J Mol Evol (1983) 19:255-271

Journal of
Molecular Evolution
Q 8pringer-Verlag 1983

Radiation of Human Mitochondria DNA Types

Analyzed by Restriction Endonuclease Cleavage Patterns

M.j. Johnson, D.C. Wallace, S.D. Ferris 1 , M.C. Rattazzi, and L.L. Cavalli-Sforza

l)ePartrnent of Genetics, Stanford School of Medicine, Stanford, California 94305, USA

Sumnaary. Human mitochondrial DNA (mtDNA) re- Introduction

striction endonuclease fragment patterns were analyzed
using total blood cell DNA isolated from 200 individuals Human mtDNA is a circular molecule of 16569 base
representing five different populations. Thirty-two frag- pairs (bp) which has been completely sequenced (Ander-
ment patterns (morphs) were observed with the enzymes son et al. 1981). Because there is substantial sequence
blpa I, Barn HI, Hae II, Msp I and Ava II yielding thirty- variation among individuals (Brown and Goodman
five different combinations of fragment patterns (mr 1979; Brown 1980; Denaro et al. 1981; C a n n e t al.
bNA types). The major ethnic groups exhibit quanti- 1982) mitochondrial DNA (mtDNA) has become a
tative as well as qualitative differences in their mtDNA useful tool in elaborating evolutionary relationships
!YPes, all of which are related to each other by a tree among closely related species (Ferris et al. 1981 a and
In Which the closely related mtDNA types cluster ac- b) and among individuals within species (Avise et al.
Cording to geographic origin. 1979a and b; Brown et al. 1981). Studies of variation
Three mtDNA types are postulated to be 'central' in different mammalian species suggest that mtDNA
to ethnic radiations due to their high frequencies, evolves at a faster rate than does the nuclear genome
their appearance in more than one ethnic group, or (Brown et at. 1979; Brown 1981 ; Ferris et al. 1981a and
their presence in other primate species. Genetic dis- b; Brown et al. 1982), but whether or not this phenom-
tances among populations were computed and employed enon is the result of a higher mutation rate and/or more
irt Construction of an average linkage tree. If one of rapid fixation of mutations is unknown. We have pre-
the three central mtDNA types is the root of the tree, viously reported Southern blot analysis of Hpa I cleav-
aifferences in evolutionary rates among the branches age patterns of mtDNAs from white blood cells of
~eeorae apparent. In particular, the Bushmen appear 235 individuals (Denaro et al. 1981). In the present
o have a higher evolutionary rate for mtDNA than the paper, we have extended this study to include cleav-
Other four populations. Comparisons with nuclear age patterns of four additional endonucleases, revealing
gene frequencies suggest that this higher evolutionary many additional polymorphic restriction sites which
rate naay be the product of an elevated mutation rate correlate with the ethnic origin of the sample. In one
or fixation of mutations in mtDNA. instance the nucleotide change responsible for gener-
ating a new cleavage pattern has been deduced from the
~ey Words: mtDNA - Restriction endonucleases - published human mtDNA sequence. These data demon-
~ _ Ethnic variation - Genetic distance strate that different human mtDNA types are related
" bivergence time by a series of mutations and that reconstruction of this
sequence of events can provide insight into the biological
history of man.

OfJbrint requests to: M.J. Johnson Materials and Methods

1 CUrrent address" Department of Biochemistry, University of Populations. Blood samples were of the following origins: 50
California, Berkeley, California, USA Caucasians (U.S. Whites and Europeans); 46 Asians (Taiwan,
256

Mainland China and Japan); 34 San (Bushmen) from the follow- searched for the specified 4 to 6 bp nucleotide recognition
ing locations in Botswana: Xaixai, Xunxa, Mahuahue, Tsodilo; sequence of each enzyme. From this program all possible sites
40 Bantu-speaking individuals (mostly Zulu) collected in Jo- only one base pair away from being a particular recognition
hannesburg, and 30 Warao Indians from Venezuela. site - later referred to as 'potential sites' (Adams and Rothman
Since mtDNA is maternally inherited (Giles et al. 1980a; 1982), were located for each enzyme. Identification of such
Case and Wallace 1982), all 200 individuals were chosen to be potential sites was based on size of the new fragment generated
unrelated on the mother's side. With the exeeptinn of the by a mutation. When the new fragments were > 5 kb in size
American Indians all individuals had been previously examined potential sites within (-+) 500 bp were considered; if > 2 kb but
for Hpa I digestion pattern variation (Denaro et al. 1981). The < 5 kb, -+ 150 bp were considered; if > 1 kb but < 2 kb, -+
Warao Hpa I pattern is analyzed in the present study. Total I00 bp; if < 1 kb, 4:50 bp.
DNA was extracted from the bully coats of whole blood from
Caucasian, Asian and African samples. Warao DNA was extracted
from approximately 2 x 106 white blood cells per individual
which had been stored in 7% DMSO and frozen in liquid nitro- Results
gen for three years. We have previously demonstrated that
individual mitochondria R.E. patterns are the same regardless
of whether DNA is isolated from buffy coats, platelets, or BAM HI: When individual DNAs from the five popula"
placental tissue (Denaro et aL 1981). tions were [Link] Bam HI (GGATCC), three dif-
ferent cleavage patterns were observed (Fig. 1). Morph
DNA Isolation. Total human DNA was extracted from buffy 1 has been described previously (Brown and Goodman
coats by the methods of Kan et al. (1970) with the addition
1979; Brown 1980) and corresponds to the published
of overnight dialysis against 10mM Tris-HCl (pH 7.5/1 mM
EDTA following the first phenol extraction. m t D N A sequence (Anderson et al. 1981) with a single
Bam HI restriction site at bp 14258. This produces a sin"
Restriction Analysis. Two micrograms of total DNA were used gle hybridization b a n d equivalent to the linear mtDblA
for each digestion. The conditions and buffers used were those molecule.
recommended by the manufacturer except that bovine serum
Morph 2 has two fragments o f approximately 14.4
albumin was omitted. All restriction enzymes were purchased
from Bethesda Research Laboratories - (Rockville, Md.). and 2.1 kb. Double digests with Pvu II and Barn HI gene"
Samples were digested with a three-fold excess of enzyme for rate three bands o f 11.6, 2.8, and 2.1 kb, indicating
five hours and then heated at 65~ for 10' to inactivate the that the new Bam HI site is at or near bp 16350. This
enzyme. A dye marker of bromocresol purple was added along site occurs in the D-loop an untranslated region of
with glycerol to a final concentration of 10% (vol/vol). The
digested fragments were separated on horizontal agarose slab gels the mitochondrial genome.
ranging from .7% to 1.8% depending on the expected size of Morph 3 was also seen in a previous study of 21
the restriction fragments. Size standaxds employed were iambda, individuals (Brown 1980) and arises as the result of
double digested with Eco RI and Hind III or human mtDNA of two Bam HI cleavage sites, the original and a new site
an established pattern double digested with Barn HI/Pvu II. within approximately 800 bp. Double digests with Ba~
Gels were run overnight at 1.5 v/era.; the fragments were then
transferred (Southern 1975) to a nitrocellulose filter HI and Pvu II .generate three bands o f 10.8, 4.9, and 0.8
(MiUipore). Hybridization was in 5x SSPE (1 x SSPE = 0.18 M kb localizing this site near bp 13460 - in an unidentified
NaCI, I0 mM NaH2PO4, lmM Na2EDTA, pH 7.0) at 42~ for reading frame - U R F 5.
48 h, using mtDNA purified from human tissue culture cells With Barn HI only the Caucasian population w~
(Giles et al. 1980b) as a probe, nick-translated (Rigby et al. polymorphic, all three Barn HI patterns being foun
1977) to a specific activity of > 1 x 108cpm/ug. Filters were
washed for 1 h in 2 x SSPE, .1% SDS and fragments were visu- in that group (Table I). In the Asian, American Indian,
alized after overnight autoradiography. and African populations only m o r p h 1 was found.
The three Barn HI morphs can be related to each otlaer
Restriction Mapping. Our analysis ofmtDNA variability assumes b y single site changes (Fig. 1), with m o r p h 1 the cen"
that different morphs are the result of point mutations which tral and presumably ancestral pattern. ChimpanZees
either create or destroy a specific restriction endonnelease re- and Orangutans appear to have the same Barn HI site
cognition site. Since the mitochondrial genome has been com- at 14258 ( m o r p h 1) but each has an additional site
pletely sequenced, the locations of these sites were easily detec-
which is not seen in humans (Ferris et al. 1981). An
ted by double digestions. The work of Greet and Kroon (1979)
and Castora et al. (1980) have demonstrated that variability ancestral pattern cannot therefore be unambiguously
in restriction patterns of mammalian mtDNA is not due to inferred.
methylation of cytosine residues. We have also analyzed a HAE II: Hae II (PGCGCQ, where P is a purine and
small subset of individuals with Msp I (insensitive to meth- Q a pyrimidine) was the second restiction endonuclease
ylation) and Hpa II (which recognizes the same site but is
sensitive to methylation) and found no difference in digestion used in characterizing the m t D N A of the CaucasiVa'
patterns between these two enzymes (data not shown). Asian, and African samples. Seven morphs were de"
The human mtDNA sequence (Anderson et al. 1981) was tected in these four populations as seen in Fig. 2.
used extensively in our analysis of variant patterns, along with Morph 1 corresponds to the published rntDblh
computerized DNA sequence analysis programs available through sequence (Anderson et al. 1981). Autoradiogra#~s
the MOLGEN PROJECT of the Stanford University Medical
Experiment in Artificial Intelligence in Medicine (SUMEX-AIM). o f this pattern show bands o f 4.7. 4.5, 4.1, 1.4, 1.5,
To determine possible mutation sites which may be responsible and 0.4 kb (a 0.1 kb band is not observable under
for the variable patterns observed, the mtDNA sequence was these conditions).
 257

