The differentiation of iron-reducing bacterial community and iron reduction activity between riverine and marine sediments in the Yellow River estuary

Iron(Ⅲ)-reducing bacteria (IRB) play a pivotal biogeochemical role in the iron redox cycle, which is coupled with anaerobic organic matter degradation (Lovley, 2006) and ammonium oxidation (Li et al., 2015). IRB are widely distributed in many freshwater and marine environments, as well as across a broad range of chemical and physical conditions. However, different communities may play different roles in different habitats. Studies have shown that IRB are usually restricted to specific habitats; for instance, the well-known clade of IRB in the genus Geobacter comprises a larger proportion of iron-reducing populations in river sediments (Kim et al., 2012; Ding et al., 2015; Zheng et al., 2015) whereas Shewanella, which is mostly found in deep-sea environments (Bowman et al., 1997; Stapleton et al., 2005; Gao et al., 2006; Roh et al., 2006). The Fe(Ⅲ) reducers, together with the Fe(Ⅱ) oxidizers, contribute to Fe redox cycling in hypersaline aquatic environments (Halobaculum gomorrense, Desulfosporosinus lacus, etc.) (Emmerich et al., 2012), in a groundwater seep environment (Rhodoferax, Aeromonas, etc.) (Roden et al., 2012) and marine sediments (Shewanella, Deferribacter, Geoglobus, etc.) (Laufer et al., 2015). However, the diversity and community composition of marine and riverine IRB remain uncharacterized, particularly in areas where rivers and sea interact.

The estuarine-marine environment is an area where freshwater and marine waters mix (i.e., a brackish water zone). Here, there are steep gradients in iron, nitrogen, sulfur, oxygen, and other elements (MacDonald et al., 2014). The variable river flow may result in these environments having a high variability in the concentrations of electron acceptors and electron donors (Zhang et al., 2017), which are the main controlling factors of IRB community structure. These gradients, which are associated with water movement, may affect the iron-reducing bacterial community structure and their biogeochemical processes.

The Yellow River estuary has one of the most active areas of river-sea interaction in the world (Li et al., 2009). The Yellow River has a very high concentration of suspended particulate matter, that contains nitrogen, iron, and other elements. It is also has the second largest sediment load flux to the ocean of any river (Poulton and Raiswell, 2002; Pan et al., 2013). Previous studies (Liu et al., 2003; Qiao et al., 2007; Gong et al., 2015b; Sheng et al., 2015; Zhang et al., 2017) have shown that the concentrations of ferrous iron (Fe(Ⅱ)), ferric iron (Fe(Ⅲ)), total iron (Fe) and ammonium (NH₄⁺-N) were higher in the coastal marine sediment than in the Yellow River sediment. Whereas the concentration of nitrate (NO₃^--N) was higher in the riverine sediment than in coastal marine sediment. Patterns of IRB abundance and diversity can indicate their contribution to the biogeochemical cycles, while changes in community structures are associated with spatial variations in nutrients (i.e., NO₃^--N, NH₄⁺-N, Fe(Ⅱ), Fe(Ⅲ)) in different habitats. However, there has been less research (Mcbeth et al., 2013) on how shifts in composition and abundance of IRB with ecosystems, i.e. from rivers through to coastal and oceanic environments, relate to environmental factors.

This study aimed (i) to determine the diversity, abundance, and spatial distribution of putative IRB in the Yellow River estuary and marine coastal sediments from typical coastal areas affected by river-sea interactions, and (ii) to evaluate the influence of biogeochemical variables on the relative abundance and biogeographic patterns of putative IRB in sediments from 6 sampling stations and 18 samples in the Yellow River and Bohai Sea (Fig. 1).

Figure 1.  Locations and sample sites in the Yellow River estuary and the adjacent Bohai Sea (China)

All samples (estimated at the 97% similarity level) were normalized to the same sequencing depth (4999 randomly selected reads per sample for all sites). The OTU richness of putative IRB (identified according to references and shown in Fig. 2) showed clear variations between the samples. The OTU richness of the IRB was calculated from all putative IRB (Fig. 2). Higher values of community richness were present in the Yellow River ecosystem compared to the marine ecosystem (Fig. S1). No significant difference was found between the three Yellow River sites; however, a significant difference was found between the three Bohai Sea sites. The richness in the samples from P1 and P2 ranged from four to eight. These values were considerably lower than those in the samples from BHB02, which ranged from 29-37 (Table S1). Of the 134 OTUs observed for putative IRB in all samples, 23 OTUs (17.2%) were shared between the two aquatic ecosystems (Fig. S2).

