RESULTS AND DISCUSSION

Distribution of SNPs: We sequenced 50 noncoding segments in nine bonobos and 17 chimpanzees from East, Central, and West Africa. The total number of nucleotide sites sequenced, after exclusion of deletions and insertions (mostly single nucleotide indels), is ∼23,500.

A total of 186 single nucleotide polymorphisms (SNPs) were found in the 17 chimpanzee samples (34 sequences); 51 of them were observed only once (i.e., singletons) and 15 only twice (doubletons). The number of variant sites found was 54 in the 12 West African chimpanzee (P. t. verus) sequences, 101 in the 10 Central African chimpanzee (P. t. troglodytes) sequences, and 39 in the 4 East African chimpanzee (P. t. schweinfurthii) sequences. Thus, many more variants were found in the Central African subspecies than in the West and East African subspecies, indicating a much higher DNA diversity in the Central African subspecies. The numbers of singletons were 14, 58, and 27 in the West, Central, and East African chimpanzee sequences, respectively. Thus, in the Central and East African subspecies, more than half of the variants were singletons, whereas in the West African subspecies less than one-third of the variants were singletons with an equal number of doubletons. A total of 63 SNPs were found in the 18 bonobo sequences; there were 21 singletons and 11 doubletons. Clearly, bonobos are less polymorphic than each of the three chimpanzee subspecies.

Adequacy of the sample sizes: As our sample sizes are relatively small, we need to consider the problem of sampling bias. For this purpose, we consider the effect of sampling on nucleotide diversity (π) because π is the quantity of our primary interest in this study; π is defined as the number of nucleotide differences per site between two randomly chosen sequences in a population. As noted in Yu et al. (2002), an estimation bias may be detected by comparing within-individual π values (π_w) with between-individual π values (π_b). Ideally, each sequence in a sample should be taken randomly from the population, but we have included the two sequences within each of the individuals sampled. The two sequences in an individual are not completely independent if the individual is “inbred” to some extent. Anyway, the within-individual π values (π_w) should tend to be smaller than the between-individual π values (π_b) and their inclusion should tend to give an underestimate of π. However, if the average π_b and π_w values are similar, then the sampling scheme would seem largely adequate and the inclusion of π_w values in the estimation of π should produce no substantial bias. To simplify the analysis, we concatenate the segments in an individual in a random manner into two continuous sequences and these sequences are then used to compute the π_b and π_w values.

For the Central African chimpanzee sequences, the distribution of π_b values, which ranges from 0.098 to 0.174%, is only somewhat wider than that of the five π_w values, which ranges from 0.102 to 0.153%. Since the average π_b (0.131%) is <10% higher than the average π_w (0.121%), the sampling bias should not be strong. A similar comment applies to the West African chimpanzee sample. There is one very low π_w value (0.051%) and the average π_w value (0.072%) is ∼10% lower than the average π_b value (0.082%). The average π_w without the outlier becomes 0.077%, which is not far from the average π_b (0.082%). This comparison suggests that the estimated π value (0.082%) in this subspecies is probably somewhat biased downward. For the East African chimpanzee sample, the average π_w value (0.088%) is close to the average π_b value (0.092%). This may suggest no substantial bias. However, because only two individuals were sampled, the estimate may not be reliable and should be taken with caution.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

—Distributions of the within-individual (π_w; □) and between-individual (π_b; ♦) nucleotide diversity values in bonobos. (a) All of the 18 sequences are included. (b) One individual (Bosondjo) and one sequence (randomly chosen) from each of three individuals (Kakowet, Lody, and Linda) are excluded.

For the bonobo sample, the distributions of π_w and π_b values are given in Figure 1a. Several π_w and π_b values are <0.04%, whereas most of the others are considerably >0.04%. This observation suggests that some of the individuals are fairly closely related to each other or inbred, although they were originally chosen to be independent. Indeed, the four π_b values between Bosondjo and Maringa (studbook numbers 64 and 60, respectively) were only 0.034, 0.034, 0.038, and 0.047%. We therefore excluded Bosondjo from comparison. Moreover, the π_w value is only 0.030% for Kakowet (studbook no. 34) and 0.038% for Lody and Linda (studbook nos. 68 and 23). We therefore excluded one sequence (randomly chosen) from each of these three individuals. After the exclusion of these sequences, only 13 sequences remain and the new distributions of π_w and π_b values are given in Figure 1b. The exclusion of the above sequences increases the average π value from 0.075 to 0.078%. The latter value will be used as our estimate.