1 2 3

16.5)=..
14.41~ -~15.7

bJ

2.1~,.
1 Fig. 1. Autoradiogram of the three
Barn HI cleavage patterns. Morphs are
denoted by numbers at top of lanes;
fragment sizes are designated in kilo-
bases. Also shown is a phylogeny of
these Bam HI morphs, arrowheads in-
-qt0.8 dicate constant sites while small letters
denote those sites which are lost or
gained

Table 1. Number of individuals and frequencies of all morphs listed by enzyme for each ethnic group sampled

Caucasian Oriental Bantu Bushmen African Amer. Indian

a~ple size 50 46 40 34 74 30
Al~ No. Freq. No. Freq. No. Freq. No. Freq. No. Freq. No. Freq.

BanlHI -I 43 86.0 46 100.0 40 100.0 34 100.0 74 100.0 30 100.0

-2 4 8.0 0 0.0 0 0.0 0 0.0 0 0,0 0 0.0
-3 3 6.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0
Iiae II -1 38 76.0 37 80.4 39 97.5 34 100.0 73 98.6
-2 7 14.0 6 13.0 1 2.5 0 0.0 1 1.4
-3 3 6.0 0 0.0 0 0.0 0 0.0 0 0.0
-4 0 0.0 1 2.2 0 0.0 0 0.0 0 0.0
-5 0 0.0 2 4.3 0 0.0 0 0.0 0 0.0
-6 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0
-7 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0
~tsp I -1 46 92.0 45 97.8 35 87.5 6 17.6 41 55.4 30 100.0
-2 0 0.0 0 0.0 5 12.5 18 52.9 23 31.1 0 0.0
-3 0 0.0 0 0.0 0 0.0 9 26,5 9 12.2 0 0.0
-4 4 8.0 I 2.2 0 0.0 0 0.0 0 0.0 0 0.0
-5 0 0.0 0 0.0 0 0.0 1 2.9 1 1.4 0 0.0
~va II -1 37 74.0 44 95.7 16 40.0 4 11.8 20 27.0 30 100.0
-2 2 4.0 0 0.0 5 12.5 20 58.8 25 33.8 0 0.0
-3 1 2.0 0 0.0 15 37.5 1 2.9 16 21.6 0 0.0
-4 0 0.0 0 0.0 0 0.0 2 5.9 2 2.7 0 0.0
-5 4 8.0 0 0.0 2 5.0 7 20.6. 9 12.2 0 0.0
-6 0 0.0 2 4.3 0 0.0 0 0.0 0 0.0 0 0.0
-7 1 2.0 0 0.0 1 2.5 0 0.0 1 1.4 0 0.0
-8 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0
-9 3 6.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0
-10 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0
-11 0 0.0 0 0.0 I 2.5 0 0.0 1 1.4 0 0.0
258

1 2 3 4 5 6 7

-~3.9

2.2]l~
1.9Z=,-

1 4b,..
1:2~. "/miim -911.3 a

0.41~

Fig. 2. Autoradiogram and phylogeny of the seven Hae II morphs found in this study. Numbers at top of lanes and within circles
correspond to each morph. The phylogeny shows that all morphs can be interconverted through morph 1 - presumed to be central
to the other site changes observed. Letters indicate the sites which change between any two morphs

Morph 2 arises by loss of the 4.5 and 4.1 bands about bp 9250. Four potential sites exist in this re"
and the appearance of a new band at 8.6 kb. This gion -+ 100 bp, and all occur in the CO III coding region'
is due to an alteration of a recognition site starting The last variable pattern - morph 7 - occurs when
at bp 9052, in the region coding for ATPase 6. Both the 1.2 and 0.14 kb contiguous bands fuse to forna
morphs I and 2 were described previously (Giles et a 1.34 kb band. Double digests with Bam HI and Hae II
al. 1980a). cut this newly formed band giving two bands of 0.66
and 0.74 kb. This pattern is due to a change in the
Morph 3 is generated by replacement of the 1.4 and recognition site beginning with bp 14858, and occUrS
4.5 kb bands with a new band of 5.9 kb. On the mtDNA within the cytochrome b coding region.
map the 1.4 and 4.5 kb bands are contiguous, with The American Indian population could not be ~a-
the new pattern probably arising as the result of a lyzed with Hae II. This DNA was isolated from aPPr~
single bp mutation within the 4529-4534 bp Hae II ximately 2 x 106 lymphocytes/individual, frozen in
recognition site found in URF 2. 7% DMSO and stored for 3 years in liquid nitrogen'
Morph 4 was described previously (Wallace et al. Even after several repurifications of these samples
1982), and results from a new Hae II site within the we were unable to obtain even partial digests using
4.1 kb band cleaving it into two fragments of 2.2 and enzymes purchased from 2 different suppliers (BRL
1.9 kb. Double digests indicate that the site occurs at and BioLabs). When Lambda DNA was added to the
approximately bp 10952, where only one potential site reaction mixture it also remained undigested, while
is found, beginning with bp 1 i001. An A to G tran- other cell samples - stored under almost identical
sition at nucleotide 11002 would create the new site circumstances and from which DNA was isolated in the
in URF 4 without altering the amino acid sequence. same manner - were fully digested with Hae II (data
The pattern for morph 5 is replacement of the not shown). We can only conclude that some substance
4.5 kb band with a new band at 4.2 kb. Double di- found in these DNA samples specifically inhibits the
gests with Hae II and Bst Eli indicate that a new Hae II enzymatic activity of Hae II restriction endonucle"
site is generated within the 4.5 kb band at approxi- ases.
mately bp 4830, generating fragments of 4.2 and 0.3 kb. With the exception of one individual with m ~
All potential sites occur in URF 2. 2 (Table 1) the two African populations are moaO"
Morph 6 is generated by loss of the 4.1 kb band morphic for morph 1, but the Caucasian and Asiva
and appearance of a 3.9 kb band. Double digestion populations are highly polymorphic. Five naorphS
with Hae II/Sst I suggests that the site should lie at were seen in the Caucasians surveyed: 76% were naorpla
 259

1, 14% morph 2, 6% morph 3, while morphs 6 and 7 are fused in morph 2, implying that it lacks two adjacent
Were found exclusively in Caucasians but at a low sites found in the published sequence. This could how-
frequency of 2%. The Asian population contained ever, be interpreted in two ways. Either morph 1 is
four morphs: 81% morph 1, 13% morph 2 and rare identical to the sequenced mtDNA (in which ease
instances of morphs 4 and 5. These later two morphs two mutations are required to get from morph 1 to 2)
Were found only in Asians. or our morph 1 has only one site in this region while
A Phylogeny of the seven Hae II morphs can be the sequenced mtDNA may have an additional site.
COnstructed (Fig. 2) wherein all morphs can be de- Both sites occur in the CO I1 gene.
rived from morph 1 by the loss or gain of a single Morph 3 is similar to morph 2, having the same
l-Iae II site. The presence of morph 1 in all four popula- pattern but with the addition of a new site within
tions at high frequencies and the fact that all morphs can the 2.396 band generating two new bands of 2.15 and
be derived directly from it suggests that it may have 0.255 kb. In double digests with Msp I and Hpa I,
been central to mtDNA radiations. the 2.15 kb fragment disappears and is replaced by
MSP I: The mtDNAs of our populations were also a 0.72 and 1.42 kb fragments suggesting that it was
a~alyzed with Msp I. Twenty three Msp I (CCGG) cut into three pieces (1.42, 0.72, and 0.255). This
sites are found in the published human mtDNA se- indicates the presence of a new site at approximately
qtlenee (Anderson et al. 1981) and all fragments above bp 11440. There are three potential sites within -+ 50 bp
0.3 kb could be resolved by separation on 1.8% aga- of this site; all found in URF 4.
rose gels followed by autoradiography. Visible bands Msp I morph 4 is similar to morph 1 but it has
range in size from 2.39 kb to 0.309 kb. Five different a new fragment of approximately 2.7 kb. Double digests
r~~ were found in our samples (Fig. 3) beginning using Msp I and Barn HI reveal that the 2.213 kb and
with morph 1 which has the same pattern as would 0.528 kb bands have fused due to a site loss. A single
be expected from the published sequence using our bp mutation at the Msp I site (beginning with bp 15925)
nlethods. would occur within the region coding for threonine
In morph 2 the 1.1 and 0.9 kb fragments of morph 1 tRNA.
are replaced by a new fragment of approximately 2.04 Msp I morph 5 is a derivative of morph 2. In addition
kb. The available mtDNA sequence contains two restric- to the fusion of the 1.1 and 0.9 kb fragments there is
tion sites between these framgents; separated by 0.038 also a new site in the 1.24 kb fragment creating bands
kb fragment; one at 8112 and one at 8150. Double of 0.95 and 0.29 kb. Double digests with Msp I and
digests with Msp I and Pst I show that these fragments Hind III cut within the 0.95 fragment generating 0.5

1 2 3 4 5
Yf "'. .

