Environmental DNA (eDNA) is a promising biomonitoring tool for marine ecosystems, but its effectiveness for North Pacific coastal fishes is limited by the inability of existing barcoding primers to differentiate among rockfishes in the genus Sebastes. Comprised of 110 commercially and ecologically important species, this recent radiation is exceptionally speciose, and exhibits high sequence similarity among species at standard barcoding loci. Here, we report new Sebastes-specific metabarcoding primers that target mitochondrial cytochrome B. Amongst the 110 Sebastes species, 85 unique barcodes (of which 62 are species-specific) were identified in our amplicon region based on available reference sequences. The majority of the remaining barcodes are shared by only two species. Importantly, MiSebastes yield unique barcodes for 28 of 44 commercially harvested species in California, a dramatic improvement compared to the widely employed MiFish-U 12S primers which only recover one of 44. Tests of these primers in an aquarium mesocosm containing 16 rockfish species confirms the utility of these new primers for eDNA metabarcoding, providing an important biomonitoring tool for these key coastal marine fishes.

To test the utility of the MiSebastes primer set for eDNA metabarcoding, we collected water samples from the kelp forest tank at the California Science Center (700 Exposition Park Drive, Los Angeles, CA 90037), a 700,000 L, 10 m deep aquarium exhibit containing 16 Sebastes species. The water for the aquarium is sourced from a coastal site off of Catalina Island and transported by barge and truck to the aquarium. The transit time of the water is often longer than 24 hours, and the water is UV sterilized through the aquarium’s filtration system. We collected triplicate biological replicates of 1 L at both the surface and bottom (one meter off the bottom of the tank) of the well-mixed aquarium (total = 6 samples). We filtered water samples through a 0.2 mm Sterivex (MilliporeSigma, Burlington, MA, USA) filter using a gravity filtering method and stored filtered eDNA at -20°C for transport back to the lab at UCLA (Miya et al. 2016). We extracted eDNA from the filters using the Qiagen DNeasy Blood and Tissue kit (Qiagen, Valencia, CA, USA) following the modifications of Spens et al. (2017), in which ATL buffer and proteinase K are added directly to the sterivex filter cartridge. Cartridges were incubated overnight and then transferred to 1.7 mL tubes using a 3 mL syringe. Equal amounts of sample, AL buffer, and 0 ˚C ethanol were added. The protocol then followed the manufacturer’s instructions. We stored extracted eDNA at -20ºC until PCR amplification.

We amplified extracted eDNA with the MiSebastes primers employing the same thermocycling parameters, above, but tripled the volume of template DNA from 1 mL to 3 mL to improve PCR sensitivity and performance. We decided to triple the volume of template DNA following the results of our first PCR tests for this primer set using only 1 mL of starting template, which resulted in very faint bands after running PCR product through electrophoresis on a 2% agarose gel. The PCR reaction volume was thus as follows: 3 mL of template DNA, 1 mL of each primer (10 mM) 12.5 mL of QIAGEN Multiplex Taq PCR 2x Master Mix, 2.5 mL of Q solution, and 5 mL of nuclease-free water. To minimize PCR dropout, we had three technical replicates for each sample by conducting all PCR reactions in triplicate (Doi et al. 2019), and all PCR reactions included negative and positive controls. Following PCR, we confirmed successful amplification and correct product size through electrophoresis on a 2% agarose gel.

To compare MiSebastes to MiFish-U primers (Miya et al. 2015), we repeated the above using the MiFish-U primers (Table 1) on the three surface samples. The bottom samples were not sequenced with the MiFish-U primers. We used a similar touchdown PCR protocol, but the touchdown annealing temperatures began with an initial temperature of 69.5°C for 30 s and subsequently decreased by 1.5°C every cycle until 50°C was reached. The reaction volume was as follows: 1 mL of template DNA, 5 mL of each primer (2 mM), 12.5 mL of QIAGEN Multiplex Taq PCR 2x Master Mix, and 1.5 mL of nuclease-free water. Template volume for MiFish-U PCR differed from MiSebastes so that both yielded PCR products of similar strength.

