The recent african physical characteristics by which they are classified necessarily reflect the models the paleontologists wish to test If one classifies, say, a pelvis as human Genesis of humans because it supported an upright pos ure, then one is presupposing that bi pedalism distinguished early from apes. Such reasoning tends to cir Genetic studies reveal that an African woman of tuve the ethr e cloeoatols g st s perspea 200,000 years ago was our common ancestor that limits its power of observation As such biologists trained in mod- ern evolutionary theory must reject the by Allan C. wilson and Rebecca L Cann notion that the fossils provide the most direct evidence of how human evolu tion actually proceeded. Fossils help to n the quest for the facts about hu- lived about 200,000 years ago, proba- fill in the knowledge of how biological man evolution, we molecular geneti- bly in Africa. Modern humans arose in processes worked in the past, but they cists have engaged in two major one place and spread elsewhere should not blind us to new lines of evi- debates with the paleontologists. Argu- Neither the gene tic information of liv- dence or new interpretations of poorly ing from their fossils, most paleontolo- ing subjects nor the fossilized remains understood and provisionally dated gists had claimed the evolutionary split of dead ones can explain in isolation chaeological materials between humans and the great apes how, when and where populations orig- occurred as long as 25 million years inated But the former evidence has a 11 the advantages of our field ago We maintained human and ape crucial advantage in determining the stood revealed in 1967. when enes were too similar for the schism structure of family trees: living genes Vincent m. sarich. working in to be more than a few million years must have ancestors, whereas dead fos- Wilsons laboratory at the University of old. After 15 years of disagreement, we sils may not have descendants. Molec- California at berkeley challenged a fos won that argument, when the paleon- ular biologists know the genes they are sil primate called Ramapithecus. Pale ologists admitted we had been right examining must have been passed ontologists had dated its fossils to and they had been wrong through lineages that survived to the about 25 million years ago. On the ba- Once again we are engaged in a de- present; paleontologists cannot be sure sis of the enamel thickness of the mo- bate, this time over the latest phase of that the fossils they examine do not lars and other skeletal characteristics, human evolution. The paleontologists lead down an evolutionary blind alley. they believed that Ramapithecus say modem humans evolved from their The molecular approach is free from peared after the divergence of the hu- r c cs ay can be traced along ma- ly tree it hopes to describe. It is not vi- distance between humans and chim- haic forebears around the world over several other limitations of paleontolo-. man and ape lineages and that it was parsons convince us that all sils or tools from each part of the fami- Sarich measured the evolutionary umans ternal lines of descent to a woman who tiated by doubts about whether tools panzees by studying their blood pro- found near fossil remains were in fact teins knowing the differences reflected made and used by the population those mutations that have accumulated since he late ALLAN C. WILSON and Rl-1 remains represent. Finally, it concems the species diverged. (At the time, it BECCA L. CANN applied the tools of itself with a set of characteristics that was much easier to compare proteins netics to paleontology during many of is complete and objective for subtle differences than to compare a genome, or full set of genes, is com- the genetic sequences that encode the 91, Wilson was professor of biochem- plete because it holds all the inherited proteins ) To check that mutations had istry at the University of California, I biological information of an individual. occurred equally fast in both lineages Berkeley. a native of New Zealand, he Moreover, all the variants on it that ap- he compared humans and chimpanzees |/限h四mm四 eceived his undergrads Berkel other-can be studied as well, so specif- Sarich now had a molecular clock; the son also worked at the Weizmann ic peculiarities need not distort the next step was to calibrate it. He did so stitute of Science, the University of Nai-I interpretation of the data. Genomes are by calculating the mutation rate in oth robi and Harvard University. Cann is as- objective sources of data because they er species whose divergences could be sociate professor of genetics and present evidence that has not been de- reliably dated from fossils. Finally, he lar biology at the John A. Bums School I fined, at the outset, by any particular applied the clock to the at Manoa. She received both her bache- evolutionary model, Gene sequences are human split, dating it to between five lor's degree in The fossil record, on the other hand, At first, most paleontologists clung with Wilson and at the University of Cal- is infamously spotty because a handful to the much earlier date. But new fossi San Francisco. As one of her cur- of surviving bones may not represent finds undermined the human status of cts. cann is using mitochon- he majority of organisms that left no Ramapithecus: it is now clear Rama to assay the senetic diversity trace of themselves. Fossils cannot, in pithecus is actually Sivapithecus, a crea- principle, be interpreted objectively: the ture ancestral to orangutans and not to 68 SCIENTIFIC AMERICAN April 1992
The Recent African Genesis of Humans Genetic studies reveal that an African woman of 200,000 years ago was our common ancestor by Allan C. Wilson and Rebecca L. Cann I n the quest for the facts about human evolution, we molecular geneticists have engaged in two major debates with the paleontologists. Arguing from their fossils, most paleontologists had claimed the evolutionary split between humans and the great apes occurred as long as 25 million years ago. We maintained human and ape genes were too similar for the schism to be more than a few million years old. After 15 years of disagreement, we won that argument, when the paleontologists admitted we had been right and they had been wrong. Once again we are engaged in a debate, this time over the latest phase of human evolution. The paleontologists say modem humans evolved from their archaic forebears around the world over the past million years. Conversely, our genetic comparisons convince us that all humans today can be traced along maternal lines of descent to a woman who The late ALLAN C. WILSON and RIBECCA L. CANN applied the tools of genetics to paleontology during many of their collaborations. Until his death In 1991, Wilson was professor of biochemistry at the University of California, Berkeley. A native of New Zealand, he received his undergraduate degree from the University of Otago in 1955, his master’s from Washington State University and his doctorate from Berkeley. Wilson also worked at the Weizmann Institute of Science, the University of Nairobi and Harvard University. Cann is associate professor of genetics and molecular biology at the John A. Bums School of Medicine of the University of Hawaii at Manoa. She received both her bachelor’s degree in genetics and her Ph.D. in anthropology from Berkeley. As a postdoctoral fellow, she worked at Berkeley with Wilson and at the University of California, San Francisco. As one of her current projects, Cann is using mitochondrial DNA to assay the genetic diversity of birds In the Hawaiian Islands. 