When Did Neanderthals and Modern Humans Diverge?

The availability of sequences from the mtDNA hypervariable region I and II of a specimen of the Neanderthal,1,2 Homo neanderthalensis, permits estimation of the time of divergence of Neanderthals and modern humans. Based on substitution rates derived from comparison of human and chimpanzee, Krings et al. provided an estimate of 550–690 KY1 and 317–741 KY2 before present for the split between the Neanderthal and contemporary human mtDNA. These estimates are dates for the split of the genes, and its confidence intervals are based on the uncertainty of the dating of the human-chimpanzee divergence (assumed to be 4–5 MY ago). The details of the estimation of the gene divergence time are only vaguely described in the seminal presentations of these Neanderthal findings. Vital information, such as values of the substitution rate and its shape parameter of the mutation rate distribution, is missing. Our reanalysis of the Neanderthal-modern human gene divergence resulted in gene divergence times in a range of 631–789 KY. We used a mutation rate ! of 0.0634 per site per million years from the human-chimpanzee divergence time of 5 MY, and of 0.0793 for the 4 MY estimate, and a substitution rate variation shape parameter " of 0.46 (cf. Excoffier and Yang,3 Wakeley,4 and our own analysis). We concentrate on the HVR I because we believe that there are some difficulties with the alignment in the HVR II. The interpretation of the three thymidine residues after site 340 as an insertion as presented by Krings et al.2 needs further investigation. This gene divergence time is informative, but the biologically more relevant parameter is the time of population divergence, which necessarily occurs after the observed gene divergence when there is any polymorphism in the ancestral population.5 Furthermore, the gene divergence between modern human and Neanderthal mtDNA is likely to be recent enough so that the difference in time between gene and population divergence may be substantial. In the light of fossil evidence,6 the estimates of Krings et al.1,2 may seem rather old, but morphological analyses7 suggest an even longer separation between modern human and Neanderthal lineages. Here, we provide an estimate and approximate confidence intervals of the population divergence time of Neanderthals and modern humans, using coalescence theory for single-locus data. Our analysis (Fig. 1A) and that of Krings et al.1 of many contemporary human mtDNA HVR I sequences and the single Neanderthal sequence strongly suggest that, going back in time, the modern human lineages coalesced with each other before coalescing with the Neanderthal sequence, and that subsequently this ancestral population coalesced with the chimpanzees. This simple sequence of events facilitates the estimation of population divergence time, making it unnecessary to use elaborate methods, such as those developed by Nielsen8 that take the uncertainty of the gene genealogy into account, because monophyly of contemporary human mtDNAs is not in question. When descendant lineages have achieved reciprocal monophyly, however, there is no information in a single locus left to provide information on the population size of the ancestors of Neanderthals and modern humans. With only a single locus, there is no information from the variance in coalescence time among loci that can be used to estimate ancestral population Peter Beerli is at the Department of Genome Sciences, Box 357730, University of Washington, Seattle, WA 98195. E-mail: [email protected] Scott V. Edwards is at the Department of Zoology, Box 351800, University of Washington, Seattle, WA 98195. Grant sponsor: National Institutes of Health; Grant numbers: GM 51929-02, GM 01989-02; Grant sponsor: National Science Foundation; Grant numbers: BIR 9527687, DEB 9707458, DEB 9815800.