Interactive comment on “ Distribution of typical denitrifying functional genes and diversity of the nirS-encoding bacterial community related to environmental characteristics of river sediments ” by S .

The biological mechanism of denitrification is of great ecological significance since this process is the most dominant source of N2O emission, which is proved recently to be the single most important ozone-depleting substance. Huang’s paper entitled ‘Distribution of typical denitrifying functional genes and diversity of the nirS-encoding bacterial community related to environmental characteristics of river sediments’ investigated the functional groups of the denitrification by analyzing the abundance of functional genes and found the spatial distribution of the denitrifiers is corralated with the dissoved inorganic N. These data are of ecological significance for understanding the mechanism of


Introduction
Agricultural and other anthropogenic activities result in increasing amount of nitrogen compounds entering the freshwater and marine systems, causing eutrophication problems.In the microbial process of denitrification, nitrate (NO − 3 ) is converted to gaseous forms then lost from the systems (Canfield et al., 2010).Therefore, the denitrification process should help to minimize the eutrophication problems.Since there is a large amount of denitrifiers and nitrogen compounds in eutrophication sediment, sediment is an important environment for nitrogen removal from an aquatic system through denitrification.
Bacterial denitrifiers are taxonomically diverse due to the fact that each bacterial species may participate in only one step of the denitrification process (Burgin and Hamilton, 2007).Denitrification consists of four sequential steps and involves four enzymes.The first step is to reduce NO − 3 to nitrite (NO − 2 ) catalyzed by a membrane-bound NO − 3 reductase (nar) or periplasmic NO − 3 reductase (nap), which are encoded by the narG or the napA genes, respectively.Denitrifying bacteria contain one or both of the NO − 3 reductase (i.e., narG and napA), in which narG is considered to be more extensive and representative (Deiglmayr et al., 2004;Smith et al., 2007;Reyna et al., 2010).In the second step, NO − S. Huang et al.: Distribution of typical denitrifying functional genes reduced to nitrous oxide (N 2 O), catalyzed by NO reductase (encoded by the norB gene).N 2 O is an obligate intermediate, some of which ultimately escapes to the atmosphere.The reduction of N 2 O to nitrogen gas (N 2 ) is the last step, which is catalyzed by N 2 O reductase encoded by the nosZ gene present in the periplasm (Throbäck et al., 2004;Chon et al., 2009).These reductase genes are generally used as biological makers of the denitrification process because an approach involving 16S rRNA is not suitable to investigate the diverse community of denitrifying bacteria (Zumft, 1997;Braker et al., 1998).
The distribution of denitrifying genes in sediments are affected by many factors, including concentration gradient of the dissolved inorganic nitrogen (DIN) (including NO − 3 , NO − 2 and NH + 4 ), organic matter content, dissolved oxygen, and redox potential (Dong et al., 2009;Reyna et al., 2010).As the depth increases, oxygen concentrations decrease rapidly, microorganism changed from facultative aerobic phylotypes to strictly anaerobic phylotypes, and different denitrifiers (base on nirS diversity) change with the depth (Tiquia et al., 2006).Concentrations of denitrification substrates, among which the DIN and organic matter are most important, also change with the sediment depth (Dalsgaard et al., 2005;Meyer et al., 2008).More gene copies of narG and nirS were observed in the river sediment with higher NO − 3 concentrations (Chon et al., 2009).Availability of organic matter and carbon/nitrogen ratio determine bacterial biomass and NO − 3 reduction rates as well as pathways (Burgin and Hamilton, 2007;Dodla et al., 2008).Some dissimilatory NO − 3 reducing communities compete for NO − 2 with denitrifying bacteria.These NO − 3 reducing pathways such as anammox (anaerobic ammonium oxidation) and DNRA (dissimilatory nitrate reduction to ammonium), might appear in different sediment depths according to various carbon/nitrogen ratios (Burgin and Hamilton, 2007).When sediment conditions favor dissimilatory NO − 3 reducing processes, complete denitrification of NO − 3 to N 2 should be reduced or inhibited.One shortcoming of the previous investigations on this topic is their focus on shallow sediments (0-10 cm depth).Competitions among the different denitrification pathways and communities in the sediment profile are still poorly understood.Therefore, further research is needed to study the denitrifying genes and potentially interacting environmental factors in deeper sediments.
The aims of the current study were twofold: (1) to investigate the spatial distributions of denitrifying functional genes (narG, nirS, nirK, nosZ) as well as nrfA, a gene marker for DNRA (Dong et al., 2009) and diversity of denitrifier community along the DIN gradient in the Pearl River sediment, and (2) to determine the relationship between sediment chemistry and denitrifying functional genes in the sediment.

