Protist diversity and community assembly in surface sediments of the South China Sea

Abstract Protists are pivotal components of marine ecosystems in terms of their high diversity, but protist communities have been poorly explored in benthic environments. Here, we investigated protist diversity and community assembly in surface sediments in the South China Sea (SCS) at a basin scale. Pyrosequencing of 18S rDNA was performed for a total of six samples taken from the surface seafloor at water depths ranging from 79 to 2,939 m. We found that Cercozoa was the dominant group, accounting for an average of 39.9% and 25.3% of the reads and operational taxonomic units (OTUs), respectively. The Cercozoa taxa were highly diverse, comprising 14 phylogenetic clades, six of which were affiliated with unknown groups belonging to Filosa and Endomyxa. Fungi were also an important group in both read‐ (18.1% on average) and OTU‐derived (9.3% on average) results. Moreover, the turnover patterns of the protist communities were differently explained by species sorting (53.3%), dispersal limitation (33.3%), mass effects (0%), and drift (13.3%). In summary, our findings show that the basin‐wide protist communities in the surface sediments of the SCS are primarily dominated by Cercozoa and are mainly assembled by species sorting and dispersal limitation.

Little is known about how protist communities are assembled in deep-sea sediments from a metacommunity perspective (Leibold et al., 2004;Vellend, 2010). Petro, Starnawski, Schramm, and Kjeldsen (2017) proposed four major processes of microbial community assembly in marine sediments: selection (i.e., species sorting), dispersal, diversification, and drift. As the predominant process (Petro et al., 2017), species sorting may be imposed by sediment differences such as water depth, pressure, and the properties of sediment particles. Moreover, dispersal limitation (derived from low dispersal), rather than mass effects (representing high dispersal), accounts for the importance of microbial dispersal in marine sediments because the microbial dispersal is passive and largely limited at a large spatial scale (e.g., the basin scale). Diversification (i.e., speciation) is supposed to have little influence within a metacommunity with individual dispersal (Stegen et al., 2013). Drift (acting alone), resulting from stochastic changes in birth and death rates, can be the dominant mechanism in extremely uniform habitats, which is not the case in marine sediments (Jacob, Soltwedel, Boetius, & Ramette, 2013). Therefore, we hypothesized that compositional turnover in protist communities in marine sediments at a basin scale would be mainly governed by a combination of species sorting and dispersal limitation.
The South China Sea (SCS) is one of the largest marginal seas located in the western Pacific Ocean, but the protist diversity across the basin-wide SCS sediments remains unclear. The SCS is characterized by a wide water depth range spanning over 5,000 m accompanied by distinct types of sediments (Liu et al., 2013). These sediments with contrasting characteristics have been shown to contribute to the compositional turnover in benthic microbial communities (Zhu, Tanabe, Yang, Zhang, & Sun, 2013). In addition, the semiclosed SCS is strongly influenced by the regulation of surface circulations by the East Asian monsoon system (Liu et al., 2002), which can also influence the seafloor microbial communities (Hamdan et al., 2013).
This influence is partially due to seasonal monsoons that contribute to the transport of fluvial sediments in the SCS Schroeder, Wiesner, & Liu, 2015).
The goal of this study was to investigate protist diversity and community assembly in surface sediments of the SCS. We investigated six sites (79-2,939 m depth) that represented common habitat types in the SCS seafloor and performed pyrosequencing of the V1-V2 region of 18S rDNA. We revealed the underlying processes that regulated community patterns of benthic protists using null model analysis and tested the hypothesis that species sorting and dispersal limitation are the two key driving forces. Overall, this study provides baseline information on the protist diversity and assembly in surface sediments of the SCS.

| Sample collection
A total of six sediment samples were collected from the surface seafloor using a grab sampler in the SCS during 28th April-21st May in 2010 ( Figure 1). This sampling design included one station (ST76) from the shallow coast (water depth = 79 m) and five stations located in the deep basin (water depths >880 m) (Table 1). Surface sediment samples (0-20 cm) were immediately collected and stored at −20°C until further analyses. Hydrodynamic profiles (i.e., temperature and salinity with water depth) of the upper waters at each station were obtained with an SBE-911 instrument (Sea-Bird Electronics, USA).

| DNA extraction and pyrosequencing
For each sediment sample, the top 0-1 cm segment was used for molecular analyses. Total DNA was extracted using an UltraClean Soil DNA Isolation Kit (MO BIO Laboratories, USA) according to the manufacturer's instructions, during which samples were homogenized for 60 s at 4 m/s using a FastPrep-24 instrument (MP Biomedicals, USA). The DNA extracts were quantified using a NanoDrop ND-1000 spectrophotometer (Nanodrop Technologies, USA). PCR amplification was performed for the V1-V2 region of 18S rDNA (approximately 420 bp) using the primers SSU_ F04 (5'-GCTTGTCTCAAAGATTAAGCC-3') and SSU_R22 (5'-GCCTGCTGCCTTCCTTGGA-3') (Bik et al., 2012). The PCR program consisted of an initial denaturation step at 95°C for 2 min; 30 cycles of 95°C for 30 s, 53°C for 30 s and 72°C for 30 s; and a final extension at 72°C for 5 min. The amplification products were then purified using an AxyPrep DNA Gel Extraction Kit (Axygen, USA).
Pyrosequencing was carried out on a 454 GS FLX Titanium system F I G U R E 1 Locations of the six samples (