68 lived about 200,000 years ago, probably in Africa. Modern humans arose in one place and spread elsewhere. Neither the gene tic information of living subjects nor the fossilized remains of dead ones can explain in isolation how, when and where populations originated. But the former evidence has a crucial advantage in determining the structure of family trees: living genes must have ancestors, whereas dead fossils may not have descendants. Molecular biologists know the genes they are examining must have been passed through lineages that survived to the present; paleontologists cannot be sure that the fossils they examine do not lead down an evolutionary blind alley. The molecular approach is free from several other limitations of paleontolo-. gy. It does not require well-dated fossils or tools from each part of the family tree it hopes to describe. It is not vitiated by doubts about whether tools found near fossil remains were in fact made and used by the population those remains represent. Finally, it concerns itself with a set of characteristics that is complete and objective. A genome, or full set of genes, is complete because it holds all the inherited biological information of an individual. Moreover, all the variants on it that appear within a population-a group of individuals who breed only with one another-can be studied as well, so specific peculiarities need not distort the ‘interpretation of the data. Genomes are objective sources of data because they present evidence that has not been defined, at the outset, by any particular evolutionary model. Gene sequences are empirically verifiable and not shaped by theoretical prejudices. The fossil record, on the other hand, is infamously spotty because a handful of surviving bones may not represent the majority of organisms that left no trace of themselves. Fossils cannot, in principle, be interpreted objectively: the physical characteristics by which they are classified necessarily reflect the models the paleontologists wish to test. If one classifies, say, a pelvis as human because it supported an upright posture, then one is presupposing that bipedalism distinguished early hominids from apes. Such reasoning tends to circularity. The paleontologist’s perspective therefore contains a built-in bias that limits its power of observation. As such, biologists trained in modern evolutionary theory must reject the notion that the fossils provide the most direct evidence of how human evolution actually proceeded. Fossils help to fill in the knowledge of how biological processes worked in the past, but they should not blind us to new lines of evidence or new interpretations of poorly understood and provisionally dated archaeological materials. 11 the advantages of our field Astood revealed in 1967, when Vincent M. Sarich, working in Wilson’s laboratory at the University of California at Berkeley, challenged a fossil primate called Ramapithecus. Paleontologists had dated its fossils to about 25 million years ago. On the basis of the enamel thickness of the molars and other skeletal characteristics, they believed that Ramapithecus appeared after the divergence of the human and ape lineages and that it was directly ancestral to humans. Sarich measured the evolutionary distance between humans and chimpanzees by studying their blood proteins, knowing the differences reflected mutations that have accumulated since the species diverged. (At the time, it was much easier to compare proteins for subtle differences than to compare the genetic sequences that encode the proteins.) To check that mutations had occurred equally fast in both lineages, he compared humans and chimpanzees against a reference species and found that all the genetic distances tallied. Sarich now had a molecular clock; the next step was to calibrate it. He did so by calculating the mutation rate in other species whose divergences could be reliably dated from fossils. Finally, he applied the clock to the chimpanzeehuman split, dating it to between five and seven million years ago-far later than anyone had imagined. At first, most paleontologists clung to the much earlier date. But new fossil finds undermined the human status of Ramapithecus: it is now clear Ramapithecus is actually Sivapithecus, a creature ancestral to orangutans and not to
any of the African apes at all. More- from the observ ation that coated double helix the mitochondrial age of some sivapitheclne fos- netic change from point DNA comes in small, two-strand rings sils was downgraded to only about six (changes in individual dNA Whereas nuclear dNa encodes an esti million years. By the early 1980s al- were so steady over long mated 100.000 genes-most of the in most all paleontologists came to accept one could use them to time divergences formation needed to make a human be Sarich's more recent date for the sepa- from a common stock ing-mitochondrial DNA encodes only ration of the human and ape lines 7. In this handful of genes, every one e could begin to apply these is essential: a single adverse mutation en reauc ed to arguing that methods to the reconstruction in any of them is known to cause some Sarich arrived at the right answer purely of later stages in human evolu- severe neurological diseases tion only after 1980, when DNA restric. For the purpose of scientists studyir Two novel concepts emerged from tion analysis made it possible to ex. when lineages diverged, mitochondrial the early comparisons of proteins from plore genetic differences with high res- DNA has two advantages over nuclear of inconsequential, or neutral, muta- Wes Brown, Mark Stocking and us, ap- drial dNA that interest us accumulate tions. Molecular evolution appears to bc plied thetechnique to trace the mater. mutations rapidly and steadily, accord dominated by such mutations, and they nal lineages of people sampled from ing to empirical observations. Because accumulate at surprisingly steady rates around the world many mutations do not alter the mito in suniving lineages In other words, The dNa we studied resides in the chondrion's function, they arc effec evolution at the gene level results mitochondria, cellular organelles that tively neutral, and natural selection mainly from the rclcntlcss accumula- convert food into a form of energy the, does not eliminate them tion of mutations that seem to be nel- rest of the cell can use Unlike the dna This mitochondrial dNA therefore ther harmful nor beneficial. The second of the nucleus, which forms bundles of behaves like a fast-ticking clock, which concept, molecular clocks, stemmed long fibers, each consisting of a protein. is essential for identifying recent genet- 190 目 AFRICAN D ASIAN▲ AUSTRALIAN▲ NEW GUINEAN■ CAUCAS|AN AFRICAN ORIGIN for all modem humans is indicated by the genet- ic evidence. A genealogy based on 182 current mitochondrial DNA types (outer edges points to the existence of a common female an. cestor who lived recently in Africa. The arrows on the maps(center 6 DIVERGENCE IN DNA indicate the minimum number of unrelated females who colonized DIVERGENCE IN DNA SEQUENCE(PERCENT) major geographic areas, as inferred from the branching pattern. SCIENTIFIC AMERICAN April 1992 69