Sample collection
The Pearl River is located in the subtropical region in the Guangdong Province, Southern China.This river represents one of the most important ecosystems linking the highly developing land area and the South China Sea.In recent years, the Pearl River has experienced high anthropogenic nitrogen loading due to rapid population increase and economic development.
Sediment samples were collected on 6 April 2010 at three locations along the Pearl River near Guangzhou city, China (23 • 8 N, 113 • 17 E), i.e., from the upstream at Zhujiang Bridge (ZB), midway down the river at Huadi (H), and to the downstream at Ersha (E) (Table 1).The three sample locations were chosen based on the contrasting land use, pollution sources, and aquatic conditions.There is a municipal refuse disposal station near location ZB, resulting in a large amount of polycyclic aromatic hydrocarbons deposited in the sediment.Several small factories are around location H, resulting in chemicals, such as chlorinated compounds and metals, detected in the sediment.Also, a large amount of water lettuces float on the surface at location H, causing a hypoxia condition in the water body.Location E is close to residential areas, which input sewage to the river.The water depths at the sample locations were about 3 m and bottom water temperatures were about 23 • C. Salinity was low in this river section, between 1-12 ‰ depending upon the tide.Dissolve oxygen level was 0.30 mg L −1 in the water-sediment interface and decreased rapidly to 0.01 mg L −1 within 10 cm sediment (Huang et al., 2011).Three sediment cores were collected at each site within a 2 m diameter area.Each sediment core was 8 cm diameter and 30 cm long.Sediment cores were transported in ice boxes to the laboratory.In the laboratory, each sediment core was sectioned into 5 cm slices (0-5, 5-10, 10-15, and 20-25 cm) and then put into sterile containers.About half of the sediment sub-samples were used for chemical analysis, which was performed within 24 h of the sample collection and on 3 replicates per depth.The remaining sediment sub-samples were stored at -80 • C for molecular analysis.

Analysis of environmental parameters
Sediment pH values were determined using a pH meter (FED20, Mettler, Toledo).Water content and grain sizes of the sediment samples were analyzed using the gravimetric method and pipette method, respectively.By drying a sediment sample overnight at 105 • C to determine the dry weight, the organic matter content was obtained from the subsequent loss of weight after continuously drying the sample at 550 • C for more than 2 h.Total carbon content (TOC) was calculated based on the organic matter content and the total nitrogen (TN) content was determined using a Foss Kjeltec 2300

DNA extraction
Three independent samples of nucleic acids for each sediment slice were extracted from 500 mg sediment samples using the Fast DNA spin kit for soil as described in the manufacturer's instructions (Bio 101, Qbiogene Inc., CA, USA).Extraction of DNA was confirmed by gel electrophoresis.The concentration of the extracted DNA was measured using a Nano-drop spectrophotometer (Nano-Drop Technologies, USA) at 260 nm.A test of replicate extractions was performed to ensure reproducible yields from the DNA extractions.

