In circumstances not involving recombination, reduced genetic variation is usually seen as a consequence of stabilizing selection. Thus reduced variation is expected in genes coding for essential proteins. I think genetic recombination can also affect local variation, by changing the effective population size. But at the USS sites I think this is the result of biased uptake plus (unbiased) recombination. The post-doc coauthor of the paper has a nice explanation:
"The simplest interpretation of this reduced variation is that chromosomal sites containing USSs tend to recombine with a pool of internalized fragments containing relatively few mutations within the USS, presumably because of selection for strong USS by the DNA uptake machinery at the cell surface."We don't have any direct evidence of how often cells take up DNA in their natural environment (on respiratory tract epithelium if they're H. influenzae), nor how much of this DNA comes from other H. influenzae cells rather than from unrelated bacteria or the human host, nor how biased each uptake event is, given the pool of available DNAs, nor whether unidentified sequence factors affect the probability that a homologous fragment will replace the chromosomal copy. But the reduced variation may be telling us that the net effect of these unknown factors dramatically reduces the rate at which existing USSs diverge.