any of the African apes at all. bloreo\‘er, the age of some sivapitheclne fossils was downgraded to only about six million years. By the early 1980s almost all paleontologists came to accept Sarich’s more recent date for the separation of the human and ape lines. Those ivho continue to reject his methods have been reduced to arguing that Sarich arrived at the right ansiver purely, b\. chance. T\vo no\?1 concepts emerged from the early comparisons of proteins from different species. One was the concept of inconsequential, or neutral, mutations. bIolecular c\,olution appears to bc dominated by such mutations, and the), accumulate at surprisingly stead), rates in suniving lineages. In other rvords, evolution at the gene 1~~~~1 results mainly from the rclcntlcss accumulation of mutations that seem to be neither harmful nor beneficial. The second concept, molecular clocks, stemmed from the obsen.ation that rates of genetic change from point mutations (changes in individual DNA base pairs) were so steady over long periods that one could use them to time divergences from a common stock. Iv e could begin to apply these methods to the reconstruction of later stages in human e\,olution onl}, after 1980, when DNA restriction analysis made it possible to esplore genetic differences with high rcsolution. kk’orkers at Berkclc); including Lk’cs Brobin, Mark Stoncking and us, applied the tcchniquc to trace the matcrnal lineages of pcoplc sampled from around the world. The DNA WC studied rcsidcs in the mitochondria, cellular organcllcs that convert food into a form of energy the, rest of the cell can use. Unlike the DK.4 of the nucl&s, which forms bundles of long fibers, each consisting of a protcincoated double helix, the mitochondrial DIV.A comes in small, two-strand rings. LYhereas nuclear DNA encodes an estimated 100,000 genes-most of the information needed to make a human being-mitochondrial DNA encodes only 37. In this handful of genes, e\er). one is essential: a single ad\,erse mutation in an)’ of them is knorkn to cause some sct’ere neurological diseases. For the purpose of scientists studying ivhen lineages diverged, mitochondrial DN.4 has t\\‘o advantages o\‘er nuclear DX-\. First, the sequences in mitochondrial DN.4 that interest us accumulate mutations rapidl}, and steadill,, according to empirical obscn,ations. Because many mutations do not alter the mitochondrion’s function, the), arc effcctivcly neutral, and natural sclcction does not eliminate them. This mitochondrial DNA therefore beha\,es like a fast-ticking clock, kvhich is essential for identifying recent genet- 0 AFRICAN DASIAN A AUSTRALIAN A NEW GUINEAN n CAUCASIAN AFRICAN ORIGIN for all modem humans is indicated by the genetic evidence. A genealogy based on 182 current mitochondrial DNA , . . _ _ . I I I I I I I 0 0.2 0.4 0.6 DIVERGENCE IN DNA SEQUENCE (PERCENT) types (outer edges1 points to the existence of a common female ancestor who lived recently in Africa. The arrows on the maps (cented indicate the minimum number of unrelated females who colonized major geographic areas, as inferred from the branching pattern. 0 . 6 DIVERGENCE IN DNA SEQUENCE (PERCENT) SCIENTIFIC AMERICAN April 1992 69
The Inheritance of Mitochondrial dna EGG FERTILIZED EGG 37 GENE MITOCHONDRIAL DNA MITOCHONDRION NUCLEAR DNA Ⅳ ost of an individuals genes are located on DNA mole- cules in the cell nucleus. Mitochondria, the specialized structures that provide cells with energy, also carry some genes for their own manufacture on a ring of DNA. When sperm and an egg cell unite, they contrit DNA in the nucleus of the resulting cell. All the mitochondn and the dNa they contain, however, derive from the egg Studies of mitochondrial dna can reveal an individuals aternal ancestry. ges. Any two humans a static population that always con- cal trees could give rise to any set of randomly from anywhere on th tains 15 mothers. Every new generation genetic data. To recognize the one that are so alike in most of their must contain 15 daughters, but some is most probably correct, one must ap quences that we can measure evolution mothers w-ill fail to produce a daugh- ply the parsimony ple, which re- in our species only by concentrating on ter, whereas others will produce two or quires that subjects be connected in the the genes that mutate fastest. Genes more Because maternal lineages die out simplest possible way. The most effi- controlling skeletal characters do not whenever there is no daughter to carry cient hy pothetical tree must be tested data to Second, unlike nuclear DNA, mito. but one lineage disappears. In a stable whether it is consistent with them. If chondrial dNa is inherited from the population the time for this fixation of the tree holds up it is analyzed for evi- mother alone, unchanged except for the maternal lineage to occur is the dence of the geographic history inher chance mutations. The father's contri- length of a generation multiplied by ent in elements bution ends up on the cutting-room twice the population size In 1988 Thomas d. Kocher of Berke. floor, as it were. The nuclear genes, to ley(now at the University of New Hamp- which the father does contribute. de- ne might refer to the lucky shire) applied just such a parsimonious scend in what we may call ordinary lin- woman whose lineage survives interpretation to the interrelatedness eages, which are of course important to as eve. Bear in mind however. of the mitochondrial dna of 14 huma the transmission of physical character- that other women were living in Eve's from around the world. He determined istics. For our studies of modern hu- generation and that Eve did not occupy 13 branching points were the fewest man origins, however, we focus on the a specially favored place in the breed- that could account for the differences mitochondrial, maternal lineages ing pattern. She is purely