Real-time PCR analysis
Table 2 lists the information on the primers selected for amplification of the different genes encoding 16S rRNA, narG, nirS, nirK, nosZ, and nrfA.Each assay contained a standard curve, independent triplicate sediment DNA templates for each sediment slice, and triplicate no template controls (NTC).Experimental Q-PCR triplicates for each DNA sample were then averaged to obtain a single gene copy number.Real-time PCRs were carried out in LightCycler480 with Sequence Detection Software v1.4 (Applied Biosystems, USA).Each PCR mixture (10 µL) was composed of 5 µL of SYBR Premix Ex TaqTM II (2×), 0.4 µL 10 nM of each forward and reverse primers, 3.2 µL dH 2 O and 1.0 µL of template DNA (TaKaRa Biotechnology, Japan).PCR amplification and detection were performed in LightCycler480 Multiwell (384-well) reaction plates with optical cap (Applied Biosystems, USA).The PCR temperature program was initiated with 30 s at 94 • C, followed by 40 cycles of 5 s at 94 • C, 30 s at the specific annealing temperature (Table 2), and 30 s at 70 • C. A melting curve analysis for SYBR Green assay was prepared after amplification to distinguish the targeted PCR product from the non-targeted PCR product.The potential presence of PCR inhibitors in sediment DNA extracts was tested by running a real-time PCR assay on serial dilution of sediment DNA extracts.No inhibition was detected in any case.Standard curves for real-time PCR were produced based on a serial dilution of known copies of PCR fragments of the respective functional genes generated using M13 PCR from clones.All the standard curves showed excellent correlations between the DNA template concentration and the crossing point with high coefficients of determination (R 2 > 0.99).The PCR efficiency was estimated as follows (Bustin et al., 2009) where E is the PCR efficiency and S is the slope of the line of standard curve obtained from the tenfold dilution series.The Q-PCR efficiency values for 16S rRNA, narG, nirS, nirK, nosZ, and nrfA were 0.964, 1.114, 0.863, 0.987, 1.020, and 1.258, respectively.The ratios of denitrifying genes to 16S rRNA gene extracted from total bacteria were used to evaluate the relative abundances of various denitrifies compared to total bacteria.

Phylogenetic analysis
Datasets of nirS sequences were analyzed using the Bayesian inference (BI) (Huelsenbeck et al., 2001), implemented with MrBayes version 3.1.2(Ronquist and Huelsenbeck 2003).A best fitting model of nucleotide substitution was identified using the Akaike information criterion (AIC) (Akaike, 1973) as implemented in MrModelTest 2.3 (Nylander, 2004).Bayesian analysis was carried out using GTR+I+G model selected by MrModelTest 2.3, in which model parameters were treated as unknown and estimated through the BI.The following settings were applied: implementing two Markov chain Monte Carlo (MCMC) runs, running four simultaneous Markov chains for 19 million generations, and sampling the Markov chains every 100 generations.Tracer V1.5 (Rambaut and Drummond, 2009) was used to judge convergence of the Bayesian Markov chain Monte Carlo runs.The first 14 000 sampled trees were discarded as burn-in.A consensus tree as show in Fig. 1 was constructed from the remaining sampled trees.

Statistical analysis
The coverage of each clone library (C) was calculated by (Mullins et al., 1995) where n is the number of unique OTUs and N the total number of clones in a library.Indices of gene diversity (Shannon-Wiener H and Simpson D) and evenness (E) were calculated using the methods of Schloss and Handelman (2005).Rarefaction analysis and two nonparametric richness estimators, the abundance-based coverage estimator (S ACE ) and the bias-corrected Chao1 (S Chao1 ), were calculated using DOTUR (Schloss and Handelman, 2005).Correlations between the gene distributions or nirS-encoding cluster assemblage vs. the environmental factors were analyzed with the canonical correspondence analysis (CCA) (Dang et al., 2008(Dang et al., , 2009)).In the analysis, the OTU percentage frequency data in each library were used as the species input and the environmental variables were normalized via Z transformation (Magalhães et al., 2008).Some large deviation of species might appear due to high occurrence of unique sequences.Considered the low coverage, we used the uni-modal CCA analysis to minimize the potential errors.Analyses of Spearman rank, multivariate and stepwise linear regressions of environmental and gene abundance variables were carried out using SPSS v12 on both raw and log-transformed data.All data were analyzed using one way analysis of variance (ANOVA) with the significant level P < 0.05.a The values represent the mean and standard error (n = 3).

