Mitochondrial DNA (mtDNA) may offer the resolving power needed to ...... Extrachromosomal DNA, ICN-UCLA Symposia on Cellular Bi- ology, eds. Borst, P.
Proc. Nati. Acad. Sci. USA Vol. 78, No. 4, pp. 2432-2436, April 1981
Evolution
Evolutionary tree for apes and humans based on cleavage maps of mitochondrial DNA (restriction endonucleases/variable and conserved sites/rearrangement/branching order/knuckle-walking)
S. D. FERRIS, A. C. WILSON, AND W. M. BROWN* Department of Biochemistry, University of California, Berkeley, California 94720
Communicated by Sherwood L. Washburn, December 22, 1980
attain such a view, a sensitive and precise method of analysis must be used to compare mtDNA of different species. This paper explores the possibility that the mapping of restriction endonuclease cleavage sites can give the required sensitivity and precision. We present restriction maps containing approximately 50 cleavage sites per mtDNA for the primates mentioned above as well as for the orangutan and a gibbon. Comparison of these maps enables a tentative branching order ofthe lineages to be discerned. In addition, the map comparisons indicate that rearrangement, as well as base substitution, has occurred in the evolution of mtDNA in the higher primates. These comparisons also provide information about the distribution of positions in the mitochondrial genome at which evolutionary change occurs.
ABSTRACT The high rate of evolution of mitochondrial DNA makes this molecule suitable for genealogical research on such closely related species as humans and apes. Because previous approaches failed to establish the branching order of the lineages leading to humans, gorillas, and chimpanzees, we compared human mitochondrial DNA to mitochondrial DNA. from five species, of ape (common chimpanzee, pygmy chimpanzee, gorilla, orangutan, and gibbon). About. 50 restriction endonuclease cleavage sites were mapped in each mitochondrial DNA, and the six.maps were aligned with respect to 11 invariant positions. Differences among the maps were evident at 121 positions. Both conserved and variable, sites are widely dispersed in the mitochondrial genome. Besides site differences, ascribed to point mutations, there is evidence for one rearrangement: the gorilla map is shorter than the others owing to the deletion of 95 base pairs near the origin of replication. The parsimony method of deriving all six maps from a common ancestor produced a genealogical tree in which the common. and pygmy chimpanzee maps are the most closely related pair; the closest relative of thispair is the gorilla map; most closely related to this trio is the human map. This tree is only slightly more parsimonious than some alternative trees. Although this study has, given a magnified view of the genetic differences among humans and apes, the possibility of a three-way split among the lineages leading to humans, gorillas, and chimpanzees still deserves serious consideration.
Our understanding of human evolution would advance if we could find out how humans are related to their closest kin, the gorillas and chimpanzees. In the absence of an adequate fossil record for gorillas and chimpanzees (1), we must rely on the comparison of living species to elucidate the branching order of the lineages in the evolutionary tree. Yet, in spite of much comparative work, uncertainty about this branching order persists. Anatomists have long thought of gorillas and chimpanzees as the most similar pair in the trio, but phylogenetic analysis has never shown convincingly that this pair is the closest genealogically (2, 3). Nor is there agreement among chromosome workers (4, 5) as to the shape of the evolutionary tree. Furthermore, biochemical research (6-8) makes it appear that the three lineages diverged simultaneously. The inability of previous biochemical approaches to resolve the tree could be a consequence of the smallness of the nucleotide sequence differences found among the nuclear DNAs of the three types of creatures (9). Mitochondrial DNA (mtDNA) may offer the resolving power needed to elucidate- the branching order of these lineages because it evolves faster than nuclear DNA (10), thereby giving a magnified view of the genetic differences among species. To
