Human and Ape Molecular Clocks and Constraints on Paleontological Hypotheses R. L. Stauffer, A. Walker, O. A. Ryder, M. Lyons-Weiler, and S. Blair Hedges Although the relationships of the living hominoid primates (humans and apes) are well known, the relationships of the fossil species, times of divergence of both living and fossil species, and the biogeographic history of hominoids are not well established. Divergence times of living species, estimated from molecular clocks, have the potential to constrain hypotheses of the relationships of fossil species. In this study, new DNA sequences from nine protein-coding nuclear genes in great apes are added to existing datasets to increase the precision of molecular time estimates bearing on the evolutionary history of apes and humans. The divergence of Old World monkeys and hominoids at the Oligocene-Miocene boundary (approx- imately 23 million years ago) provides the best primate calibration point and yields (a time and 95% confidence interval of 5.4 ؎ 1.1 million years ago (36 nuclear genes for the human-chimpanzee divergence. Older splitting events are estimated as 6.4 ,million years ago (gorilla, 31 genes), 11.3 ؎ 1.3 million years ago (orangutan 1.5 ؎ genes), and 14.9 ؎ 2.0 million years ago (gibbon, 27 genes). Based on these 33 molecular constraints, we find that several proposed phylogenies of fossil homi- noid taxa are unlikely to be correct. Fossils of the earliest hominoids (21 mil- preservation biases (Fleagle 1999). All of lion years ago) and the cercopithecoids these factors make it difficult to impose (Old World monkeys; 19 million years ago) time constraints on the origin of living are known from the early Miocene (Gebo species of hominoids. et al. 1997; Lewin 1999; Miller 1999; Pil- With such uncertainty in the hominoid beam 1996). Between then and the end of fossil record, considerable attention has the Miocene (approximately 5 million been focused on molecular clocks during years ago), hominoids decreased and cer- the last three decades. During the first half copithecoids increased in diversity in the of the 20th century, anthropologists as- fossil record (Fleagle 1999). Relating the sumed that the great apes formed a single Miocene apes to living species has proven evolutionary group distinct from the hu- to be problematic (Pilbeam 1996). There man lineage, with a divergence time of ap- is no fossil species that is clearly a close proximately 30 million years ago (Lewin relative of the gorilla, chimpanzee, or gib- 1999). However, the first applications of bon. It has been debated whether Sivapi- molecular techniques to this problem thecus (8–13 million years ago) or other showed that humans are closer to African From the Department of Biology and Institute of Mo- Eurasian fossil apes are close relatives of apes than to Asian apes (Goodman 1962) lecular Evolutionary Genetics (Stauffer, Walker, Lyons- Weiler, and Hedges) and Department of Anthropology the orangutan lineage (Pilbeam 1996; and the human-African ape divergence oc- (Walker), Pennsylvania State University, University Ward 1997). Although the skull of one par- curred only 5 million years ago (Sarich Park, PA 16802, and the Center for Reproduction of En- ticular Sivapithecus species from 8 million and Wilson 1967). Many molecular studies dangered Species, Zoological Society of San Diego, San Diego, CA 92112-0551 (Ryder). We thank Mary T. Silcox years ago is orangutan-like, postcranial have been published since then (Easteal for comments. This research was supported by a grant features and the morphology of the cheek et al. 1995) and have clarified the branch- from the Innovative Biotechnology Research Fund of teeth have suggested affinities with archa- ing order ((((human, chimpanzee) gorilla) the Biotechnology Institute, Life Sciences Consortium, Pennsylvania State University (to S.B.H. and A.W.). Ad- ic hominoids (Pilbeam 1996). With this un- orangutan) gibbon). However, divergence dress correspondence to S. Blair Hedges, Department certainty, the orangutan divergence is of time estimates have varied considerably of Biology, 208 Mueller Laboratory, Pennsylvania State University, University Park, PA 16802, or e-mail: limited value as a calibration point for mo- (Figure 1). If the ratios of the distances or [email protected]. This paper was delivered at a sympo- lecular time estimates. The absence of time estimates are considered, the results sium entitled ‘‘Primate Evolutionary Genetics’’ spon- Plio-Pleistocene fossil apes from Africa are more consistent among studies. This sored by the American Genetic Association at Town and Country Resort and Convention Center, San Diego, contrasts strongly with the rich hominid suggests that variation in time estimates CA, USA, May 19–20, 2001. fossil record during that same period and is largely attributable to the calibration ᭧ 2001 The American Genetic Association 92:469–474 is most likely explained by ecological and used in each study. 