Endophytic bacterial diversity in roots of Phragmites australis in constructed Beijing Cuihu Wetland (China)


  • Editor: Paolina Garbeva

Correspondence: Yan Hong Li, College of Life Science, Capital Normal University, Xisanhuan Northroad 105, Haidian district, Beijing 100048, China. Tel: +86 10 6890 1692; fax: +86 10 6890 2328; e-mail: liyh@mail.cnu.edu.cn


The community structure and diversity of endophytic bacteria in reed (Phragmites australis) roots growing in the Beijing Cuihu Wetland, China was investigated using the 16S rRNA library technique. Primers 799f and 1492r were used to amplify the specific bacterial 16S rRNA fragments successfully and construct the clone library. In total, 166 individual sequences were verified by colony PCR and used to assess the diversity of endophytic bacteria in reed roots. Phylogenetic analysis revealed that 78.9% of the clones were affiliated with Proteobacteria and included all five classes. Other clones belonged to Firmicutes (9.0%), Cytophaga/Flexibacter/Bacteroids (6.6%), Fusobacteria (2.4%), and nearly 3.0% were unidentified bacteria. In Proteobacteria, the Alpha and Gamma subgroups were the most abundant, accounting for approximately 34.4% and 31.3% of all Proteobacteria, respectively, and the dominant genera included Pleomorphomonas, Azospirillum, and Aeromonas. In addition, nearly 13.6% of the Proteobacteria were very similar to some genera of sulfate-reducing bacteria (SRB) such as Dechloromonas, Desulfovibrio, and Sulfurospirillum. The bacteria in these genera are considered to play important roles in the metabolism of nitrogen, phosphorus, sulfur, and some organic compounds in wetland systems. Hence, this study demonstrates that within the diverse bacterial communities found in reed roots, endophytic strains might have a strong potential to enhance phytoremediation by reed wetlands.


Endophytic bacteria are defined as those bacteria that can be isolated from surface-disinfected plant tissues or extracted from within the plant and that are not observed to harm the host plant (Hallmann et al., 1998). They are found in most, if not all, plant species, span a wide range of bacterial phyla, and are known to play a role in plant growth-promoting and pathogen-control activities (Hallmann et al., 1997; Hallmann & Berg, 2006; Ryan et al., 2008). Many factors, such as plant rotations, soil conditions, and phytopathogen populations, are known to influence the population structures of endophytic bacteria (Graner et al., 2003). Recent research suggests that these beneficial impacts may, in the case of plants growing at contaminated sites, extend to the degradation of xenobiotic compounds and may thus play an important role in phytoremediation (Germaine et al., 2006).

So far, most information on endophytic bacterial diversity has been obtained using culture-dependent approaches. Both Gram-positive and Gram-negative bacterial endophytes have been isolated from several types of tissues from numerous plant species (Kobayashi & Palumbo, 2000). Recent studies of plant endophytic bacteria have focused on their roles within plants in relation to plant nutrition (Dalton et al., 2004), pollutant catabolism (Moore et al., 2006), stress or defense responses, and invading pathogens (Graner et al., 2003). However, due to the unknown growth requirements of many bacteria and the presence of cells that are in a viable, but noncultivable state (Tholozan et al., 1999), the proportion of microbial diversity that has been identified using conventional cultivation techniques is <1% of the bacterial species present (Amann et al., 1995). These methodological constraints have seriously limited our knowledge regarding endophytic bacteria. More recently, the genetic diversity among endophytic populations of crop plants has been monitored successfully using PCR-based techniques (Sessitsch et al., 2002; Sun et al., 2008).

Common reed (Phragmites australis Cav. Trin.) is one of the most widely distributed plant species on earth and is restricted mainly to marshy areas and swamps. It is considered to have high detoxification and phytoremediation potential and has been used widely to treat industrial wastewater containing heavy metals (Jean & De, 1997) in wetland systems including wetlands constructed specifically for wastewater treatment (Vymazal & Kropfelova, 2005). Earlier studies have focused on cell counts and the activity of bacteria in the reed rhizosphere using cultivation-based techniques (Borsodi et al., 2003). Others have focused on the community structure and diversity of bacteria associated with the reed rhizosphere in freshwaters using molecular methods (Borsodi et al., 2007; Ravit et al., 2007; Rusznyak et al., 2007; Vladar et al., 2008), but no study has examined the endophytic bacteria associated with reed roots and their possible roles in phytoremediation mediated by reed wetland.

