Bacterioplankton are a crucial part of the marine food web and mediate major biogeochemical cycling in the ocean (Azam and Malfatti, 2007). They play an important role in organic matter decomposition (Herndl and Reinthaler, 2013), C and N fixation (Hügler and Sievert, 2010; Turk-Kubo et al., 2014), nitrification (Van Kessel et al., 2015), and denitrification (Ward et al., 2009). Microbial function in ecological and biogeochemical processes depends on the activity or behavior associated with their functional genes (Robinson et al., 2010). In in situ environments, the function of microbes is regulated by gene transcription as well as environmental variables. Meanwhile, the presence/absence and diversity of microbial functional genes suggest an associated functional potential in the ecosystem. Therefore, understanding the diversity of functional genes and uncovering the microbial functional potential are important for linking the microbial community to ecological and biogeochemical processes. In turn, the microbial community is inevitably influenced by the environment it inhabits. Some researchers have found associations of microbial communities with strong anthropogenic activity and natural gradients in complex coastal areas (Nogales et al., 2011; Ngugi et al., 2012). The close relationship between phylogenetic diversity of the microbial community and environmental factors such as temperature, salinity, and nutrients has been demonstrated in many marine ecosystems such as shelf and coastal areas (Baltar et al., 2015; Tinta et al., 2015; Lucas et al., 2016). Changes in environmental properties will influence microbial functional traits and consequently ecosystem function, which has been demonstrated in soil and marine environments (Thompson et al., 2016; Zhang et al., 2016). However, little is known about the functional gene composition and metabolic potential of bacterioplankton in estuaries areas, environmentally complex but ecologically important ecosystems.

The ECS is a marginal sea in the northwest Pacific Ocean. The KC flows along its eastern edge and brings its high-temperature water from the western equatorial Pacific. The TWC flows from the south along the eastern side of the sea, while the Yellow Sea Coastal Water flows from the north along the northwest coast of the sea (Tsai et al., 2010). The ECS is greatly influenced by the Changjiang River (YZR), which empties into the ECS from the west with decreasing salinity and conveys high levels of nutrients (Zhang et al., 2007), and is also influenced by oligotrophic, high salinity oceanic waters from the east (Jiao et al., 2005). Depending on the hydrographic stages of the YZR, nutrient and chlorophyll a concentrations generally decrease seaward from estuarine and coastal areas (Zhang et al., 2007). The hydrographic complexity of the ECS and the concomitant heterogeneous temperature, nutrient, and salt concentrations result in dynamic environmental conditions, which have profound impacts on the microbial communities in this environment (Schultz et al., 2003; Kirchman et al., 2005; Fuhrman et al., 2006).

Studies have indicated that the changes in temperature, salinity, and nutrients from the YZR discharge and KC result in different distribution patterns of Prochlorococcus, Synechococcus, and heterotrophic bacteria in the ECS (Jiao et al., 2005), while freshwater influx and oligotrophic conditions impact the phylogenetic diversity and structure of bacterial and archaeal communities (Zeng et al., 2006; Hu et al., 2011, 2013). However, little is known about how the changes in environmental parameters affect the functional potential of the ECS microbial communities.

In this study, we analyzed microbial communities from different depths of the ECS using a functional gene microarray (GeoChip) to characterize differences in the functional potential based on location and depth. Specifically, we intended to address the following questions: (i) are there differences in the functional potential of the microbial community among the different water masses? (ii) Which functional processes change among water masses? And (iii) are there correlations between the functional potential of the microbial community and the properties of the water masses? Significant differences were observed in the overall functional diversity and in specific functional categories, such as carbon fixation, carbon degradation, and other key metabolic pathways, which were likely driven by temperature, salinity, and Chl a differences. This study provides new insights into our understanding of the functional diversity, structure, and metabolic potential of microbial communities in the ECS ecosystems.

