Fig. 2.

A paralog-swapping experiment addressing whether improvement to model fit can be attributed to IGC or to artifacts. Both Scenarios A and B specify the correct rooted phylogeny between L. kluyveri and the paralogs of S. castellii and S. cerevisiae . Scenario A shows the biologically correct situation that has IGC between paralogs in the same genome. In Scenario B , IGC homogenization events involve one paralog from S. castellii and one from S. cerevisiae . Because Scenario A corresponds to how observed data are generated, Scenario A should fit better than Scenario B if IGC is actually being detected. Note that this paralog-swapping experiment would not be possible if only 1 postduplication species was used and would not be effective with more than two postduplication species.