This paper describes the diversity and community structure of endophytic bacteria in reed roots growing in a constructed wetland. We used the 16S rRNA library technique, a culture-independent method, with the goal of understanding the role of bacteria within reed roots in enhancing the phytoremediation of eutrophic water mediated by reed-constructed wetland.

Materials and methods

Reed root sampling and surface disinfection

Reed roots were obtained from the common reed (P. australis Cav. Trin.) zone of Beijing CuiHu Wetland, China, in July 2008. The wetland was used to treat a mixture of domestic wastewater from the surrounding area and water from Shangzhuang reservoir. In this study, one treatment region with marshy plants (mainly reed) and one control region (without any plants) were chosen to measure the water quality, in order to determine the effect of reed on the water body. The control region shared the same water source with the reed planted region, but was 50 m away from it. The physicochemical characteristics of the water in the treatment region were as follows: pH 7.34, 1.37 mg L−1 total nitrogen (N), 0.13 mg L−1 total phosphorus (P), and 27.85 mg L−1 organic matter. In the control region, the water quality indexes were as follows: pH 7.56, 3.11 mg L−1 total nitrogen, 0.25 mg L−1 total phosphorus, and 31.90 mg L−1 organic matter. The observations and sampling took place in July 2008.

The reed roots were sampled from 15 cm below the water surface within the treatment region. Three samples of 1 g fibrous roots were taken from three different locations with a distance of about 10 m. They were immediately mixed and transported to the laboratory. Reed roots were first washed three times with tap water to remove attached soil. Subsequently, the roots were immersed in 70% ethanol for 3 min, washed with a fresh sodium hypochlorite solution for 5 min, rinsed three times with 70% ethanol for 30 s, and finally washed five times with sterile-distilled water as described in Sun et al. (2008). To confirm that the disinfection process was successful, aliquots of the sterile-distilled water used in the final rinse were set on Luria–Bertani (LB) medium plates. The plates were examined for bacterial growth after incubation at 30 °C for 3 days. Reed root samples that were not contaminated as determined by the culture-dependent disinfection test were used for subsequent analyses.

DNA extraction and amplification of the bacterial 16S rRNA

About 0.5 g of the surface-disinfected reed roots were frozen with liquid nitrogen and ground to a fine powder in a sterilized and precooled mortar. Then, the hot cetyltrimethylammonium bromide (CTAB) procedure (Xie et al., 1999) was used to extract the total DNA. The DNA was then resuspended in 25 μL of sterile Milli-Q water. The pair of primers 799f (5′-AACAGGATTAGATACCCTG-3′) and 1492r (5′-GGTTACCTTGTTACGACTT-3′) (Chelius & Triplett, 2001) was selected to amplify the DNA of reed endophytic bacteria. The 50-μL PCR mixture contained 100 ng of DNA extract, 5 μL 10 × Taq reaction buffer (including 1.5 mM MgCl2), 10 pmol of each primer, 200 μM each dNTP, and 1.5 U of Taq DNA polymerase (Takara Co.). After initial denaturation at 94 °C for 5 min, each thermal cycling was as follows: denaturation at 94 °C for 1 min, annealing at 53 °C for 1 min, and elongation at 72 °C for 1 min. At the end of 30 cycles, the final extension step was at 72 °C for 15 min. Products of four parallel PCRs were combined and separated electrophoretically. A band approximately 700 bp in size in the electrophoresis pattern was excised from a 1% agarose gel and purified using the Gel Extraction Kit (Omega Co.) as described by the manufacturer.

