Abstract
In mammals sequences that are either late replicating or highly recombining have high rates of evolution at putatively neutral sites. As early replicating domains and highly recombining domains both tend to be GC rich we a priori expect these two variables to covary. If so, the relative contribution of either of these variables to the local neutral substitution rate might have been wrongly estimated owing to covariance with the other. Against our expectations, we find that sex-averaged recombination rates show little or no correlation with replication timing, suggesting that they are independent determinants of substitution rates. However, this result masks significant sex-specific complexity: late replicating domains tend to have high recombination rates in females but low recombination rates in males. That these trends are antagonistic explains why sex-averaged recombination is not correlated with replication timing. This unexpected result has several important implications. First, although both male and female recombination rates covary significantly with intronic substitution rates, the magnitude of this correlation is moderately underestimated for male recombination and slightly overestimated for female recombination, owing to covariance with replicating timing. Second, the result could explain why male recombination is strongly correlated with GC content but female recombination is not. If to explain the correlation between GC content and replication timing we suppose that late replication forces reduced GC content, then GC promotion by biased gene conversion during female recombination is partly countered by the antagonistic effect of later replicating sequence tending increase AT content. Indeed, the strength of the correlation between female recombination rate and local GC content is more than doubled by control for replication timing. Our results underpin the need to consider sex-specific recombination rates and potential covariates in analysis of GC content and rates of evolution.
Highlights
In mammals autosomal regions differ in the rate of evolution of putatively neutral sites [1,2]
Is it robustly found that replication time and the local recombination rate, defined multiple ways, both correlate with the intronic substitution rate? Second, is it true that recombination and replication time covary as we presume? If the second is true the former results would need to be analysed under a covariate controlled model
All recombination rate datasets that involved an element of smoothing resulted in stronger correlations with Ki than the gene-focused curation methods such as overall rates, weighted, base pair and interval averages (Table 1)
Summary
In mammals autosomal regions differ in the rate of evolution of putatively neutral sites [1,2]. Several reports have supported the possibility that genomic domains have characteristic replication times through the cell cycle, that these timings are evolutionarily conserved and that early replicating sequence, for reasons unknown, have low neutral rates of evolution [8,9,10]. Genomic domains have characteristic and conserved (on the megabase scale) recombination rates, with high rates being associated with high rates of evolution at putatively neutral sites [11,12,13,14,15]. Due to biases in the mismatch repair process [19], the latter process tends to favour fixation of G/C over A/T and has been suggested as a mechanism for the origin or maintenance of isochores ([20] and references therein, [21]) and can increase rates of evolution that are not at equilibrium [22]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