Figure 2.  Relative abundance (n=3) of putative iron-reducing bacteria in bacterial communities at the genus level at 6 sites

The relative abundance of putative IRB at the genus level is summarized in Fig. 2. These show that significant changes in community composition occurred with changes in the redox conditions in the riverine to marine ecosystem gradient. Twenty genera of IRB were present in the samples from YR1 and YR2, 22 genera were present in the samples from the mixing zone (YR3 and BHB02), and 14 genera were present in the samples from P1 and P2. Pseudomonas (Naganuma et al., 2006) (1.70% of total reads), Thiobacillus (Sand, 1989) (0.81%), Geobacter (Coates et al., 1996) (0.19%), Rhodoferax (Kim et al., 2012) (0.11%), and Clostridium (Xu et al., 2014) (0.11%) were the most abundant genera in the sediments from the Yellow River estuary, while, Vibrio (Jones et al., 1984) (0.76%), Shewanella (Bowman et al., 1997) (0.27%), Thiobacillus (0.14%), Clostridium (0.06%), and Bacillus (Kanso et al., 2002) (0.05%) were most abundant in the sediments from the Bohai Sea ecosystem.

A meta-analysis of data on the relative abundance of putative IRB in the sediments from Yellow River and Bohai Sea ecosystems was conducted to identify the relationship between specific genera and the type of sample. The analysis revealed that 15 bacterial groups of putative IRB, such as Geothrix (Lovley et al., 1999), Pseudomonas, and Geobacter were enriched in riverine sediments, while, five genera, namely Shewanella, Vibrio, Pelosinus, Desulfuromonas, and Desulfobacter were overrepresented in Bohai Sea relative to the Yellow River (Fig. 3).

Figure 3.  Response Ratio (RR) of iron-reducing bacteria in 2 contrasting systems, the Yellow River (YR), and Bohai Sea (BH). The vertical line is drawn at RR=0

To evaluate the potential effects of environmental factors on putative IRB, correlations between the putative IRB community structures and environmental factors were determined by RDA. The spatial patterns in physicochemical characteristics across the sites changed markedly from the riverine ecosystem to the marine ecosystem (Table S2). From the variance inflation, tested by 999 Monte-Carlo permutations, three significant factors (P < 0.05) were identified: TN, NH₄⁺-N and NO₃^--N (Fig. 4). In the Yellow River ecosystem, the IRB were most influenced by the NO₃^--N concentration. In the Bohai Sea ecosystem, IRB were positively influenced by the concentration of TN and NH₄⁺-N.

Figure 4.  Redundancy analysis (RDA) compares the putative iron-reducing bacterial community structures in sediment samples and environmental factors (arrows), including total N (TN), ammonium (NH₄⁺-N), and nitrate (NO₃^--N)

To identify active iron-reducing bacteria, bacterial communities were cultured in iron(Ⅲ)-reducing enrichment media. Different iron-reducing bacterial communities were enriched in the riverine and marine cultures. Six genera of IRB were detected in both riverine and marine enrichment cultures: Desulfotomaculum (Dalla et al., 2014), Geothermobacter (Emerson, 2009), Clostridium, Desulfosporosinus (Bertel et al., 2011), Pseudomonas, and Bacillus (Fig. 5). Deferrisoma spp. (Slobodkina et al., 2012), which were not detected in the field samples, were most abundant in the enrichment samples taken from P1 and P2 (31.13% and 43.15% of total bacteria, respectively), and were also abundant in samples from YR1 and YR2. The highest relative abundance (14.99% of total bacteria) of the genus Desulfotomaculum was obtained from the enrichment samples from BHB02. Deferrisoma, Desulfotomaculum, and Geothermobacter were the dominant genera in iron(Ⅲ)-reducing enrichment cultures from riverine sediments.

Figure 5.  Relative abundance of iron-reducing bacteria in iron(Ⅲ)-reducing enrichment cultures with sediments at 6 sites. Note: e denotes samples in enrichment cultures