, b "'.

i ~ "~ orb e "... 9

,,

2.73
lib "912.15
"( 2.04 ....

1.4 b,.
1'2 b,,,. "'"'.ta
e
or b i
"'J

,9 1~.' ~ ~ -~ .95

'g2 ....

Rig
93. Msp I autoradiogram and phylogeny of cleavage patterns. These patterns are related in a phylogeny which assumes that the
~ 1 of our study is not the same as that of the published sequence of Anderson et al. (t981) but rather has lost either size a or b.
e morph corresponding to the published sequence is designated by a star, and was presumably not seen in this study
260

and 0.44 kb fragments, indicating that the new site the most common pattern in four groups: Caucasians
is at about bp 13070. Potential sites all occur within (92%); Asians (98%); American Indians (100%); and
URF 5. Bantu-speakers (88%). Morph 1 was found in onlY
As seen in Table 1 these morphs cluster according 18% of the Bushmen surveyed. A significant difference
to the geographic origin of the samples. Morph 1 was is observed in the frequency of morph 2, which is

1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 110 111

9.8

[Link] qlW IIilP

12.0

, ~ID 1.3
1.0 ~ im

o. ~ ;~,i~ <.~ _

f f

Fig. 4. Autoradiogram and phylogeny of Ava II cleavage patterns. All morphs are aligned individually against morph 1 for comparison"
The phylogeny shows all possible alternatives for relating the eleven morphs, although only one alternative is necessary to derive ot~r
morph from another. As in the other phylogenies, the numbers indicate the morph, arrowheads designate sites which do not chatl/P'
and small letters with arrows indicate those sites which are either lost or gained
 261

Present in 53% of the Bushmen and 13% of the Bantus site changes. It could come from morph 9 by fusing
but not found in Caucasian; Asian; or American Indian the 3.7 kb band (seen in morph 9) with the 0.836 kb
POpulations. band to make a new fragment of 4.5 kb. Conversely,
Morph 3 is found exclusively in Bushmen, and it may derive from morph 3, which has a 3.836 kb band
accounted for 27% of the sample. Morph 4 is found (formed by fusion of 3.023 and 0.836). Loss of the
in 8% of Caucasians surveyed, is rare in Asians (2%) Ava I1 site at 13367 in URF 5eseparates the 3.023 band
and is not found in either Bantu or Bushmen, while from the 0.7 kb band.
the rare morph 5 is found exclusively in Bushmen. Morph 8 is related to morph 1 by a site gain in
A phylogeny of these Msp I morphs (Fig. 3) shows the largest (9.853 kb) fragment generating two new
that all morphs can be interconverted by single site fragments of 7.85 and 2.0 kb. Double digests with
changes if we assume that our morph 1 is not the same Sst I and Ava II gave new bands of 4.9 and 2.95, cutting
as the published sequence. in the 7.85 kb fragment, indicating that morph 8 is
AVA II: Digesting with Ava 1I (GG[A/T]CC) eleven due to a new site at approximately bp 4776, in URF 2.
different patterns were observed within the five popula- Morph 9 differs from morph 1 by the fusion of
tions surveyed. Autoradiograms and restriction maps are the two contiguous fragments of 3.023 and 0.738 to
shown in Fig.4. Morph 1 corresponds to the published form a new band of 3.761. This morph could arise
Sequence, with ei_ght recognition sites generating visible in 2 ways,either from morph 1 by a site loss at bp 13367
fragments of 9.853, 3.023, 1.09, 0.836, and 0.738 kb. or from morph 7 by a site gain at 16390 separating
banMdOrph 2 is the result of a site gain in the 9.853 kb the 0.836 kb fragment from the fused 3.023 and 0.738
which generates two bands of approximately kb fragments. Base pair 13367 is within URF 5 while
5.45 and 4.4 kb. Double digests with Sst I and Ava II bp 16390 is in the D-loop.
cut Within the smaller of these two fragments localizing Morph 10 is a site gain in the 9.853 fragment gen-
the new site to bp, 8229. Numerous potential sites erating two new fragments of 8.7 and 1.1 kb. Double
OCcur near this location, all are found within the CO II digests with Sst I and Ava II cut the larger 8.7 kb frag-
COdingregion. ment into two fragments of 3.0 and 5.7 kb. This im-
Morph 3 is related to morph 1 by a site loss at bp plies that the new site which generates morph 10 occurs
16390 so that two contiguous segments (3.023 and at bp 3876, in URF 1.
0.836) are fused to form a 3.859 kb fragment in Ava II The final pattern seen with Ava II - morph 11 -
digests. This mutational event occurs within the D-loop. differs from morph 1 by having two additional sites. As
Morph 5 seems to be a combination of morphs 2 in morph 2, morph 11 has a new site at about 8229
and 3 having gained a site at bp 8275 (as in morph 2) creating 5.45 and 4.4 kb fragments, and like morph
and having lost one at Ava II recognition site 16390 6, another new site at 15870 generating 2.5 and 0.5 kb
(as in morph 3). For reasons which we will discuss fragments.
later, we suggest that these site losses at bp 16390 Caucasians were highly polymorphic with this en-
are independent events occurring at a mutational hot zyme (Table 1). Morph 1 accounts for 74% of those
spot in the D-loop of the mitochondria. surveyed, morph 2-4%, morphs 3,7,8 and 10 each
Morph 4 is related to morph 5 with an additional 2%, morph 5-8% and morph 9 with a frequency of 6%.
Site in the 3.859 kb fragment producing 2 new frag- Morphs 8-10 were found exclusively in Cau-
ments of approximately 2.5 and 1.3 kb. Double digests casians.
With Barn HI and Ava II cut within the 2.5 kb frag- Asians were less polymorphic, having primarily
rnent indicating that the site generating morph 4 at morph 1 (96%). Only one other morph, morph 6,
approximately bp 15890 could occur either within was found in 4% of those surveyed, and not seen in
the threonine tRNA, the proline tRNA, cytochrome b, any other population. American Indians surveyed were
or Within the D-loop. monomorphic for morph 1.
Morph 6 is related to morph 1 by another site gain Bantus were primarily morph 1 (40%). with morph
within the 3.023 fragment, yielding two new fragments 2 at 12.5% and morph 3 at 37.5%. Morphs 5,7, and
of 2.5 and 0.5 [Link] digests with Barn HI and 11 were also seen although at low frequencies. Bush-
AVa II give new bands of 1.6 and 0.9 - cutting in- men were the only population which was not predom-
Side the 2.5 kb fragments. This again implicates a site inately morph 1 (only 12%). Instead, 59% were morph
! or near bla 15890 (as in momh 4). Without sequen- 2, and 20% were morph 5, while morphs 3 (3%) and
th.~ however, it is impossible to tell if they are really 4 (6%) were also found. A phylogeny of these Ava II
= sanae mutation or mutations in different sites which morphs is shown in figure 4 indicating that there are
are in close proximity. many possible ways to interrelate these morphs.
thM~ 7 differs from morph 1 by the fusion of In this study we noted considerable variation in
e 3.023 fragment to the 0.836 and 0.738 kb frag- the information provided by individual enzymes. Table 2
ments to yield a new fragment of 4.59 kb. This morph illustrates results from a Chi Square test for significance
cOUld be derived from either morphs 3 or 9 by single of the different patterns observed in the five popula-
262

