Triplett EW. Geographic and environmental sources of variation in lake bac- Environ Microbiol 2010;76:2129–35. terial community composition. Appl Environ Microbiol 2005;71:227–39. Van Horn DJ, Sinsabaugh RL, Takacs-Vesbach CD, Mitchell KR, Dahm CN. Response of het- Yergeau E, Sanschagrin S, Waiser MJ, Lawrence JR, Greer CW. Sub-inhibitory concentrations erotrophic stream biofilm communities to a gradient of resources. Aquat Microb Ecol of different pharmaceutical products affect the meta-transcriptome of river biofilm 2011;64:149–61. communities cultivated in rotating annular reactors. Environ Microbiol Rep 2012;4: Villéger S, Blanchet S, Beauchard O, Oberdorff T, Brosse S. Homogenization patterns of the 350–9. world's freshwater fish faunas. Proc Natl Acad Sci U S A 2011;108:18003–8. Yusoff MZM, Hu A, Feng C, Maeda T, Shirai Y, Hassan MA, et al. Influence of pretreated ac- Vishnivetskaya TA, Mosher JJ, Palumbo AV, Yang ZK, Podar M, Brown SD, et al. Mercury tivated sludge for electricity generation in microbial fuel cell application. Bioresour and other heavy metals influence bacterial community structure in contaminated Technol 2013;145:90–6. Tennessee streams. Appl Environ Microbiol 2011;77:302–11. Zhang X, Zhang DD, Zhang H, Luo ZX, Yan CZ. Occurrence, distribution, and seasonal var- Vörösmarty CJ, McIntyre PB, Gessner MO, Dudgeon D, Prusevich A, Green P, et al. Global iation of estrogenic compounds and antibiotic residues in Jiulongjiang River, South threats to human water security and river biodiversity. Nature 2010;467:555–61. China. Environ Sci Pollut Res 2012;19:1392–404. Wang SY, Sudduth EB, Wallenstein MD, Wright JP, Bernhardt ES. Watershed urbanization Zheng SL, Qiu XY, Chen B, Yu XG, Liu ZH, Zhong GP, et al. Antibiotics pollution in Jiulong alters the composition and function of stream bacterial communities. PLoS ONE River estuary: Source, distribution and bacterial resistance. Chemosphere 2011;84: 2011;6:e22972. 1677–85. Weijters MJ, Janse JH, Alkemade R, Verhoeven JTA. Quantifying the effect of catchment Zwart G, Crump BC, Agterveld MPKV, Hagen F, Han SK. Typical freshwater bacteria: an land use and water nutrient concentrations on freshwater river and stream biodiver- analysis of available 16S rRNA gene sequences from plankton of lakes and rivers. sity. Aquat Conserv 2009;19:104–12. Aquat Microb Ecol 2002;28:141–55.