To investigate the Fe(Ⅲ) reduction activity of sediment samples from different sites, the Fe(Ⅱ) concentration in iron(Ⅲ)-reducing enrichment cultures was measured. Fe(Ⅱ) was produced in all enrichment samples, with amorphous Fe(Ⅲ) oxides (AmoFe) as the electron acceptor and acetate as the electron donor (Fig. 6a). The Fe(Ⅱ) concentration of the sediment samples from the Yellow River markedly increased and reached 29.96 ± 0.8-30.43 ± 0.8 mmol L^-1 at Day 30. The sample eYR1 (e: enrichment) produced the highest Fe(Ⅱ) concentration (34.08 ± 0.3 mmol L^-1) at Day 40. Like the relative abundance and diversity of the in situ samples from BHB02, the Fe(Ⅱ) concentration of the eBHB02 sample, reached 32.33 ± 0.52 mmol L^-1 by Day 40; it increased more rapidly than either the eP1 and eP2 samples. The Fe(Ⅱ) concentration of the eP1 and eP2 samples by Day 40 were 19.05 ± 1.40 mmol L^-1 and 23.22 ± 0.38 mmol L^-1, respectively. The control cultures, i.e. autoclaved sediments (CK samples), did not show Fe(Ⅲ) reduction, showing it was a biogeochemical process.

Figure 6.  Fe(Ⅱ) production and acetate consumption of enrichment cultures. Time courses of Fe(Ⅱ) production (a) and acetate consumption (b) in enrichment cultures with 33 mmol L^-1 of acetate in the presence of 66 mmol L^-1 of Fe(Ⅲ) oxides. Sterile sediment sample (CK) treated with Fe(Ⅲ) oxides was used as the negative control. Data were presented in triplicate, and standard deviation was shown for each data point

The decrease of acetate was more rapid in the enrichment samples from the Yellow River than in the samples from the Bohai Sea. The rate of decrease in acetate corresponded to the rate of increase in ferrous iron during the enrichment period (Fig. 6b).

The sequence data related to putative IRB at the genus level was estimated in accordance with published reviews, acknowledging that this estimation method might have led to an overestimation of OTUs for IRB because of the limitation of sequence length from the MiSeq platform and the lack of universal functional gene markers for IRB. Furthermore, limited knowledge of IRB communities may cause an underestimation of the diversity of IRB. The results presented here show that the relative abundance and diversity of IRB in sediments varies with the aquatic ecosystem. More OTUs of IRB were generally found in riverine sediments than in marine sediments. Greater abundances were also observed in the samples obtained from YR3 and BHB02.

Illumina high-throughput sequencing verified the occurrence of substantial changes in the community structure of putative IRB in sediments in the transect from the Yellow River to the sea. The average relative abundance and diversity in the river system were much greater than those in the marine system; this is in agreement with results found in the sediments from other river and coastal marine zones of Bohai Sea (data not shown). This could potentially be attributed to the high concentration of sulfate in marine sediments. Iron oxides are the most important anaerobic electron acceptors in freshwater sediments (Roden and Wetzel, 1996), while sulfate reduction is the most important anaerobic pathway for organic matter degradation in marine sediments (Henrichs and Reeburgh, 1987).

At the genus level, the dominant communities of putative iron-reducing bacteria in the riverine sediments were distinct from those in the marine sediments. The high relative abundances of Pseudomonas, Thiobacillus, Geobacter, Rhodoferax, and Clostridium are thus proposed as the most important mediators of iron reduction in the riverine sediments of the Yellow River estuary; this is consistent with results obtained in previous studies of freshwater systems (Weber et al., 2006b; Haaijer et al., 2012; Kim et al., 2012; Peng et al., 2016).

The dominant IRB genera in marine sediments were Vibrio, Shewanella, and Thiobacillus. As previously discussed, these genera are commonly detected in the marine system (Emerson et al., 2010; Esther et al., 2015), suggesting their participation in the biogeochemical process of iron reduction in coastal marine environments. The abundance of Shewanella in marine sediments here is consistent with the data from the Black Sea (Nealson et al., 1991; Perry et al., 1993).

Among of the most significant changes found was the increase in the relative abundance of putative iron-reducing bacteria in the samples from YR3 and BHB02, which are located near to the estuary and markedly affected by the river-sea interaction. This is consistent with estuarine sediments providing an intermediate salinity zone, where freshwater and seawater iron-reducing bacteria mingled (Mcbeth et al., 2013), and suggests that IRB communities can be affected by river-sea interactions.

In the present study, the RDA analysis showed that concentrations of TN and NH₄⁺-N significantly and positively influenced IRB communities in coastal marine sediments. By comparison, the IRB community of riverine sediments was closely related to the concentration of NO₃^--N. This implies that sediment NH₄⁺-N and NO₃^--N contents are important parameters in determining the IRB in riverine and coastal marine sediments. This relationship may be attributable to the coupled redox cycling of iron and nitrogen. For example, some members of the dominant genera Pseudomonas and Thiobacillus (e.g., Pseudomonas stutzeri, Thiobacillus ferrooxidans) are also known to be capable of nitrate-dependent Fe(Ⅱ) oxidation (Sugio et al., 1994; Straub et al., 1996; Coby et al., 2011). Some Geobacter species are also capable of nitrate-reducing Fe(Ⅱ) oxidation with the reduction of NO₃^- to NH₄⁺ (Weber et al., 2006b; Coby et al., 2011). Shewanella can also reduce nitrate (Gao et al., 2009). In addition, ammonium is a major end product of anaerobic nitrate-dependent Fe(Ⅱ) oxidation and might participate in Fe(Ⅲ) reduction as an electron donor.