Table 2. Statistical significance of individual enzymes are indicated o n a linear m a p o f the m t D N A mole-
cule. Polymorphic sites are indicated b y asterisks and
Enzyme Cauc.- Cauc.- Asian- Bt- Cauc.- Asian- Afr.- these occur in m a n y different coding sequences, tRNAs,
Asian Afr. Afr. Bsh A. Ind. A. Ind. A. Ind.
rRNAs and w i t h i n the D-loop.
Hpal ++ +++ +++ 4+ - + +++
BamHI ++ ++ . . . . .
Haell + +++ ++ - nd nd nd P h y l o g e n y of m t D N A T y p e s
Mspl - +++ +++ +++ - - -
Avail +++ +++ +++ +++ +++ - +++
By c o m b i n i n g the restriction endonuclease morphS
Levels of significance of X2 tests between frequencies of morphs observed for each individual we f o u n d 35 distinct
represented in the five populations: - : P > 5%; +: 0.1% < P < m t D N A types which are listed by f r e q u e n c y in Table 3.
5%; ++: 0.1% < P < 1%; +++: P < 0.1%; nd: no data available. A m e r i c a n Indians are listed in the last c o l u m n as mtDNA
Comparisons are made between all populations: Cauc. = Cau- type ( 2 - 1 - ? - 1 - 1 ) due to the u n c e r t a i n t y o f their Hae
casian; Afr. = African; Bt = Bantu; Bsh = Bushmen; A. Ind. =
II m o r p h . [An additional table which gives all R.E. sites
American Indian
for each m t D N A type is included in the appendix.] The
i n f o r m a t i o n in Table 3 was t h e n used to construct a
tions. Bam HI which generated the fewest morphs, was p h y l o g e n y o f h u m a n m t D N A types. This phylogeny
the least informative o f all e n z y m e s e m p l o y e d in this (Fig. 6) was derived b y relating the existing mtDNA
study. Hpa I and Ava II were particularly useful - show- types through single site changes. O n l y 5 missing inter"
ing significant differences b e t w e e n most p o p u l a t i o n s mediates (represented in parentheses) need to be
tested while Msp 1 c o n t r i b u t e d most significantly to postulated to show the relationships among the 35
the distinction b e t w e e n the two African populations. m t D N A types. Our p h y l o g e n y was compared to several
In Fig. 5 all sites observed with these five e n z y m e s p a r s i m o n i o u s (Wagner) trees (Farris 1972) and was

* *

T
11 11 II ~ 9,

12s RNA 165 R N A URF 1

0%
]O~ II~ II If I 25%
H F V L

"' Ill !! !!!11 III 9* o

URF 2 CO 1 CO li
t I!1 II IIII/ I,I/
25% i t ..-"/I t N\ 50%
QM WANOLCY SO
Fig. 5. Linear Map of Hunaar~
.K- 4(- * mtDNA Molecule. The 16.5 kilo"
TT T bases of the human mitochondri"
"" I I II II I II II ... al molecule are converted to map
units of 0 to 100% (i.e. 1 map
6 unit = 165 base pairs) and these
IURF A6L CO t i t URF3 URF4L LIRF 4
50% I~ * I I,I II I II I l l are represented in four equal seg"
~ ~ I 75% ments. All restriction sites anal"
K ATPase 6 E R HSL
yzed in this study are marked,
those found to be polymorphiC
are designated by an asterisg'
Conventional coding sequence at0-
I I1,, II1 breviations are used: URF = [Link]-
6 identified Reading Frame; CO I.
URF 5 URF 6 Cyt b III - Cytochrome e oxidase sub"
75% ] I Itl
E
/I tl
TP
I1oo units; Cyt b = Cytochrome b. The
tRNAs are abbreviated using the
one letter amino acid code, origi0S
T""" SM.., ;A..,, of replication are O H for the heaVY
strand and O L for the light stratad
= Polymorphir Sites origins
 263

Table 3. MtDNA types. Enzyme morphs are listed in parenthesis in the order: Hpa I, Barn HI, Hae II, Msp I, and Ava II. American
Indians are listed in the last row as type ( 2 - 1 - ? - 1 - 1 ) due to the uncertainty as to their Hae II morph. Asterisks denote samples
Which include the Warao, for computation of genetic distance. F, indicated at the bottom of the table, is a measure of homogeneity in
the POPulations as discussed in the text

Caucasian Oriental Bantu Bushmen African Amer. Indian World

araple size 50 46 40 34 74 30 170/200"

~ntDNA type No. Freq. No. Freq. No. Freq. No. Freq. No. Freq. No. Freq.

1(2-I-1-1_1) 29 58.0 32 69.6 10 25.0 1 2.9 11 14.9 0 0.0 72/102"

2 (3-1-1-1_3) 0 0.0 0 0.0 13 32.5 1 2.9 14 18.9 0 0.0 14
3 (3-1-1-2_2) 0 0.0 0 0.0 3 7.5 9 26.5 12 16.2 0 0.0 12
4 (3-1-1-3_2) 0 0.0 0 0.0 0 0.0 9 26.5 9 12.2 0 0.0 9
5 (3-1-1-2_5) 0 0.0 0 0.0 1 2.5 7 20.6 8 10.8 0 0.0 8
6 (2-I-2-1_1) 6 12.0 1 2.2 0 0.0 0 0.0 0 0.0 0 0.0 7
7 (3-1-1-1_1) 0 0.0 0 0.0 4 i0.0 2 5.9 6 8.1 0 0.0 6
8 (1-1-1-1_1) 0 0.0 2 4.3 2 5.0 0 0.0 2 2.7 0 0.0 4
9 (1-I-2-1_1) 0 0.0 4 8.7 0 0.0 0 0.0 0 0.0 0 0.0 4
10 ( 3 - 1 - 1 _ 1-2) 0 0.0 0 0.0 2 5.0 2 5.9 4 5.4 0 0.0 4
11 ( 2 - 2 - 3 - 1 _ 5 ) 3 6.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 3
12 ( 4 - 1 - 1 - 1 _ 1 ) 0 0.0 2 4.3 0 0.0 0 0.0 0 0.0 0 0.0 2
13 ( 2 - 1 - 5 - 1 _ 1 ) 0 0.0 2 4.3 0 0.0 0 0.0 0 0.0 0 0.0 2
14 ( 3 - 1 - 1 - 2 - 0 0.0 0 0.0 0 0.0 2 5.9 2 2.7 0 0.0 2
15(2_1_1_1
. . . . 8))10) 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
16(2 1 2 1 - 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
17 (2-1~1-1_9) 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
}1~ ( 2 - 3 - 1 - 4 _ 9 ) I 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
~ (2-3-1-4_7) 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
20 ( 2 - 3 - 7 - 4 - 9 ) 1 2.0 0 0.0 0 0.0 0 0,0 0 0.0 0 0.0 1
21 ( 2 ~ 1 - 1 - 1 - 2 ) 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
22(2-1-1_1_5) 1 2.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1
~3~2-2_1_1_3) ~ 20 0 00 0 o0 o 00 0 0o 0 00 1
~(G-1-1-~) 1 ~.0 0 oo 0 0.0 0 oo 0 oo 0 oo 1
~,(2 L L 4 _ 1 2.0 0 0.0 0 00 0 0.0 0 0.0 0 oo 1
~(2-1-6_1_1) 1 2.0 0 0.0 0 0.0 0 oo 0 00 0 0.0 1
27 ( 2 - 1 - 4 - 1 - 6 ) 0 0.0 1 2.2 0 0.0 0 0.0 0 0.0 0 0.0 1
~(~-~-1-4-1) 0 oo 1 2.2 0 oo 0 0.0 0 0.0 0 0.0 1
29 ( 6 - 1 ~ 2 - 1 - 6 ) 0 0.0 1 2,2 O 0.0 0 0.0 0 0.0 0 0.0 1
30 ( 3 - 1 - 1 - 1 - 15)
31 (3-1~1-1 1) 0 0.0 0 0.0 1 2.5 0 0.0 I 1.4 0 0.0 1
32 (3_i ~ -1~ 0 0.0 0 0.0 1 2.5 0 0.0 1 1.4 0 0.0 1
33 r . - 1 - ~ - ~ J 0 0.0 0 0.0 0 0.0 1 2.9 1 1.4 0 0.0 1
~. ~ a ~ 1 - 1 - 2 - 3 ) 0 0.0 0 0.0 1 2.5 0 0.0 1 1.4 0 0.0 1
~4 (3~1~2_1_3) 0 0.0 0 0.0 1 2.5 0 0.0 1 1.4 0 0.0 1
J5 (3~1~1_1_7) 0 0.0 0 0.0 1 2.5 0 0.0 1 1.4 0 0.0 1

-•(2-1-?-1-1) 0 0.0 0 0.0 0 0.0 0 0.9 0 0.0 30 100.0 0.0

0.359 0.499 0.192 0.195 0.122 1.00 0.277

identical to one such tree. This phylogeny however, branches, while in two instances it was necessary
Permits separation o f the geographically distinct groups to postulate independent acquisitions of the same
onto different branches while maintaining a minimal site (Ava II sites c and e). Site e could be located within
number o f mutations. the D-loop which has been shown to be highly var-
The most divergent phylogeny which has equal iable by nucleotide sequencing (Aquadro and Green-
n ~ b e r s o f mutations, but which does not take account berg 1982).
of geographic constraints, derives the branch begin- Several conclusions can be made on the basis o f
~ing With m t D N A t y p e 21 (which contains types found Fig. 6 and Table 3. First, the phylogeny is centered
Only in Caucasians) from the African t y p e 10 by a around one specific m t D N A type - 1 ( 2 - 1 - 1 - 1 - 1 )
blpa I site change rather than from m t D N A type 1 from which more branches radiate than from any
by an Ava II site change. Other possible phylogenies other type. One o f these branches forms a distinct
are listed in the legend to Fig. 6. In some instances African lineage in which m t D N A type 1 is found in
(also in legend to Fig. 6) the phylogeny assumes that lower frequency (18%), while other - shorter - lin-
the Same site loss occurs independently in different eages are seen in the Caucasian and Oriental m t D N A
264

~k, African (Bantu& I~ln-,en)

9 Bantu

23
[ ] Caucasian []
( ~ Oriental 1"-119
24

IZ2
35

26
r-
15

%.