Environment conditions of Pearl River sediments
The sediments were mainly soft silt, and had a light hydrogen sulfide smell.As shown in Table 1, on average, DIN contents were rather high, especially NH + 4 , compared to other river sediments (Richardson et al., 2004;Laverman et al., 2010).DIN concentrations of the sediment surfaces (0-5 cm) increased from locations ZB to E, such that NO − 3 , NO − 2 , and NH + 4 increased from 8.60, 0.34, and 153 mg kg −1 to 19.7, 1.28, and 1613 mg kg −1 , respectively.DIN concentrations at location E were much higher than those at the other sampling locations, which were attributable to the inputs of domestic wastewater.Tides might be another factor to cause the pollutant accumulation.NO − 3 and NH + 4 increased with the sediment depth at locations ZB and H. Nitrate concentrations at locations ZB and H, respectively, changed from 8.60 and 8.71 mg kg −1 in the sediment surface to 12.0 and 17.7 mg kg −1 in the bottom layer, while NH + 4 concentrations changed from 153 and 194 mg kg −1 to 1195 and 1218 mg kg −1 , respectively.Nitrate and NH + 4 concentrations at location E were the highest among sample locations, while NO − 3 and NH + 4 concentrations at location E showed insignificant variations with the sediment depth.In contrast, the vertical change of NO − 2 at location E was significant, and the concentrations were highest at the sediment surface and decreased with depth of 0-15 cm.TOC was abundant in the sediment and the maximum content appeared in E10 (10 cm depth at location E) layer (60.9 g kg −1 , Table 1).TOC values increased gradually from locations ZB to E.

Relative abundances of different denitrifying genes
Distributions of the denitrifying genes (narG, nirS, nirK, nosZ, and nrfA) varied between the three sampling locations.Although the highest concentrations of DIN and TOC appeared at location E, the relative abundance of denitrifying genes was the lowest at this location.The lowest abundance of 16S rRNA was founded at location H, whereas the relative abundance of denitrifying genes was the highest at this location (Table 3).The relative contributions of nirS gene to total bacteria were the highest among the denitrification genes at all locations.Abundances of 16S rRNA decreased with the sediment depth at the three sampling locations.
The relative abundance of narG gradually increased with the sediment depth except for location ZB with the lowest proportion at 25 cm depth.For nirS and nirK, the relative abundance was high within the depth of 0-10 cm and decreased rapidly below 10 cm.Since the proportion of nir genes can represent the potential denitrification rate (Dong et al., 2009;Graham et al., 2010), the result indicated that nitrite reduction process should mainly occur above 10 cm in the sediment.For nosZ, the highest proportions appeared in the depth of 5-15 cm.Interestingly, different from other denitrifying genes, relative abundance of nosZ was higher at location ZB.For nrfA, the relative abundance also gradually increased with sediment depth except for location H.It was noted that the total nrfA numbers actually decreased with the depth although the relative abundance of nrfA increased with the depth.www.biogeosciences.net/8/3041/2011/Biogeosciences, 8, 3041-3051, 2011

