Diatom metabarcoding and microscopic analyses from sediment samples at Lake Nam Co, Tibet: The effect of sample-size and bioinformatics on the identified communities

Abstract

Abstract Diatoms (Bacillariophyceae) are characterized by silicified cell walls that favor their long-term preservation in sediments, therefore widely used as bioindicators of present and past water conditions. Alongside with traditional morphological analyses, metabarcoding has become a valuable tool to study the community structures of various organisms, including diatoms. Here, we test whether the quantity of sediment sample used for DNA extraction affects the results obtained from high-throughput sequencing (metabarcoding) of the diatom rbcL region by isolating DNA from 10\xa0g and 0.5\xa0g (wet weight) of lake surface sediment samples. Because bioinformatics processing of metabarcoding data may affect the outcome, we also tested the consistency of the results from three different pipelines: 1) ESVs (exact sequence variants) pipeline; 2) clustering sequences at 95% sequence identity to form OTUs (operational taxonomic units; 95% OTUs); and 3) 97% OTUs pipeline. Additionally, the agreement between metabarcoding data and morphological inventories of corresponding samples were compared. Our results demonstrate highly uniform patterns between the diatom rbcL amplicons from 10\xa0g and 0.5\xa0g of sedimentary DNA (sedDNA) extracts (HTS 10 and HTS 0.5, respectively). Furthermore, after the careful curation of the sequencing data, metabarcoding results were highly consistent among the data sets produced by different bioinformatics pipelines. Comparing results from metabarcoding and microscopy, we identified some taxonomic mismatches: morphological analyses identified 59 diatom genera, whereas metabarcoding 49 to 54 genera. These mismatches are related to incompleteness of the sequence databases, but also to inconsistencies in diatom taxonomy in general and potential dissolution effects of diatom valves caused by high alkalinity of the investigated lake waters. Nevertheless, multivariate community analysis revealed consistent results between data sets identified by microscopy and metabarcoding – water depth and conductivity as the most significant variables in driving diatom communities in Lake Nam Co – further confirming that metabarcoding is a viable method for identifying diatom-environment relationships.