the beneficia- he found. Taking the geographic consid- Maternal lineages are closest among ry of chance. Moreover, if we were to erations into account, he then conclud- siblings because their mitochondrial reconstruct the ordinary lineages for ed that Africa was the ultimate human DNA has had only one generation in the population, they would trace back homeland: the global distribution of which to accumulate mutations. The de- to many of the men and women who mitochondrial DNA types he saw could fir e moves along the pedigree, from tion geneticists Daniel L. Hartl of Wash- sult of no more than three migration ternal grandmother, to second cousins and Andrew G. Clark of Pennsylvania A crucial assumption in this analysis descended from a common matenal State University estimate that as man) is that all the mitochondrial lineages reat-grandmother and so on. The far-, as 10,000 people could have lived then. evolve at the same rate. For that reason, the circle of maternal relatives becomes leading-she is not the ultimate source son of the human mitochondrial dnas until at last it embraces everyone alive. of all the ordinary lineages, as the bibli- he also included analogous sequences Logically, then, all human mitochon- cal Eve wa from four chimpanzees. If the human drial dNA must have had an ultimate From mitochondrial dNa data, it is lineages had differed in the rate at common female ancestor. But it is easy possible to define the maternal lin. which they accumulated mutations to show she did not necessarily live in eages of living Individuals all the way then some of the 14 human sequences a small population or constitute the back to a common ancestor. In theory, would be significantly closer or farther only woman of her generation. Imagine a great number of different genealogi away from the chimpanzee sequences CIENTIFIC AMERICAN April1992
ic changes. Any two humans chosen randomly from anywhere on the planet are so alike in most of their DNA sequences that we can measure evolution in our species only by concentrating on the genes that mutate fastest. Genes controlling skeletal characters do not fall within this group. Second, unlike nuclear DNA, mitochondrial DNA is inherited from the mother alone, unchanged except for chance mutations. The father’s contribution ends up on the cutting-room floor, as it were. The nuclear genes, to which the father does contribute, descend in what we may call ordinary lineages, which are of course important to the transmission of physical characteristics. For our studies of modern human origins, however, we focus on the mitochondrial, maternal lineages. Maternal lineages are closest among siblings because their mitochondrial DNA has had only one generation in which to accumulate mutations. The degree of relatedness declines step by step as one moves along the pedigree, from first cousins descended from the maternal grandmother, to second cousins descended from a common maternal great-grandmother and so on. The farther back the genealogy goes, the larger the circle of maternal relatives becomes, until at last it embraces everyone alive. Logically, then, all human mitochondrial DNA must have had an ultimate common female ancestor. But it is easy to show she did not necessarily live in a small population or constitute the only woman of her generation. Imagine 70 S~~ENTIFI~~~ERI~AN April1992 The Inheritance of Mitochondrial DNA EGG FERTILIZED EGG MITOCHONDRIAL DNA NUCLEAR DNA ost of an individual’s genes are located on DNA molecules in the cell nucleus. Mitochondria, the specialized Qa ? structures that provide cells with energy, also carry some ‘-4 ) genes for their own manufacture on a ring of DNA. When a sperm and an egg cell unite, they contribute equally to the SPERM DNA in the nucleus of the resulting cell. All the mitochondria and the DNA they contain, however, derive from the egg. 4 Studies of mitochondrial DNA can reveal an individual’s maternal ancestry. a static population that always contains 15 mothers. Every new generation must contain 15 daughters, but some mothers w-ill fail to produce a daughter, whereas others will produce two or more. Because maternal lineages die out whenever there is no daughter to carq’ on, it is only a matter of time before all but one lineage disappears. In a stable population the time for this fivation of the maternal lineage to occur is the length of a generation multiplied b). twice the population size. 0 ne might refer to the luck). woman whose lineage survii’es as Eve. Bear in mind, however, that other women were living in Eve’s generation and that Eve did not occup)- a specially favored place in the breeding pattern. She is purely the beneficiary of chance. Moreover, if we were to reconstruct the ordinary lineages for the population, they would trace back to many of the men and women who lived at the same time as Eve. Population geneticists Daniel L. Hart1 of Washington Universib School of Medicine and Andrew G. Clark of Pennsylvania ! State Universiw estimate that as man) as 10,000 people could have lived then. ’ The name “Eve” can therefore be misleading-she is not the ultimate source of all the ordinary lineages, as the biblical Eve was. From mitochondrial DNA data, it is possible to define the maternal lineages of living Individuals all the wa)p back to a common ancestor. In theov, a great number of different genealogical trees could gi\*e rise to an)’ set of genetic data. To recognize the one that is most probably correct, one must apply the parsimony principle, which requires that subjects be connected in the simplest possible \\.a)‘. The most efficient hypothetical tree must be tested by comparison Mith other data to see whether it is consistent with them. If the tree holds up, it is analyzed for evidence of the geographic historI. inherent in elements. In 1988 Thomas D. Kocher of Berkeley (now at the Uni\,ersity of New Hampshire) applied just such a parsimonious interpretation to the interrelatedness of the mitochondrial DNA of 14 humans from around the world. He determined 13 branching points were the fejvest that could account for the differences he found. Taking the geographic considerations into account, he then concluded that Africa was the ultimate human homeland: the global distribution of mitochondrial DNA t)-pes he saw could then be esplained most easily as the result of no more than three migration events to other continents. A crucial assumption in this analysis is that all the mitochondrial lineages evolve at the same rate. For that reason, when Kocher conducted his comparison of the human mitochondrial DNAs, he also included analogous sequences from four chimpanzees. If the human lineages had differed in the rate at which they accumulated mutations, then save of the 14 human sequences would be significantly closer or farther away from the chimpanzee sequences