MATERIALS AND METHODS Tissues and Cell Lines. mtDNA was purified from cultured cells, blood samples from living animals, or organs of animals that died of natural causes as follows: human HeLa. cells, common chimpanzee (Pan troglodytes) leukocytes, pygmy chimpanzee (Pan paniscus) liver, lowland gorilla (Gorilla gorilla) liver, orangutan (Pongo pygmaeus) liver, and white-handed gibbon (Hylobates lar) liver. The gorilla, chimpanzee, and orangutan samples were from the Yerkes Primate Center, Atlanta, GA, and the gibbon sample was from the Veterinary School, University of California, Davis, CA. Preparation and Cleavage Mapping of mtDNA. Our method of preparing mtDNA from tissues and cultured cells has been described (10). The 19 restriction endonucleases employed, with their single letter codes, are listed in. the legend to Fig. 1. All enzymes were obtained from New England BioLabs (Beverly, MA) I and used according to the supplier's directions. DNA fragments were labeled at the ends with 32p according to Brown's description (11). The methods for gel electrophoresis of DNA fragments and for mapping the cleavage sites have been described (10, 11). In most cases, the smallest routinely scored fragment was approximately 150 base pairs (bp). Details of the cleavage maps will be published elsewhere. Orientation of the Cleavage Maps. Cleavage maps of mtDNA were aligned with respect to the origin and direction of replication. The rationale and assumptions implicit. in this alignment are discussed elsewhere (10, 12). For several of the. enzymes employed, the cleavage maps of human and common chimpanzee mtDNA relative to the origin and direction of replication have been established (10, 13); sites detected with the remaining enzymes were oriented relative to these. The cleavage maps for the remaining species were aligned with respect to the 11 cleavage sites common to all six species.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
Abbreviation: bp; base pair(s). * Present address: Division of Biological Sciences, University of Michigan, Ann Arbor, MI 48109.
2432
Evolution: Ferris et al.
Proc. Natl. Acad. Sci. USA 78 (1981)
firms our placement of the HincII and Xba I sites in this region. Sites Containing Sequence Ambiguities. Those restriction enzymes that recognize more than one sequence can be very informative in mapping, when used in conjunction with "semiisoschizomers," i.e., enzymes recognizing specific subsets of the larger array of possible sequences. For example, HincII recognizes G-T-T-G-A-C, G-T-C-A-A-C, G-T-T-A-A-C, and GT-C-G-A-C. The latter two sequences are also recognized, respectively, by Hpa I and Sal 1 (15). Such relationships have been useful in assessing the homology of sites and in mapping sites that are very close to each other. For example, a side-by-side comparison of single enzyme digests of common chimpanzee mtDNA reveals that the Hpa I fragment produced by cleavage of the sites at 40 and 54 units is 260 bp larger (i.e., 1.5 units larger) than the corresponding HincIl fragment from this region. There must, therefore, be an intervening HincII site at either 41.5 or 52.5 units. A highly conserved Pst I site is located at 46 units. Side-by-side comparisons of double digests of chimpanzee mtDNA revealed that the fragment made by cleavage of