31

32 ~ ~ I 12
13

b2?
29

Fig. 6. Phylogeny o f m t D N A Types. This tree was derived b y parsimonious m e t h o d s as described in the text. In several instances'
recurrent sites losses are postulated in order to relate the m t D N A types by a minimal n u m b e r o f m u t a t i o n s . These are: a) Hae 11 c,
8 --, 9, 2 ~ 34, 1 --, 6 ; b ) Msp 1 a or b, 2--, 33, 3 ~ 1 0 ; c ) Msp I e, 1 7 ~ 18, 1 ~ 28, 21 ~ 2 4 ; d ) A v a i l c, 3 ~ 5 , 2 - - , 7 , 1 0 ~ 31, 18-"
19, 21 -~ 22; e: Ava II d, 1 ~ 17, 2 ~ 35. In two instances a recurrent site gain was postulated, again in order to relate the rn tDNA
types b y a minimal n u m b e r o f m u t a t i o n s . These are: a) Ava II e, 1 ~ 21, 7 --* 10; b) Ava II e, 5 --, 14, 9 ~ 29, 1 ~ 27, 1 0 ~ 30. Tlae
three postulated central m t D N A types (types 1, 6 & 8) are enclosed within a dashed line. Other possible phylogenies requiring e l t ~
n u m b e r s o f m u t a t i o n s would derive a) m t D N A type 6 from 9 by a Hpa I c+ rather t h a n Hae II c- site change; b) m t D N A type 24 tfo
28 by an Ava II c+ change rather t h a n an Msp I e- change

populations. These peripheral Caucasian and Oriental The question of which of these three types (or ao~
mtDNA types are found in lower frequencies for these other) is ancestral to the others can only be answere
populations are predominantly type 1. MTDNA type in terms of probability. Formally, mtDNA types rai~ t
1 is also the only one found in all the major groups be considered as alleles at a single locus. Under con&
studied and is the most common type of the entire tions of neutrality the most common allele is likely
sample. Two other mtDNA types separated from type 1 to be the oldest, with a probability equal to its fre"
by a single mutation - type 6 ( 2 - I - 2 - 1 - 1 ) and quency (Watterson and Guess, 1977). This woold
type 8 (1 - 1- 1-1 - 1) - are the only other types found make mtDNA type 1 ( 2 - 1 - 1 - 1 - 1 ) the most eligible
in more than one ethnic group. Strictly speaking, only for this position. Hovever, comparisons with HPa
type 1 occupies a central position in the phylogeny, morphs of other hominoids (Denaro et al. 1981) h~
but in Figure 6 we have enclosed types 6 and 8 as already indicated that Hpa I-1 was likely to be ancestr
well within a dashed line to emphasize the fact that to Hpa I-2. Sequencing of the sites has substantiateu
these are the only types shared by more than one this interpretation (Blanc et al. 1982). FurthermOre'
ethnic group. the nucleotide change differentiating Hpa I morphs 1
The simplest interpretation of this phylogeny is and 2 has been found to alter the amino acid sequer~Ce
that the three central types are among those most of URF 5, raising the possibility that these two morplas
likely to have been present prior to the formation may have been under some selective pressure. Of tlae
of the extant human races, and mutations giving rise mtDNA types having Hpa I - m o r p h 1, only tYPe 8d
to the other 32 types have most likely occurred since !1-1-1,1-1) is found among the three postulateU
that time. All of these 'newer' types are found in only original types and therefore would be the anceStr~
one major ethnic group, and with the exception of type of this phylogeny.
American Indians, these 32 mtDNA types are distri- In the last row of Table 3 we give a measure of
buted almost equally among the three groups. homogeneity of the population - F (the sum of the
 265

Squares of the relative frequencies of each 'allele'). nisms, not even necessarily sexual organisms, so that
F is highest (equal to 1) for complete homogeneity they apply to mtDNA. A problem may arise because
as in the Warao and lowest (F = 0.19) for Bushmen and our representatives of the human species were not
Bantus. We feel that the lack of variation among the collected as a random sample. In particular, two sub-
Warao may be due to the fact that they have only groups (Warao and Bushmen) are from very restricted
recently undergone a population expansion. There- areas. However, removing either or both groups does
fore, genetic drift may have increased the frequency not substantially improve the fit of the model of equi-
of a few mtDNA types so that only one type has given librium under neutrality to the data set. Explanations
rise to a predominant portion of the recent population for this deviation other than non-random samples could
(Wilbert and Layrisse 1980). Furthermore, this Indian be: 1) n o n - h o m o g e n e i t y within the species with respect
group is believed to have had no admixture with Cre- to mutation rates; 2) natural selection for some o f
oles thereby precluding introduction of new m t D N A the morphs; 3) or lack of equilibrium between muta-
types into the population by gene flow (Layrisse 1976; tion and drift, which may not have been reached because
Layrisse Z 1980). Since, by our methods the mtDNAs of of the major increase in population size that has taken
a single individual seem to be identical, the F value is place in the last 10.000 years (Coale 1974) with the
equivalent to homozygosity in the diploid condition. adoption of agriculture. As in the Warao, there could
The world F value is 0.277. A test suggested by hardly have been sufficient time for many new muta-
EWens (1970) and Watterson (1978) allows us to com- tions to occur and spread through the whole world.
Pare the observed F value with that expected on the Consequently the F value could not yet have decreased
hypothesis of equilibrium for neutral alleles, given our to the lower equilibrium value expected with the higher
35 alleles, and a sample size of 200. From a table in population size.
EWens (1970) one can see that under present conditions
the 2.5% and 97.5% significance levels of the F value are
0.04 and 0.10. Our observed F value is outside this
interval, and thus too large to support the hypothesis of Genetic Distances
equilibrium under neutrality.
It may be argued that Ewen's and Watterson's The results obtained with the various restriction en-
tests apply only to diploid organisms in randomly zymes were also used to compute genetic distances
mating populations. However, the models upon which between population pairs. In Table 4 we give the sums of
the tests are based really imply basically haploid orga- the indices computed separately for each enzyme

Table 4a. The mean numbers of restriction site differences between two populations,
Calculated as in (1), are represented below the diagonal. On the diagonal are the intra-
.P~ heterogeneities computed using the same formula with p = q, but the result
Is entered after multiplying it times n/(n-1) where n is the number of individuals in the
sanaPle. Above the diagonal are the distances between two populations computed by
SUbtracting, from Vxy, the arithmetic mean of Vxx and Vyy

Caucasian Oriental Amer. Indian Bantu Bushmen

~aucasian 1.5576 0.0589 0.0612 0.6250 2.2708

Oriental 1.2783 0.8812 0.0377 0.7445 2.5174
hrner. Indian 0.8400 0.4783 0.0000 0.7308 2.4848
ltantu
I~ttshnaen 2.2730 2.0543 1.6000 1. 7385 1.0421
3.8588 3.7673 3.2941 2.7206 1.6185

Table 4b. The mean number of restriction site differences are calculated as in (1), but
.uSing as the value of vij the number of differences between mtDNA types as diagrammed
1~
ti the phylogeny of Fig. 6. The distances axe shown below the diagonal. Intra-popula-
o~n heterogeneities and distances subtracting this value are shown on the diagonal and
~ e diagonal respectively, as in 4a

