Leveraging eDNA metabarcoding to characterize nearshore fish communities in Southeast Alaska: Do habitat and tide matter?

Abstract

AbstractNearshore marine habitats are critical for a variety of commercially important fish species, but assessing fish communities in these habitats is costly and time‐intensive. Here, we leverage eDNA metabarcoding to characterize nearshore fish communities near Juneau, Alaska, USA, a high‐latitude environment with large tidal swings, strong currents, and significant freshwater input. We investigated whether species richness and community composition differed across three habitat types (sand beaches, eelgrass beds, and rocky shorelines) and between high and low tides. Additionally, we tested whether replication of field samples and PCR reactions influenced either species richness or composition. We amplified a 12S mitochondrial locus in our samples and identified 167 fish amplicon sequence variants, which were grouped into 24 unique taxa based on sequence similarity, with approximately half of these taxa resolved to single species. Species richness and composition inferred from eDNA differed substantially among habitats, with rock habitats containing fewer taxa and fewer overall detections than sand and eelgrass habitats. The effect of tide was not significant on its own, but a significant habitat‐tide interaction was documented, with the most pronounced differences in taxa between tides found in sand habitats. Power analyses indicated that additional field sampling is useful to detect small changes in species richness such as those due to tide. PCR replicates typically identified few additional taxa. Our results provide important information that can be used to guide future studies, most notably, that the influence of tide on eDNA results appears to be minimal and potentially isolated to certain habitats. This suggests that replication across tides may not be vital for future eDNA studies and that additional replication across space—particularly across heterogeneous environments—likely is a better allocation of sampling effort.