Limitations of MiFish primers
High numbers of unidentified ASVs are a common feature of barcoding and metabarcoding studies (e.g. Leray & Knowlton, 2017). These unidentified ASVs are typically attributed to incomplete reference databases (Curd et al., 2019; Ransome et al., 2017; Schenekar et al., 2020) and/or novel biodiversity (Barber & Boyce, 2006; Boussarie et al., 2018). However, given that FishCARD includes 82.8% of all California Current fishes, and the remaining 15.6% are mostly rare species unlikely to be found in a kelp forest, it was extremely surprising that half of all ASVs and a quarter of all sequences generated in our eDNA test datasets could not be assigned.
The vast majority of these sequences and ASVs did not belong to vertebrates, but instead uncultured marine bacteria, specifically matching to 16S , rather than 12S loci. Since mitochondria represent the capture of microbial endosymbionts by ancient eukaryotes (Roger, Muñoz-Gómez, & Kamikawa, 2017) and that this capture occurred in the sea, it perhaps is not surprising that primers designed to target vertebrate 12S might also capture marine prokaryotes. Similarly, the homology between vertebrate 12S and prokaryotic and bacterial16S genes is well known (Crews & Attardi, 1980) suggesting capturing microbial 16S with vertebrate 12S primers is also not surprising. However, this particular feature of the MiFish primer set was previously unreported, potentially impacting the interpretation of unidentified ASVs in fish metabarcoding studies.
These findings highlight the importance of accurate universal metabarcoding primer design, especially in outlining both target and non-target sequences. In the design of the MiFish Teleost 12Sprimers, uncultured marine microbe 16S sequences were not considered as potential alternative targets for the primer set, resulting in the selection of a metabarcoding locus with a high degree of non-target amplification (Miya et al., 2015). This is an important finding for the marine vertebrate eDNA community which has recently converged on the MiFish 12S primers (Closek et al., 2019; O’Donnell et al., 2017; Valsecchi et al., 2019; Yamahara et al., 2019) as these results suggest that the MiFish 12S primer set will generate substantial quantities of non-target eDNA reads. At best, this non-target amplification will lead wasted sequencing effort, as every microbial sequence generated reduces the number of vertebrate sequences. At worst, it could result in incorrect interpretation of unidentified ASVs. This problem is of particular concern in environments with high relative abundance of marine bacterial communities and low relative abundance of vertebrate biomass such as in some pelagic midwater and deep sea habitats where recent eDNA sample collection efforts have struggled to detect vertebrate sequences (K. Pitz personal communication).