Nucleotide diversity: For the 50 DNA segments we obtained, the range of π is from 0 (14 segments) to 0.39% in the West African chimpanzee sample, from 0 (7 segments) to 0.46% in the Central African chimpanzee sample, from 0 (23 segments) to 0.45% in the East African chimpanzee sample, and from 0 (3 segments) to 0.46% in the entire chimpanzee sample (Table 1). The range of π is from 0 to 0.36% in the bonobo sample and from 0 to 0.30% in the human sample (Table 1). Such large fluctuations are not surprising because the nucleotide diversity in a short DNA region is subject to strong stochastic effects. In addition, variation in π may also arise from variation in mutation rate among genomic regions.

Table 2 shows that Central African chimpanzees have the highest average π value (0.130%), followed by East African chimpanzees (0.092%), and then West African chimpanzees (0.082%). As mentioned above, the π value estimated from the East African chimpanzee sample may not be reliable because of a small sample size and that from the West African chimpanzee sample might be biased downward. We note further that in previous studies the East African chimpanzee had a greater π at APOB and PABX but a smaller π at the HOXB6 intergenic region and at the mtDNA ND2 and the mtDNA control region than did the West African chimpanzee (Table 2; Wiseet al. 1997; Deinard and Kidd 2000; Stoneet al. 2002). Therefore, further data are needed to see whether the π value for the East African chimpanzee is actually higher than that for the West African chimpanzee.

Surprisingly, the average π value in bonobos (0.078%) is somewhat lower than that in humans (0.088%; Table 2). The observation that bonobos have lower nucleotide diversity than humans is in agreement with the Xq13.3 data, but contrary to the HOXB6, DRD4, DRD2, and NRY regions. Furthermore, the average π values in the East and West African chimpanzee subspecies (0.092 and 0.082%) are similar to that in humans. The Central African chimpanzee is the only subspecies that has a π value (0.130%) higher than that in humans and the difference is only 50%. We note further that even the highest π_b value for the concatenated sequences in the Central African chimpanzee sample is only 0.174% (see above), which is only two times the average π value in humans. When all chimpanzee sequences are considered together, the average π value (0.132%) is again only 50% higher than that in humans. Actually, if we consider African humans only, the π value for the 50 DNA segments becomes 0.115% (Yuet al. 2002), which is only 13% lower than that for chimpanzees.

Previous reports have suggested that chimpanzees have two to four times greater amounts of π than humans at several autosomal nuclear loci, including the noncoding intergenic regions near HOXB6 and DRD4 (Deinard and Kidd 1999; Jensen-Seamanet al. 2001), the intronic regions of DRD2 and AHD1 (Deinard and Kidd 1998; Jensen-Seamanet al. 2001), and synonymous sites at six protein coding loci (Satta 2001). Similar results were also observed at the X chromosomal locus Xq13.3 (Kaessmannet al. 1999) and the Y chromosomal NRY locus (Stoneet al. 2002). However, since our data set has a much wider genomic representation, it should be more reliable. The stochastic effects of examining only a small number of loci can be seen, for example, in that in one-third (17 of 50) of the segments that we examined chimpanzees had more than three times higher π than humans (Table 1), which is approximately the value that was found by others when looking at a single locus (Table 2; Deinard and Kidd 1999; Kaessmannet al. 1999). Thus, the difference in nucleotide diversity between humans and chimpanzees is considerably smaller for nuclear DNA than for mtDNA data (Table 2). The disparity in using mtDNA vs. nuclear DNA in comparisons between humans and chimpanzees was originally pointed out by Wise et al. (1997), who in further comparisons to other nonhuman primates suggested that humans, not chimpanzees, were unusual in possessing such low levels of mtDNA diversity relative to that of the nuclear genome.

Another measure of genetic variability is the number of segregating alleles in the sample. However, because the sample sizes are different for different populations, we consider θ= 4N_eu, where N_e is the effective population size and u is the mutation rate per site per generation. The θ values estimated from the numbers of segregating sites by Watterson’s estimator (Watterson 1975) are given in Table 2. We note that, compared to the difference in π (0.088 vs. 0.078%) between humans and bonobos, the difference in θ (0.123 vs. 0.082%) is even larger (Table 2), probably reflecting an increase in low-frequency alleles due to a recent population expansion in humans. The highest θ value for the three chimpanzee subspecies (0.152%) is that for Central African chimpanzees, but it is only 24% higher than that for humans. When all chimpanzee sequences are considered together, θ becomes 0.194%, which is only 50% higher than that in humans.

