I am analysing PacBio Sequel II full-length 16S rRNA CCS reads (~1450 bp) using the DADA2 long-read workflow and observing an unusually high number of unique sequences.

Almost all reads appear unique (e.g., ~11,200 unique reads from ~12,300 total reads). After denoising (learnErrors, dada, chimera removal), only a small number of reads remain.

Is such a high unique/read ratio normal for PacBio full-length 16S CCS data? Could this be related to sequence orientation, primer trimming, or filtering parameters?

Any suggestions for diagnosing or resolving this issue would be appreciated.