Information on diversity indices and abundance of individual species is crucial for the assessment of ecosystem health, especially for endangered ecosystems such as coral reefs. The application of environmental DNA (eDNA) to monitor coral biodiversity is, however, just beginning to come into focus for marine biologists. In this study, an eDNA metabarcoding approach of seawater samples in three different reefs on Koh Pha-ngan, Thailand, was compared with simultaneously collected visual census data. In addition, differences in read abundance and number of genera detected between daytime and nighttime eDNA samples were examined, and a local coral barcode reference database (n=23 genera; COI gene) was constructed to improve assignment of eDNA reads to the genus level. As a technical extension of existing assays, two methods for library construction were compared: a commercial kit and in-house developed fusion primers.

Combining eDNA metabarcoding and visual data, 29 different genera of scleractinian corals from 14 families were detected. In addition, a log-linear correlation was found between the abundance of eDNA reads and visually determined relative coral cover at the genus level, suggesting a predictive relationship between eDNA reads and coral cover. Results also showed diurnal variation between day and night samples in the number of eDNA reads, purported to relate to the activity phases of corals. The use of uniquely labeled fusion primers gave comparable results to a commercially available library preparation kit. Especially with frequent use, fusion primers can be very cost-effective, and therefore a consideration for large-scale studies. Using a custom reference database of 89 sequences from coral tissue samples of 23 different coral genera produced better results than querying against NCBI GenBank, highlighting the importance of locally optimized databases. We consider these results important for establishing eDNA as a complementary tool to visual surveys to track changes in coral diversity and cover.

Visual surveys, coral tissue collection and seawater sampling for eDNA metabarcoding were conducted from July to September 2019 at the three sampling sites, Haad Khom, Haad Salad and Haad Yao, along the northwestern shore of Koh Pha-ngan, in the Gulf of Thailand. For visual surveys at each of the three sites, six 30-m transects were aligned perpendicularly to the shore at 5–10 m intervals. Photos of a frame (0.25 m²) were taken every 2 m on alternating sides along a transect. Thus, 16 quadrats per transect were photographed. Scleractinia were visually identified from the photographs to genus level using the Indo Pacific Coral Finder (Kelley, 2016) and analyzed using 50 random points per quadrat (for a total of 800 points per transect) in CPCe (Kohler & Gill, 2006) for an estimate of coral cover and biodiversity. Statistical analyses were computed in RStudio (RStudio Team, 2020).

To build a genetic reference database, tissues and photos were taken of 138 different coral colonies (34 genera; 15 families) from five sampling sites (see Publication). Corals were identified to genus level using the Indo Pacific Coral Finder (Kelley, 2016) and the Corals of the World online database (Veron, Stafford-Smith, Turak, & DeVantier, accessed 2021: http://www.coralsoftheworld.org); taxonomic nomenclature was assigned based on the currently accepted nomenclature in the World Register of Marine Species (WoRMS; http://www.marinespecies.org, accessed November 2021). Genomic DNA of the coral tissues was extracted using the Qiagen DNeasy Blood & Tissue Kit (Quiagen, Hildesheim, Germany) according to the manufacturer's protocol. COI reference sequences of around 400 bp (HICORCOX_F1: 5’-GAACAAGGRGCKGGBAC-3’ and HICORCOX_R2: 5’-GCAACAAAAGTYGGKATTAT-3’, Nichols & Marco, 2019) were amplified using 6.25 µl VWR Go Taq Master Mix (VWR, Darmstadt, Germany), 1 µl extracted DNA template, 3.75 µl nuclease-free water and 0.25 µl of the forward and reverse primer (10 pmol/µl). Subsequently, the unincorporated primers and nucleotides were removed with ExoSAP (GE Healthcare, Solingen, Germany) digestion and the samples were sequenced at the Ruhr University Bochum Sequencing Service (Bochum, Germany).

Water samples were collected between 24 July 2019 and 3 August 2019 (midday and night-time) and again from 28 August 2019 to 1 September 2019 (midday). Collection was performed three times per sampling site: at noon, 12 hours later on the same day and four weeks later during daytime. Samples of 6 L each were collected with sterile containers at depths between 1–6 m by scuba divers, on one of the survey's transects per site. During the day (at noon), one sample was taken at the middle of the inner transect, one at the shallow end of the transect and one at the deep end of the transect (Figure 1). At night, three samples were taken at the middle of the transect. Immediately after collection, seawater samples were placed in the dark on ice and processed within 2 hours of collection. Collection blanks of 1-liter tap water were additionally taken into the field and afterwards handled like all other samples to exclude contamination during transport (Turner, Uy, & Everhart, 2015).

DNA from the filters was extracted using the Qiagen DNeasy Blood & Tissue Kit plus an additional bead beating step and minor adjustments (Nichols & Marko, 2019) to the extraction protocol. HICORCOX_F1 and HICORCOX_R2 (Nichols & Marko 2019) were used to amplify a ~400 bp sequence from the COI mitochondrial gene and PCR products were adjusted to 80 ng each. Libraries were prepared using either the NEBNext® Ultra II DNA Library Prep Kit for Illumina® (New England Bio Labs, Frankfurt a. M., Germany) or with custom-fusion primers. All libraries were measured on QIAxcel Advanced (Qiagen, Hildesheim, Germany) to assess size distribution of DNA samples (560–700 bp). DNA concentrations were finally measured with a Qubit fluorometer (Thermo Fisher Scientific) and samples were equimolar pooled for sequencing. In total, 81 libraries, 54 prepared with NEBNext Ultra II and 27 prepared with fusion primers were equimolar pooled and sent to Macrogen (Seoul, South Korea) for sequencing on an Illumina MiSeq platform.

We provide here the photos of the coral colonies sampled for the reference database as well as the alignment of consensus sequences of all genera we have been able to successfully amplify and sequence. Further photographs of the visual transects as well as the results of the CPCe analysis are given. Sequence data of demultiplexed unfiltered reads have been deposited in the NCBI Sequence Read Archive (BioProject ID: PRJNA894349).