*2.2. Sampling and Data Collection*

The survey was conducted in June and September 2018, respectively. The total number of sampling sites was fifteen, and encompassed the extensive area from the Seomjin River estuary to the outer Gwangyang Bay (Figure 1). The water samples were collected vertically from sediments to surface (depth: 10–50 m). For marine plankton sampling, a 200 μm mesh-sized net was used. The corresponding water volume (ca. 7560 L; 7.56 m3) was calculated by a flow-meter equipped in front of the net inlet. Zooplankton samples were identified and counted under a dissecting microscope (SV11, Zeiss and SZ60, Olympus, Tokyo, Japan), according to Chihara and Murano [30]. Water temperature and salinity were measured on site using a portable probe (Professional Plus, YSI, Yellow Springs, OH, USA). Nutrient and chlorophyll *a* concentrations (Chl-*a*) were analyzed in the lab using the collected water samples. Specifically for the measurement of phosphorus, nitrogen, and Chl-*<sup>a</sup>*, automatic water quality analyzer (AutoAnalyzer 3 HR, Seal Analytical Inc., Mequon, WI, USA) was used, and we adapted the standard analytical methods proposed by the Korea Ministry of Oceans and Fisheries (downloadable from http://www.mof.go.kr/jfile/readDownloadFile.do?fileId=MOF\_ARTICLE\_5689&fileSeq=1). For Chl-*a* measurement and eDNA metabarcoding, the water samples (1 L per sample) were immediately filtered in the lab, using a 0.45 μm pore-size membrane (MFS membrane filter, Advantec, Irvine, CA, USA). The membrane for Chl-*a* was then, homogenized after acetone extraction prior to the spectrophotometry. The membrane for eDNA was preserved at −80 °C. Organic and inorganic carbon concentrations were measured using a carbon analyzer (vario TOC cub, Elemetar, Langenselbold, Germany) on the basis of 850 °C combustion catalytic oxidation methods.

### *2.3. DNA Extraction and Metagenomic Sequencing*

Genomic DNA was extracted by means of PowerSoil® DNA Isolation Kit (Cat. No. 12888, MO BIO, Germantown, MD, USA) in compliance with the manufacturers' protocol. Extracted DNA for sequencing was prepared according to the Illumina 18S Metagenomic Sequencing Library protocols (San Diego, CA, USA). DNA quantity, quality, and integrity were measured by PicoGreen (Thermo Fisher Scientific, Waltham, MA, USA) and VICTOR Nivo Multimode Microplate Readers (PerkinElmer, Akron, OH, USA). For our study, the 18S rDNA V9 barcode was used, because it has often been applied to semi-quantitatively estimate relative abundances within a sample [31–33]. More specifically, we obtained the primer information from a study by Guo et al. [33], which also followed the universal

primers for 18S V9 region designed by Amaral-Zettler et al. [34]. The primer sequences are as follows: 18S V9 primer including adaptor sequence (Forward Primer: 5 TCGTCGGCAGCGTCAGATGTGT ATAAGAGACAG**CCCTGCCHTTTGTACACAC** 3, Reverse Primer: 5 GTCTCGTGGGCTCGGA GATGTGTATAAGAGACAG**CCTTCYGCAGGTTCACCTAC** 3, the primers are in bold). The PCR master mixture of 25 μL (Macrogen Inc., Seoul, Korea) comprised 2 μL of genomic DNA (1 ng/μL), 1.25 μL of each primer (5 μM), 5 μL of 5 × Herculase II Reaction Buffer, 0.25 μL of dNTP mix (100 mM), 0.5 μL of Herculase II Fusion DNA polymerase (Agilent, Waldbronn, Germany), and 14.75 μL of PCR Grade water. To amplify the target region attached with adapters, as a first PCR process, the extracted DNA was amplified by 18S V9 primers with one cycle of 3 min at 95 ◦C, 25 cycles of 30 s at 95 ◦C, 30 s at 55 ◦C, 30 s at 72 ◦C, and a final step of 5 min at 72 ◦C for amplicon PCR product. As a second process, to produce indexing PCR, the first PCR product was subsequently amplified with one cycle of 3 min at 95 ◦C, 8 cycles of 30 s at 95 ◦C, 30 s at 55 ◦C, 30 s at 72 ◦C, and a final step of 5 min at 72 ◦C. A subsequent limited-cycle amplification step was performed to add multiplexing indices and Illumina sequencing adapters (Figure 2). The final products were normalized and pooled using the PicoGreen (ThermoFisher Scientific, Waltham, MA, USA), and the size of the libraries was verified using the LabChip GX HT DNA High Sensitivity Kit (PerkinElmer, Waltham, MA, USA).

**Figure 2.** Analytical procedure of environmental DNA (eDNA) extraction and metagenomic sequencing.

A sequencing library is prepared by random fragmentation of the DNA or cDNA sample, followed by 5 and 3 adapter ligation. Alternatively, "tagmentation" combines the fragmentation and ligation reactions into a single step that greatly increases the efficiency of the library preparation process. Adapter-ligated fragments are then PCR amplified and gel purified. The PCR products were sequenced using the MiSeq™ platform (Illumina, San Diego, CA, USA) from commercial service (Macrogen Inc., Republic of Korea). In total, filtered 6,151,975 paired-end reads from the 30 samples were generated on the platform, of which 97.11% passed Q30 (Phred quality score > 30) in this study. Raw reads were trimmed with CD-HIT-OTU and chimeras were identified and removed using rDNATools. For paired-end merging, FLASH (Fast Length Adjustment of Short reads) version 1.2.11 was used. Each sample yielded paired-end reads ranging from 21,101–299,305 reads (mean: 180,940 reads), and all samples exhibited the saturation of the number of operational taxonomic units (OTUs) by rarefaction curve analysis (see Appendix A). Merged reads were processed and were clustered into OTUs using a bioinformatic algorithm, UCLUST [35], at a 97% OTU cutoff value (352 OTUs in gamma-diversity). The resulting 552 OTUs were classified into 19 genus-level taxonomic groups (those representing < 0.04% abundance were not plotted). Taxonomy was assigned to the obtained representative sequences with BLAST (Reference DB: NCBI—18S) [36] using UCLUST [35]. For the aforementioned processes of BLAST and UCLUST, we used an open-source bioinformatics pipeline for performing microbiome analysis, QIIME version 2 [37].
