Duplex versus simplex nanopore 16S reads for eDNA-based impact assessment

Salmon aquaculture is expanding globally, increasing the need for efficient monitoring of benthic impacts caused by organic enrichment beneath production sites. Traditional assessments based on macrobenthic taxonomy and AMBI scoring are reliable but slow, costly, and dependent on expert identification, motivating the development of eDNA-based alternatives. Microbial community profiling via 16S rRNA sequencing has emerged as a sensitive indicator of aquaculture driven sediment change, yet most studies rely on short read Illumina data with limited taxonomic resolution. Advances in Oxford Nanopore Technologies, particularly duplex sequencing with Q30 level accuracy, offer the potential for high resolution, full length 16S rRNA profiling, though low duplex yields raise concerns about sequencing depth and feature stability in complex sediments. In this study, we benchmarked Random Forest models trained on Q30 filtered Nanopore duplex reads against models trained on substantially deeper simplex datasets from the same samples. Using macrobenthic AMBI scores as ground truth, we evaluated whether the higher per read accuracy of duplex data can compensate for reduced throughput and whether duplex datasets can match or surpass simplex performance in predicting benthic environmental quality under salmon aquaculture.