Interpreting phylogenetic placements for taxonomic assignment of environmental DNA.

Taxonomic assignment of operational taxonomic units (OTUs) is an important bioinformatics step in analyzing environmental sequencing data. Pairwise alignment and phylogenetic-placement methods represent two alternative approaches to taxonomic assignments, but their results can differ. Here we used available colpodean ciliate OTUs from forest soils to compare the taxonomic assignments of VSEARCH (which performs pairwise alignments) and EPA-ng (which performs phylogenetic placements). We showed that when there are differences in taxonomic assignments between pairwise alignments and phylogenetic placements at the subtaxon level, there is a low pairwise similarity of the OTUs to the reference database. We then showcase how the output of EPA-ng can be further evaluated using GAPPA to assess the taxonomic assignments when there exist multiple equally likely placements of an OTU, by taking into account the sum over the likelihood weights of the OTU placements within a subtaxon, and the branch distances between equally likely placement locations. We also inferred the evolutionary and ecological characteristics of the colpodean OTUs using their placements within subtaxa. This study demonstrates how to fully analyze the output of EPA-ng, by using GAPPA in conjunction with knowledge of the taxonomic diversity of the clade of interest.

Colpodean ciliate phylogeny and reference alignments for phylogenetic placements.

The Colpodea form a major clade of ciliates that are often found in environmental DNA sequencing studies. They are united by similar somatic ciliature, but differentiated by complex oral structures. Although there are four well supported colpodean subclades, there is disagreement in molecular phylogenetic inferences about their branching order. Using available nuclear SSU-rRNA sequences, we evaluated if the bursariomorphids or the platyophryids are sister to the remaining colpodeans. We inferred the "platyophryids-early" topologies using different alignment and masking methods, but constrained analyses could not reject the "bursariomorphids-early" topology. Both bursariomorphids and platyophryids clades have a similar number of nucleotide positions shared with the outgroup, and both are interconnected with the outgroup in phylogenetic networks. Based on these discordant results, it is hard to determine which clade branched off first, although the "platyophryids-early topology" is also supported by mitochondrial SSU-rRNA data. We also offer different reference alignments that can be used to phylogenetically place short- and long-read data from environmental DNA sequencing studies, and we propose some tentative evolutionary and ecological interpretations of those placements.