Phylogenetic diversity of nirS-encoding bacteria
The nirS gene libraries from the layers of ZB5, ZB10, ZB15, H5, H10, and H15 resulted in totally 305 nirS gene sequences from clones.These nirS gene sequences were assigned to 51 OTUs based on a 5 % cutoff.The coverage of these nirS clone libraries was 86 %, 79 %, 72 %, 75 %, 71 %, and 65 % in the layers ZB5, ZB10, ZB15, H5, H10, and H15, respectively.The diversity indexes of these libraries suggested a slightly increasing diversity of the denitrifying bacterial community with increasing depth (Table 4).The rarefaction curve suggested a similar trend (data not shown).
Nitrite reductase gene sequences from the Pearl River sediment were similar to those of other typical estuarine and marine environments, as well as sewage treatment systems.A phylogenetic tree indicated that the 51 nirS OTUs could be classified into nine clusters (Cluster-A to Cluster-I) base on their class and function (Fig. 1).Cluster-A dominated in all clone libraries with 57 %, 80 %, 52 %, 63 %, 61 %, and 65 % in the layers of ZB5, ZB10, ZB15, H5, H10, and H15, respectively.This cluster belonged to the family Rhodocyclaceaewhich contained mainly aerobic and denitrifying rodshaped bacteria with very versatile metabolic capabilities (Martins et al., 2010;Wongwilaiwalin et al., 2010).One OTU was closely related to the family Comamonadaceae.Both these families were usually found in aquatic sediment habitats, constructed wetland, and activated sludge (Ruiz-Rueda et al., 2007;Spain et al., 2007).Sequences of Cluster-B were similar to species Ralstonia eutropha strain (78-90 % indentify),belonging to the family Burkholderiaceae.R. eutropha H16 was previously known as Alcaligenes eutrophus and originally isolated from sludge.Cluster-C was closely affiliated with the species Thiobacillus denitrificans.This is a chemoautotrophic bacterium using sulfide as electron donor (Beller et al., 2006).The bacterium was only detected in deeper layers at location ZB with 3.3 % and 8.7 % in ZB10 and ZB15.Cluster-D had high sequence similarity (88-94 % identity) with an environmental clone from Chesapeake Bay sediment (Bulow et al., 2008), accounting for 7.1 %, 6.7 %, 13.0 %, 6.3 %, 13.0 %, and 20.0% in layers of ZB5, ZB10, ZB15, H5, H10, and H15, respectively.Cluster-E was detected in the sediment of location H with 12.5%, 17.4 %, and 10 % of the total number of clones in H5, H10, and H15, respectively.Clones in this cluster were related to species Dechloromonas aromatic.Cluster-F was related to species Aromatoleum aromaticum (80-92 % identity) and mainly appeared in ZB sediment with the largest percentage (40 %) for layer ZB10.Cluster-G and Cluster-H were only detected in deeper layer at location H, accounting for 4.3 % and 4.4 % in layer H10, 10.0 % and 5.0 % in layer H15, respectively.Sequences of Cluster-G were related to Magnetospirillum sp.(84 % identity), which was a Gram-negative α-proteobacterium.This is a group of facultative anaerobic bacteria related to iron oxidation (Matsunaga et al., 2005).Cluster-H was related with Herbaspirillum sp. and belonged to the family Burkholderiaceae.These bacteria were usually found in the rhizosphere of soil and sediment environments.They were considered to participate in nitrogen fixation (Jung et al., 2007).Cluster-I belonged to the family Candidatus Accumulibacter, which was ecologically significant because it was used to remove phosphorous from waste water (Fukushima et al., 2007).Cluster-I appeared in ZB5, ZB10, ZB15, H5, and H10, accounting for 14.3 %, 13.3 %, 4.4 %, 12.5 %, and 8.7 %, respectively.

