18S-V9 DNA metabarcoding detects the effect of water-quality impairment on stream biofilm eukaryotic assemblages

DNA metabarcoding is rapidly expanding as a new approach to biodiversity assessments and biomonitoring and is especially valuable for characterizing microbial communities in aquatic habitats. When applied to eukaryotic organisms, metabarcoding is usually targeting specific taxonomic groups, such as macroinvertebrates, fungi, diatoms, or other protists. The goal of this study was to explore the potential use of metabarcoding of entire biofilm eukaryotic assemblages for the purpose of stream biomonitoring. We sampled 14 stream sites in New Jersey, USA along an impairment gradient and characterized rock biofilm assemblages using Illumina Mi-Seq sequencing of the V9 hypervariable region of 18S rDNA following the Earth Microbiome Project (EMP) protocol. We also enumerated diatoms from the same samples to compare DNA metabarcoding results with morphological assessments. Among the 5866 unique rDNA sequence variants, the fungal and holozoan sequences were the most diverse, while diatom sequences were the most abundant in most sites. Among-site variability of assemblage composition was significantly higher than within-site variability of field and lab replicates, which indicates an acceptable level of reproducibility of the method. Different taxonomic groups of eukaryotes exhibited similar, but not identical patterns of assemblage variation in response to underlying environmental gradients. Both morphological and metabarcoding approaches recovered strong relationships between diatom assemblage composition and water quality impairment. Several other groups of eukaryotes, such as fungi, peronosporomycetes, green algae, and holozoans had only slightly weaker response to water quality impairment than diatoms. These findings suggest that molecular characterization of biofilm eukaryotic assemblages can be an effective tool for monitoring stream biota and its responses to disturbance even if the taxonomic assignments of sequences are only partially resolved.