Construction of the 16S rRNA clone library

The purified PCR products were ligated into the pMD18-T vector (Takara Co.). Escherichia coli Top10 competent cells (Tiangen Co.) were transformed with the ligation products and spread onto LB agar plates with ampicillin (100 mg L−1) for standard blue and white screening (Sambrook et al., 1989). Randomly selected colonies were screened directly for inserts by performing colony PCR with primers RV-M (5′-GAGCGGATAACAATTTCACACAGG-3′) and M13-47 (5′-CGCCAGGGTTTTCCCAGTCACGAC-3′) for the vector (Takara Co.). A total of 180 clones containing inserts of the correct size were sequenced using an ABI PRISM 3730 automatic sequencer (Shanghai Sangon Co. Ltd).

Phylogenetic analysis

After being trimmed by removing the vector sequences using the editseq program in the dnastar package (Burland, 2000), clones with >97% sequence identity were grouped into one operational taxonomic unit (OTU) by sequencher 4.8 (Gene Codes, Ann Arbor, MI). All the nucleotide sequences, approximately 700 bases, were compared with the NCBI database using blastn or aligned by the identify analysis of EzTaxon server 2.1 (Chun et al., 2007). Sequences with >97% similarity were assigned to the same species and those with >95% similarity were assigned to the same genus. The sequences were aligned using clustal w (Thompson et al., 1994), and tree constructions were performed with the mega 3 program package (Kumar et al., 2004) using the neighbor-joining method. Bootstrap analysis was performed using data resampled 1000 times. The trees were constructed by calculating Kimura distances (Kimura, 1980).

Estimation of the size of the clone library

To estimate the representation of the library, the clone coverage was calculated with the following equation based on the sequencing results: C=(1−n1/N) × 100%, where n1 is the number of single clones and N is the total number of clones in the clone library. The diversity of the clone library was investigated by rarefaction analysis. Rarefaction curves were calculated using ecosim 7.0 software (Gotelli & Entsminger, 2001).


Total DNA extracted from surface-disinfected reed roots was used to amplify the bacterial 16S rRNA fragments using primers 799f and 1492r. The amplified DNA displayed only one distinct band, approximately 700 bp, on the agarose gel. Thus, the primers 799f and 1492r were deemed sufficient for specific amplification of the bacterial 16S rRNA fragments and satisfactorily excluded any contamination from reed mtDNA. The purified PCR products were used to construct a 16S rRNA clone library of reed endophytic bacteria.

One hundred and sixty-six individual sequences derived from 180 positive clones were verified by colony PCR and submitted to GenBank (accession no.: GU178822GU178836, GU178838GU178862, GU178864GU178880). They were used to identify the bacterial endophyte diversity in the roots of P. australis. Phylogenetic analysis of all sequences revealed that the majority of clones were affiliated with Proteobacteria (131 clones, 78.9%). Other clones belonged to Firmicutes (15 clones, 9.0%), Cytophaga/Flexibacter/Bacteroides (CFB) (11 clones, 6.6%), Fusobacteria (four clones, 2.4%), and nearly 3% (five clones) of the sequences showed a high similarity to unidentified bacterial sequences. Details of all OTUs in the clone library are listed in Table 1.