-,~ Caucasian Oriental Amer. Indian Bantu Bushmen

CaUcasian 1.6408 0.0854 0.0596 0.7429 2.5713

Oriental 1.3522 0.8928 0.0319 0.7000 2.5524
Arner. Indian 0.8800 0.4783 O.0000 0.6833 2.5205
uantu 2.4800 2.0630 1.6000 1.8333 1.2891
I~UShmen 4.2241 3.8312 3.3529 3.0382 1.6649
266

using a formula of Nei and Tajima (1981). Their index bury 1974) and 40,000-100,000 for mtDNA (Nei
defined as mean number of restriction site differences is: 1982). The basic archeological evidence indicates that
the major continents (Africa, Australia, Europe, and
Vxy = Ei ~ Piqjvij Asia) do not show the presence of Homo sapiens before
40-70,000 years ago. Evidence from Africa, though not
satisfactory indicates first dates of 90,000 to slightly
where v i. is the number of restriction site differences be- over 100,000 years ago (Trinkhaus). Thus, the order of
tween t~e ith and jth morphs, and pj is the relative magnitude of the estimated times is correct, but the
frequency of the ith morph in population x and qj variation in distances between the various groups irtdi"
is that of the jth morph in population y. For compu- cares some source of heterogeneity which will be dis"
tation of this index American Indians were presumed cussed further.
to have the most common Hae II morph - - 1 since
they were identical to the most common Caucasian/
Asian morphs with the other four enzymes. The in- Evolutionary Rates
dex was computed in two ways: in Table 4a by summing
over all the enzymes the V values obtained for The distances in Table 4a were then used to construct
each, and, in Table 4b the valueYv., |j was determined by (by average linkage) the tree seen in Fig. 7. AlthoUP
counting the number of muational steps separating it is particularly difficult to 'root' trees, by assurairig
the mtDNA types within the phylogeny diagrammed equal rates of evolution in the various branches the
in Fig. 6. Calculations using mutational distances root would be placed between Bushmen and all other
from the tree tend to accentuate the distance of the populations. This placement of the root does not,
African populations from the others and this is most however, take account of other information previoUS"
noticeable for the Bushmen. ly discussed wherein there are only three mtDNA
types common to more than one ethnic group.
Assuming that the molecular data have correctly
Time of Divergence identified the root as mtDNA types 1,6 or 8, by takifag
each of these types in turn to be the root, one can
It is possible to estimate the time of divergence of compute distances from it for each individual arid
human ethnic groups by taking the average genetic then average individual distances for the various pop"
distance (1.12 -+ 0.33) between those groups given ulations. Means and standard errors of 'distances fror~
in Table 4a (considered as representatives of popula- the root' are shown in Table 5. There are some dif"
tions from four continents out of five), and comparing it ferences depending on which root is used, but the
with the average distance between species for which time rank order of the population distances is always the
separation is known. One has to use, for reference, same. If all the central mtDNA types had been present
species that have a separation of less than 20 million prior to ethnic radiation - and if their proportions
years because mtDNA nucleotide differences seem to were known - the most appropriate measurement
reach saturation by that time (Brown et aI. t982). Using of evolutionary divergence from the common root
data that compare Man (M), Chimpanzee (C), and would be an average of those shown, weighted for
Gorilla (G) for 896 nucleotides of mtDNA. (Brown et al. frequency. In any case, the table evaluates the number
1982) one finds 79 differences between M-G; and 95 of mutational substitutions in every branch originating
between C - G ; making an average of 89 differences or from the central types. The root itself would be placed
9.9% of the nucleotides involved. In our sample the between Africans and other groups as shown in Fig. 7.
average difference between ethnic groups is 1.12 and this As the numbers of substitutions are assumed to have
refers to 517 nucleotides (see appendix for estimation), arisen in the same time period, they also indicate evo"
a proportion of changed nucleotides equal to 0.00217. lutionary rates. Table 5 shows that Bushmen have
To estimate time divergence we need the ratio of the had an evolutionary rate twice that of Bantus, and
logarithms of the proportion of unchanged nucleotides 3 - 5 times higher than Asians or Caucasians. A differ"
(loge ( 1 - 0.00217)/log e (1-0.099) = 0.0208) between ence in evolutionary rates could also be a reasonable
the human ethnic groups and between humans and explanation for results obtained by Cann et al. (1982)i~
the apes, as well as the estimate for separation between which Austalian aborigines were the most divergent
man and the two nearest primates which many authors of the populations which they studied. Both Bushr~e~
place at 5 - 7 million years (Sarich and Cronin 1976). and Australians are hunter/gatherers which have very
The divergence of human ethnic groups is thus esti- low population densities, and who live in extre6ae
mated at 0.020 x 5 x l0 t; = 104,000 years, to be com- environments. It is possible that these conditions
pared with estimates of 25-100,000 for protein and somehow contribute to the observed effect. Alth ~
immunological markers (Cavalli-Sforza 1969). 50,000- the frequencies of mtDNA types differs from the other
120,000 years for protein markers (Nei and Roychoud- populations, only a few types are exclusive to BushmeN'
 267

i w !

Oriental - ~
Fig. 7. The evolutionary tree was constructed by average
Am, Indian - . . Location of root based linkage and shows two possible roots: one based on the
CaUcasian pes Root of average assumption of equal rates of evolution; and the second
linkage tree based on the assumption of identifying the root as one
Bantu ~" of the three central rntDNA types. The units represent
ar~uming = rates
f3UShmen ~" of evolution genetic distances taken from Table 4a. Bushmen diverge
at 2.4 units, Bantus at 0.7, Caucasians at 0.06 and
, 9 I .... I |
1 2 3
Asians diverge from American Indians at 0.038 units of
genetic distance

Table 5. Mean distance and standard error are given for each different data sets shows that the distances between
P~ for the three 'central' mtDNA types. One or all are
~ ~ to be the root of an evolutionary tree seen in Figs.
Bushmen and the other populations compared with all
other distances not involving Bushmen, are about 4 - 5
and 7. Distances are estimated as number of mutational
s.~bstitutions times higher for mtDNA, but only slightly higher for
nDNA as can be seen from the following averages.
mtDNA Type 1 mtDNA Type 6 mtDNA Type 8 In general, the relatively greater number of muta-
(2-1-1-1-1) (2-1-2-1-1) (1-t-1-1-1) tional substitutions for mtDNA vs nDNA in Bushmen
can be explained by drift, selection (i.e. differences
Caucasian 0.84 • 0.18 1.56 +-0.20 1.84 • 0.18 in rate o f fixation) or mutation (i.e. production rates).
Oriental 0.48 • 0.13 1.24 +-0.07 1.17 • 0.08
Under conditions of neutrality (i.e. drift and mutation
A~er. Indian 0.00 +-0.00 1.00 -+0.00 1.00 -+0.00
Bantu alone), accumulation o f new mutations in a closed
l~shnaen 1.60 + 0.18 2.55 +-0.18 2.50 ~: 0.20
3.29 -+0.52 4.29 -+0.19 4.29 • 0.19 population is proportional only to mutation rate and
is independent o f population size (Kimura 1968).
Drift alone is unlikely to be a factor, moreover, Bush-
men have only very recently faced extreme reductions
Of these, only type 4 (which is distinguished by an in population size (Hiernaux 1975). Also, any effect
l~lsp I site) is found in high frequency. Therefore, caused by drift should be proportional between nDNA
it is not impossible that drift is responsible for the and mtDNA, while Table 7 indicates that tiffs is not the
differences between Bushmen and other populations. case. The difference in transmission entirely maternal
Alternatively, increased mutation rate or selection for mtDNA (Giles et al. 1980a) and bisexual for nDNA
on this Msp I morph may also be responsible. does not seem to explain the relatively greater evo-
It is relevant to compare mtDNA divergence of these lutionary rate o f Bushmen mtDNA. Both female and
~ with those observed for nuclear DNA male dispersals seem to be the same in Bustmaen and
DNA). Distance measurements were computed for the in Pygymies (Cavalli-Sforza, unpublished data),
Same ethnic groups based upon nuclear gene frequencies Differences in directional selection could be respon-
of 14 different loci, and these are shown in Table 6. In sible for the increased number of substitutions in Bush-
Order to minimize possible effects due to methods o f men mtDNA. If a non-negligible fraction o f new
COmputing genetic distances, the nDNA distance data mutations is even moderately advantageous, the evolu-
~Vere Computed with the same formula used for mtDNA, tionary rate measured by the number of mutational
except that vi,'s which are not known for the nuclear substitutions is higher than under strict neutrality,
gerte alleles en~ployed, were set equal to I whenever i c j; The simulations by Bodmer and Cavalli-Sforza ( 1 9 7 I )
nd to 0 when i = j. This internal comparison o f the show some numerical examples. Thus, while drift

~[Link]. 6. Genetic distances based on nuclear (nDNA) gone frequencies. The data are from
~".ya!ti-Sforzaet al. (1969); Mourant et ai. (1978); Nurse and Jenkins (1977) and from un-
u/~l,ished frequency data from this laboratory on Bushmen. Loci used were ABO, Rh, Hp,
e~l 3, Tf, Gm, Kidd, Gc, AcPh, PGM, 6GPD, AK, PepA and PepD. Distances were cal-
, ~ e d as described in the text