Spatial distributions of denitrifying genes and the nirS-encoding bacterial assemblage
Spatial distributions of denitrifying genes and nirS-encoding bacterial assemblages might be influenced by nutrient conditions in the Pearl River sediment.The CCA analysis of denitrifying gene abundance and nirS-encoding bacterial assemblages in response to environmental variables confirmed the influence.In Fig. 2a, the first two CCA axes (CCA1 and CCA2) explained 73 % of the total variance in the denitrifying gene abundance and 83 % of the cumulative variance of the gene-environment relationship.Dissolved inorganic nitrogen had significant effects on distributions of denitrifying genes.The CCA analysis showed that CCA1 represented the vertical distributions of denitrifying genes.Nitrite reductase genes mainly assembled in 0-10 cm in the sediment and nosZ distributed in 5-15 cm, while a large number of narG and nrfA genes gathered below 15 cm.Distributions of denitrifying genes had obvious stratification in the Pearl River sediment (Fig. 2a).CCA2 represented the nutritional level of the sediment.Results show that TOC was not limited to the distribution of denitrifying genes (Fig. 2a).Concentrations of NH + 4 were significantly affecting nrfA distributions (P = 0.025), and nirS and nirK abundance was correlated with NO − 2 (P = 0.032).Distributions of narG were affected by the combined effects of DIN.However, nosZ was negatively related to various sediment nutritional factors, which indicated that oligotrophic conditions might promote the activity of nosZ (Fig. 2a).
In Fig. 2b, the first two CCA axes (CCA1 and CCA2) explained 71 % of the total variance in the nirS-encoding bacterium composition and 81 % of the cumulative variance of the bacteria-environment relationship.CCA1 and CCA2 distinguished the nirS-encoding bacterial assemblage of layer ZB15 from those of the other layers.TOC and NO − 2 contributed more to the bacteria-environment relationship than the other environmental factors (Fig. 2b).The distribution of Cluster-D was positively correlated with NH + 4 and Cluster-E was related to TOC and NO − 3 .However, the distributions of Cluster-F and Cluster-I were negatively correlated with the nutrient-rich conditions.Along the most important CCA1 axis, Cluster-D, -E, -G, and Cluster-H might correspond to the nutrient-rich conditions in the sediment, whereas Cluster-F and Cluster-I corresponded to the less eutrophic conditions (Fig. 2b).Various denitrifier communities were found among the different locations and sediment depths.