| Sequence processing
The pyrosequencing data were processed using the Quantitative Insights Into Microbial Ecology (QIIME v. 1.9.1) pipeline (Caporaso et al., 2010). Briefly, the quality of reads was checked using a 50-bp sliding window and an average Phred threshold of 25, and short reads (<200 bp) were discarded. The remaining reads were run through DeNoiser (Reeder & Knight, 2010) to reduce pyrosequencing errors.
The resulting sequences were grouped into operational taxonomic units (OTUs) using UCLUST (Edgar, 2010) with a minimum identity of 97%. The representative sequence per OTU was selected, and chimeras were checked using ChimeraSlayer (Haas et al., 2011). The assignment of the representative sequences was determined using the PR 2 database (Guillou et al., 2013) with a BLAST E-value of 10 −6 and a minimum percent similarity of 90% (Zhang, Schwartz, Wagner, & Miller, 2000). Singletons (OTUs with only a single sequence in the entire data set) and OTUs with sequences detected in only a single sample were removed. Metazoans, as multicellular animals, were also removed because this study focused on single-celled protists. Consequently, OTUs assigned to metazoans were removed from further analyses.

| Phylogenetic analysis of Cercozoa
Considering the large percentage of Cercozoa sequences detected in sediment protist communities, we performed detailed phylogenetic analyses of the benthic Cercozoa. We carefully checked all representative sequences affiliated with the Cercozoa to ensure the performance of the phylogenetic analysis. The raw reads were generated from the orientation of the forward primer, while only sequences containing the accurate reverse primer (no mismatches) were retained in the subset of Cercozoa. All resulting sequences were aligned using MAFFT with the E-INS-i method, and the reverse primer was excluded. Each sequence was then manually checked using BLAST against the GenBank database. If a sequence had a similarity lower than 90% with the GenBank top hit and was rare (relative abundance <1% in all samples), we removed it from the data set. Reference sequences were added to perform phylogenetic analyses, and the whole sequences were aligned using the E-INS-i method. We manually trimmed positions with >95% gaps in each aligned column. A maximum-likelihood phylogenetic tree was constructed using PhyML (Guindon et al., 2010) with 1,000 bootstraps and the GTR + G + I model.

| Statistical analysis
Rarefaction analyses were performed to examine the degree of sampling saturation. To compare the OTU richness among the six sediment samples, we calculated nonparametric richness estimators (Chao1 and Shannon indexes). Chao1 and Shannon indexes were estimated based on the standardized data of 5,792 sequences per sample using the vegan package (Oksanen et al., 2014). To compare community dissimilarities, we performed phylogenetically informed beta diversity analyses using the weighted UniFrac distance metric (Lozupone & Knight, 2005) implemented in the QIIME pipeline (based on a standardized OTU  Stegen et al. (2013). First, the between-community variation in βMNTD was calculated based on the rarified OTU table (5,792 sequences per sample) using the picante package (Kembel et al., 2010). The degree to which the observed βMNTD deviated from a null model expectation was quantified after 999 randomizations.

| Water column environment
Vertical hydrographic profiles of the upper waters indicated that the sampling sites were characterized by low temperature (e.g., 5.8°C at a 796 m depth at ST65; Figure A1a) and high salinity (e.g., 34.5 psu at a 795 m depth at ST61; Figure A1b), except for the coastal site ST76 (21.7°C and 34.2 psu at a 61 m depth). However, detailed in situ environmental variables were unavailable for sediments.

| Benthic diversity
Pyrosequencing recovered a total of 74,091 quality-filtered reads (5,792-21,009 reads per sample) that were grouped into 269-408 OTUs per sample ( Table 1). The rarefaction curves of the observed OTUs showed unsaturated sampling profiles for all six samples

| Cercozoa dominate benthic diversity
Sequences of Cercozoa were clustered into 180 OTUs belonging to 14 phylogenetic groups (Figure 4a), suggesting a striking diversity of benthic Cercozoa. Remarkably, a few phylogenetic groups belonged to unknown clades (e.g., Unknown Filosa Groups I, II, III, and IV; Unknown Endomyxa Groups I and II), indicating that they might be novel taxa. Ascetosporea, Euglyphida, and Thecofilosea, as the top 3 groups, contributed an average proportion of 32.1%, 11.4%, and 10%, respectively, to the total Cercozoa OTUs ( Figure 4b and Table A1).