469 tween the origin of hominids themselves quence was available for Homo and at and bipedal locomotion (the first major least one other ape genus (Pan, Gorilla, hominid adaptation) and environmental Pongo, Hylobates), and (2) at least one cal- change. It also is possible that extinction ibration species (from Cercopithecidae, rather than speciation events are correlat- Artiodactyla, or Rodentia) and a mamma- ed with climate change (Foley 1994). lian or avian outgroup species sequence was available for relative rate testing. Fur- thermore, all Pan and Gorilla sequences Materials and Methods that were identical to the corresponding Portions of complementary DNAs (cDNAs) Homo sequence were deemed uninforma- from the following nine nuclear genes tive and were therefore eliminated. All were amplified and sequenced for Gorilla analyses were performed on both the gorilla and Pongo pygmaeus: acyl-coA: cho- group of rate-constant genes only and on lesterol acyltransferase I, alcohol dehydro- the entire dataset. genase 1, beta-glucuronidase, Cd 46, CMP- The relatively low pairwise distances for N-acetylneuraminic acid hydroxylase, in- most protein coding genes in these com- Figure 1. Molecular divergence time estimates for ␣ apes and human. The results of selected studies pub- terleukin- 1, prostaglandin D2 synthase, parisons of closely related species, com- lished during the last four decades are shown, where chemokine receptor 2, and muscarinic bined with limited sequence lengths, fa- an Old World monkey (cercopithecoid) also was in- acetylcholine receptor 5. A cDNA pool for vors the more variable nucleotide data (all cluded. Left panel shows the ratio of the human-ape divergence time divided by the hominoid-cercopithe- each species was created by reverse tran- three codon positions) instead of amino coid divergence time. Right panel shows the actual scription polymerase chain reaction (RT- acid data. For time estimation, the Kimura divergence times. Symbols represent the following di- vergences: human-chimpanzee (open circles), human- PCR) (Perkin-Elmer RNA Core kit). RNA (1980) two-parameter with gamma model gorilla (closed circles), human-orangutan (open was extracted using the RNAqueous kit was used, which accounts for rate varia- squares), human-gibbon (closed squares), and human- (Ambion, Inc., Austin, TX) from fibroblast tion among sites. The gamma parameter cercopithecoid (open triangles). The data are from the following studies: 1 (Sarich and Wilson 1967), 2 (Sibley cell cultures established and character- was estimated by maximum likelihood es- and Ahlquist 1987), 3 (Bailey et al. 1992), 4 (Easteal ized at the Zoological Society of San Diego timation ( Yang 1997) for each gene. Be- and Herbert 1997), 5 (Takahata and Satta 1997), 6 (Ku- (www.sandiegozoo.org/cres/frozen.html). tween-group distance estimation was mar and Hedges 1998), 7 (Arnason et al. 1998), 8 ( Yod- er and Yang 2000), and 9 (Page and Goodman 2001). Primers were designed from conserved re- made using PHYLTEST (Kumar 1996), and gions of the cercopithecoid and human se- two methods of time estimation were quences in the public databases. Gene used. The multigene method uses the To gain better and more precise esti- fragments were amplified (PCR) and com- mean (or mode) of single-gene time esti- mates of hominoid splitting we have col- plimentary strands were sequenced. Gene mates (Hedges et al. 1996; Kumar and lected new sequence data from nine nucle- fragments for each gene were combined Hedges 1998). The average distance meth- ar protein-coding genes in selected apes. and aligned using CLUSTAL W (Thompson od is similar, but averages the concatenat- Analyses of these data, along with all oth- et al. 1994). All primer sequences, align- ed distances, each weighted by sequence er available sequence data, have helped to ments, and sequence accession numbers length (Lynch 1999; Nei et al. 2001). Rate constrain hypotheses concerning the phy- for this project are available at http:// tests (Takezaki et al. 1995) were made for logenetic placement of important fossil www.evogenomics.org/publications/data/ all comparisons using PHYLTEST. hominoids. One major element of uncer- primate/. We used the hominoid-cercopithecoid tainty is the time of the human-chimpan- The other nuclear genes analyzed were divergence, set at 23.3 million years ago, zee divergence. Although the hominid fos- 5-hydroxytriptamine receptor 1a, alpha 1,3 as the primate calibration point. It is a fos- sil record is relatively good, there are no galactosyltrasferase, alanine: glyoxylate sil calibration point, because the earliest undisputed Pliocene fossils of African aminotransferase, atrophin, beta-nerve fossils of each lineage are known from 19– apes (chimpanzees and gorillas) and no growth factor, blue opsin, carbonic anhy- 21 million years ago (see above). The spe- Miocene ape fossils that clearly constrain drase, c-myc oncogne, cytochrome oxi- cific date used (23.3 million years ago) is a lower limit to that divergence. An advan- dase subunit 4, DDX5 (p68 RNA helicase), the geologic boundary between the Oli- tage of molecular time estimates is that decay accelerating factor, dopamine 4 re- gocene and Miocene epochs (Harland et they measure the mean time of separation ceptor, dystrophin, eosinophil-derived al.