the Pst I site at 46 units and the HincIl site at 54 units is 260 bp shorter than that formed by the cleavage of the corresponding Pst I and Hpa I sites. Because cleavage of the Pst I site at 46 units and either the Hpa I or the HinclI site at 40 units yielded identically migrating fragments, the intervening HincII site must be at 52.2 units. Gorilla mtDNA has a Deletion. That gorilla mtDNA has a deletion became evident during mapping of FnuDII sites near the origin of replication. All six species have FnuDII sites at 85, 92, 95, and 0.5 units (Fig. 1). However, the fragment mapping from 95 to 0.5 units was 95 bp shorter in gorilla than in human (or chimpanzee) mtDNAs (Fig. 2a). No FnuDII fragment of this size was observed upon electrophoresis in 3.5% polyacrylamide, which can allow detection of fragments as small as 16 bp (11). Studies with other restriction enzymes confirmed this interpretation. Every small fragment containing the origin of replication proved to be shorter by about 95 bp in the case of gorilla mtDNA (see, for example, Fig. 2b). Hence, the deletion is in the region between 95 map units and the origin of repli-
RESULTS Genome Size. Electron microscopy reveals that human and common chimpanzee mtDNA are 16,500 ± 400 bp long (10). This estimate agrees with that made by electrophoretically measuring the sizes of fragments produced by restriction enzymes (10). The mtDNA of pygmy chimpanzee, orangutan, and gibbon are identical in size to human and common chimpanzee mtDNA according to the electrophoretic method.t In contrast, gorilla mtDNA is shorter than the other five mtDNAs by about 95 bp (see below). Cleavage Maps. The cleavage maps for gorilla, common chimpanzee, human, orangutan, and gibbon mtDNA appear in Fig. 1. The 19 restriction enzymes, each of which is designated by a single letter, cleaved each genome at an average of about 50 sites. The precision with which a site may be mapped is illustrated for the case of the Pvu II sites (abbreviated as h on the maps), which occur at 59.5 map units in gorilla and chimpanzees. Because there is a HindIII (abbreviated as b on the maps) at 63 units in all 6 species, it was used as a reference point to assess the possible homology ofthe Pvu II sites. From the HindIII and Pvu II cleavage maps in Fig. 1, one predicts that the double digest should produce a fragment of about 600 bp, arising by cleavage of the HindI11 site at 63 units and the Pvu II site at 59.5 units (Fig. 1). If this fragment were 10 bp longer in one species, its electrophoretic mobility would be detectably different from that in the other species. In a side-by-side comparison of double digests, the mobilities of these fragments were identical. Thus, it is highly probable that the Pvu II sites at 59.5 units are homologous in gorilla and common chimpanzee. Further evidence for the accuracy of the maps is provided by comparing them with the published base sequence of the human mtDNA region from 50 to 54 map units (14), which conabsolute size estimates obtained from gel electrophoresis have an estimated SD of 5%, relative size estimates are considerably more accurate. The side-by-side comparisons of the mtDNA from the six species provided such relative estimates.
t Although
Deletion f zj k
rGorilla 11 1%.r -
I
...%a
dmz
a
b jxbgf
mxoz
II 11 11
I
~
00
Hl l
I
~
ei gwojowoxeeyh
C
II
I
b
c 0
h
oo
hc a x e kzo
I' 1 I1 11I I I
9ILIfe
IYt
I
C
km m fmexbf
z
Chimp
II I
a
I m
-HumanilLI
I
bxbgwh
11111 1 I I fL
iI l
zj kk
c oz eyzg
oz
e
xb
I 11111 III11
I
zjo k m
Orang
LL
m
Gibbon 11 I-f I LJ LO %.f I I
m
oh bf
m
I
I
I
I
d a f myb I-
I
0
mooxo
I11II11II11
I yp
m
zb
20
i
m I
e
j I
zg xeoox
LiJj,
I
i
I
zo
I
c
axo
kzo
1I I1 icr'
c yxbzyz j. I Lil W I
k axybe kz I -tf id
C ec
.
I
.
bxfo
Id
m m
I 40
zw
hed
1 se
.
.
..