than others In fact. all 14 human se- uences are nearl distant from the mba e sequences which implies FATHER MOTHER the rates of change among humans are The chimpanzee data also illustrated how remarkably homogeneous humans are at the genetic level: chimpanzees nly show as much as 10 times fact alone suggests that all of modern humanity sprang from a relatively small stock of common ancestors working at Berkeley with Stone w'e expanded on Kocher's work amining a larger genealogical tree made up from 182 distinct types of mitochon-MALE drial DNA from 241 individuals. The FEMale multiple occurrences of mitochondrial people from the same continent and MITOCHONDRIAL DNA SOURCE usuallyin persons who lived within pEDigree of one individual illustrates the difference between the patterns of nu- 100 miles of one another. Because the clear and mitochondrial inheritance. All 32 ancestors from five generations ago tree we constructed had two main contributed equally to his nuclear DNA. His mitochondrial lineage(blue line)leads branches, both of which led back to back to only one person in every generation sis that Africa was the place of origin for modern humans ike biological species. We propose that light skin color of many whites One point that jumps out of our the anatomic characteristics uniting ably a late development that study is that although geographic bar- New Guineans were not inherited from in Europe after that continent riers do influence a population's mito- the first settlers. They evolved after nized by Africans chondrial DNA, people from a given people colonized the island, chiefly as During the 980s, when we were continent do not generally all belong to the result of mutations in nuclear constructing analogical tree. we the same maternal lineage. The New genes spread by sex and recombination had to rely or Americans as sub Guineans are typical in this respect. throughout New Guinea. Similarly, the stitutes for Africans, whose mitochon- remarkable variety of language fami lies-generally classified as papuan oken on this one island [see"The 2 Austronesian Dispersal and the Origin by Pet FIc AMERICAN, July 19911. On our genealogical tree, New Guineans showed up on several different branches, which 5 proved that the common female ances tor of all New guineans was not some. one in New Guinea. The population of New guinea must have been founded by many mothers whose maternal lin- eages were most closely related to 29 those in asia That finding is what one would ex- pect if the African origin hypothesis were true: as people walked east out of 12 Africa, they would have passed through Asia. Travel was probably slow, and 13 during the time it took to reach New 14 Guinea, mutations accumulated both in the lineages that stayed in Asia and in those that moved on e who are apparently re- lated by membership in a common ge- UNIVERSAL MATERNAL ANCESTOR can be found for all the members of any pop- graphic race need not be very closely ulation. The example traces the lineages of 15 females in a stable population. In related in their mitochondrial dNa ach generation, some maternal lineages proliferate and others become extinct. tochondrially speaking, races are not Eventually, by chance, one maternal lineage replaces all the others SCIENTTFIC AMERICAN April 1992
than others. In fact. all 14 human sequences are nearly equidistant from the chimpanzee sequences, which implies the rates of change among humans are fairly uniform. The chimpanzee data also illustrated how remarkabl). homogeneous humans are at the genetic level: chimpanzees ’ commonly show as much as 10 times the genetic variation as humans. That fact alone suggests that all of modern humanit). sprang from a relativeI), small stock of common ancestors. 1Vorking at Berkeley, rlith Stoneking. \vc espanded on Kochcr’s bvork bl, cxamining a larger genealogical tree made up from 182 distinct t)-pes of mitochondrial DNA from 241 indhiduals. The multiple occurrences of mitochondrial DNA t)‘pcs %verc ahva).s found among people from the same continent and usuall~~ in persons \\,ho lived rsithin 100 miles of one another. Because the tree we constructed had two main branches, both of which led back to Africa, it, too, supported the hypothesis that Africa was the place of origin for modern humans. One point that jumps out of our study is that although geographic barriers do influence a population’s mitochondrial DNA, people from a given continent do not generally all belong to the same maternal lineage. The New Guineans are typical in this respect. Their genetic diversity had been suspected from linguistic analyses of the remarkable variety of language families-generally classified as Papuanspoken on this one islsnd [see “The Austronesian Dispersal and the Origin of Languages,” by Peter Bellwood; SCIENTIFIC AMERICAN, July 19911. On our genealogical tree, New Guineans showed up on several different branches, which proved that the common female ancestor of all New Guineans was not someone in New Guinea. The population of New Guinea must have been founded by many mothers whose maternal lineages Lvere most closely related to those in Asia. That finding is what one would expect if the African origin hypothesis were true: as people walked east out of Africa, they would have passed through Asia. Travel was probably slow, and during the time it took to reach New Guinea, mutations accumulated both in the lineages that stayed in Asia and in those that moved on. Thus, people who are apparently related by membership in a common geographic race need not be very closely related in their mitochondrial DNA. Mitochondrially speaking, races are not MALE l FEMALE 4 SON - MITOCHONDdL DNA SOURCE PEDIGREE of one individual illustrates the difference between the patterns of nuclear andmitochondrial inheritance. All 32 ancestors from five generations ago contributed equally to his nuclear DNA. His mitochondrial lineage (blue line) leads back to only one person in every generation. like biological species. We propose that the anatomic characteristics uniting New Guineans were not inherited from the first settlers. They evolved after people colonized the island, chiefly as the result of mutations in nuclear genes spread b). sex and recombination throughout New Guinea. Similarly, the light skin color of many whites is probably a late development that occurred in Europe after that continent was colonized b). Africans. During the early 198Os, when we were constructing our genealogical tree, we’ had to rely on black Americans as sub: stitutes for Africans, whose mitochonUNIVERSAL MATERNAL ANCESTOR can be found for all the members of any population. The example traces the lineages of 15 females in a stable population. In each generation, some maternal lineages proliferate and others become extinct. Eventually, by chance, one maternal lineage replaces all the others. SCIENTIFIC AMERICA Apt-d 1992 71
