Evidence for chimp-bonobo-human incomplete lineage sorting.

*Each divergent site class is designated by a string of 0's and 1's, the bases seen in chimp/bonobo/human/orangutan/macaque. The macaque allele is defined as state “0”.
†Under a model of incomplete lineage sorting, our EM analysis (Text S3) obtains a good fit between observed and expected, with nominal χ2 = 9, and a prediction that 27% of CH sites and 30% of BH sites are due to recurrent mutation. These results are concordant with our observation (Fig. 4) that very close to CB sites, the rate of CH and BH sites is reduced to 26±6% of the average. If we only allow a model with genealogies clustering chimpanzees and bonobos, the best fit has a nominal χ2 = 673.
‡We examined all 18,985 alignments, looking for ones where genealogical trees clustering CH or BH are favored over those clustering CB (>20,000:1 likelihood ratio), and in which the counts fit the proposed genealogies well (Text S8). Although these alignments should be treated with caution as they are extremes from a distribution, they are strong prospects for loci where chimpanzees and bonobos not being most closely related (alignment details are at genepath.med.harvard.edu/∼reich/Data%20Sets.htm).