The cruise was carried out in the South China Sea with the Shiyan 3 from August to September 2011 ( Figure 1 ). A conductivity-temperature-depth (CTD) system (SeaBird SBE-911 Plus, US) was deployed to acquire hydrographic parameters. Seawater samples were collected at different depths (0, 25, 50, 75, 100, 150, and 200 m) with CTD 12-L Niskin bottles (General Oceanics, Inc., Miami, FL). Once collected, the samples were immediately filtered using polycarbonate membranes (EMD Millipore, US) with a pore size of 0.22 μm. The filter was immediately placed in a 1.5-ml sterile centrifuge tube and stored in liquid nitrogen for further DNA extraction. Sample DNA were extracted using the DNA Extraction Kit (OMEGA, US) according to the protocol instructions. Nitrate (NO₃), phosphate (PO₄), silicate (SiO₃), and chlorophyll (Chl a) concentrations were measured according to the protocols of “the specialties for oceanography survey” (GB17378.4-2007, China).

Two dmdA primers were used to amplify the different Roseobacter subclades (A/1 and A/2) (Varaljay et al., 2010) for the DMSP demethylase gene, and dddP primers targeting the Roseobacter group were used to detect the DMSP lyase genes (Levine et al., 2012). The reactions were performed in the iQ5 Real-Time PCR Detection System (Bio-Rad, US). Quantification was based on the increasing fluorescence intensity of the SYBR green dye during amplification. QPCR standards were made from PCR products amplified from environmental samples using the pMD19-T cloning kit (TaKaRa, Japan). The real-time PCR assay was performed in a 20-μL reaction volume with SYBR Premix Ex Taq II (TaKaRa, Japan). All qPCRs were run in triplicate for each sample. QPCR conditions were as follows: predenaturation at 95°C for 30 s, 35 cycles at 95°C for 5 s, and annealing at 60°C for 30 s. Tenfold serially diluted standard and no-template controls were run in triplicate for each reaction.

The primer pairs, dmdA282F (5′-TGCTSTSAACGAYCCSGT-3′) and dmdA591R (5′-ACRTAGAYYTCRAAVCCBCCYT-3′) (Zeng et al., 2016), were used for Roseobacter-like dmdA gene amplification. PCR conditions were as follows: 94°C for 3 min, 35 cycles at 94°C for 1 min, 54°C for 30 s, extension at 72°C for 30 s, and a final extension at 72°C for 10 min. The PCR comprised a 20-μL reaction volume containing 4 μL of 5× FastPfu Buffer, 2 μL of 2.5 mM dNTPs, 0.5 μL of FastPfu polymerase, 1.0 μL of primers (5 μM), and 10 ng of template DNA. A positive control and non-template control samples were run to validate PCR. PCR products were purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, US), and quantified using QuantiFluor™-ST (Promega, US). An Illumina MiSeq platform (Illumina, US) was used for paired-end sequencing (2 × 300) according to the standard protocols.

For pair-ended reads obtained by Illumina sequencing, barcodes and primers were trimmed and then assembled using FLASH (V1.2.7). Reads that contained Ns were shorter than 50 bp or had primer mismatches which were also excluded. Sequences were compared with RDP reference database using VSEARCH (1.9.6) to detect chimeric sequences. Then sequences were grouped into OTUs (operational taxonomic units) using UPARSE (v7.0.1001), and pre-clustered at 97% sequence identity. The highest OTU frequencies were selected as representative OTU sequences. The taxonomy of each dmdA gene sequence was analyzed by RDP classifier algorithm against the NCBI non-redundant (nr) database using a confidence threshold of 70%.