| Benthic community structure and assembly
Principal coordinates analysis plots using UniFrac dissimilarities showed that protist communities from different water depths were well separated ( Figure 5), which suggested that water depth played an important role in shaping the benthic protist communities. Specifically, a linear regression using water depths and PCoA 1 values was significant and yielded an r 2 statistic of 0.77 (Pearson's coefficient, p < 0.05). This outcome that water depth shaped beta diversity was also supported by the result of the Mantel test, showing a significant correlation between water depths and the weighted UniFrac distances (r = 0.52; p < 0.05; permutations = 720).
The results of the null model analysis showed that species sorting, dispersal limitation, mass effects, and drift accounted for 53.3%, 33.3%, 0%, and 13.3% of protist community pairs, respectively.

| Diversity of benthic protists
First of all, our results uncovered the dominance of Cercozoa in protist communities of the surface sediments of the SCS (Figure 3). The dominance of Cercozoa suggests distinct microbial webs in surface sediments compared with planktonic ecosystems in the SCS, where protist communities are commonly dominated by Syndiniales (in pelagic waters) (Strassert et al., 2018;Wu, Huang, Liao, & Sun, 2014) and Radiolaria and Polycystinea (in bathypelagic waters) (Xu et al., 2017). In European coasts, the prevalence of Cercozoa generated a major difference in community composition between planktonic and benthic protists (Forster et al., 2016). However, Cercozoa failed to show dominance in estuarine sediments in Sydney Harbor (Chariton, Court, Hartley, Colloff, & Hardy, 2010) and the East China Sea (Jiang, Wang, Yu, & Liu, 2016). These disagreements support the idea that deep-sea sediments harbor different protists than coastal and shallow-sea sediments; thus, water depth can strongly influence benthic protist communities (Gong et al., 2015). However, it remains unclear whether the primer pair used in this study targeting the V1-V2 region, rather than the most often V4 and V9 regions, biases protist community patterns, which imposes potential effects on the dominance of Cercozoa.
A number of other groups, in addition to Cercozoa, made considerable contributions to the protist communities ( Figure 3). Fungi stages (Piredda et al., 2017). Again, since DNA signatures were used in this study, we cannot rule out the possibility that these species were from the upper waters (Capo, Debroas, Arnaud, & Domaizon, 2015).
Some studies based on rRNA sequencing confirm the existence of active protists in marine sediments (Bernhard et al., 2014). For example, Bacillariophyceae rRNA sequences can even be detected in subsurface sediments, suggesting that rRNA may be more stable than previously considered in benthic environments (Orsi, Biddle, & Edgcomb, 2013

| Community assembly of benthic protists
Protist communities in the basin-wide surface sediments of the SCS are mainly shaped by species sorting and dispersal limitation. This finding supports the idea that species sorting and dispersal limitation are the two key drivers of microbial community assembly in marine sediments (Petro et al., 2017). Moreover, the relative importance of species sorting indicates that benthic habitats are strongly different.
Water depth may act as an important factor shaping benthic protist communities. The relationship between community dissimilarity and water depth agrees with the so-called depth decay in marine sediments (Jacob et al., 2013). However, it should be noted that water depth may have been a proxy of a set of associated environmental variables that were unmeasured in this study. That is, benthic protist communities may be structured by something other than the water depth itself. Marine sediments represent extreme energy-limited     barrier. It has been reported that benthic bacteria can show steeper distance-decay curves than both surface-sea and deep-sea bacteria can (Zinger, Boetius, & Ramette, 2014). This difference may mainly result from the difference in the extent of dispersal potential of microorganisms between benthic and planktonic habitats.
In contrast, Chen et al. (2017) showed that protist communities in intertidal sediments were strongly governed by spatial processes, potentially because the passive dispersal of protists contributed by water currents is very intense (i.e., mass effects) in shallow sediments relative to deep-sea sediments. Again, disentangling protist communities can be obscured by the limitation that sedimentary DNA may be from numerous planktonic groups (Capo et al., 2015) that are not part of the indigenous and active protist community.

| CON CLUS ION
Our results provide baseline information on the diversity and community assembly of benthic protists in the subtropical-tropical SCS. We show that the highly diverse Cercozoa group dominates the protist communities at the basin scale, and species sorting and dispersal limitation represent the two main forces that drive the community assembly of the benthic protists. Finally, we propose that more efforts, such as RNA-based surveys, are needed to unveil the hidden diversity and function of protists in marine sediments.

ACK N OWLED G M ENTS
This work was also supported by the National Key Research and Development Program (2016YFA0601201). W.W. was supported by funds (09530-18070002 and 42000-18841200) provided by Sun Yat-sen University. We thank H.M. at the South China Sea Institute of Oceanography (CAS) for the hydrographic data.

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interest.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
The raw sequence data were deposited in the Sequence Read TA B L E A 1 Summary of operational taxonomic unit (OTU; 97% similarity level) assignments of the benthic Cercozoa recovered in this study. For each OTU, we provide the closest relative in GenBank with the accession number (20-Mar-2016 database), sequence percent similarity, BLAST score, query/subject ratio, relative abundance (%) in each sample, and taxonomy