To increasethe in situ availability of IRB and its differentiation between the two ecosystems in culture, it is necessary to add AmoFe in relatively large amounts to the anaerobic iron(Ⅲ)-reducing bacteria, although this experimental result may lead to a discrepancy arising from the field conditions. A total of 16 genera of well-known iron-reducing bacteria were identified in the iron(Ⅲ)-reducing enrichment cultures; these corresponded with the putative iron-reducing bacterial communities found in the in situ samples. Notably, Deferrisoma spp. that were not detected in the in situ samples exhibited the highest relative abundance in iron(Ⅲ)-reducing enrichment cultures of sediments from P1 and P2. Thus, some iron-reducing populations with low abundance that couldn't be detected in the field samples, could be stimulated by the presence of AmoFe in enrichment cultures. However, the Fe(Ⅱ) production of riverine sediments was higher than that of marine sediments (but not BHB02 from the mixing zone). Thus, the samples from river and BHB02, which have a greater IRB diversity, exhibited increased activity of Fe(Ⅲ) reduction in enrichment cultures. Deferrisoma spp., which have been previously observed in shallow-water or deep-sea hydrothermal vents (Slobodkina et al., 2012; Pérez-Rodríguez et al., 2016), had the highest relative IRB abundance in riverine sediments enrichment cultures (YR1 and YR2). Ethanol is an important intermediate metabolite of acetate and can also be used as an electron donor for IRB. Thus, Fe(Ⅱ) is still produced even when acetate is exhausted in the enrichment cultures (Fig. 6b).

16S rRNA gene analysis showed the differences in iron-reducing bacterial community compositions and indicated that the diversity and distribution of IRB in sediments, from typical coastal areas with river-sea interaction, were mainly determined by hydrology and habitat type. Shifts in the diversity of iron-reducing bacteria indicated a novel distribution pattern in the Yellow River estuary and the adjacent Bohai Sea. High relative abundances and diversities of iron-reducing bacteria were observed at the coastal transitional sites. The results of the enrichment culture studies indicate that IRB in marine sediment can be stimulated by Fe(Ⅲ) reduction activity in the presence of AmoFe. In addition, TN, NH₄⁺-N, and NO₃^--N concentrations contributed to variations in iron-reducing bacterial communities, demonstrating the relationship between iron redox biogeochemical cycling and nitrogen cycling. The results presented here show that there is a variation in the structure of iron-reducing bacterial communities and this indicates the presence of a biogeographic pattern in coastal ecosystems.

The Yellow River sediment samples were collected in August 2014 from the Yellow River to the Bohai Sea (37-45'N-38-5'N, 118-48'E-119-30'E) (Fig. 1), which is located in the most active river-sea interaction coastal zone. Six sites were selected based on the distance from the river mouth, including 3 sites (YR1, YR2, and YR3) in the Yellow River (YR) and 3 other sites (BHB02, P1, and P2) in the coastal marine system (Bohai Sea, BH). YR3 is near the river mouth, and BHB02 is a transitional site. Each site contained 3 duplicate plots, and 18 (6 sites×3 replicates) sediment samples were collected.

The riverine and marine sediment and pore water samples were collected using a grab sampler, and the surface-layer subsamples (top: 0-5 cm) were sub-cored with a custom-made corer (inner diameter: 1.5 cm) and then homogenized. Core material was transferred into cryovials and then stored immediately in liquid nitrogen for DNA extraction. Sediment subsamples were placed in sealed containers and archived at 4℃ for physicochemical analysis within 7 d.