Two subclades (A/1 and A/2) of dmdA genes and dddP genes were quantified using qPCR ( Figure 2 ). The copy numbers of dmdA and dddP genes and their distribution in the northwestern SCS varied greatly among sampling sites and depths. This result was similar with the study in the Pacific Ocean (Cui et al., 2015), indicating that the abundance of DMSP degradation genes had great variability of abundance in the ocean. Overall, the abundance of dmdA and dddP genes located in the northwest SCS was higher than that located in the northeastern SCS. As seen from a spatial scale, sites (S1, S2, S3, S8, and S9) showed higher copy numbers of dmdA and dddP genes than other sites (S5, S6, S12, and S13). In these five stations, two sites (S1 and S2) are close to the shore, and the remaining three (S3, S8, S9) are in the middle of the survey area. Similar to previous studies in the SCS (Ling et al., 2012; Sun et al., 2015), bacterial community in the northwestern SCS had higher diversity than that in the northeastern SCS. The abundance difference is likely due to water temperature. As seen in Figure 1 , the SST of the sites to the left of longitude 117°E is higher than that of the sites to the right. This result was similar to other reports that temperature is an important factor in determining the abundance of dmdA and dddP in surface water (Levine et al., 2012; Varaljay et al., 2012; Cui et al., 2015).

DmdA and dddP genes from the Roseobacter group were particularly enriched in surface waters with an order of magnitude difference in their abundance relative to deep waters ( Figure 2 ), which indicated that the abundance of dmdA and dddP genes was strongly separated by water depth. Copy numbers of dmdA and dddP genes were higher above 75 m water than those below 100 m water, and the highest abundance of these genes was observed in the surface layer. Variation of gene abundance of dmdA and dddP may be closely related to the DMSP concentration in vertical depth. Previous studies reported that DMSP and DMS concentrations in SCS were markedly high in the surface seawater, and decreased gradually with increasing depth (Yang, 2000; Zhai et al., 2020). Other studies also reported that distribution patterns of dmdA and dddP were roughly consistent with the distribution characteristics of DMSP concentration, and were mainly influenced by the Chl a concentrations, depth, salinity, and temperature (Howard et al., 2011; Varaljay et al., 2012; Cui et al., 2015).

In addition, the copy numbers of the dddP gene were far lower than those of the dmdA gene at almost all sites and depths ( Figure 2 ), even if only the dmdA A/2 clade was considered. The copy number ratios (dmdA/dddP) of these genes ranged from 2 to 156 times. The copy numbers of the dmdA A/2 gene were higher than those of the dmdA A/1 gene at almost all sites and depths, which suggested that subclade A2 was the main group of Roseobacter group in the demethylation pathway. The other study reported that two Roseobacter group dmdA gene subclades (A/1 and A/2) showed opposite depth distributions in the summer (Levine et al., 2012). Interestingly, high ratios of dmdA/dddP and dmdA2/dmdA1 were mainly observed in deep waters (below 75 m), even if these three genes had relatively low abundance, indicating that demethylation is the main pathway of DMSP degradation in the water. Previous studies suggested that 80% of DMSP degradation is processed through the demethylation pathway, and only 20% is cleaved to DMS (Kiene et al., 1999). Marine bacteria keep the ability of DMSP demethylation to a suitable evolutionary pressure, which can explain the consistently stable and high dmdA gene frequencies in the ocean (Varaljay et al., 2012). In marine waters, nutrients and organic sulfur such as DMSP are important because cells need increased sulfur demand for growth, causing more sulfur to be incorporated into cell protein (Kiene et al., 1999).

A total of 231,246 valid reads and 688 OTUs were obtained from the 19 samples through Illumina MiSeq sequencing analysis. Each of the samples contained 9155 to 17,955 reads, with OTUs ranging from 12 to 55. The coverage was more than 0.999, which suggested that sequencing data had favorable coverage for dmdA diversity. Diversity indices, including Shannon, Chao, Ace, and Simpson, are demonstrated in Supplementary Table S1 . The result indicated that the dmdA gene had higher community diversity above 50 m than in deep waters (100 and 200 m).

The study revealed that the composition of the dmdA gene varied significantly among the sites and depths (ANOVA, p < 0.01). Overall, the dmdA gene was mainly affiliated with Alphaproteobacteria and Gammaproteobacteria ( Figure 3 ). In most sites, the dmdA gene was dominated by Alphaproteobacteria, whereas a few of the samples (S2_0, S6_0, S6_50, and S9_25) were dominated by Gammaproteobacteria. In addition, low abundance of Acidithiobacillia-like dmdA gene was detected (1.37%). High abundance of dmdA (2.64%–87.95%), which had very low similarity with the amino acid identity of uncultured bacteria, was found in some samples. Moreover, 21.78% of all sequences had 73%–79% amino acid similarity with Gammaproteobacteria. This result provided further evidence that ocean water contains a high diversity of dmdA genes, which has not yet been unearthed in the tropical ocean.