oboz ah e
n
C
0
h
0
I obol o11 I IyLx kz ,,,,,irl ,, , , .Ih,
60
80
common chimpanzee. The scale is in
map units with the origin ofreplication
h x
for the FnuDIl sites in the region from
11 I
the position of one HincII site in orangutan mtDNA, which has been arbitrarily put at 36 units but may instead be at 39 units. The small letters represent sites of cleavage by the following restriction enzymes: a, EcoRI; b, HindIII; c, Hpa I; d, Bgl II; e, Xba I; f, BamHI; g, Pst I; h, Pvu II; i, Sal I; j, Sac I;I;k,o, Kpn I; m,x,Ava I; 1,w,Xho y, Bcl I;I; n, BstEII; Sma HincII;
at 0 units and the direction of replication to the right. Each map is for a single individual. The positions shown
96 to 99 units in the mtDNA of orangutan and gibbon are tentative, as is
ze z
a h oc gywbyo a w a x e kzo Zeoz I I i r' III? 11 I I I
c
mxwo Is
jgk
I I1 I1 I11 I
b I
1 Ii
LI 111 11
I11
C
yfm
d zx
xjoo )yhmbyo oh
c
ac
fm oombo
c
e e
FIG. 1. Cleavage maps of mitochondrial DNA from five species of higher primates. "Chimp" refers to the
h ze z 11 I
0
~
2433
zeozzz
I
'IP,
m n
x
If..I I'
f
100
Bgl I, and z, FnuDll. The mtDNA for pygmy chimpanzee (not shown) differs from mtDNA for common chimpanzee by having cleavage sites at an additional 11 positions (3w, 19o, 25b, 27z, 34z, 37y, 64h, 66a, 73z, 94y, and 99w) and lacking cleavage sites at 8 positions (4k, 17x, 45y, 51j, 54co, 59y, 60m, and 79o).
Evolution: Ferris et al.
2434
H
Proc. Natl. Acad. Sci. USA 78 (1981) H
G(
b
a
U
b
a
G
g
Gorilla Chimps Humaor
...
Orong Gibbon
o.
2905 `2810 -760 -665
FIG. 2. Electrophoretic detection of a deletion in gorilla mtDNA. (a) Restriction digests ofhuman (H) and gorilla (G) mtDNA were prepared with the enzyme FnuDH and subjected to electrophoresis. The gorilla mtDNA digest differs from human by having a fragment 665 bp long, instead of760 bp. This fragment spans the region from 95 units to 0.5 unit on the maps. (b)Kpn I digests ofhuman and gorilla mtDNA. The gorilla mtfNA~digest differs from that ofhuman by having a 2810bp fragment instead of the 2905-bp fragment. This fragment spans
from 84.5 units to4 units on the map (Fig. 1). Hence, the 95-bp deletion must have occurred between 95 units and 0.5 unit on the map.
cation (see Fig. 1). The deletion is evident inmtDNA from every one of four gorillas examined. Comparison of Maps. Of the six cleavage maps compared; those of the two chimpanzees are most similar. The number of sites mapped is 48 for the common chimpanzee and 51 for the pygmy chimpanzee. Of those sites, 40 occur at identical positions in the two species. By contrast, the two most dissimilar maps, those of the orangutan and gibbon, share only 19 sites. From these results, we estimated the probable percentage difference in base sequence for each pair of mtDNAs with the method of Nei and Li (16), using their equation 15. The estimated sequence differences range from 4% to 23%. These values exceed those obtained from annealing studies with singlecopy nuclear DNA (6, 9), confirming for the Hominoidea that mtDNA evolves 5-10 times faster than, nuclear DNA (10). Evolutionary Tree. The minimal number of mutations required to produce the six maps from an ancestral map was calculated. The tree that would allow the observed maps to evolve with the fewest mutations appears as a in Fig. 3. The method of choosing the tree and doing the calculation is based on the parsimony principle (17). This tree-building method, it should be emphasized, does not assume that the rate of molecular evolution is constant. The first step in the calculation is to list all 42 of the cleavage positions that are phylogenetically informative (Table 1). f The second step is to assume a particular tree and calculate for each position listed in Table 1 the minimal number ofmutations needed to produce the variation observed. For tree a in Fig. 3 and the first position listed in the table (16e), only one mutation is necessary to account for the e site's presence in the two chimpanzees and its absence in the other four species. At position 17x, two mutations would account for an x site being present only in the common chimt Not included in Table 1 are the 79 "unique" positions-i.e.,
positions for which the number of phylogenetically inferred changes is independent of the tree topology.