The GeoChip results indicated the complexity of the functional gene structure of the microbial community in the ECS. A previous study observed the effect of the currents, including TWC, KC, and YZR input, on the distribution of picoplankton in the ECS (Jiao et al., 2005). However, the current study did not observe a distinct influence from these currents on the microbial community functional potential in the surface water. Temperature is known to be an important factor affecting chemical and biochemical processes, and subsequently the diversity of microbial communities in marine ecosystems (Pomeroy and Wiebe, 2001). Higher temperature always facilitates a richer microbial community structure on a geographical scale (Pommier et al., 2007; Fuhrman et al., 2008). Previous studies have also demonstrated that salinity is a crucial determinant for microbial communities (Lozupone and Knight, 2007), and that the microbial phylogenetic distribution significantly varies along a salinity gradient in coastal waters (Campbell and Kirchman, 2013). The low temperature and salinity association of G1 resulted from the YZR input, while G2 with high temperature and salinity was strongly influenced by the TWC, and the Kuroshio-branched water intruded into the ECS shelf via the bottom layer (Lian et al., 2016). The temperature and salinity were spatially similar near the seafloor (Wang and Oey, 2016), suggesting a more stable environment in the deep seawater compared with the surface water. The positive correlation between salinity and GeoChip data suggested that, similar to the phylogenetic diversity, the functional gene diversity of the microbial communities increased as the salinity increased from coastal to open ocean. In addition, the abundance of heterotrophic bacteria in the surface and deep seawater did not vary much (Jiao et al., 2005). Together, temperature and salinity, which are determined by the water mass, showed a considerable influence on the functional gene structure of water microbial communities in the ECS.

Additionally, our data indicates that the functional potential of water microbial communities has a close relationship with primary production. Two high-Chl a areas were observed in our study, one was near the mouth of the YZR, another was offshore in the northeast, which was mostly consistent with previous observations (Guo et al., 2014). It is likely that high primary production reduced the functional gene diversity of microbial communities in the surface seawater. The decreased diversity of microbial communities caused by increased dissolved organic matter with the bloom has been demonstrated in several studies (Larsen et al., 2004; Landa et al., 2016). Here, we found that the functional gene evenness based on all detected genes in the bottom samples showed a significantly positive correlation with Chl a in the bottom seawater. Thus, the high primary production indicated by high Chl a might benefit the functions of microbial communities with low abundance in the bottom seawater. In contrast, there was no significant correlation between Chl a and functional gene evenness in the surface seawater samples (Pearson correlation P-value of J and Si, 0.098 and 0.085 separately). This different response to Chl a between surface and bottom microbial communities could be due to the heterogeneity of the microenvironment and composition of microbial communities in the bottom water compared with the surface water. The distinct structure of microbial communities between surface and bottom seawater have been demonstrated in various coastal areas, such as Pearl River estuarine (Liu et al., 2015), South China Sea (Zhang et al., 2014b), and Oregon and Washington coastal (Fortunato et al., 2012). Here, the greater degradation potential of the relatively recalcitrant C and aromatics was observed in bottom samples, especially in G4, which had high Chl a (Figure 3). Furthermore, higher proportion of bacterial groups that utilized the high-molecular-mass substrates such as Gammaproteobacteria, Cytophaga, and Flavobacteria (Cottrell and Kirchman, 2000; McCarren et al., 2010; Sosa et al., 2015) were often observed in the bottom seawater (Zhang et al., 2014b; Liu et al., 2015). Therefore, the increased functional gene evenness of microbial communities might be associated with the microbes able to utilize the high-molecular-mass substrates, which may be stimulated by primary production.

An eddy usually generated by the detachment from the YZR diluted water southwest of Jeju Island (∼125°E, 32°N), and the stratification was completely destroyed in the entire water column (Moon et al., 2010). In the current study, low temperature and salinity were observed at three sites (DH1-6, DH2-6, and DH3-6) at the northeast of the ECS near Jeju Island (Figure 1B), which was consistent with the detachment demonstrated previously (Moon et al., 2010). The resultant upwelling drew nutrients from the deep seawater and sediment, consequently increasing the amount of Chl a. Consistently, a high Chl a concentration was observed in this area in the present study (Figure 1B). Comparing the surface samples with bottom samples at these three sites, DH3-6 with low salinity showed a similar functional gene structure of water microbial communities at both depths (Supplementary Figure S4), which suggested a vertical mixing in this area. Overall, Chl a, the distribution of which was associated with the YZR discharge that caused the eddy, had a crucial impact on the functional gene diversity and evenness of the microbial community.