Table 1.   Distribution of 16S rRNA clones detected from endophytes in reed roots
GroupNo. of OTUsNo. of clones% Total clonesClosest NCBI matchIdentity (%)
 63.6Pleomorphomonas oryzae (AB159680)99.9
 31.8Pleomorphomonas oryzae (AB159680)97.1
 10.6Pleomorphomonas koreensis (AB127972)93.7
 95.4Azospirillum picis (AM922283)97.5
 42.4Agrobacterium vitis (D14502)99.3
 21.2Rhodoplanes roseus (D25313)93.5
 21.2Rhodoplanes elegans (D25311)99.7
 21.2Rhizobium daejeonense (AY341343)98.8
 10.6Rhizobium daejeonense (AY341343)98.6
 10.6Magnetospirillum magnetotacticum (AAAP01003867)96.4
 21.2Magnetospirillum gryphiswaldense (Y10109)97.1
 10.6Kaistia soli (EF592609)93
 10.6Sinorhizobium chiapanecum (EU286550)98.6
 31.8Brevundimonas alba (AJ227785)95.9
 21.2Prosthecomicrobium mishustinii (FJ560749)95.8
 21.2Bosea minatitlanensis (AF273081)94.2
 10.6Filomicrobium insigne (EF117253)96.6
 10.6Devosia sp. (EF575560)94.5
 10.6Telmatospirillum siberiense (AF524863)92.7
 21.2Dechloromonas hortensis (AY277621)93.9
 53.0Dechloromonas hortensis (AY277621)96.6
 53.0Rhodoferax antarcticus (AF084947)98.4
 31.8Hydrogenophaga bisanensis (EF532793)98.8
 31.8Janthinobacterium lividum (Y08846)99.3
 21.2Piscinibacter aquaticus (DQ664244)99.6
 21.2Aquincola tertiaricarbons (DQ656489)98.2
 21.2Roseateles depolymerans (AB003626)99.7
 21.2Coccomonas naphthalovorans (AY166684)99.9
 10.6Ideonella dechloratans (X72724)98.3
 2615.6Aeromonas bivalvium (DQ504429)100
 63.6Aeromonas hydrophila (AJ508766)98.8
 31.8Dickeya dadantii (AF520707)97.9
 31.8Beggiatoa alba (AF110274)93.1
 21.2Enterobacter cloacae (Z96079)97.5
 10.6Pseudomonas gessardii (AF074384)96.4
 53.0Pelobacter propionicus (X70954)99.3
 53.0Desulfomicrobium norvegicum (NR_025407)100
 31.8Desulfobacterium catecholicum (AJ237602)97.4
 10.6Desulfomonile limimaris (NR_025079)99
 10.6Desulfovibrio putealis (AY574979)99.9
 31.8Sulfurospirillum halorespirans (AF218076)99.9
 31.8Clostridium hylemonae (AB023972)93.6
 31.8Clostridium aminovalericum (X73436)90.7
 21.2Clostridium alkalicellulosi (AY959944)94.3
 10.6Acetobacterium malicum (X96957)100
 21.2Acidaminobacter hydrogenoformans (AF016691)91.1
 21.2Clostridium tertium (Y18174)97.1
 21.2Cohnella thermotolerans (AJ971483)97.9
 31.8Bacteroides graminisolvens (AB363973)99.4
 21.2Flavisolibacter ginsengiterrae (AB267476)92.9
 21.2Prolixibacter bellariivorans (AY918928)92.4
 21.2Wandonia haliotis (FJ424814)90.2
 10.6Prevotella paludivivens (AB078827)97.9
 10.6Paludibacter propionicigenes (AB078842)91.5
 4 Ilyobacter insuetus (AJ307980)91.2
Uncultured bacterium25   
 21.2Uncultured bacterium (FJ535011)94
 31.8Uncultured bacterium (EU234315)99

The sequences related to Proteobacteria made up the largest fraction of the clone library, which included Alpha, Beta, Gamma, Delta and Epsilon classes.

Of 131 clones affiliated with Proteobacteria, 45 and 41 clones exhibited a high similarity to Alphaproteobacteria and Gammaproteobacteria, respectively. The number of clones grouped into Beta, Delta and Epsilon classes was 27, 15, and three, respectively. Thus, the most abundant classes were Alpha- and Gammaproteobacteria, which accounted for 34.4% and 31.3% of the Proteobacteria, respectively.

Forty-five clones in the class Alphaproteobacteria comprising 19 OTUs were related to three orders of bacteria, which included Rhizobiales, Rhodospirillales, and Caulobacterales (Fig. 1a). Among them, 28 clones were grouped into order Rhizobiales and these included nine genera (Bosea, Pleomorphomonas, Sinorhizobium, Rhizobium, Rhodoplanes, Agrobacterium, Devosia, Filomicrobium, and Prosthecomicrobium); the most abundant genus was Pleomorphomonas. Fourteen sequences were grouped into order Rhodospirillales and belonged to three genera (Telmatospirillum, Magnetospirillum, and Azospirillum). Nine of these 14 sequences were similar to Azospirillum picis (97.5% sequence identity). In addition, three clones were similar to Brevundimonas in Caulobacteraceae of Caulobacterales (95.9% sequence identity) (Table 1).

Figure 1.

Figure 1.