_
FIG. 3. Evolutionary trees for mitochondrial DNA of six higher primates. Tree a is based on the cleavage maps in Fig. 1 and requires a minimum of 67 mutations at the 42 positions listed in Table 1. The tree may be written in network notation as CP, G; H; OGi. (The abbreviations are explained in Table 1.) Other branching orders require more mutations, as shown by the following examples: CP, H; G; OGi (68); CP, G; HO; Gi (69); CP, GH; OGi (70); CP, H, G; OGi (71; this corresponds to tree b); CP, G; 0; HGi (71); HP, CG; OGi (79); CPGHOGi (98).
panzee and the gorilla. Altogether, a minimum. of 67 mutations suffices to produce the observed distribution of sites at the 42 positions tabulated, when tree a is used. This tree was found to be the most parsimonious, and it agrees exactly in branching order with that proposed by Simpson (18) on the basis of a qualitative analysis of morphological evidence. § When other trees are assumed, more mutations (68-98) are necessary (see legend to Fig. 3). Some of the alternative trees require only 1-4 more mutations than does tree a. For example, the tree supported by previous biochemical research and containing a three-way split among the gorilla, chimpanzee, and human lineages (see tree b, Fig. 3) requires 71 mutations, 4 more than does tree a, By contrast, most ofthe alternative trees require at least 80 mutations. We also note that all of the best trees-i.e., those having.67-71 mutations-put the two chimpanzees together as the most closely related pair among the six species, consistent with other biological evidence. Trees that separate these two species require at least 9 more mutations than do the trees that unite them. This result attests to the phylogenetic utility of the mtDNA mapping method. Distribution of Site Changes. From the tree analysis, it is inferred that a minimum of 147 mutations were fixed during the evolutionary divergence of the six maps. Sixty-seven of these mutations were at the 42 positions listed in Table 1 and the remaining 80 were at unique positions (see § footnote). These mutations are distributed among 121 variable positions, scattered widely in the genome (Fig. 4). The 11 invariant positions are also scattered widely (Fig. 4). No large region is exempt from evolutionary change, as is evident from the results of the numerical analysis in Table 2. The region coding mostly for ribosomal RNA may, however, be slightly more conservative than most other regions. DISCUSSION
Distribution of Site Changes. Although the positions at which there is site variation seem to be distributed almost randomly in the mitochondrial genome (Fig. 4, Table 2), we know that mutations are not fixed at random in this DNA. Thermostability studies with heteroduplex mtDNA indicate a wide range of susceptibility to evolutionary change in the mitochondrial genome (10). The nonlinear relationship observed between the extent of evolutionary divergence at restriction sites and the time since divergence (10) also indicates that there are § Other methods of tree construction also gave the result in Fig. 3a. In addition, individual variation has been examined within the human species (11) and within ape species (unpublished work). Tree a is supported by including all the variants.
Table 1. Phylogenetically informative variation at 42 positions in mtDNA of Hominoidea*
Position and nature of sitet 16e, 18b, 27b, 35h, 44e,53o,54e 17x,51j
32g,60h,95h
Site presentt C
P
+ + +
+ +
55x
19o 41z,45z,50x 29x 86co 21o
G + + + +
H
+
+
+
+
+
+
o
+
+ o
+
4k,59y 15f,65y 35o,38x,66z,81y 23m,61w 47w 95o 85h 1j,82e 33w,95x 20f,70h
+ +
+
+ + +
+
+ +
+ + +
+ +
+ + + +
+ +
Minimal no. of events
1
+
+
Gi
+ + +
+ +
6m
0
7 4 3 2 2 6
+
52w
+ + +
+ +
+
+ + + + + + + +
2 2 3 2 4 4 4 4 2 1 2 2 4 4
2 m Total 67 * The hominoid species examined are abbreviated as follows: P, pygmy chimpanzee; C, common chimpanzee; G, gorilla; H, human; 0, orangutan; and Gi, gibbon. t Each site is designated by a number for its position on the map and by one or two letters standing for the nature of the cleavage site. See legend of Fig. 1 for the single-letter code. t Presence of a cleavage site is indicated by a + or by a letter. In the latter case, the site is cleaved by only one ofthe two enzymes referred to in the first column.
