Long vs. short read sequencing for microbial ecology of sedimentary environments: a case study from Lake Arnon, Switzerland

The subsurface biosphere remains poorly characterized, with many resident microorganisms uncultured and lacking genomic references. Despite the growing accessibility of shotgun metagenomics, 16S rRNA gene sequencing remains a standard tool for microbial community profiling, often relying on sequence similarity to reference databases such as SILVA to infer taxonomy and potential function. However, in environments with low biomass and high proportions of unknown lineages, such as deeper sedimentary environments, the accuracy of these inferences and our ability to capture rare taxa remain uncertain. A better inference of these rare taxa may now be possible with the advent of accurate long-read applications that have recently become available. Here we provide a comparison of long-read (PacBio) and short-read (Illumina NextSeq) 16S rRNA approaches for microbial communities from a sediment core of Lake Arnon (Switzerland). We compared community composition in environmental samples and mock controls to evaluate the strengths and limitations of each method. While sequencing technology significantly influenced observed community structure, sediment depth had an even stronger effect. Taxonomic profiles were broadly consistent across methods for most bacterial groups, but archaeal diversity was underrepresented in the long-read data, likely due to primer mismatch. When detected, long-read sequencing offered more accurate taxonomic resolution, often down to the species level, enabling better inference of metabolic potential. Beta diversity patterns were similar at broad taxonomic levels between methods, though more detailed metrics such as species contributions to beta diversity (SCBD) and co-occurrence networks showed enhanced resolution and specificity in long-read datasets. Our results highlight the critical importance of primer design, in particular for capturing archaeal taxa that play important roles in the deep biosphere. With improved primer coverage and continued cost reductions, long-read sequencing holds strong potential for advancing our understanding of subsurface microbial identity, structure and function.