Multilocus sequence analysis (MLSA)

Orthologs were identified using Proteinortho5 (Lechner et al., 2011) with the “-selfblast” option (which enables paralog-detection) enabled. Only genes present exclusively in single copy in all compared genomes were selected for MLSA. Alignments of the respective gene product amino acid sequences were generated individually for each ortholog group using MUSCLE v3.8.31 (Edgar, 2004a,b) and subsequently concatenated. Unalignable regions, caused by e.g., unique N-terminal or C-terminal sequence overhangs, were filtered from the concatenated alignment using Gblocks v.0.91b (Castresana, 2000). Phylogenetic relationships were inferred from the remaining unambiguous alignment positions by Neighbor Joining clustering with 1000 bootstrap iterations, using ARB 6.0.5 (Ludwig et al., 2004) and by Maximum Likelihood calculation using RAxML v. 8.0.26 (Stamatakis, 2014).