Sequence processing

The resulting raw reads from sequencing were joined, quality filtered, clustered and had chimeras removed using the DADA2 package (Callahan et al., 2016). Sequences were trimmed to excise forward and reverse primer sequences in methods utilized by Honeyman, Day & Spear (2018) in which the first 40 nucleotides of the forward primer sequences and the last 20 of the reverse primer were removed and then trimmed using a quality score of 2. 18S sequences were analyzed separately from 16S where the forward and reverse reads were combined by concatenating the reads directly, despite the absence of an overhanging region of 18S gene sequenced by both the forward and reverse reads. Both 18S and 16S taxonomic assignments were created using Silva v128 (Pruesse, Peplies & Glöckner, 2012). Sequences for each sample were filtered into either Eukaryota (18S) or Bacteria (16S). Chloroplasts and Mitochondria were removed from all 16S sequences. After processing and quality filtering for the 16S genes, 796,847 sequences were obtained across 69 samples with sequence depth ranging from 18 –19,498 sequences per sample and were rarefied to 5,638 sequences. The yield for 18S sequences with these primers was lower at 34,009 across 69 samples. Sequence depth ranged from 0 to 2,950 sequences per sample and were rarefied to 200 sequences. All subsequent analysis except differential abundance testing was based on the rarefied sequence tables.

All downstream analysis was conducted in R (v1.0.153), using the Phyloseq (v1.26.1) (McMurdie & Holmes, 2013), vegan (Oksanen et al., 2019), DESeq2 (Love, Huber & Anders, 2014), and ampvis2 (Andersen et al., 2018) packages.