3.5 Minimum spanning tree
Next, we used the core-SNP matrix of all M. bovis isolates to
infer a distance matrix (Appendix Figures 2 and 3, Appendix Table 2) and
a minimum spanning tree of the dataset (Figure 5). Traditionally, SNP
distances between genomes have been used to infer transmission ofM. tuberculosis between individuals. Thresholds varying from 5
and 12 SNPs have been proposed to determine if two genomes are from the
same M. tuberculosis strain (Meehan et al., 2019). By using this
same SNP threshold, at least four transmission links in animals from the
safari park, involving five deer, were detected (Figure 5). The SNP
distance between these genomes were: 4 (n= 2), 8 (n= 1) and
10 (n= 1) (Table 2). No transmission links between llamas and deer
were detected using this approach. This finding is further corroborated
by the fact that M. bovis from deer and from llamas carry
distinct spoligotype patterns (SB1401 and SB0295), which are highly
divergent in terms of the presence of spacer sequences (Appendix Figure
4). Multiple other transmission links (n= 28) were detected
involving cattle from vicinity herds, but no recent transmission could
be inferred between the safari park and cattle herds (Figure 5 and Table
1). The two closest M. bovis genomes of cattle to M. bovisgenomes of the safari park differed by 92 and 64 SNPs, which are 5 to 7
times higher than the SNP cut-off to infer recent transmission (Figure
5). Along with the dating estimates, these results indicate that
multiple M. bovis strains were circulating in the safari park at
the time of sample collection, resulting from at least three different
past introductions.