The molecular clock hypothesis posits that DNA substitutions accumulate at a constant rate through time, and that this evolutionary rate is homogeneous among lineages. However, this assumption is usually violated in practice, with lineages evolving at different rates. The causes of rate variation can be divided into lineage effects and residual effects. Lineage effects include the factors that cause a universal change across a genome, such as a change in underlying mutation rate or a change in generation time. In contrast, residual effects are heterogeneous across the genome and include factors such as selection.

The Universal Pacemaker (UPM) Model of genome evolution ascribes all important rate variation to lineage effects. According to this model, all genes share the same pattern of among-lineage rate variation. At the other end of the spectrum, the Degenerate Multiple Pacemaker (DPM) model proposes that each gene has a distinct pattern of among-lineage rate variation. Between these extremes is the Multiple Pacemaker (MPM) model, whereby genes are clustered into distinct pacemakers.

We developed a computational framework to test among these pacemaker models of genome evolution. We used it to analyse 426 genes from a broad taxonomic range of mammal species. Our results favour the MPM with 3 to 18 pacemakers, with poor support for the UPM and the DPM. We also investigated whether the pacemakers are associated with gene function and family. Although the exact number of pacemakers will vary among data sets, we suggest that genomic rates of evolution are governed by several discrete pacemakers, implying that the MPM should generally have higher support than the UPM and the DPM.