Overall, dmdA genes in the nSCS showed greater variation near surface water than in deep waters, which is related to the vertical structure of water ( Figure 3 ). At the genus level, high abundance of dmdA genes below 100 m belonged to Phaeobacter, Roseicitreum, Ruegeria, Tropicibacter, and Shimia. The upwelling site (S1-25) was characterized by Tateyamaria (43.11%) and Pelagimonas (8.40%). These findings indicated that coastal upwelling had different dmdA taxonomy of the Roseobacter group with open water. S5_0 and S9_0 sites were dominated by Leisingera (>20%) and Nioella (>46%). S3_0 and S8_0 sites were characterized by highly abundant Roseovarius (>80%). Site S6_25 was mainly composed of Roseobacter (58%) and Shimia (9.56%). Site S7_25 was mainly composed of Roseobacter (38%), Roseovarius (12.32%), and Planktotalea (4.97%). The dmdA gene of Donghicola mainly dominated in the 25- and 50-m water layer. Moreover, dmdA genes of Leisingera, Planktotalea, Roseovarius, and Ruegeria were dominant in the 25- or 50-m water layer.

Principal component analysis (PCA) illustrated that the dmdA gene of the bacterial community had distinct differences among different sites and depths ( Figure 4 ). The results indicated that the dmdA gene of the Roseobacter group in shallow water (0 m, 25 m, and 50 m) and deep layer (100 and 200 m) had clustered together separately. The dmdA gene of the deep water was separated by a large distance from the shallow layer. Overall, the results indicated the clearly distinct structure among sampling sites and depths (ANOVA, p < 0.01). Veen analysis showed that dmdA gene diversity had 14 and 11 common OTUs between the surface water and 25-m depth of the horizontal scale. Considering the vertical depths, no common OTU was found from the surface to 200 m water depths ( Figure 5 ).

The CCA analysis constructed a correlation between environmental factors and dmdA gene diversity ( Figure 6A ). The first ordination axis accounted for 32.1% of the cumulative percentage variance in the matrix, while the second axis accounted for 31.2%. DmdA gene in the upwelling (S1-25) was mainly positively related to Chl a and silicate concentration (p<0.05). DmdA gene below 100-m depth was negatively defined by water depth (p<0.01), while shallow water samples (0 m and 25 m) were strongly determined by water temperature (p<0.01). Other environmental factors, such as salinity, nitrate, and phosphate, had no obvious effects on the diversity of dmdA genes.

In the CCA map ( Figure 6B ), cluster A, mainly composed by Tateyamaria, Pelagimonas, and Marivita, had significant correlation (p < 0.05) with Chl a and silicate concentration, which was influenced by the upwelling. High abundance OTUs of cluster B, which was afflicted with Leisingera, Nioella, Roseovarius, and Roseobacter, were mainly positively related to temperature. High abundance of OTUs (cluster C), which was afflicted with Roseobacter, Shimia, Planktotalea, and Donghicola, were mainly related to depth, phosphate, and salinity. Cluster D and E, mainly composed by Phaeobacter, Roseicitreum, Ruegeria, and Tropicibacter, were mainly positively related to depth, indicating that these Roseobacter OTUs were suitable to environment change with deep water. As described above, a high ratio of dmdA/dddP was found in deep water, which indicated that the Roseobacter group had more nutrition demand and greater degree of demethylation than DMSP cleavage. Different ecological strategies of DMSP degradation may be used by members of the bacterial community harboring demethylation and/or cleavage genes (Reisch C. et al., 2011). In previous studies, the Roseobacter group is the dominant microbial taxa in the offshore, upwelling, and mesoscale eddy of the South China Sea (Zhang et al., 2016; Sun et al., 2020; Sun et al., 2022), and diversity of DMSP degradation potential has obvious significance for sulfur material cycling and transformation.