Caucasian Oriental Amer. Ind. Bantu/ Bushmen Australian

W..Afr.
CaUcasian
tientaj 5.925 0.912 1.160 1.415 1.588 2.177
6,500 5.251 0.624 1.760 1.930 1.157
a%er. Ind 6,385 5.513 4.526 1.624 1.981 1.314
~antu/w. Aft. 6.524 6.533 6.033 4.299 0.413 3.224
~shr~en 6.613 6.618 6.307 4.623 4.125 3.177
~ian 7,307 5.950 5.744 7.538 7,407 4.335
268

Table 7

mtDNA nDNA
vij r 1 vij = 1 vii = 1
(as in Table 4a)

Average distances between all other 2.08 -+0.35 1.56 +-0.32 1.48 +-0.37
populations with Bushmen
Average distances between all popu- 0.38 _+0.15 0.38 -+0.14 1.25 -+0.18
lations without Bushmen

Ratio of above averages 5.5 4.1 1.2

alone is unlikely to be the only factor responsible and to compute evolutionary rates in each o f the in"
for the discrepancy, to decide between mutations dependent branches. We observe that there are variations
and selection we need knowledge o f the selective in the evolutionary rates o f the different brancheS,
value o f new mutations as welI as population size. and in particular that there m a y have been a higher
One o f the mutations accumulated in Africans is known accumulation o f mutations in Bushmen than in the
to be silent (i.e. Hpa I - 2 to Hpa 1 - 3 ) (Denaro et al. other ethnic groups examined. Comparisons with nuclear
1981), b u t for others information is not yet available. DNA gene frequency data indicate that the differences
Brown et al. (1982) have suggested possible reasons involving genetic distances to the Bushmen are much
why the mutation rate o f m t D N A could be higher more pronounced in mtDNA. There are various possible
than that o f nDNA, including oxidative damage, an explanations for this discrepancy, perhaps the most
error prone system o f replication, a less efficient editing acceptable o f which, at this time, is a local variatiO~
or repair system, and a higher rate o f turnover. One in evolutionary rates, perhaps due to increased mutatiO0
or more or these reasons might also explain why m t D N A rates.
in Bushmen and some other populations appear to have
a higher m u t a t i o n rate.
Our previous estimate o f 104,000 years for diver- Acknowledgements. This work was supported by NIH grants
gence o f the human ethnic groups assumes that all GM20467, GM284228, NSF grant PCM 80-21871 and MarCh
groups separated from each other simultaneously. of Dimes Basic Research Grant 1--778. The authors would like
to recognize the technical assistance of B. Young, M. D,Arnore
An alternative possibility is to consider that the split
and C. Adams as well as J. Huang. We would like to tha~l~
leading to Bushmen occurred prior to others. Using E. Wijsman and R. Cassin for helpful discussion and to express
the order o f separation suggested b y the data (Table our gratitude to those who supplied blood samples: T. jenkins,
4a) and a time scale o f 5 x 106 years for the h u m a n - E. Wilmsen, K.H. Chert, Z. Layrisse and H. Cann.
ape separation, we arrive at the following divergence
times: Bushmen diverge at 220,000 y.a.; Bantus 65,000
y.a.; and Asians/American Indians diverging from Cau- Appendix
casians 5,500 y.a. Both the first and last o f these values
are incompatible with present archeological evidence and Computations of distances or other quantities like muation rates
more compatible with the idea that there can be substan- and divergence times, based on restriction enzyme data req~irr
an evaluation of the opportunities for mutation. These ate
tial variation in evolutionary rates in the various human determinded by the number of nucleotides per restriction
ethnic groups. site and the number of sites in the genome (actual and potea"
In conclusion, using five restriction enzymes and tial). In Nei and Tajima (1981) computations are done bY
a sample o f 200 individuals, we find that there is a theoretical formulas designed to evaluate the number of nuc ,
high correlation between m t D N A types and the ethnic tide differences per nucleotide site, or 'nucleotide diversl
(Nei and Li 1979), on the basis of the mean number of re"
origin o f an individual. This is particularly striking striction site differences. Their formulas however do not ~akr
in our two African populations, where a distinct lineage use of relevant information that has recently become available'
separates these populations from both Caucasian and For instance, since the human mitochondrial DNA is rlo~
Oriental m t D N A types. fully sequenced (Anderson et al. 1981) the number of pot~
tlal sites Is known empirically. Moreover, the distributlO, .i
9 ' - ' n v-

Constructing a phylogeny from our frequency data

potential sites (those which are one base away from actual [Link]~
allows us to identify central types which, due to their . not .m good. agreement
is . .
with .
a random distribution (A d a l J ~,
,~

relatively higher frequencies, their representation in and Rothman 1982). Finally in mitochondria, tranSl~_os
more than one ethnic group, and in one instance, the form almost the totality of base pair sequence dlffere -d
presence o f an identical m o r p h in another hominoid between closely related species: 92% (Brown et al. 1982) ao
96% (Aquadro and Greenberg 1982).
species, we believe are more likely to be the ancestral
We have used this information in the following naan,~e~i
types. This in turn allows us to give a root to the tree The probability of loss of a restriction site made up of m nuC~
 269

tides, is equal to 1 - ( 1 - U l - U 2 )m = m(u 1 + u2) approximately The probability of gain of a site will be estimated ignoring those
~rhen Ul = probability of transition, and u 2 = probability of potential sites which are more than one nucleotide away from
ransversion. If there are n sites per genome (averaging over being an actual site. Given that there is a difference in rates
the various morphs) the expected n u m b e r of site losses per of transitions and transversions, the n u m b e r of potential sites
generation will be: which can give rise to a site b y a transition, (nts), is estimated
separately from the sites which require a transversion, (ntv). The
runs (u t + u2) (I) probability of gain of a site is then

Table 1

Observed n u m b e r of sites Expected n u m b e r Factor for computation

(published seq uence/present study) of sites of ~t
EnZYme and Actual Potential Avg. no. (n s) Nucleotides in Nucleotides in
recognition sites sites sites in potential sites observed sites
site our study
n
----. s nts ntv
Flpal GTTAAC 3 34 42 3.5 5.56 33.4 + 21.0
I~am HI GGATCC 1 6 28 1.1 1.07 6.4 + 6.6
Flae 1I PGCGCQ 6 49 135 6.2 8.32 49.9 + 31.0
Msp I CCGG 23 135 298 22.7 33.92 135.7 + 90.8
Ava lI GGA/TCC 8 122 160 5 23.98 119.9 + 22.5
345.3 + 171.9 = 517.2

Table 2. This table details the presence or absence o f the individual restriction sites in each mtDNA type for the five enzymes used in
this Study

MtDNA Hpa 1 Bam HI Hae 1I Msp 1 Ava II

TYpes
-... a b c d e f g a b a b c d e f a/b c d e a b c d e f

1(2-1-1-1_1) + + + + - - + - - - - + + + - - - - - - + - - +

2(3-1-1_1._3) + + + + + - - + + + + +

3(3-1-1_2_2) + + + + + - - + + + - - - - + + - - +

4(3-1-1_3_2) + + + + + - - + + -- + - - + - - - - + + - - +

5(3-1-1-2_5) 4- + + + + - - 4 - 4- 4- - - - - 4 - +

6(2-1-2-1_1) + + 4- 4- 4- 4- 4- -- _ _ 4- -- 4-

7(3-1-1_1_1) + + + + 4- -- 4- + + + -- _ -- + -- 4-

8 (1-1-1-1_1) 4- + 4- -- + 4- 4- + -- _ -- + -- 4-

9 (1-1-2-1-1) + + 4- 4- 4- 4- + -- 4-

10(3-1-1_1_2) + + + + 4---4- 4- 4- 4- - - - - + 4 - - - +
11(2-2- 3_ _ + + + -- 4-
+ 4- + 4- - - - - + +

12(4-1-1_~_511 4- 4- 4- - 4- 4---4- + + 4- - - - - - - 4 - - - 4 -

13(2-1-5_1_1) + + + + + 4 - - - - - 4 - + + - - - - - - 4 - - - 4 -

14(3_1_1_2_4)) + - - + 4- + - - - - + + 4 - - -

15(2-1-1-1-0) + + 4- + 4- _ + + + + -- 4- -- 4- -- 4-

16(2_1_2_1_ 1 + + + + + + 4- 4- + -- +

17(2-1-1_1_9) + + + 4- -- 4- + 4- 4- +

1B(2-3-1_4_9) + + + 4- -- + -- + 4- + +

19(2-3-1~4_7) + 4- + + -- 4- -- + 4- +

21](2-3-7_4_9) + + + 4- -- 4- -- + + 4-

21(211 1_ + + + + -- 4- 4- 4- 4- + -- +
2 2 ( 2 - 1 1 1 25)) + + + 4- -- 4- 4- + 4- -- -- + +
23(2-2-1-1_3) + + + -- + + - - . + - - - - + 4- + 4-