drial DN. 4 was difficult to obtain in the of Africa; the), too, found that the mi. trol we arrived at a more an required quantities. Fortunately, the re- tochondrial lineages led back to Africa. cient date for the common mother cent development of a technique called We estimate the odds of their arriving That less certain, however, be the polymerase chain reaction has at that conclusion accidentally were cause questions remain about how to eliminated that constraint. The reaction only four in 100. Although these statis- correct for multiple mutations that oc- makes it possible to duplicate DNA se- tical evaluations are not strong or rig- cur within the control region quences easily, ad infinitum; a small orous tests, they do make it seem likely One might object that a molecular starting sample of DNA can espand that the theory of an African origin for clock known to be accurate over five into an endless supply [see" The Un- human mitochondrial DNA is now fair- million years could still be unreliable usual Origin of the Poly merase Chain ly secure for shorter periods. It is conceivable Reaction, by Kary B. Mullis: ScIENTIrIC for example, that intervals of genetic AMERICAN, April 19901 ccause our comparisons with the stagnation might be interrupted by The poly merase chain reaction en chimpanzee data showed the hu- short bursts of change when, say, a abled Linda vigilant, now at Pennsylva B man mitochondrial dna clock new mutagen enters the environment nia State University, to redo our study has ticked steadily for millions of ycars, or a virus infects the germ-line cells, or using mitochondrial DN.4 data from we knew it should be possible to calcu- intense natural selection affects all seg- parts of the sub-Saharan region. Vigi- manity lived. Wc assumed the human sibility that the clock might run by fits lant traced a genealogical tree whose and chimpanzee lineages diverged five and starts, we ran a test to measure 1+ dccpcst branches leadexclusively to million years ago, as Sarich's work had how much mitochondrial DNA has Africans and whose 15th branch leads shown. Wc then calculated how much evolved in populations founded at a to both africans and non -Africans The humans had diverged from one anoth- known time non-Africans lie on shallow secondary er relative to how much they had di- The aboriginal populations of New branches stemming from the 15th verged from chimpanzees-that is, we Guinea and Australia are estimated to branch. Considering the number of found the ratio of mitochondrial dna have been founded less than 50,000 to African and non-African mitochondrial divergence among humans to that be- 60,000 years ago. The amount of evolu- DNAs surveyed, the probability that tween humans and chimpanzees tion that has since occurred in each of lusively African is one in 10,000 for a determined the ratio was less than that shown by the whole human species tree with this branching order 1: 25. Human maternal lineages there- Accordingly, we can infer that Eve lived Satoshi Horai and Kenji Hayasaka of fore grew apart in a period less than three times 50,000 to 60,000 years ago the National Institute of Genetics in / 25th as long as five million years, or or roughly 150,000 to 180,000 years Japan analogously surveyed popula- less than 200,000 years. with a third ago. All our estimates thus agree the tion samples that included many more set of data on cl s in a section of split happened not far from 200,000 Asians and individuals from fewer parts the mitochondrial DNA called the con. years ago Those estimates fit with at least one H line of fossil evidence. The remains of anatomically modern people appear first in Africa, then in the Middle east and later in Europe and east Asia. An- thropologists have speculated that in east Africa the transition from anatom- ically archaic to modem people took lace as recently as 130,000 years ago see "The Emergence of Modern Hu On the other hand a second line of evidence appears to conflict with this view. The fossil record shows clearly that the southern parts of Eurasia were occupied by archaic people who had mi- grated from Africa to Asia nearly a mil- AFRICAN Tion years ago. Such famous fossils as NON AFRICAN PEOPLE Java man and beijing man are of this type 口 PYGMY CHIMPANZEES hat the archaic Eurasian population un derwent anatomic changes that made them resemble more modem people led y to the m multiregional evolutio similar evolutionary changes in sepa- ate geographic regions converted the INTERRELATEDNESS of 14 huma d four chimpanzees inhabitants from archaic small-brained was inferred from similarities in ochondrial dna se. to modem big-brained types. when various divergences in the human lineages occurred e uences. The chimpanzee data Huge levels of gene flow between 3 continents. however, would be neces- 72 SCIENTiFIC AMERICAN April 1992
drill DN.4 was difficult to obtain in the required quantities. Fortunately, the recent development of a technique called the pol).merase chain reaction has eliminated that constraint. The reaction makes it possible to duplicate DNA sequences easily, ad infinitum; a small starting sample of DNA can espand into an endless supply [see “The Unusual Origin of the Pollmerase Chain Reaction,” by Kary B. Mullis; Scrr~rtrrc RWERICAN, April 19901. The pol},mcrasc chain reaction cnabled Linda vigilant, now at Pcnns),lvania State Univcrsit),, to redo our stud). using mitochondrial DN.4 data from 130 Africans, representing six dh,crsc parts of the sub-Saharan region. Vigilant traced a genealogical tree whose 1 -L dccpcst branches lead csclusivcly to Africans and whose 15th branch leads to both Africans and non-Africans. The non-Africans lie on shallow secondar) branches stemming from the 15th branch. Considering the number of African and non-African mitochondrial DM4s surveyed, the probability that the 14 deepest branches would be exclusively African is one in 10,000 for a tree with this branching order. Satoshi Horai and Kenji Hayasaka of the National Institute of Genetics in Japan analogously surveyed population samples that included many more Asians and individuals from fewer parts I’ l--l u of Africa; the),, too, found that the mitochondrial lineages led back to Africa. We estimate the odds of their arrikjng at that conclusion accidentally tvere only four in 100. Although these statistical evaluations are not strong or rigorous tests, they do make it seem like11 that the theory of an African origin for human mitochondrial DNA is now fair- 1). secure. B ecause our comparisons bc-ith the chimpanzee data shorved the human mitochondrial DNA clock has ticked steadily for millions of years, ivc knew it should be possible to calculate kvhcn the common mother of humanity lived. WC assumed the human and chimpanzee lineages diverged fi\,c million years ago, as Sarich’s bvork had shown. WC then calculated how much humans had diverged from one another relative to