Filter pore size influences taxonomic composition of retained eDNA from seawater samples—evidence from shotgun sequencing

Environmental DNA (eDNA) is increasingly used in aquatic research and applied biodiversity monitoring, mainly through a metabarcoding approach. This allows detailed analyses of specific taxa of interest but involves amplification by PCR, which can lead to significant taxonomic bias. On the other hand, broad tree-of-life approaches such as shotgun sequencing are easily dominated by prokaryotic DNA, failing to detect less abundant taxa such as metazoans. However, the proportion of eukaryotic DNA could potentially be increased through water filtration with larger pore sizes. In this study, we investigate how filter pore size influences the taxonomic composition of retained aquatic eDNA. We conducted high-throughput shotgun sequencing and 18S rDNA metabarcoding on 15 seawater eDNA samples collected at the same locality but filtered with different pore sizes spanning from 0.2 to 8.0 µm. We could assign only 0.78% of the total shotgun reads to the superkingdom level, but smaller pore sizes (0.2 and 1.2 μm) yielded significantly higher proportions of bacterial reads compared to eukaryotic reads (63 vs. 28%), while larger pore sizes (5.0 and 8.0 μm) yielded significantly higher proportions of eukaryotic reads compared to bacterial reads (49 vs. 31%). Of the 19 metazoan phyla detected with shotgun sequencing, all but one were more abundant in larger pore sizes. Most of the eukaryotic phyla detected (39 of 54) were shared between shotgun and metabarcoding analyses, indicating similar performance of the methods regarding presence/absence estimation at high taxonomic levels. Finally, we compared the reliability of genus-level detections between the two methods for selected groups of metazoans to evaluate the potential of shotgun sequencing for biomonitoring. Our study suggests that the prokaryotic dominance in shotgun sequencing data from aquatic eDNA can be reduced by increasing filter pore size. As reference databases expand, the taxonomic coverage and resolution of shotgun sequencing should improve, likely enhancing the potential of shotgun sequencing for future eDNA research.