Salinity and pH in the overlying water at each site were measured with an electronic probe (Hydrolab MS5, HACH, USA). Nitrate (NO₃^--N), ammonium (NH₄⁺-N), total organic carbon (TOC) and total nitrogen (TN) in sediment samples, and the concentration of sulfate (SO₄^2-) in the sediment pore waters were determined as described in Zhang et al., (2017). Dissolved ferrous iron (Fe(Ⅱ)) was measured using a ferrozine-based assay (Stookey, 1970; Lovley and Phillips, 1987). Total reactive hydroxylamine-reducible iron (Fe(Ⅲ)) was extracted from the sediments and then measured in accordance with our previous report (Zheng et al., 2015). Low-frequency magnetic susceptibility (χ_LF) was measured with a Bartington MS2B sensor. The concentration of acetate was measured by high-performance liquid chromatography, using a1260 Infinity HPLC (Agilent Technologies, USA) with a Hi-plex H column equipped with a refractive index detector, using 5 mmol L^-1 H₂SO₄ as the eluent.

Iron-reducing bacteria in riverine sediments and marine sediments were enriched in both the freshwater and seawater enrichment medium. The basic components of the freshwater and seawater enrichments were prepared according to Lovley and Phillips (1986). Amorphous Fe(Ⅲ) oxides at a final concentration of 66 mmol L^-1 and acetate at a final concentration of 33 mmol L^-1 were added into the media, with the former as the electron acceptor and the latter as the electron donor (Lovley and Phillips, 1986; Zheng et al., 2015). Amorphous Fe(Ⅲ) oxides was formed by neutralizing a 0.4 mol L^-1 solution of FeCl₃ to a pH of 7 with NaOH (Lovley and Phillips, 1986).

RNA extraction of enrichment cultures and cDNA synthesis were conducted as described in a previous study (Zheng et al., 2015). The V4-V5 hypervariable regions of 16S rRNA were amplified using the universal primer set 519f (5'-CAGCMGCCGCGGTAATWC-3') and 907r (5'-CCGTCAATTCMTTTRAGTTT-3') (Feng et al., 2015) for bacteria.

Total DNA was extracted from 0.5 g of sediment by using a FastDNA SPIN Kit for soil (MP Biomedicals, Santa Ana, CA, USA) in accordance with the manufacturer's instructions. PCR amplification was conducted using the universal primer set 519f/907r targeting the V4-V5 hypervariable regions of bacterial 16S rRNA. The purified PCR products with different barcodes were normalized in equimolar amounts and then prepared using TruSeq™ DNA Sample Prep LT Kit and sequenced using MiSeq Reagent Kit (500-cycles-PE) following the protocols provided by the manufacturer.

Raw deep sequencing data were processed using the Quantitative Insights Into Microbial Ecology (http://www.qiime.org, QIIME version 1.7.0; (Caporaso et al., 2010)) with the default parameters unless otherwise noted. After the low-quality sequences and chimeric sequences were removed, qualified sequences were clustered into operational taxonomic units (OTUs) at the 97% sequence identity level, and the most abundant sequence from each OTU was chosen as a representative sequence for that OTU. Taxonomic classification of each OTU was assigned using the Ribosomal Database Project classifier (Li et al., 2016).

16S rRNA gene-based sequences related to putative IRB at the genus level were selected according to published reviews (Lovley et al., 2004; Lovley, 2006; Weber et al., 2006a; Emerson et al., 2010; Esther et al., 2015) owing to the lack of universal functional gene markers for IRB.

To assess the relationship between the iron-reducing bacterial communities and the environmental factors, redundancy analysis (RDA) was conducted using the CANOCO 4.5 software. A Monte Carlo permutation test (999 random unrestricted permutations) was performed to test the statistical significance of the environmental variables. Meta-analysis of data on the relative abundance of iron-cycling bacteria in riverine and marine sediments was conducted in accordance with previous study (Luo et al., 2006).

Raw sequence reads of bacterial 16S RNA and 16S rRNA genes have been submitted to the Sequence Read Archive (SRA) with accession number PRJNA342373.

We would like to thank Prof. Jianhui Tang for sharing marine sediment samples. This research was supported by the National Natural Science Foundation of China (No. 91751112, 41807325 and 41573071); the senior user project of RV KEXUE (No. KEXUE2018G01) and the Key Research Project of Frontier Science (No. QYZDJ-SSW-DQC015) of Chinese Academy of Sciences; the Natural Science Foundation (No. JQ201608 and ZR2018MD011) and the Young Taishan Scholars Program (No. tsqn20161054) of Shandong Province.

HZ, FL and SZ contributed to the presented idea and design. HZ implemented the computational and statistic analyses and took the lead in writing the manuscript. XZ assisted with data analysis. FL and SZ supervised the findings of this work. All authors provided critical feedback and helped to conduct the research, analysis and manuscript.

Conflict of interest The authors declare that they have no conflict of interest.

Animal and Human Rights Statement This article does not contain any studies with human participants or animals performed by any of the authors.