eDNA metabarcoding
Environmental DNA was extracted from approximate 0.25 g of sediment using the DNeasy PowerSoil Kit (QIAGEN), with one DNA extraction control for every 11 sediment samples. For metabarcoding, the V1 and V2 regions of the 18S rRNA gene were amplified using primers (F04mod: GCTTGWCTCAAAGATTAAGCC; R22mod: CCTGCTGCCTTCCTTDGA) (Blaxter et al., 1998). We used this 18S rRNA marker rather than a COI marker for eDNA metabarcoding because: (1) this marker has been used successfully to characterize benthic metazoans (e.g. Fonseca et al., 2010; Sinniger et al., 2016); and (2) it amplifies eukaryotes only, whereas amplicon sequencing targeting eukaryotes using COI markers typically returns a high percentage of off-target bacterial reads due to non-specific amplification (Collins et al., 2019; Weigand & Macher, 2018). To multiplex, each synthesized primer starts with two or three Ns, followed by an eight‐base tag and the gene‐specific primer sequences (He, Sutherland, Pawlowski, & Abbott, 2019). PCR protocol and program followed He et al. (submitted). Three PCR replicates were used to amplify each eDNA sample (using the same tagged primers) and replicates were combined after PCR. PCR products from 21 eDNA samples, one DNA extraction control and one PCR negative control were combined into a mixture that was then purified using Agencourt AMPure XP beads (Beckman Coulter) and quantified using SYBR Green I (Thermo Fisher Scientific) with a Tecan Infinite M200PRO microplate reader. In total, 12 PCR product mixtures were made. Then, 1 µg of purified PCR mixture was used for library preparation using the TruSeq DNA PCR-Free Library Prep Kit (Illumina). After library preparation and quantification, four libraries were pooled for sequencing using one MiSeq Reagent Kit v3 (600-cycles) (Illumina). In total, three MiSeq Reagent Kit v3 (600-cycles) were used to sequence the 252 sediment samples.