Chromosome assembly and phylogeny reconstruction
Illumina reads were first trimmed for Illumina MiSeq adapter sequences using Trimmomatic v. 0.38 (Bolger, Lohse, & Usadel, 2014) before being assembled using SPAdes v. 3.13.0 (Bankevich et al., 2012), which has been shown to be the best option for de novo assemblies ofBorrelia genomes (Becker et al., 2020). Pacific Bioscience sequences were obtained for three B. bavariensis isolates (PBi, A104S, and NT24) (Becker et al., 2020) and three B. gariniiisolates (PHeI, PBr, and NT31; see Suppl. Met.). Additionally, threeB. afzelii chromosomes were downloaded from GenBank for use as references and inclusion in all analyses: PKo (CP009058.1), K78 (CP002933.1), and ACA-1 (NZ_ABCU00000000.2). SPAdes contigs were then mapped to reference chromosomes using NUCmer v. 3.23 from the package MUMmer (Delcher, Phillippy, Carlton, & Salzberg, 2002; Kurtz et al., 2004). Final chromosomes were produced according to the mapping protocol outlined in Becker et al. (2020) (see Suppl. Met.). Three additionalB. bavariensis chromosomes were downloaded from GenBank and used in further analyses: SZ (CP007564.1), BgVir (CP003151.1), and NWJW1 (CP003866.1).
Final assembled chromosomes were aligned using MAFFT v. 7.407 (Katoh, Misawa, Kuma, & Miyata, 2002; Katoh & Standley, 2013). Recombination is known to be low on the Borrelia chromosome (Gatzmann et al., 2015) but as recombinant regions could bias the phylogenetic signal, we searched for areas of the chromosome violating the four-gamete condition (Richard R Hudson & Kaplan, 1985) (as described in Gatzmann et al. (2015); see Suppl. Met.). Regions with strong violation of the four-gamete condition were assumed to be recombinant and were removed from the final alignments (final alignment length: 936908bp). Phylogeny reconstruction was done in MrBayes v. 3.2.6 (Huelsenbeck & Ronquist, 2001; Ronquist et al., 2012) with ploidy set to haploid and a GTR (Tavaré, 1986) substitution model with gamma distributed rate variation. Three independent runs were launched and ran for 5 million generations at which point convergence of parameters was checked with Tracer v. 1.7.1 (Rambaut, Drummond, Xie, Baele, & Suchard, 2018). Consensus trees were built using the sumt command from MrBayes using a respective burn-in of 25%. Convergence to a single topology in all three independent runs was checked manually in FigTree v. 1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/) which was also used to plot the tree shown in Figure 2. Trees were midpoint rooted on the longest branch, which corresponded to the well-established delineation betweenB. afzelii and the monophyletic group containing B. bavariensis and B. garinii (Becker et al., 2016; Gabriele Margos, Vollmer, Ogden, & Fish, 2011).