101m
m
m
m
2435
Proc. Natl. Acad. Sci. USA 78 (1981)
Evolution: Ferris et al.
+
+
many positions in this genome at which evolutionary change is intolerable. These positions must be scattered widely, because there is little evidence for strong conservatism of any large region in the genome. The large region coding for ribosomal RNA (Fig. 4) was expected to be a prime candidate for evolutionary conservatism (21). The nuclear ribosomal genes are far more conservative than the average single-copy DNA sequence (22). The conservative nature of ribosomal genes in higher primates is confirmed by the work of Arnheim et al. (23). These investigators found no evolutionary change at 10 positions in the nuclear ribosomal genes, whereas we found 17 site changes at 13 positions in the mitochondrial ribosomal genes of the same six species. This dramatic difference in substitution rate emphasizes, for specific genes, the rapidity of mtDNA evolution in regions where strong conservation would be expected. Rearrangements. Although there is no evidence for gross rearrangement of the mitochondrial genome in any of the six species, considerable evidence exists for a small deletion in gorilla mtDNA near the origin of replication. Sequence rearrangements near the origin of replication have also been observed among individual sheep (24) and among Drosophila species (25); the rearrangements appear to be confined to this region in animal mtDNA (12). Such rearrangements can result in erroneous estimates of sequence similarity when these estimates are based solely on the fraction of shared restriction fragments. All fragments encompassing the region of the rearrangement will have their size
Table 2. Distribution of site changes in five regions of the mitochondrial genome Site changes No. of Site Genomic per position positions changes region, (a/b) (b) (a) map units 1.21 19 23 0-19.9 1.03 33 34 20-39.9 1.27 33 42 40-59.9
60-79.9 80-99.9
27 21
25 22
1.08 0.95
0-100
147
132
1.11
and hence their electrophoretic mobility altered. The fragment size comparison method used in recent studies (26) is thus subject to a source of error that the map method avoids. Resolving Power. Our cleavage mapping study of mtDNA provides the evolutionary biologist with many more genetic traits than are available from studies with proteins. Consider the case of mutational differences between humans and common chimpanzees. The observed number of site differences between these two species is 44 (Fig. 1). This number contrasts with that obtained from protein studies. The proteins whose amino acid sequences have been compared in the two species are myoglobin, fibrinopeptides, carbonic anhydrase, cytochrome c, and the a, 13, y, and 8 polypeptides of hemoglobin (9). For the 963 amino acid positions compared in these proteins, only three substitutions have been detected (9). Electrophoresis is another method of comparison that is valuable for detecting protein differences among closely related species. In an electrophoretic study of 23 loci in the same six species with which we are concerned, humans and chimpanzees were found to differ at an average of 7 loci (8). Thus, our map study of mtDNA provides 6 times more resolving power than conventional protein electrophoresis. Another class of biochemical methods measures genetic distance on a continuous scale rather than in mutational units. These are the methods of protein immunology (7, 27) and nu-
clear DNA hybridization (6, 9). According to these procedures, the human, gorilla, and chimpanzee lineages are equally divergent from one another (as shown in tree b, Fig. 3). Compared to the small distances among these species, the errors in estimating these distances are large (±25%). Hence, if the gorilla-chimpanzee divergence were 25% smaller than the ape-human divergence, these methods could fail to reveal this, simply because of the large error in measurement. The magnified view of the genetic differences among apes and humans provided by mtDNA indicates that the gorilla-chimpanzee divergence could be more recent than the divergence between the human and African ape lineages. Nevertheless, those alternative trees that are nearly as parsimonious as tree a focus our attention on the possibility that the gorilla, chimpanzee, and human lineages diverged almost simultaneously. In the absence of adequate statistical tests for comparing alternative trees, this possibility merits continued consideration. Another observation consistent with this possibility is that most of the positions at which restriction enzymes cleave appear to have experienced multiple substitutions (Table 1); this is indicative of a high incidence of parallel and back mutations during the evolution of hominoid mtDNA. The validity of the tree inferred from mtDNA maps can be tested by further studies. We speculate that at least 10 times more genetic information will be required to resolve the branching order for the gorilla, chimpanzee, and human lin-
OAII ,136
Evolution: Ferris et al.
