The application of environmental DNA technologies is a promising new approach to rapidly audit biodiversity across large-scale, remote regions. Here, we examine the efficacy of a dual-assay eDNA metabarcoding approach for sessile benthic bioassessments in the turbid waters of the Lalang-garram Marine Parks, in the inshore Kimberley region, north-western Australia. We ask three principal questions: 1. Is the eDNA released by sessile benthic taxa (i.e. hard and soft corals, sponges and tunicates) locally detectable? 2.  What level of taxonomic resolution is afforded by eDNA metabarcoding using the ITS2 region? and 3. How well does eDNA metabarcoding compare to conventional benthic survey techniques, such as belt and point-intercept transects? We report that a dual-assay eDNA metabarcoding approach is capable of detecting approximately 70% of the local benthic taxa (i.e. at a species, genus level etc) identified at the surveyed locations. It is, however, not as effective at the individual/population level, detecting only approximately 40% of unique amplicon sequence variant (ASV) signals released by an array of individual benthic organisms at the surveyed locations. In examining the efficacy and resolution of the applied ITS2 metabarcoding markers for bioassessments, we report large gaps in the variety of publicly available benthic ITS2 reference sequence data, limiting our ability to provide robust taxonomic assignments. These findings highlight the need to extend ITS2 databases for greater regional representation. Until this is adequately addressed, we recommend that investigating taxonomic assignments to a genus-level is the most robust approach to benthic monitoring using eDNA. Lastly, we found eDNA metabarcoding and conventional belt-transect surveys each detected numerous unique hard coral genera, indicating that a combined approach provides the most effective way to audit benthic biodiversity. Furthermore, eDNA metabarcoding had the power to distinguish similar diversity trends between sites to that determined by the belt-transect methodology, validating the application of eDNA metabarcoding as either a stand-alone, or complementary technique for assessing sessile benthic taxa.

For this study, seven intertidal reef sites were surveyed in Lalang-garram marine parks of the Southern Kimberley region in October 2018 (see Figure 1 and Table S1 in manuscript). At each site, 10 x 500 ml water replicates were sampled in addition to the opportunistic collection of 1-16 tissue from sessile benthic organisms, such as hard corals (order Scleractinia), soft corals (order Alcyonacea), sponges (phylum Porifera) and tunicates (subphylum Tunicata). Specimens were collected at spring low tide and stored in 100% ethanol. Water samples were individually filtered through 0.45 mm cellulose filter membranes using a Pall Sentino® Microbiology pump (Pall Corporation, Port Washington, USA) within 4-6 hours of collection. Between the filtration of each replicate, all filtering equipment was soaked in 10% bleach for a minimum of 10 min and further rinsed with desalinated and filtered water. This was to prevent cross contamination of eDNA between replicates and sites. Two bleach and desalinated tap water samples were taken at the end of each filtering day to serve as filtration controls.

Seawater-borne DNA was extracted from half of each respective membrane (including filtration controls) using a DNeasy Blood and Tissue Kit (Qiagen; Venlo, Netherlands) following the manufacturers protocol and additional modifications as described in Alexander et al. (2020). Benthic tissue samples were also extracted using a DNeasy Blood and Tissue Kit. DNA extracts were then stored at -20°C.

Two PCR metabarcoding assays targeting the nuclear ribosomal internal transcribed spacer 2 (ITS2) region were employed for this study: CoralITS2 (Brian, Davy, & Wilkinson, 2019) which amplifies a range of scleractinian taxa (exempting the genus Acropora), and CoralITS2_acro (Alexander et al., 2020) to amplify Acropora (see Table 1 in manuscript). Fusion-tagged amplicons were sequenced on 500 cycle MiSeq® V2 Standard Flow Cells on an Illumina MiSeq platform (Illumina, San Diego, USA), housed in the TrEnD Laboratory at Curtin University, Western Australia. Sequences were demultiplexed using the insect package (Wilkinson, Davy, Bunce, & Stat, 2018), and quality filtered (minimum length=60, maximum expected errors=2, no ambiguous nucleotides), dereplicated, denoised (pool=TRUE), merged and filtered for chimeras using the DADA2 pipeline (Callahan et al., 2016) in RStudio (v1.1.423; Team, 2015). Because of the multi copy nature of ITS2 and variable length of the gene, no strict trimming was performed as recommended by the developers pipeline (Callahan et al., 2016). Resulting amplicon sequence variants (ASVs) for each assay, i.e. unique sequences that can separated by one or more nucleotide differences (Callahan, McMurdie, & Holmes, 2017), were then queried against NCBI’s GenBank nucleotide database (Benson, Karsch-Mizrachi, Lipman, Ostell, & Wheeler, 2005; accessed in 2020) and the addition of a coral ITS2 database (Dugal et al. in review). Taxonomic assignments of ASVs were curated using a lowest common ancestor (LCA) approach (https://github.com/mahsa-mousavi/eDNAFlow/tree/master/LCA_taxonomyAssignment_scripts; Mousavi-Derazmahalleh et al. in review).

We provide here, the demultiplexed (unfiltered) fastq data for each sample (water and tissue) amplified via the ITS2 and ITS2_acro assays. We also provide the resulting read count matrices (post-filtering, blasting and LCA curation) at both an ASV and taxonomic level for both assays.