To compare the extent of diversity among humans and the great apes, we calculated Watterson′s estimate of the parameter theta5, θ w , which is based on the number of variable positions and takes the sample size analyzed in each species into account. The results (Table 1) show that θ w in gorillas is about twice that in humans, whereas in chimpanzees and orang-utans it is about 3.1 and 3.5 times greater, respectively. One may argue that the more extensive diversity in the great apes is the result of more extensive population subdivision in these species. Although mtDNA data indicate genetically distinct subspecies6,7, nuclear DNA shows the subspecies of all the great apes to be intermixed1,8,9 (data not shown), indicating a rather recent separation of the subspecies. Thus, it appears to be justified to compare overall levels of diversity within the great ape species with those of humans. It is notable, however, that θ w is 1.6 to 4.0 times greater compared with the human estimate even when it is calculated for each of the ape subspecies separately (Table 1). The only exception is western African chimpanzees, for which only zoo samples, which may not represent the entire diversity of this subspecies, were available.

Table 1: Sequence diversity at Xq13.3 Full size table

The more extensive diversity in the great apes is reflected in the longer times to the most recent common ancestors (MRCA) in the ape species. The age of the MRCA for humans is about 540,000 years; for chimpanzees, 1,900,000 years; for gorillas, 1,160,000 years; and for orang-utans, 2,120,000 years (Table 1). A maximum likelihood10 tree (Fig. 1) relating the great ape and human DNA sequences illustrates the occurrence of deep branches within all great ape species, whereas humans constitute only a small cluster with short branches.

Figure 1: Phylogenetic tree10 of human (n=70; ref. 2), chimpanzee (n=30), bonobo (n=5; ref. 1), gorilla (n=11) and orang-utan (n=14) Xq13.3 sequences. Gorilla DNA samples (n=11) were obtained from zoos and primate research institutes. Orang-utan samples were from skin fibroblast cell lines (n=9; collected using remote biopsy darts) of wild orang-utans from both Borneo and Sumatra4 as well as from zoos and primate research institutes (n=5). We performed PCR amplification and sequencing as described1. We used a gibbon sequence as an outgroup. The maximum likelihood tree reconstruction was performed with PUZZLE 4.0 (ref. 10) assuming a Tamura-Nei model with γ-distributed rates. Full size image

Studies of mitochondrial DNA (mtDNA) have shown a three- to fourfold higher nucleotide diversity in chimpanzees compared with humans1, as well as a greater extent of mtDNA sequence variation in gorillas and orang-utans11. Thus, both mtDNA and Xq13.3, the only two loci for which comparative intra-specific DNA sequence information is available from humans as well as the great apes, yield a congruent picture in which humans are unique compared with the great apes in having little genetic variation. Although it is possible that this is due to selection, this is unlikely because the putative selective factors would have had to affect both loci in a similar way in all these species. Rather, these data indicate that the population history of humans differs from that of our closest evolutionary relatives. For example, one may speculate that archaic human forms such as Neandertals, who disappeared about 30,000 years ago12, carried additional Xq13.3 lineages not present in modern humans. Thus, had the Neandertals or other archaic humans survived until today, contemporary humans would perhaps have been more like the great apes in terms of genetic diversity.

To elucidate whether signals of past expansions in population size can be seen in the Xq13.3 sequences, we used Fu and Li′s D* test13, which compares the number of singletons and the total number of mutations. Assuming that the noncoding Xq13.3 sequences evolve selectively neutral, this test can be used to detect past population growth. Although the hypothesis of constant population size cannot be rejected for any of the great ape species, it is rejected in humans (P<0.02; Table 1). Thus, humans differ from the great apes in having both a low level of genetic variation and a signal of expansion at Xq13.3. To estimate the date of the beginning of the expansion, we used two different approaches, one based on the coalescent14 and the other based on mismatch distributions15. The maximum likelihood value from the coalescent analysis indicates that the human population started to expand approximately 190,000 years ago from an initial effective population size of about 3,700, whereas the mismatch approach indicates 160,000 years ago as the start of the expansion.

Mitochondrial DNA sequences indicate an expansion of modern humans 40,000 to 50,000 years ago16,17, a date that is associated with a change in human behaviour as indicated by a transition to more advanced and varied tool industries and the appearance of art. Due to its recent coalescence to one common ancestor, mtDNA may have `captured' only this more recent population expansion, whereas Xq13.3, which has a MRCA approximately 540,000 years ago, may reveal an earlier expansion in the history of modern humans starting 160,000 to 190,000 years ago. Because it is not possible to obtain true confidence intervals around these estimates with current methods, we cannot rigorously exclude that mtDNA and Xq13.3 reflect the same human population expansion.

We note that three other studies of nuclear DNA sequence variation in humans18 have failed to detect a population expansion. These DNA sequences are from transcribed genes that carry alleles implicated in disease and are therefore likely to be influenced not only by demographic phenomena but also by selection. It is possible that Xq13.3, which is noncoding, may be more suitable for elucidating historical demography. Two other recent studies of non-coding loci on chromosome 1 and 22, for which multiple human sequences of similar length were determined, also indicate a substitutional pattern consistent with a population expansion in humans19,20. It will be extremely important to study long DNA sequences from additional nuclear loci in humans as well as the great apes to elucidate whether a reduced diversity and a tendency to expansion relative to the great apes is typical for the human genome.