Effective population sizes: To estimate effective population size (N_e) we estimate the mutation rate per nucleotide site per generation (u) by using the sequence divergence (d) between species (Table 3) and assuming that the divergence time between the human and chimpanzee-bonobo lineages is 6 MY (Brunetet al. 2002; Vignaudet al. 2002). Since we are interested in the long-term effective population size, we use Tajima’s estimator, π= 4N_eu (Tajima 1983), and we assume that the generation time is 15 years for chimpanzees and bonobos and 20 years for humans. The N_e for humans is estimated to be 10,500 (Table 2), which is similar to the commonly used value (10,000) in the literature (Nei and Graur 1984; Takahataet al. 1995; Zhaoet al. 2000), while that for bonobos (12,400) is only slightly larger and that for the Central African chimpanzees (20,900), or for the entire species of chimpanzees, is about twice as large.

Causes for different patterns of nuclear DNA and mtDNA diversity: As noted above, for nuclear DNA the nucleotide diversity in humans is only 50% lower than that in chimpanzees, whereas previous studies have found the mtDNA nucleotide diversity in humans to be at most only one-third of that found in chimpanzees (Wiseet al. 1997; Stoneet al. 2002). What are the causes for this sharp contrast? A highly plausible cause is a reduction in the effective population size in the human lineage since the human-chimp divergence. There is strong evidence for this putative reduction (Ruvolo 1997; Harpendinget al. 1998; Chen and Li 2001). As noted by Fay and Wu (1999), a reduction in N_e causes a larger decrease in nucleotide diversity for mtDNA than for nuclear DNA. This follows from the theory that the effect of a bottleneck on π is proportional to T/N₁, where T is the time since the bottleneck and N₁ is the new effective population size, and from the fact that the effective population size for mtDNA is usually smaller than that for nuclear DNA (Takahata 1993). An additional factor is that there has been an increase in generation time in the human lineage, which leads to a higher mutation rate per generation and also to a higher expected π value in the case of nuclear DNA but to little increase in the case of mtDNA because of maternal inheritance and limited germ cell divisions per generation in the female germline.

Another possibility is that population subdivision might have been stronger in humans than in chimpanzees or the female migration rate might have been lower in humans than in chimpanzees, so that the effective population size for mtDNA is considerably smaller in humans than in chimpanzees (Birkyet al. 1989). Evidence to support this argument may come from the observation that chimpanzee females uniformly disperse from their natal troop, in contrast to the typical mammalian (and primate) pattern of female philopatry with male dispersal (Greenwood 1980). The potential for high rates of female migration in chimpanzees is seen in the long-distance (>900 km) sharing of mtDNA haplotypes (Morinet al. 1994; Goldberg and Ruvolo 1997).

Divergences between subspecies and species: The average sequence divergence (d; average nucleotide differences per site) between chimpanzee subspecies is the smallest between the West and East African chimpanzees (Table 3), although in terms of geographic distance it should be the largest. The smaller divergence occurs because the levels of nucleotide diversity in the East and West African chimpanzees are smaller than that in the Central African chimpanzees. For the same reason, the d value is the largest between the Central and East African subspecies. The between-subspecies d values are only slightly higher than the within-subspecies π values, indicating a small separation between subspecies.

The sequence divergence between the bonobo and the chimpanzee (0.373%) is about three times higher than the nucleotide diversity within the chimpanzee species (0.132%), indicating a much longer separation time between the two species than between the chimpanzee subspecies. The sequence divergence between the human and the chimpanzee is 1.22% (Table 3), which is similar to the values (∼1.2%) obtained on the basis of >2 million bp (Chenet al. 2001; Ebersbergeret al. 2002; Fujiyamaet al. 2002). The sequence divergence between the human and the bonobo (1.30%) is somewhat higher than that between the human and the chimpanzee, probably because the bonobo has a smaller effective population size than the chimpanzee and so has been subject to a stronger effect of random drift.

These data also provide the largest and most comprehensive estimate of divergence time between chimpanzees and bonobos, estimated here to be 1.8 MYA, assuming a 6-MYA Homo-Pan split (Brunetet al. 2002; Vignaudet al. 2002). This estimate is the same as that by Stone et al. (2002), based on their data from the nonrecombing region of Y, and is intermediate between published data using mtDNA (2.5 MYA; Gagneuxet al. 1999) and X chromosomal DNA (0.9 MYA; Kaessmannet al. 1999). Also, the date of the formation of the Congo River, which currently prevents contact between these species, has been estimated from geological and limnological evidence to have formed ∼1.5 MYA (Beadle 1981). Perhaps the formation of the Congo River initiated the separation between chimpanzees and bonobos. This was also a time of potentially large climatic changes in the African climate and changes in vegetation patterns, proposed by some to have catalyzed the origins of several hominid species and human innovations (Vrba 1995).