Previous studies of the ECS found that the phylogenetic diversity in the bottom bacterial community was higher than that in the surface bacterial community, especially along the outer edge of the shelf (Dong et al., 2014), although there was no significant difference in bacterial abundance between the surface and bottom seawater (Jiao et al., 2005). In contrast, we found the higher functional gene diversity in the surface samples than that in the bottom samples (Table 2). These results, however, suggested that a higher phylogenetic diversity did not accompany higher functional gene diversity. The inconsistency between the functional and phylogenetic diversities might suggest the functional redundancy of the microbial community within the ecosystem. According to the functional redundancy theory, different species in similar niches might play similar functional roles in biogeochemical cycling, and overlapping niches may increase the functional redundancy of the ecosystem. Indeed, another study showed that the phylogenetic diversity only partially correlated with functional diversity (Flynn et al., 2011). Furthermore, stable ecosystems generally have lower diversity (Wohl et al., 2004), which may in turn indicate a low functional diversity. The complex hydrography in the ECS area, where the YZR, KC, and TWC provide an influx of various organic matter, could disturb the stability of the ecosystem (Tsai et al., 2010; Yu et al., 2012; Dong et al., 2014). As such, surface bacterioplankton communities might have developed diverse functions to fully utilize the available heterogeneous organic matter and nutrients. Compared with the complex surface seawater environment, the bottom environment is more stable (Jiao et al., 2002, 2005). Despite the relatively higher concentration of nutrients in the bottom seawater (Yu et al., 2012), the relatively low concentration of labile substrate for bacterioplankton might reduce the functional gene diversity. In addition, the different copy numbers of some functional genes in bacterial genomes, such as the amo gene in different ammonia-oxidizing bacteria (Hommes et al., 2001), might also contribute to this inconsistency.

In the ECS marginal shelf, a large amount of labile and recalcitrant organic compounds is discharged by YZR runoff all year round, especially during the summer (Zhang et al., 2003). In addition, phytodetritus and zooplankton fecal pellets provide another source of organic matter in this system (Bertilsson and Jones, 2003; Polz et al., 2006). Hence, the vertical distribution of organic matter may influence the metabolic preference of microbes in the ESC. The genes involved in C degradation showed different distributions in microbial communities from surface water and bottom water (Figures 2B, 3). Since cellulose and chitin are more recalcitrant than starch (You et al., 2012; Hamre et al., 2014), these results indicate that surface bacterioplankton have a higher metabolic potential for utilizing relatively labile organic matter, while bottom bacterioplankton have a higher potential for degradation of relatively recalcitrant organic matter. The degradation of labile organic matter preferentially occurs in the surface seawater where it is generated or transferred, while the recalcitrant organic matter settles to the bottom and is buried in the sediment (Hulthe et al., 1998). Besides, the mixing along the ECS coast would introduce more of the aged sedimentary organic matter into the water column (Hulthe et al., 1998), where the bottom microbial communities are able to utilize it. Additionally, relatively aerobic conditions near the bottom would further enhance the degradation of the recalcitrant organic matter (Hulthe et al., 1998). A higher proportion of genes involved in organic remediation was also observed in the microbial communities of the bottom seawater (Figure 2A). Genes involved in aromatic metabolism (such as genes for metabolism of aromatic carboxylic acid and chlorinated aromatics) were the major contributors to this difference. Aromatics are relatively resistant to degradation, as cleavage of such a ring structure requires an energy-consuming process for bacteria, especially under anoxic conditions (Ismail and Gescher, 2012). However, high nutrient concentrations promote the growth and energy generation of bacteria. Therefore, the higher level of nutrients such as nitrate, ammonia, and dissolved silicate (Sharp, 1991; Zhang et al., 2007) in the bottom seawater, compared with surface seawaters, might be beneficial for bacteria to degrade these recalcitrant/semi-labile compounds (Zaidi and Imam, 1999; Carlson et al., 2002; Beolchini et al., 2010). Therefore, the C-degradation potential of the microbial community is closely associated with the composition of available organic compounds, which is in turn dependent on the depth and the freshwater discharged from the YZR.

