Apart from its contribution to climate change, offshore oil and gas extraction is also a potential threat to the diversity and function of marine ecosystems. Routine monitoring of the environmental status of affected areas is therefore critical for proper management. While current morphology-based monitoring is relatively time-consuming, costly and prone to identification bias, environmental DNA metabarcoding offers an attractive alternative, including for impacts of oil drilling, extraction or spills. However, to be ready for routine monitoring, its performance needs to be demonstrated, through agreement with assessments based on physicochemical measurements and current bioindicators. To this end, we applied metabarcoding to sequence the metazoan and total eukaryotic benthic components. We targeted an range of sites, from high to low impacts, located near active production platforms and reference sites, in the North and Barents Seas. Alpha diversity and community structure of both datasets correlated strongly with a physicochemical pressure index (PI), based on total hydrocarbons, PAH16, Ba and Cu. The macroinvertebrate-based Norwegian Sensitivity Index (NSI) based on the COI metabarcoding data also agreed well with morpho-taxonomy based values, and with PI. Further, we identified a set of bioindicator taxa from both metabarcoding datasets, to develop novel biotic indices and demonstrate their predictive performance using cross-validation. Finally, we compared co-occurrence networks from impacted v less disturbed sites, to improve the understanding of the ecological consequences of impacts. Our study demonstrates that metabarcoding is comparable to the morpho-taxonomic approach in terms of accuracy, and could eventually be used to replace it, given further efforts.

Metabarcoding of environmental DNA (eDNA) is an attractive complement to morphological methods for characterizing marine sediment benthic communities. However, incomplete sampling is a major concern when inferring community composition, and metabarcoding results are heavily dependent on methodology. Using 18S V1-V2 and COI markers, we investigated the effect on observed alpha- and beta diversity of (A) homogenization intensity during sediment DNA extraction, (B) extraction replicates vs larger sediment extraction volume, and (C) pre- and post-PCR extract pooling. We show that an intermediate Precellys homogenizer program for DNA extraction can significantly improve sediment metabarcoding results in terms of captured diversity and inter-replicate homogeneity compared to vortexing only. This effect was stronger than that of increased sediment extract volume. Pre-PCR pooling of DNA extraction replicates increased observed rarefied richness compared to single extract medians, but not to the extent of amplifying or sequencing extraction replicates individually before pooling, i.e. post-PCR, or in silico pooling, respectively. We argue that this discrepancy was due to both an increased number of PCR artifacts and reduced PCR drift. Inter-sample heterogeneity was considerably higher for the COI metazoan dataset, compared to the total eukaryotic 18S dataset, likely due to a combination of metazoan eDNA distribution, stochastic effects due to less conserved primer sites, and a high degree of COI non-target amplification. Based on our results, extraction replicates of smaller sediment volumes, in combination with firm but intermediate homogenization and pre-PCR pooling, is a cost-effective way of maximizing sediment eDNA metabarcoding sample coverage, compared to increased extraction volume.