how much they had diverged thorn chimpanzees-that is, we found the ratio of mitochondrial DNA divergence among humans to that between humans and chimpanzees. Using two different sets of data, we determined the ratio was less than 1:25. Human maternal lineages therefore grew apart in a period less than ‘/,:th as long as five million years, or less than 200,000 )-ears. With a third set of data on changes in a section of the mitochondrial DNA called the conrv+: 4\ n AFRICAN NON AFRICAN PEOPLE n CHIMPANZEES q PYGMY CHIMPANZEES trol region, we arrived at a more ancient date for the common mother. That date is less certain, however, because questions remain about holv to correct for multiple mutations that occur within the control region. One might object that a molecular clock known to be accurate over fi1.e million years could still be unreliable for shorter periods. It is conceii,able, for example, that inten.als of genetic stagnation might be interrupted b). short bursts of change ivhen, say, a new mutagen enters the emironment, or a virus infects the germ-line cells, or intense natural selection affects all segments of the DNA. To rule out the possibilit), that the clock might run by fits and starts, we ran a test to measure how much mitochondrial DNA has evolved in populations founded at a knobin time. The aboriginal populations of New Guinea and Australia are estimated to have been founded less than 50,000 to 60,000 years ago. The amount of evolution that has since occurred in each of those places seems about one third of that shown by the lvhole human species. Accordingly, we can infer that Eve lived three times 50,000 to 60,000 years ago, or roughly 150,000 to 180,000 years ago. All our estimates thus agree the split happened not far from 200,000 years ago. Those estimates fit riith at least one line of fossil evidence. The remains of anatomically modern people appear first in Africa, then in the Middle East and later in Europe and east Asia. Anthropologists have speculated that in east Africa the transition from anatomically archaic to modem people took place as recently as 130,000 years ago [see “The Emergence of Modern Humans,” by Christopher B. Stringer; SCIENTIFIC A~IERK~, December 19901. INTERRELATEDNESS of 14 humans and four chimpanzees Ll L--r, On the other hand, a second line of evidence appears to conflict with this L-iew. The fossil record shows clearI), that the southern parts of Eurasia were occupied by archaic people who had migrated from Africa to Asia nearly a million years ago. Such famous fossils as Java Man and Beijing Man are of this Q-pe. This finding and the hypothesis that the archaic Eurasian population underwent anatomic changes that made them resemble more modem people led to the multiregional evolution model: similar evolutionary changes in separate geographic regions converted the inhabitants from archaic small-brained to modem big-brained t)pes. 3 3 I Huge levels of gene flow between ; continents, however, would be neceswas inferred from similarities in their mitochondrial DNA sequences. The chimpanzee data help researchers measure when various divergences in the human lineages occurred. 72 Scmmmc &RICAN April 1992
sarv to maintain human populations as PRESENT AFRICAN EUROPEAN EAST ASIAN AUSTRALIAN one biological species. The multiregion al evolution model also predicts that at least some genes in the modern east Asian population would be linked more closely to those of their archaic Asian 100,000| NEANDERTHAL redecessors than to those of modern DALI Africans. We would expect to find deep eages in Eurasia, especially in the far East. Yet survey s in our laboratories and in others, involving more than 1,000 2 300.000 SALDANHA PETRALONA BEIING") SAMBUNGMACHAN people from Eurasia and its mitochon- s drial DNA satellites(Australia, Oceania and the Americas), have given no hint of that result It therefore seems very unlikely that 700.000 OLDUVA EUROPEAN LANTIAN JAVA any truly ancient lineages survive un- detected in Eurasia. We simply do not see the result predicted by the regional odel. Moreover. geneticists such as OMO ERECTUS Masatoshi Nei of Pennsylvania State Uni- HUMAN GROUPS WERE REPLACED throughout the old world by modern hu versity, Kenneth K. Kidd of Yale Univer- from Africa. Archaic females do not seem to have contributed mitochondrial genes sity, James wainscot of the University to the modem people of Europe, east Asia and Australia of oxford and luigi l cavalli-Sforza of Stanford University have found support for an African origin model in their in Israel suggest Neanderthals and mod- duct a detailed genealogical analysis studies of nuclear genes em humans lived side by side for Still more progress can be expected 40,000 years, yet they left little evidence in the immediate future as molecular P Its of the multiregional biologists learn to apply their tech- evolution model emphasize the How one human population might niques to materials uncovered by our have documented a continuity of have replaced archaic humans without friendly rivals, the paleontologists. Pre- anatomic morphologies between the ar. any detectable genetic mixing is still liminary molecular studies have al chaic and modem residents of different a mystery. One of us( Cann)suspects ready been conducted on DNA from regions; they insist these morphologies infectious diseases could have con. mummified tissues found in a Florida would be unlikely to evolve indepen- tributed to the process by helping to bog and dated to 7,500 years ago. Im- dently in any invading people. For that eliminate one group. Cavalli-Sforza has I proved methods of extracting DNA argument to hold true, it must also be speculated the ancestors of modern from still older fossilized bone now ap shown that the cranial features in ques- humans may have developed some pear close at hand. With them, we may tion are truly independent of one an- modern trait, such as advanced lan- begin building the family tree from a other-that is, that natural selection guage skills, that effectively cut them i root that was alive when the human would not tend to favor certain constel- off from breeding with other hominids. family was young ations of functionally related features. This and related questions may yield Yet we know powerful jaw muscles may as molecular biologists learn how to impose changes on the mandible, the link specific genetic sequences to the brow-ridge and other points on the physical and behavioral traits those se- FURTHER READING skull; circumstances that promoted the quences influence MITocHONDRIAL DNA AND HUMAN EVO evolution of these features in one pop- Even before then, further studies of LUTION. Rebecca L Cann, Mark stor ulation might do so again in a related both nuclear and mitochondrial dNa d Allan C. wilson in Nature, Vol ulation will render more informative genetic lo. 6099, pages 31-36, January the evidence