 16S rRNA based dendrogram showing the phylogenetic relationship of clones from reed roots. Phylogenies were inferred using neighbor-joining analysis and trees were generated using mega 3 software. Numbers in parentheses represent the sequence accession numbers in GenBank. Numbers in square brackets indicate the clone number out of the total clones. Numbers at branch points indicate bootstrap values. The scale bar represents a 2% estimated difference in nucleotide sequence. (a) Alphaproteobacteria; (b) Betaproteobacteria and Gammaproteobacteria; (c) Deltaproteobacteria, Epsilonproteobacteria, Firmicutes, Fusobacteria, and CFB. Sequences of Pseudomonas gessardii (AF074384), Clostridium hylemonae (AB023972), and Agrobacterium vitis (D14502) were used as outgroup references in these three trees, respectively.

Figure 1.

Figure 1.

 16S rRNA based dendrogram showing the phylogenetic relationship of clones from reed roots. Phylogenies were inferred using neighbor-joining analysis and trees were generated using mega 3 software. Numbers in parentheses represent the sequence accession numbers in GenBank. Numbers in square brackets indicate the clone number out of the total clones. Numbers at branch points indicate bootstrap values. The scale bar represents a 2% estimated difference in nucleotide sequence. (a) Alphaproteobacteria; (b) Betaproteobacteria and Gammaproteobacteria; (c) Deltaproteobacteria, Epsilonproteobacteria, Firmicutes, Fusobacteria, and CFB. Sequences of Pseudomonas gessardii (AF074384), Clostridium hylemonae (AB023972), and Agrobacterium vitis (D14502) were used as outgroup references in these three trees, respectively.

Gammaproteobacteria were the second most abundant group of Proteobacteria. The 41 clones comprising six OTUs in this class represented bacteria in four orders (Enterobacteriales, Aeromonadales, Thiotrichales, and Pseudomonadales). A majority of the sequences (32 clones) exhibited high similarity (98.8–100% sequence identity) to bacteria of genus Aeromonas, accounting for nearly 80% of Gammaproteobacteria. The other nine sequences were related to the genera Beggiatoa, Pseudomonas, Dicheya, and Enterobacter (Table 1; Fig. 1b).

Betaproteobacteria were less abundant than Alpha and Gamma classes of Proteobacteria. Of the 27 clones in the Betaproteobacteria class (Fig. 1b), 20 were closely related to Burkholderiales (74.1% of Betaproteobacteria) and belonged to genera Roseateles, Aquincola, Ideonella, Piscinibacter, Coccomonas, Hydrogenophaga, Rhodoferax, and Janthinobacterium. An additional seven clones were grouped into Rhodocyclales and classified as Dechloromonas. Dechloromonas and Rhodoferax were the most abundant genera in this subgroup (Table 1).

Fifteen clones grouped into Deltaproteobacteria, including five OTUs, were closely related to five different species of sulfate-reducing bacteria (SRB) (97–100% sequence identity). Of these, the sequences of five clones were closely related to Desulfomicrobium norvegicum and Pelobacter propionicus, making them the dominant species of Deltaproteobacteria. In addition, other clones were assigned to Desulfomonile limimaris, Desulfobacterium catecholicum, and Desulfovibrio putealis. All three clones related to Epsilonproteobacteria showed high similarity to Sulfurospirillum halorespirans (99.9% sequence identity) (Table 1; Fig. 1c). In total, the SRB occupied nearly 13.6% of Proteobacteria.

Among non-Proteobacteria, the remaining 15 and 11 clones exhibited high similarity to the Firmicutes and CFB phyla (Fig. 1c), respectively. In Firmicutes, all 15 clones belonged to Clostridiales and the dominant genus was Clostridium (10 clones). Other genera included Cohnella (two clones), Acidaminobacter (two clones), and Acetobacterium (one clone). Of 11 clones grouped into the CFB phylum, four were closely related to genera Bacteroides (99.4% sequence identity) and Prevotella (97.9% sequence identity) in the Bacteroidales order, and others were distantly related to the genera Paludibacte, Prolixibacter, Wandonia, and Flavisolibacter (90–92% sequence identity). Finally, four clones represented sequences assigned to Fusobacteria; they were distantly related to Ilyobacter (91.2% sequence identity) in the order Fusobacteriales (Table 1; Fig. 1c).

