Newbler assembler v 2.3 (Roche) was used to batch sequen- ces based on MID-identifiers and to combine corresponding sequences derived from forward and reverse reads with a similarity of 99% and an overlap of 400 bases for the bacterial
sequences. Due to thelowsequence diversity, a similarity value of 98% and an overlap of 200 bases were sufficient for archaeal sequences. The consensus sequences were inspected for chi- mera with the help of the Bellerophon software (http://foo. maths.uq.edu.au/_huber/bellerophon.pl)andputativechimera were removed from the dataset.