Fig. 2

16S rRNA gene sequence processing steps for each of the 17 pipelines. All raw sequences began with identical adapter/primer removal by cutadapt. Paired-end sequences were then fed into each of the four trimming procedures: no trimming (N; red), length trimming by cutadapt (L; orange), quality trimming by cutadapt (Q; dark blue), and quality trimming by DADA2 (Qd; light blue). Non-trimmed paired-end sequences were piped directly into the merging/concatenating tools after primers were removed. Sequences designated for trimming were merged and/or concatenated after trimming by cutadapt or DADA2. Paired-end sequences designated for merging by DADA2 (Md) were first passed through DADA2’s quality filter and denoising procedures before being merged and chimeras removed. ASV curation immediately followed. Sequences run through the only Qd pipeline had all steps from trimming to ASV designations performed within DADA2. The majority of pipelines included length and quality trimming of paired-end sequences in cutadapt that were then piped into PANDAseq for merging (Mp) or merging and concatenating of those sequences unable to merge (Bp). Full concatenation (Cs) of paired-end sequences was done with a custom script after cutadapt trimming. Mp, Bp, and Cs sequences were fed into DADA2 as single-end where quality filtering, denoising, chimera removal, and ASV curation occurred. Non-trimmed single-end (NR1, NR2) and cutadapt length-trimmed single-end (LR1 and LR2) sequences did not go through a merging or concatenating procedure, but were submitted in the same manner to DADA2 as Mp, Bp, and Cs. Post-DADA2, taxonomic classification of ASVs was identical for all pipelines. ASVs were fed into QIIME 2’s feature-classifier classify-sklearn tool which aligned ASVs to one of two designated reference databases separately, Greengenes and SILVA. These reference database aligned files were then merged with the appropriate metadata files before taxonomic analysis in R