We rstly make a direct comparison among these tested bacteria. Then we compare these estimates to the results from previous research [69, 68, 102, 371, 388] on distantly related STs (non-SLVs), and also to the results from previous research on the STs with SLVs. The direct comparison of several bacteria shows that estimates of
ρ/θvary (listed by order from small to large): N. gonorrhoeae (0.07); K. pneumoniae
(1.75); B. cereus (1.97); S. zooepidemicus (2.54); N. lactamica (2.54); E. faecium (3.09); S. uberis (7.17); H. inuenzae (17.13); N. meningitidis (19.39); and S. aureus (3044.21). Based on previous studies [69, 68, 102, 105, 371, 388], the per-site r/m
ratio for dierent bacteria varies, and is listed from low to high: S. aureus (0.1); K. pneumoniae (0.3 [388]); B. cereus (0.7 [68, 388]; 1.3 to 2.8 [69]); E. faecium (1.1 [388]); H. inuenzae (3.7 [388]); Neisseria meningitidis (5 [69], 7.1 [388], 80 to 100 [102, 105]); N. lactamica (6.2 [388]); and S. uberis (226 [371]). This comparison shows the dierence between r/m and ρ/θ estimates, and one estimate cannot be
used to infer the other. For each bacteria, if the distribution of the number of dierences is similar to each other (or similar to the total distribution), theρ/θratio
should have the same order of magnitude. The dierent mechanisms or evolutionary dynamics among species produce dierent distributions of the number of dierences for SLVs.
In these tables (Tables 4.5 to 4.14 in section 4.6), the range is from 1.75 to 19.39, ex- cept for S. aureus (3000+) and N. gonorrhoeae (0.07). The extreme values for these two species may be due to several reasons. The main reason could be the dierent ability to recombine across dierent bacteria. N. gonorrhoeae can only be found in an infected person, or persons who have had sexual contact with infected persons. The reason for the limited ability of N. gonorrhoeae to adapt a new environment could be lack of recombination. In addition, the great ability of S. aureus to exist widely could be due to a higher ratio of recombination to mutation.
The dierences in the third column in Table 4.2 demonstrate the dierences between the estimation of the ρ/θ ratio from the average recombination and mutation rates
for each bacterium. This variation may demonstrate dierent evolutionary dynamics in dierent genes and species. Dierent bacteria have the dierent capacities for genetic exchange [101, 106, 225, 388], some bacteria can uptake of DNA from the extracellular environment and integrate the free DNA into their genomes, but some bacteria do not have this capacity [234].
Table 4.4 shows estimates in this chapter are much larger than the more recent analysis [388] on estimating the relative recombination to mutation. One explanation is that the estimates in this chapter have been calculated from closely related STs (SLVs), whereas Vos and Dedilot's analysis [388] works on distantly related ST. This comparison demonstrates that the purifying selection plays a role in the evolution of all the compared species: B. cereus, E. faecium, H. inuenzae, K. pneumoniae, N. lactamica, and Neisseria meningitidis.
A range of previous research on ρ/θ estimates show that the ratio for overall Neis-
seria is 0.7 to 1.2 [69] or 3.6 to 5 [102, 100]. The latter range is calculated on SLVs, whereas the former is calculated on more distantly related strains. This dierence already shows that the closely related strains in Neisseria have a higher ρ/θ ratio
[100, 102] than that of distantly related strains [69]. Similarly, the ρ/θ ratio for B.
cereus is 0.125 to 0.25 [68] or 0.2 to 0.5 [69]. Unlike estimate in this chapter for B. cereus (1.97), their smaller estimates have been calculated based on distantly related STs. This comparison shows that the purifying selection plays a role in the evolution of both Neisseria and B. cereus.
For analyzed bacteria, the estimates based on SLV strains are much larger than those based on the more distant strains, although slightly dierent estimates were obtained using dierent methods. The larger estimates for clonal strains indicate purifying selection plays a role for the tested species. Most of the previous research
on SLV strains are based on Feil's methods [101, 104, 105], which were originally put forward by Guttman and Dykhuizen (1994) [149]. As described in their research [101, 104, 105], Feil's analysis provides a lower bound of theρ/θ estimates. With the
increasing amount of MLST data, the per-event ratio of recombination to mutation can be calculated and rened.
Feil's method [101, 104, 105] also works on SLV data from MLST. Compared with Feil's method [101, 104, 105], in this chapter's analysis more STs are available and analysed, so the results are less biased by the sampling methods. In addition, the new method put forward in this chapter is less labour intensive compared with Feil's and other studies based on manual inspection. In addition, the estimates are calculated for each locus, rather than only one value used to represent seven genes across the genome.
The estimate in this chapter for S. aureus is consistent with some previous public- ations [62, 102, 343]. However, the larger recombination eect than mutation for S. aureus is dierent from a previous study [100]. That study [100] indicates that the mutation eects are larger than the recombination, and also identied the error in one previous publication [62]. Feil et al. [100] claim that some errors were found in the data used in Day's analysis [62]; therefore, they used the revised data to cal- culate the recombination and mutation eect for S. aureus. Both studies [62, 100] applied Feil's methods. One limitation of Feil's method is that the analysis is based on a limited number of STs and SLVs (there are 35 SLVs out of 75 unique STs). Recent research on the comparison of recombination to mutation in S. aureus [18] indicates that the role played by recombination may be quite large, but the result in that paper is not conclusive.
Combined with the results for Campylobacter, recombination occurs more frequently than mutation for the majority of tested bacteria, except for N. gonorrhoeae. The smaller ratio of recombination rate to mutation rate for N. gonorrhoeae may be due to the limited sample size. The overall results show that recombination can occur more frequently than mutation for a range of bacteria. Furthermore, purifying selection may act more stringently on recombination events than on point mutations for a range of bacteria.
As mentioned in a previous review [71], the evolution of clonality can be dierent from that of STs that are distantly related in their evolutionary history. The evol- utionary rates within clonality for several bacteria have been calculated, and the results show the evolutionary history within clonal complexes. These kind of STs have less chance of experiencing natural selection, but more chance of reecting the real evolutionary dynamics than the distantly related STs. Future work on the mechanism of recombination and selection pressure is needed.