LAUSR

Diatoms are used routinely to assess pollution level in rivers and lakes. Current methods are based on identification by light microscopy, which is laborious. An alternative is to identify species based on short DNA fragments and High-Throughput Sequencing (HTS). However a potential limitation is the incomplete coverage of species in reference barcode libraries. Usually these libraries are compiled by isolating cells, before culturing and sequencing them, which is tedious and often unsuccessful. Here we propose the use of rbcL sequences from environmental samples analysed by HTS. We set several criteria to ensure good sequence quality and correspondence with the target species observed in microscopy: the sequence needed to be abundant in the sample, and with no insertions nor deletions or stop codon, phylogenetic neighbour taxa had to correspond to neighbour taxonomic taxa expected from morphological observations. Four species from tropical rivers are given as examples, including one that is new to science.