Discussion
The spatial distributions and concentrations of DIN in the sediment clearly showed the urban river characteristics of the Pearl River.Because of a large amount of domestic wastewater inputs, various pollution characteristics appeared at different locations.Concentrations of NH + 4 in the sediment of this river section were higher than those of other rivers, and the vertical variations of NO − 3 , which accumulated at deeper layers, were different from those in other river sediments (Tiquia et al., 2006;Laverman et al., 2010).Previous studies showed that denitrifying bacterial diversity was higher at deeper sediment layers, but denitrification rates of deeper sediments could be neglected due to the absence of oxidized NO − 3 and NO − 2 (Tiquia et al., 2006).However, NO − 3 concentrations did not decrease with the sediment depth in the Pearl River and the relative abundance of narG was higher at deeper sediment layers (Tables 1 and 3).These results indicated that NO − 3 reduction rates in the deeper sediments could not be neglected in the river region.Dramatically decrease of abundances of nirS and nirK below 15 cm did not cause NO − 2 accumulation, indicating that there were other NO − 2 reduction pathways in the deeper sediment.Relative abundance of nrfA increased with the sediment depth and DNRA might become more active in the deeper sediment.However, low C/N below 15 cm inhibited the DNRA process (Burgin www.biogeosciences  and Hamilton, 2007).Therefore other metabolism pathways, such as anammox that drives N 2 production in the marine and estuarine environments, might be important processes to consume NO − 2 in this river sediment (Kuypers et al., 2003;Dalsgaard et al., 2005).The N 2 generate efficiency of anammox and its impact on denitrification needs further studies.Our results showed that denitrification processes in different sediment depths were dominated by different NO − 3 and NO − 2 reduction pathways.Therefore, it is essential to study the vertical distribution of denitrification in the river sediment.
In this study, both nirK and nirS were detected in the Pearl River sediment.However, nirK abundance was much lower than nirS abundance in the sediment because nirK only prevails in conditionally oxygen-exposed environment (Desnues et al., 2007;Knapp et al., 2009).The enzymes encoded by nirS and nirK require different substrates.Particularly environmental conditions may alter the proportion of denitrifiers with nirK or nirS.Moderate levels of NO − 3 in the river sediment increase nirS diversity (Yan et al., 2003).The nirS gene is found to be more widespread in the bacterial communities in various sediments (Priemé et al., 2002;Liu et al., 2003;Throbäck et al., 2004;Tiquia et al., 2006;Oakley et al., 2007;Dang et al., 2009).Therefore, we used the nirS gene to characterize the diversity of denitrifier communities in the Pearl River sediment.Most of the nirS sequences in this sediment had a close match with those originally detected in estuarine and marine sediments as well as sludge (Nogales et al., 2002;Tsuneda et al., 2005;Osaka et al., 2006;Ruiz-Rueda et al., 2007;Spain et al., 2007;Bulow et al., 2008), suggesting that this river has both tidal and urban characteristics.When the river is affected by both domestic pollutions and irregularly tides, diversity and distributions of denitrifying bacteria should become more complicated.In comparable environments, similar denitrifying communities may develop despite of different geographical locations (Castro-Gonázlez et al., 2005;Falk et al., 2007).Sediments from locations ZB and H were use to study the diversity of denitrifier community according to the high abundance of nirS genes in these locations.Some of the nirS gene sequences were unique with the sampling locations, and several nirS gene sequences were novel in the Pearl River, as most of them had <70 % nucleic acid identity with those from cultivated strains and <90 % identity with those from environmental clones.Cluster-G and Cluster-H were only detected at location H, corresponding to the serious pollution created by the aquatic plants and metal release at this location.Bacteria of Cluster-E and Cluster-F, which utilize different aromatic compounds, appeared only in the sediments of locations H and ZB, respectively, due to different categories of polycyclic aromatic hydrocarbons (Chakraborty et al., 2005;Tamang et al., 2009).It was very likely that at least some of these bacteria were the contaminant-degrading species in the nirS communities.Existence of many novel bacteria was another reason for the lower abundance of narG compared to nirS in this environment (Table 1).Further research is needed to cultivate the novel denitrifying bacteria, to study their ecophysiology, and to understand their roles and mechanisms in N cycling and environmental bioremediation.In addition, more specific primers should been designed and verified experimentally.
In this study, besides the abundance of denitrifying genes, the composition of denitrifying bacterial communities based on nirS gene also showed vertical variations.Denitrifier communities were dominated by a few phylotypes in the surface layers and showed higher diversity in the deeper layers.On the other hand, unique denitrifying bacterial phylotypes appeared in the different sediment depths (Table 4 and Fig. 1).Denitrifying bacteria changed from facultative aerobic phylotypes to strictly anaerobic phylotypes attributable to the decrease of dissolved oxygen with the sediment depth.Bacteria of Cluster-I was facultative denitrifying bacteria, which preferred low dissolved oxygen conditions and usually accumulated in the upper sediments (Fukushima et al., 2007).Nitrification and denitrification processes can simultaneously occur in the surface sediment, thereby increasing the reduction rates of NO − 3 and NO − 2 (Rysgaard al., 1993;Scott et al., 2008).R. eutropha H16 of Cluster-B, which preferred non-halophilic habitats and did not require oxygen, appeared in the deeper sediment (Schwartz et al., 2003).With the capability to carry out DNRA processes, bacteria of Cluster-D did not exist in the sediment surface, but the bacterial proportion became remarkable with the sediment depth.All of these results suggested that the vertical distributions of denitrifying genes in this sediment had a similar stratified characteristic, and that transformation, competition, and cooperation of different denitrifier communities were determined by the depth-dependent sediment characteristics.Based on the stratified characteristic, the sediments could be divided generally into three layers.In our sediment, at 0-5 cm, synchronic nitrification and denitrification processes dominated in this low oxygen environment (Fan et al., 2006).At 5-15 cm, nitrite reductase gene (nirS and nirK) and nosZ were concentrated, potential denitrification ability was high, and the complete denitrification process (with the final product of N 2 ) accommodated this environment (Graham et al., 2010).Below 15 cm, a strictly anoxic condition limited nitrification, thus a large amount of NH + 4 accumulated and dissimilatory NO − 3 reducing pathways became obvious.Although the exact depth range of each layer may vary for different sampling locations, the stratification pattern of denitrification should be prevalent in different sediments (Fan et al., 2006;Tiquia et al., 2006).
Among the nutrient characteristics of the Pear River sediment, the TOC and NO − 2 had the most significant impact on the nirS-encoding bacterial community structure and spatial distributions (Fig. 2b).Organic carbon was the primary electron donor for the respiratory denitrifying bacteria (Burgin and Hamilton, 2007).Although the abundance of nir genes was not restricted by the TOC concentrations, the diversity of nirS genes was closely related to TOC in the sediment (Figs.2a, b).As a direct electron acceptor of denitrification, NO − 2 concentrations had great impact on the quantity of nirS and nirK.However, none of the denitrifier clusters was limited by NO − 2 concentrations directly (Figs. 2a,b), suggesting that distributions of denitrifying bacteria in the sediment were determined by combined effects of NO − 2 and other depth-dependent factors.Environmental factors that restrict the nosZ distribution also deserve to be studied because nosZ is the key functional gene for the complete denitrification.The CCA analysis showed that abundance of nosZ was higher at the least nutrient condition in the sediment.This result was contrary to that of Laverman et al. (2010), who considered that higher numbers of nosZ appeared in nutrientrich sediment and clean river might reduce the efficiency of N 2 production.Our result implied that the distribution of nosZ was not limited by concentrations of organic carbon or nitrogen but by the ratios of them.The uncertain conclusions about NO reduction and nosZ need for further studies.