Furthermore, alignment of all 166 sequences showed that the number of single type sequences was 15, and the calculated coverage of the clone library was 90.97%. The rarefaction curve also tended to plateau (Fig. 2), indicating that this library was sufficient to detect a large majority of the endophytic bacterial diversity in the reed roots used in our research.

Figure 2.

 Rarefaction curve of the endophytic bacterial 16S rRNA clone library of reed roots.


Because plant chloroplast 16S rRNA mitochondrial 18S rRNA and bacterial 16S rRNA are homologous, applying highly specific primers to investigate reed root endophytic bacterial diversity was challenging, but crucial to our study. As reported, the pair of primers (799f and 1492r) would not amplify chloroplast 16S rRNA from 41 plants and mitochondrial 18S rRNA of six Chlorophyta plants. In this study, we obtained only one band approximately 700 bp of bacterial 16S rRNA fragments using this pair of primers. This demonstrated that the primers 799f and 1492r could specifically amplify the endophytic bacterial 16S rRNA fragments and could not amplify mitochondrial 18S rRNA in reed roots; thus, it was suitable for use in the study of reed root endophytic bacteria.

Proteobacteria were the most dominant group in our clone library and all five classes were detected, which was consistent with other studies (Chelius & Triplett, 2001; Sun et al., 2008). In the most abundant subgroup of Alphaproteobacteria, 10 clones were assigned to Pleomorphomonas oryzae and Pleomorphomonas koreensis, both nitrogen-fixing bacteria (Xie & Yokota, 2005; Im et al., 2006); nine clones were related to A. picis, which was also identified as a nitrogen fixer (Peng et al., 2006). Other Azospirillum species have been isolated from roots of numerous wild and cultivated grasses, cereals, food crops, and soils, and proved to be capable of enhancing the growth of plants through the production of phytohormones (Bashan & Holguin, 1997) and supplying nitrogen to their host plants (Dobereiner, 1980; Okon, 1985).

Another dominant subgroup was observed in the Gammaproteobacteria. A majority of the clones were highly similar to Aeromonas bivalvium 868E, which was originally isolated from bivalve mollusks (Minana-Galbis et al., 2007) and was a primary or an opportunistic pathogen in invertebrates and vertebrates including humans (Martin-Carnahan & Joseph, 2005). It was also demonstrated to be capable of reducing nitrate (NO3) to nitrite (NO2) and producing indole from tryptophan (Minana-Galbis et al., 2007). A number of sequences were very similar to bacteria in genera Beggiatoa, Pseudomonas, Enterobacter, and Dickeya. According to previous reports, species in Beggiatoa can use NO3 anaerobically as an alternative electron acceptor in place of O2 and can perform anaerobic H2S oxidation with NO3 (Kamp et al., 2006). Thus, they have a significant impact on the aquatic nitrogen and sulfur cycles. Pseudomonads are also often found in contaminated aquifers, because they are able to use a large number of substances as energy or carbon sources and can often tolerate toxic compounds (Moore et al., 2006). Some strains of Enterobacter are reported to have the ability to fix nitrogen or display antagonistic activity to phytopathogens (Hallmann et al., 1997; Tsuda et al., 2001); they have also been shown to use phytate and play an important role in phosphorus cycling (Fuentes et al., 2009).