for continuity. They argue quences on the Y chromosome that de-A. c. Wilson in the colonization of the modern populations are not linked to termine maleness and that are there- Paciic: A Genetic Trail. Edited by Adri- past ones by morphological character- fore inherited from the father alone. an V.S. Hill and Susan W Serjeantson. istics that evolved uniquely in the fossil Gerad Lucotte's laboratory at College Oxford University Press, 1989 record. Instead fossils and modem pop- de France has indirectly compared MTOCHONDRIAL DNA SEQUENCES DN SIN ulations are united by their shared re- such sequences in an effort to trace pa- GLE H AIRS FROMA SOUTHERN AFRT. teristic. The continuity seen by believ."Adam, if you will. Those prelimimary kocher Hn A ln e wisTh pased llusion The idea that modem humans could wo nal lineages may be ab cohabit a region with archaic ones and to provide an invaluable check on our SEQUENCE EVOLUTION OF replace them completely without any results for maternal lineages. Unfortu- AL DNA D HUMANS AND NZEES mixture may sound unlikely. Never Evolu less, some fossil finds do sup on useful regions of the y chromosome. tion of tife Idea. Discoveries in the caves at Qafzeh making it technically difficult to con- SCIENTIFIC AMERICAN April 1992 73
sarv to maintain human Dooulations as 1 one biological species. T<e hultiregional evolution model also predicts that at least some genes in the modern east Asian population would be linked more closely to those of their archaic Asian predecessors than to those of modern Africans. Il’e would expect to find deep lineages in Eurasia, especially in the Far East. Yet sume)a in our laboratories and in others, invohing more than 1,000 people from Eurasia and its mitochondrial DNA satellites (Australia, Oceania and the Americas), have given no hint of that result. PRESENT AFRICAN EUROPEAN EAST ASIAN AUSTRALIAN It therefore seems very unlikely that an). trul), ancient lineages survive undetected in Eurasia. We simply do not see the result predicted by the regional model. Moreover, geneticists such as hlasatoshi Nei of Pennsylvania State University, Kenneth K. Kidd of Yale Universiv, James Wainscoat of the University of Oxford and Luigi L. Cavalli-Sforza of Stanford University have found support for an African origin model in their studies of nuclear genes. P roponents of the multiregional evolution model emphasize they have documented a continuity of anatomic morphologies between the archaic and modem residents of different regions; they insist these morphologies would be unlikely to evolve independently in any invading people. For that argument to hold true, it must also be sho\%n that the cranial features in question are truly independent of one another-that is, that natural selection would not tend to favor certain constellations of functionally related features. Yet we know powerful jaw muscles may impose changes on the mandible, the brow-ridge and other points on the skull; circumstances that promoted the evolution of these features in one population might do so again in a related population. Other paleontologists also dispute the evidence for continuity. They argue modern populations are not linked to past ones by morphological characteristics that evol\.ed uniquely in the fossil record. instead fossils and modem populations are united by their shared retention of still older ancestral characteristics. The continuity seen by believers in multiregional evolution may be an illusion. The idea that modem humans could cohabit a region with archaic ones and replace them completely without any mixture may sound unlikely. Nevertheless, some fossil finds do support the idea. Discoveries in the caves at Qafzeh 100,000~ v(1] NGAN;IONG 700,000 1 OLDUVAI 1 1 EUROPEAN 1 1 LANTIAN 1 1 JAVA I I I I I I I I I Hot40 &TUS HUMAN GROUPS WERE REPLACED throughout the Old World by modern humans from Africa. Archaic females do not seem to have contributed mitochontial genes to the modem people of Europe, east Asia and Australia. in Israel suggest Neanderthals and mod- ’ duct a detailed genealogical analysis. em humans lived side by side for Still more progress can be expected 40,000 years, yet they left little evidence in the immediate future, as molecular of interbreeding. biologists learn to apply their techHow one human population might niques to materials uncovered by our have replaced archaic humans without friendly rivals, the paleontologists. Preany detectable genetic mixing is still liminary molecular studies have ala mystery. One of us (Cann) suspects ’ ready been conducted on DNA from. infectious diseases could have con- mummified tissues found in a Florida tributed to the process by helping to bog and dated to 7,500 years ago. Imeliminate one group. Cavalli-Sforza has proved methods of extracting DNA speculated the ancestors of modern from still older fossilized bone now aphumans may have developed some pear close at hand. With them, we may modern trait, such as advanced lan- begin building the family tree from a guage skills, that effectively cut them j root that was alive when the human off from breeding with other hominids. ,I family was young. This and related questions may yield as molecular biologists learn how to link specific genetic sequences to the physical and behavioral traits those sequences influence. FURTHER READING Even before then, further studies of both nuclear and mitochondrial DNA will render more informative genetic trees. Particularly enticing are the sequences on the Y chromosome that determine maleness and that are therefore inherited from the father alone. Gerad Lucotte’s laboratory at College de France has indirectly compared such sequences in an effort to trace paternal lineages to a single progenitor- “Adam,” if you will. Those preliminary results also point to an African homeland, and with further refinements this work on paternal lineages may be able to provide an invaluable check on our results for maternal lineages. Unfortunately, base changes accumulate slowly on useful regions of the Y chromosome, making it technically difficult to con- ~IITOCHONDRIALDNAANDHUMAN EvoLUTION. Rebecca L Cann, Mark Stoneking and Allan C. Wilson in Nature, Vol. 325, No. 6099, pages 31-36; January l-7,1987. ~IITOCHOMXULDNA. M. Stonekingand A. C. Wilson in The Colonization of the Pacific: A Genetic Trail. Edited by Adrian V. S. Hill and Susan W. Serjeantson. Oxford University Press, 1989. hlrrOCHO~~DNASEQLTENCESNSINGLEkiAIRS FROMASOUTHERN AFRICAN POPIJUTION. Linda Vigilant, Renee Pennington, Henry Harpending, Thomas D. Kocher and Allan C. Wilson in Proceedings of the National Academy of Sciences, Vol. 86, No. 23, pages 9350-9354; December 1989. SEQUENCEEVOLUIIONOFMTOCHONDRIALDNANHuMANsANDCHI?~PANZEES. T. D. Kocher and A. C. Wilson in Evolution of Life. Edited by S. Osawa and T. Honjo. Springer-Verlag, Tokyo, 1991. SCIENTIFIC AMERICAN April 1992 73