Conclusions
A direct relationship was established between the depthdependent environmental characteristics and the distributions of various denitrifying genes as well as the structure of nirSencoding denitrifier community in the Pearl River sediment.It was found that the diversity of the denitrifier communities was high in the sediment because the river sediment was affected by both domestic sewage inputs and irregular tides.The sediment depth and nutrient conditions were the major factors to control the distributions and diversities of denitrifying genes.Denitrification stratification in the sediment could be classified based on the vertical distribution of denitrifying genes and DIN.Many novel nirS-encoding denitrifiers were found in this typical sediment environment.The physiology and ecological roles of these novel denitrifiers in the nitrogen transformation and contaminants degradation in the environment should be of great interests in future studies.

Fig. 2 .
Fig. 2 743 744 745 Fig. 2. Ordination plots of the canonical correspondence analysis (CCA) for the first two dimensions of CCA of the relationship between the distributions of (a) denitrifying genes and (b) nirS clusters vs. the nutrients in the Pearl River sediment.Correlations between environmental variables and CCA axes are represented by the length and angle of arrows (environmental factor vectors).

Table 1 .
Characteristics of sample locations in the Pearl River.

Table 2 .
Real time PCR primers used for the amplification of denitrifying genes.
performed.During the first 10 cycles, the annealing temperature was decreased by 0.5 • C every cycle, starting at 45 • C until it reached a touchdown at 40 • C. The additional 20 cycles were performed at an annealing temperature of 43 • C. The amplification products were analyzed by electrophoresis on 1 % (wt/vol) agarose gels (Genesnap, SYNGENE) followed by a 15 min stained with Genefinder (0.02 µL mL −1 ).PCR products were purified by agarose gel extraction and cloned into a PMD19-T vector.Positive recombinant clones were identified by colony PCR and the PCR products were cleaned with ExoSap treatment and sequenced on an automated ABI3730 automatic sequencer (Applied Biosystems, USA).

Table 3 .
Copy numbers of denitrifying genes and bacterial 16S rRNA genes in the Pearl River sediment.

Table 4 .
Biodiversity and predicted richness of nirS sequences recovered from six layers in the Pearl River sediment.