Proc. Natl. Acad. Sci. USA 78 (1981)
Site changes H II 1 1
li
1
I
I
lA
Conserved sites _ K.
DpbJ4 -
-
|
I
11
1311A 121
1
I
1
1
L 0
16S
112SE I
20
40
60
80
100
FIG. 4. Distribution of evolutionary changes affecting restriction sites in mtDNA of higher primates. In the upper part of the figure, variable sites are indicated by vertical bars and invariant sites by triangles (A, four-base site; *, six-base site). There are 121 positions at which site variation occurs. All variable positions that occur within a given map unit have been grouped; the height of each vertical bar indicates the number of site changes within that map unit. The number of site changes has been inferred from tree analysis, using tree a shown in Fig. 3. The middle portion of the figure shows the positions, some tentative, of the D-loop and the genes for cytochrome b (b), three subunits of cytochrome oxidase (1, 2, and 3), and an ATPase subunit (A), as well as several genes ofunknown function and the two ribosomal RNA genes (16S and 12S). Black bars indicate tRNA genes and noncoding regions (19, 20). The distance scale at the bottom is in map units.
eages definitively. This amount of information can certainly be provided by mapping additional restriction sites and by direct nucleotide sequence analysis ofboth mtDNA and nuclear DNA. Significance of Branching Order. Knowledge of the branching order of the lineages leading to humans and African apes is a prerequisite for clear thinking about human ancestry. As long as there is uncertainty about the branching order, one cannot proceed to reconstruct the probable phenotype of the common ancestor ofhumans and African apes. Trees a and b (Fig. 3) may be used to illustrate this point. According to tree b, the common ancestor of gorillas and chimpanzees was also the ancestor of humans. In this event, many ofthe traits that gorillas and chimpanzees share were probably present in the ape-human ancestor. Knuckle-walking, for example, is unique to gorillas and chimpanzees (28). If tree b is correct, our ancestors were probably knuckle-walkers. If tree a is correct, however, knucklewalking probably arose after the human lineage had branched off. The importance of elucidating the branching order definitively is thus apparent. We thank G. Bourne, M. George, T. Kawakami, H. McClure, 0. Ryder, B. Swenson, and E. Zimmer for help in obtaining primate tissues; N. Arnheim, R. Cann, S. Carr, B. Chapman, W. Davidson, M.C. King, E. Prager, V. Sarich, A. Wang, R. T. White, and E. Zimmer for discussion; and L. Erdley for assistance in preparation of this manuscript. This research was supported by a grant from the National Science Foundation, a Center grant from the National Institute of Environmental Health Sciences, and fellowships from the Miller Institute and the Guggenheim Foundation. This paper was written while A.C.W. was on sabbatical leave in R. C. Lewontin's laboratory at Harvard
University. 1. McHenry, H. M. & Corruccini, R. S. (1980) Nature (London)
285, 397-398.
2. Oxnard, C. E. (1979) in Environment, Behavior and Morphology, eds. Morbeck, M. E., Preuschoft, H. & Gomberg, N. (Wenner-Gren Fndn., Gustav Fischer, New York), pp. 209-229. 3. Greenfield, L. 0. (1980) Am.J. Phys. Anthropol. 52, 351-365. 4. Spotorno, A. E. (1977) in Proceedings of the III Latinoamerican Congress of Genetics, Montevideo, Uruguay (UNESCO, New York), pp. 179-187. 5. Miller, D. A. (1977) Science 198, 1116-1124. 6. Benveniste, R. E. & Todaro, G. J. (1976) Nature (London) 261, 101-108.
