16S rRNA Amplicon Sequencing Pipeline

Amplicon libraries were obtained using barcoded pyrosequencing primers 909F and 1492R, pooled, purified and quantified as described before (Berry et al., 2012 (link)). Pyrosequencing was performed with Titanium reagents on a 454 genome sequencer FLX (Roche, Vienna, Austria). All 20 samples collected on 5 sampling days were subjected to 16S rRNA amplicon sequencing, but one sample from day 1 was excluded from analysis due to low sequencing depth. Reads were quality-filtered using the amplicon pipeline of the GS Run Processor (Roche) and the Pyronoise algorithm in mothur (Schloss et al., 2011 (link)). Operational taxonomic units were generated at 97% sequence identity using UCLUST, as described (Berry et al., 2012 (link)). Amplicon sequencing libraries, which had a mean number of 17 999 reads (range [4415–26 495]) of an average length of 269 nt were re-sampled at 3500 reads for α- and β-diversity analyses using QIIME (Caporaso et al., 2010 (link)).