Clones related to Betaproteobacteria represented many genera including Dechloromonas, Janthinobacterium, Hydrogenophaga, and Rhodoferax. Dechloromonas, the most abundant genus of them, has been isolated from the gut of earthworms and was shown to have the ability to produce N2O and carry out complete denitrification (Horn et al., 2005). Desulfomicrobium norvegicum was one of the dominant species of Deltaproteobacteria and is able to tolerate microaerophilic conditions. It was originally described as a member of the genus Desulfovibrio (Genthner et al., 1997), which was also detected in the reed rhizosphere and considered to be able to use carbohydrates and propanediols as carbon sources (Basso et al., 2005; Vladar et al., 2008). Pelobacter propionicus, another dominant species in Deltaproteobacteria, can use 2,3-butanediol, acetoin, ethanol, pyruvate, and lactate for growth under strictly anaerobic conditions and induce propionate formation from C2 compounds (Schink, 1984). In addition, other species detected in this research such as D. limimaris, D. catecholicum, and D. putealis reflected the diversity of SRB in reed roots, which was quite similar to that found in the rhizosphere of P. australis in Lake Velencei in Hungary (Vladar et al., 2008). Sulfurospirillum halorespirans in the Epsilonproteobacteria subgroup was detected in our library and has been reported to be capable of reducing tetrachloroethene to cis-dichloroethene in an anaerobic environment (Luijten et al., 2004). In addition, they were also able to reduce oxidized metals and to reduce and oxidize quinone moieties coupled to energy conservation (Luijten et al., 2004).

All 15 clones assigned to Firmicutes belonged to order Clostridiales. The genus Clostridium has been reported to be a ubiquitous endophytic bacterium in gramineous plants and has exhibited nitrogen-fixing capability in association with nondiazotrophic endophytes (Minamisawa et al., 2004). In addition, sequences of some clones showed low identity to the cultured bacterial genera, but a high identity to the uncultured bacteria, revealing the presence of some uncultured bacteria in the reed endophytic bacterial community.

Water eutrophication is one of the most challenging environmental problems in the world. At present, N and P input and enrichment in water are the primary factors thought to be responsible for eutrophication. Phragmites australis has been confirmed as an important plant with the capacity to degrade N and P in wetland systems. The water quality index analysis in this research showed that it contributed to removing approximately 56%, 48%, and 13% of the total N, P, and organic matter, respectively, in our study system. As reported, P. australis could absorb N and P in tissues to remove the nutrient in the water (Tian et al., 2009). In our clone library, we found many endophytic bacteria that were considered to have the capacity to fix nitrogen, such as P. oryzae and A. picis; we also detected some bacteria that might reduce nitrate to nitrite, such as A. bivalvium and Aeromonas hydrophila; some that might carry out denitrification, such as Dechloromonas hortensis; some that might play important roles in the sulfur cycle, such as D. norvegicum and S. halorespirans; and some that might remove toxic material from the water, such as bacteria in genera Beggiatoa, Desulfobacterium and Sulfurospirillum. In addition, a few of the genera detected might have the ability to utilize organic phosphorus, such as those in genus Enterobacter. In short, most of the endophytic bacteria in reed roots might have a strong potential to enhance phytoremediation, especially with regard to the nitrogen and sulfur cycles and removal of some organic matter during water purification by the reed-constructed wetland.

However, no ammonia-oxidizing bacteria (AOB), such as Nitrosomonas and Nitrosospira, and no anammox bacteria, such as CandidatusBrocadia’, ‘Kuenenia’, ‘Scalindua’, and ‘Jettenia,’ which are often detected in certain soil types and at particular depths (Humbert et al., 2010), were detected in our clone library. This suggests that the endophytic bacteria in reed roots are probably not involved in the first step of nitrification during which ammonia is converted to nitrite, and anaerobic oxidation of ammonium, but could carry out the other steps of nitrification as well as denitrification and nitrogen fixation. However, some AOB and Bacillus bacteria have been found in the rhizosphere of P. australis (Xing et al., 2008; Xie et al., 2009), indicating that bacteria in the rhizosphere and endophytic bacteria may play different roles in nutrient metabolism in wetland ecosystems.

However, because the cloning sequences cannot provide direct information on the function of the individual community members, further work is necessary to improve our understanding of the mechanisms through which endophytic bacteria of reed roots mediate water purification.


We would like to thank Daniel Keck at UC Santa Cruz for his assistance with English language and grammatical editing of the manuscript. This work was funded by the Scientific Research Program of Beijing Municipal Commission of Education.

Authors' contribution

Y.H.L. and J.N.Z. contributed equally to this work.


Nucleotide sequence data reported are available in the GenBank databases under the accession numbers from GU178822 to GU178836, from GU178838 to GU178862 and from GU178864 to GU178880.