Community metabarcoding is robust to sampling device bias

In this study we test if the community metabarcoding method is able to differentiate meiofauna communities inhabiting different sediment types in the German Wadden Sea along the coast of Schleswig-Holstein. Further we test if there is a significant difference in results obtained by Multiple Corer (9.8 cm inner diameter) samples vs Van Veen grab (0.1 m²) samples. The samples were collected in May 2020 at 25 locations representing 5 habitat types (Sandbank “SB”, Circalittoral Mud “CMUD”, Circalittoral Sands “CS”, Circalittoral Mixed Sediments “CMS”, and areas with coarse sand, gravel and dead shells “KGS”) between 8 m and 56 m depth. Organisms were extracted from the sediment by centrifugation on Levasil®, bulk DNA was extracted from the organism and the COI gene was amplified and sequenced on a MiSeq platform following a metabarcoding approach. Initial 1,635 amplicon sequence variants 'ASVs' recovered by DADA2 could be assigned to meiofauna taxa, but these were grouped into 889 GMYC taxonomic units 'GTUs' using the Generalized Yule Coalescent Model. Highest number of GTUs belong to Nematoda (456), followed by Arthropoda (217), Platyhelminthes (151) and Gastrotricha (46). At order level the highest number of GTUs belonged to Harpacticoida (144) and to the nematode orders Chromadorida (107) and Desmodorida (99). Results show significant differences between the meiofauna communities at different habitat types with the exemption of CMS and KGS. CMUD had the highest number of unique GTUs (223) followed by SB (153) and CS (145). Unexpectedly there is no significant difference in the communities retrieved using Multiple Corer and Van Veen grab samples although these gears have very different sampling efficiency and sampling volumes. Samples of the 2 gears overlap in ordination space for the different habitats, showing that there is great agreement in taxon composition and dominance structure between gears. In general grab samples were slightly but not significantly less diverse than multicorer samples. We conclude that community metabarcoding is sensitive enough to recognize significant differences between the meiofauna communities at different sediment types and therefore a valuable method for routine meiofauna monitoring. We also conclude that the method is robust against bias introduced by sampling devices as no significant differences were observed between sampling gears. Finally we address that not having identified the meiofauna organisms morphologically is a limitation of this study as we cannot conclude how good the metabarcoding method is retrieving the actual meiofauna biodiversity present in the samples.