Bioinformatic analysis
RSII subread files in BAX format were converted to the newer BAM format using “bax2bam” from PacBio SMRT tools 5.0.1, and reads were demultiplexed using “lima” from PacBio SMRT tools 7.0.1 using the options “–different” and “–peek-guess”. Sequences which were not assigned to one of the barcode pairs used in this experiment were discarded. Circular consensus sequences (CCS) were generated from the demultiplexed BAM files using “ccs” from PacBio SMRT tools 5.0.1 (the last version which supports RSII data), resulting in 49,709 reads. Sequences were oriented in the forward direction by matching the forward and reverse primer sequences using Cutadapt v.3.0 (Martin, 2011). Only reads with both a forward and a reverse primer sequence in the correct orientation (ITS1 and reverse-complemented LR5) were retained. Concatamers (Griffith et al., 2018) were identified by searching for the primer sequence pairs ITS1/reverse-complemented ITS1, LR5/reverse-complemented LR5, ITS1/ITS1, and LR5/LR5 within the forward and reverse strands of each of the reads, and if detected, the read was discarded. Remaining reads were length and quality filtered, allowing for read lengths of 50 - 2999 bp and a maximum of 12 expected errors per read. Filtering was performed within the AmpliSeq pipeline, or using VSEARCH version 2.15.1 (Rognes et al., 2016) for OTU clustering methods, which did not use AmpliSeq.