We prepared PCR products for sequencing following the CaleDNA protocol (Meyer et al. 2019). The three technical replicates for each sample were sequenced separately for the MiSebastes primer set; the technical replicates for each sample were pooled before being sequenced for the MiFish-U primer set. The difference in pooling protocol is due to the fact that the MiFish-U samples had been prepared for a previous project for which the protocol dictated that PCR products be pooled prior to sequencing. However, because MiFish-U and MiSebastes results were summed across all samples and technical replicates, this difference in methodology had a negligible effect on comparisons of overall primer taxonomic resolution. Taxonomic detections by individual technical replicates for MiSebastes samples are shown in Supplementary Figure 1.

We generated two sequencing libraries (one for the MiFish-U samples and one for the MiSebastes samples) through an indexing PCR using IDT for Illumina Nextera UD Indexes Sets A and D (Illumina, San Diego, CA, 92122) and KAPA HiFi HotStart Ready Mix (Kapa Biosystems, Wilmington, MA, USA). This second indexing PCR was performed using a 25 μL reaction mixture containing 12.5 μL of Kapa HiFi Hotstart Ready mix, 0.625 μL of primer i7, 0.625 μL of primer i5, and 10ng of template DNA, and used the following thermocycling parameters: denaturation at 95˚C for 5 min, 5 cycles of denaturation at 98˚C for 20 sec, annealing at 56˚C for 30 sec, extension at 72˚C for 3 min, followed by a final extension at 72˚C for 5 min. We ran all indexed PCR products on a 2% agarose gels to ensure successful PCR and correct product size. We then cleaned the pooled samples using Serapure magnetic beads (Faircloth and Glenn 2014) and quantified cleaned PCR product concentrations using the high sensitivity Quant-iT™ dsDNA Assay Kit (Thermofisher Scientific, Waltham, MA, USA) on a Victor3 plate reader (Perkin Elmer Waltham, MA, USA). Following quantification, the indexed samples were pooled in equimolar concentration so that the final library (volume 300 μL) had a DNA concentration of 20 ng/μL. Because the samples were sequenced together with samples for other projects, there were 57 samples in the library that contained the MiFish-U samples and 91 samples in the library that contained the MiSebastes samples (Supplementary Table 4). Lastly, we sequenced the libraries on a MiSeq PE 2x300bp at the Technology Center for Genomics & Bioinformatics (University of California – Los Angeles, CA, USA), using Reagent Kit V3 with 20% PhiX added to all sequencing runs.

We processed metabarcoding sequences from the MiSebastes primer set using the Anacapa Toolkit (Curd et al. 2019). To maximize identification of metabarcode sequences, we used CRUX to create a Sebastes-specific reference database following the standard CRUX parameters. As the Anacapa Toolkit assigns taxonomy to ASVs with a confidence score using a Bayesian Lowest Common Ancestor (BLCA) method, it can only assign one species to a particular ASV. As such, we developed an additional script to the Anacapa Toolkit to help assign taxonomy to sequences that are shared between multiple species for our metabarcode (https://github.com/markusmin/misebastes_annotator). This add-on for the Anacapa Toolkit identifies all species that share an ASV for our barcode and adds this information to the Anacapa taxonomy output file. We employed standard Anacapa Toolkit parameters and a Bayesian cutoff score of 40 to preserve the low-confidence species-level annotations caused by multiple Sebastes species sharing the same barcode for the MiSebastes primer set. Lowering the Bayesian cutoff score from the standard value of 60 to 40 was necessary to allow the MiSebastes-specific annotation script to identify instances where multiple species share the same barcode, as this script uses our in silico PCR results to add additional species that are known to share the same barcode to the taxonomic assignments from the Anacapa Toolkit.

We processed metabarcoding sequences from the MiFish-U primer set using the Anacapa Toolkit (Curd et al. 2019) following standard parameters and a Bayesian cutoff score of 60 (Gao et al. 2017). We assigned taxonomy to MiFish-U data using a California Current-specific fish 12S-specific reference barcode database (Gold et al. 2021).