Assessing differences between full-length and V4 datasets
To generate a short-read dataset from our full-length sequences, we repeated the DADA2 pipeline but used V4 universal bacterial primers (V4_F = GTGCCAGCMGCCGCGGTAA V4_R = GGACTACHVGGGTWTCTAAT ) in place of PacBio sequencing primers, then repeated all downstream bioinformatics steps. To compare inference between datasets, we produced new phylogenetic trees and repeated our beta diversity analyses using this V4 dataset. The effects of different predictors on the beta diversity of our V4 dataset were assessed via adonis PERMANOVA tests, as with full-length sequences. In addition, to determine which V4 ASVs comprised which full-length ASVs, we mapped ASVs in the full-length dataset to their matching V4-only ASVs using the command bb.split.sh , part of the BBTools (Bushnell 2014) package.