Microarray-based technology is reasonably quantitative and has a high hybridization specificity for generating results that can be species/strain specific (Tiquia et al., 2004; He et al., 2010). A total of 312 gene families are targeted in the GeoChip 4, such as C cycling (41), N cycling (17), organic remediation (184), P utilization (3), and sulfur metabolism (6) (Tu et al., 2014). Meanwhile, in the GeoChip 4, ∼22% of probes are sequence-specific and the other ∼78% of probes are group-specific, which covers 141,995 coding sequences (Tu et al., 2014). Furthermore, the GeoChip 4 targets 4,332 bacterial strains, 188 archaeal strains, 420 eukaryotic strains, and 273 bacteriophage (Tu et al., 2014). It is available for ecosystem-level studies owing to the capability of simultaneously identifying and quantifying many microbial functional genes/pathways (Zhou et al., 2015). A study demonstrated the feasibility of using the GeoChip in a spatial ecological investigation of the marine microbial community (Yergeau et al., 2007). Since it has many probes from different functional genes, the GeoChip used in this study could detect, reasonably well, some functional genes that have been studied intensively, although continuous improvement with higher gene coverage is always needed (Zhou et al., 2008). For example, previous quantitative-PCR results indicated that the abundance of amoA and nirS was higher in the bottom than in the surface water (Zhang et al., 2014a). However, we did not observe a significant difference in the proportion of these gene groups with our GeoChip analysis. These results might associate with the different primer/probe coverage between these two methods (Braker et al., 1998; Francis et al., 2005; He et al., 2010; Huang et al., 2011; Zhang et al., 2014a). In addition, the possible underestimation of functional diversity by microarray-based hybridization should be noted if the probes on the arrays do not well-represent the diversity of the genes of interest in a given microbial community (Zhou et al., 2008). In contrast, a metagenomic approach could uncover more information from rare species, which are also crucial for the functioning of the ecosystem. Despite of the advantages and disadvantages of each approach, previous study found the consistent results of microbial community using between metagenomic and GeoChip approach in soil environment responded to climate change (Xue et al., 2016), which indicated the complementation of these approaches. In addition, the local environmental factors not measured in our study, such as the dissolved oxygen, the quantity and bioavailability of organic carbon, and the concentration and composition of N and P nutrients, might be influenced by the water masses, which might impact the functioning of the microbial community as well (Jiang et al., 2014; Gan et al., 2016). Therefore, a more comprehensive study would be needed to evaluate the influence of these attributes on the functional diversity of the microbial community in the ECS.

In this study, we comprehensively examined the microbial functional gene diversity and structure in the complex ECS shelf area using GeoChip technology to understand the ecological role of water microbial communities and how that role is affected by environmental dynamics. The overall functional gene pattern of the ECS microbial communities was not consistent with their phylogenetic diversity pattern, with higher functional diversity and metabolic potential in the surface water compared with the bottom water. The microbial metabolic preference for different carbon substrates was revealed by the proportion of carbon-degradation genes. Our statistical analysis suggested that prominent environmental factors (temperature, salinity, and Chl a) associated with water mass were strong drivers in shaping the functional gene pattern of water microbial communities. These findings provide insight into the effects of the differing water masses on the ecological potential of microbes in the ECS. However, it should be noted that the presence of a functional gene or gene fragment does not necessarily imply functionality. Therefore, expression-based studies, such as mRNA-based microarray hybridization and metatranscriptome sequencing, are required to further elucidate microbial roles in this coastal sea ecosystem.

YW contributed for the experiment conduction and data analysis. YW and RZ contributed for the manuscript writing. ZH, JVN, and QZ contributed for the manuscript amendments on technology and language. RZ, JZ, and NJ contributed for the experiment design.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.