7. Sarich, V. M. & Cronin, J. E. (1977) in Molecular Anthropology, eds. Goodman, M. & Tashian, R. E. (Plenum, New York), pp. 139-170. 8. Bruce, E. J. & Ayala, F. J. (1979) Evolution 33, 1040-1056. 9. King, M.-C. & Wilson, A. C. (1975) Science 188, 107-116. 10. Brown, W. M., George, M., Jr. & Wilson, A. C. (1979) Proc. Natl. Acad. Sci. USA 76, 1967-1971. 11. Brown, W. M. (1980) Proc. Natl. Acad. Sci. USA 77, 3605-609. 12. Brown, W. (1980) Ann. N.Y. Acad. Sci. 361, 119-134. 13. Brown, W. M. & Goodman, H. M. (1979) in Extrachromosomal DNA, ICN-UCLA Symposia on Cellular Biology, eds. Borst, P., Dawid, I. B., Weissman, S. & Fox, C. F. (Academic, New York), Vol. 15, pp. 485-500. 14. Barrell, B. G., Bankier, A. T. & Drouin, J. (1979) Nature (London) 282, 189-194. 15. Roberts, R. J. (1980) Nucleic Acids Res. 8, r63-r80. 16. Nei, M. & Li, W.-H. (1979) Proc. Natl. Acad. Sci. USA 76, 5269-5273. 17. Farris, J. S. (1970) Sys. Zool. 19, 83-92. 18. Simpson, G. G. (1963) in Classification and Human Evolution, ed. Washburn, S. L. (Aldine, Chicago), pp. 1-31. 19. Barrell, B. G., Anderson, S., Bankier, A. T., de Bruijn, M.H.L., Chen, E., Coulson, A. R., Drouin, J., Eperon, I. C., Nierlich, D. P., Roe, B., Sanger, F., Schreier, P. H., Smith, A.J.H.R., Staden, R. & Young, 1.G. (1980) Proceedings of the 31st Mosbach Colloquium, eds. Bucher, T., Sebald, W. & Weiss, H. (Springer, Heidelberg, Fed. Rep. Germany), pp. 11-25. 20. Attardi, G., Cantatore, P., Ching, E., Crews, S., Gelfand, R., Merkel, C., Montoya, J. & Ojala, D. (1980) in The Organization and Expression of the Mitochondrial Genome, eds. Kroon, A. M. & Saccone, C. (Elsevier/North-Holland, Amsterdam), pp. 103-119. 21. Shah, D. & Langley, C. H. (1979) Nature (London) 281, 696-699. 22. Sinclair, J. H. & Brown, D. D. (1971) Biochenistry 10, 2761-2769. 23. Arnheim, N., Krystal, M., Schmickel, R., Wilson, G., Ryder, 0. & Zimmer, E. (1980) Proc. Natl. Acad. Sci. USA 77, 7323-7327. 24. Upholt, W. B. & Dawid, I. B. (1977) Cell 11, 571-583. 25. Wolstenholme, D., Goddard, J. M. & Faruon, C. M. (1979) in Extrachromosomal DNA, ICN-UCLA Symposia on Cellular Biology, eds. Borst, P., Dawid, I.B., Weissman, S. & Fox, C.F. (Academic, New York), Vol. 15, pp. 409-425. 26. Avise, J. C., Lansman, R. A. & Shade, R. 0. (1979) Genetics 92, 279-295. 27. Wilson, A. C., Carlson, S. S. & White, T. J. (1977) Annu. Rev. Biochem. 46, 573-639. 28. Washburn, S. L. & Moore, R. (1980) Ape into Human (Little, Brown, Boston), 2nd Ed.