24(2 1 1 4 _ + + + _ _ + _ 4 - - - + 4- + - - - - 4 - 4 - - - 4 -
2 5 ( 5 - 1 1 1 21)) + + + 4 - - - + - - - - 4 - 4- + 4 - - - +

26(2_1 6 1_ + + + 4 - - - 4 - + - - 4 - + 4- 4---4-
27(2 1 4 1 161 + + + 4 - - - § 4- + + + 4 -

28(2-1-1-4_1) + + + 4---4- 4- 4- 4 - - - +

29(6-1-2_1_6) + + + 4- + 4- 4- 4 - + 4 -

+ - - 4 - - - - - + 4- 4- 4 - + + 4 -

(3 1 1 1 5 + + + + 4 - - - + - - - - 4 - 4- + - - - - + +
32(3_1_1_~1 ~ +
4-
+
4-
+
4-
+
+ + _ + -- -- + -- -- 4- 4- 4- -- +
+ -- + -- -- + + +
54 3 ~-~" + + + + 4- + + 4- 4-
35 ( - 1 - 2 - 1 - 3 ) + + + +
-~3-1-1--1_7) + + + + + -- 4- -- -- + + 4-
270

u2 Arise JC, Giblin-Davidson C, Laerm J, Patton JC, Lansman RA

ulnts "~ ~ - ntv (2)
(1979b) Mitochondriat DNA clones and matriarchal phylog"
eny within and among geographic populations of the pocket
which is also equal to the expected number of new sites gained gopher, Geomys pinetis. Proc Natl Acad Sci USA 76:6694-
by mutation. At equilibrium this number will be equal to the 6698
number of losses. Actually the situation is perhaps not in equi- Blanc H, Chen K-H, D'Amore MA, Wallace DC (1983) Amino
librium at least for some enzymes, indicating that the situation acid change associated with the major polymorphic HinclI
may be more complicated than in the simple theory given site of Oriental and Caucasian mitochondrial DNAs. Am J
above. Table I shows the average number of sites for our 5 Human Genet 35:167-176
enzymes computed from the observed population distribution 8odmer WF, Cavalli-Sforza LL (1971) Variation in fitness and
of morphs, as well as the number of potential sites by transitions molecular evolution. Proceedings of the Sixth Berkeley
and by transversion in the human mtDNA sequence (Anderson Symposium on Mathematical Statistics and ProbabilitY,
ei al. 1981). For comparison, the expected number of sites is pp 255-275
also given in the table calculated with the following formula, ob- Brown GG, Castora FJ (1981) Mitoehondrial DNA polymOr"
tained by equating (1) and (2) and calling P = u l / ( u l + u2): phism: evolutionary studies on the genus Rattus. Ann NY
Aead Sci 135-153
(1 - P) Brown WM (1980) Polymorphism in mitochondrial DNA of
ntsP + ntv 2 humans as revealed by restriction endonuclease analySis'
n s -- Proc Nat1 Acad Sci USA 77:3605-3609
m Brown WM (1981) Mechanisms of evolution in animal ~aite"
chondrial DNA. Ann N Y Acad Sci 119-133
In Table 1 it was assumed that P = 0.95. At least for Ava II Brown WM, George M Jr, Wilson AC (1979) Rapid evolutiola
and possibly for others the observed number of actual sites of animal mitochondrial DNA. Proc Natl Acad Sci uSA
seems smaller than the expected number, indicating that some 76:1967-1971
other factor perhaps dependent on nearest neighbors may Brown WM, Goodman HM (1979) Quantitation of intrapopula"
also be at work. tion variation by restriction endonuclease analysis of human
To compute the number of nucleotides potentially involved mitochondrial DNA. ICN-UCLA Symposia on MolecUlar
in mutational studies with a given restriction enzymes, we and Cellular Biology, Volume XV, pp 485.-499
evaluated from the data shown in table 1, the quantity: Brown WM, Prager EM, Wang A, Wilson AC (1982) MitochOn"
drial DNA sequences of primates: tempo and mode of
evolution. J Mol Evol 18:225-239
(1 - P) Cann RL, Brown WM, Wilson AC (1982) Evolution of humat~
mns + ntsP + ntv 2 mitochondrial DNA: a preliminary report. Human GenetiCs,
Part A: The Unfolding Genome, Alan R. Liss, Inc, New york,
which describes the number of sites times the number of nucleo- pp 157-165
tides in a site, plus the number of nueleotides potentially sam- Case JT, Wallace DC (1982) Maternal inheritance of mito"
pied. The resulting 517.2 nucleotide value represents the mean chondrial DNA polymorphisms in cultured human fibro"
number of nucleotides sampled by the restriction enzymes blasts. Som Cell Genet 7:103-108
used in this study and can be used to calculate a mutation Castora FJ, Arnheim N, Simpson MV (1980) Mitochondtial
rate per nucleotide from the previously determined mutation DNA polymorphism: evidence that variants detected bY
frequency per individual. restriction enzymes differ in nucleotide sequence rattaef
A final table (Table 2) has been included to more clearly than in methylation. Proc Natl Acad Sci USA 77:6415~
show the combination of restriction enzyme sites that define 6419
each mtDNA type. Cavalli-Sforza LL (1969) Human diversity. Proceedings of the
12th International Congress of Genetics, ed. Oshima C
(Dai Nippon, Tokyo) Vol 3 , 4 0 5 - 4 1 6
Cavalli-Sforza LL, Bodmer WF (1971) The Genetics of Hurna~
Populations, WH Freeman and Co, San Francisco, PP 90
Cavalli-Sforza LL, Zonta LA, Nuzzo F, Bernini L, De Jong W~,
Khan PM, Ray AK, Went LN, Sinisealco M, NijenhuiS LI~,
References Van Loghem E, Modiano G (1969) Studies on African
Pygmies. I. A pilot investigation of Babinga pygmies in tBe
Central African Republic (with an analysis of genetic dis-
Adams J, Rothman ED (1982) Estimation of phylogenetic
eases). Am J Hum Genet 21:252-274
relationships from DNA restriction patterns and selection of
Coale AJ (1974) The history of the human population. Sciea"
endonuclease cleavage sites. Proc Natl Acad Sci USA 79:
tific American 231(3):40-51
3560-3564
Denaro M, Blanc H, Johnson MJ, Chert KH, Wilmsen E, Cavalli"
Anderson S, Bankier AT, Bah'ell BG, de Bruijn MHL, Coulson
Sforza LL, Wallace DC (1981) Ethnic variation in HP.a!
AR, Drouin J, Eperon IC, Nierlich DP, Roe BA, Sanger F, endonuclease cleavage patterns of human mitochortdrla~
Schreier PH, Smith AJH, Staden R, Young IG (1981) Se-
DNA. Proc Natl Acad Sci USA 78:5768-5772
quence and organization of the human mitochondrial ge- Farris JS (1972) Estimating phylogenetic trees from dista0~e
nome. Nature 290:457-474
matrices. Am Naturalist 106:645-668
Aquadro CF, Greenberg BD (1982) Human mitochondrial DNA
Ferris SD, Brown WM, Davidson WS, Wilson AC (1981a) Extea"
sequence variation and evolution. Genetics 103:287-312
sire polymorphism in the mitochondrial DNA of apes. proc
Avise JC, Lansman RA, Shade RO (1979a) The use of restriction
Natl Aead Sci USA 78:6319-6323
endonueleases to measure mitochondrial DNA sequence Ferris SD, Wilson AC, Brown WM (198 lb) Evolutionary tree f.~t
relatedness in natural populations. I. Population structure
apes and humans based on cleavage maps of mitochondr~'
and evolution in the genus peromyscus. Genetics 92:279-
DNA. Proc Natl Acad Sci USA 78:2432 -2436
295
 271

~iles RE, Blanc H, Cann HM, Wallace DC (1980a) Maternal Nei M, Tajima F (1981) DNA polymorphism detectable by
inheritance of human mitoehondrial DNA. Proc Natl Acad restriction endonucleases. Genetics 97:145-163
Sci USA 77:6715-6719 Nurse GT, Jenkins T (1977) Health and the Hunter-Gatherer.
~iles RE, Stroyonowski I, Wallace DC (1980b) Characterization S Karger, Basel, p 259
of mitoehondrial DNA in chloramphenicol-resistant inter- Rigby PWJ, Dieekmann M, Rhodes C, Berg P (1977) Labeling
specific hybrids and a cybrid. Sore Cell Genet 6:543-554 deoxyribonucleic acid to high specific activity in vitro by
Groot GSP, Kroon AM (1979) Mitochondrial DNA from various nick translation with DNA polymerase I. J Mol Biol 113:
organisms does not contain internally methylated cytosine 237-251
I-I in -CCGG- sequences. Biochim Biophys Acta 564:355-357 Sarich VM, Cronin J (1976) Molecular systematies of the
le~aux J (1975) The People o f Africa, Charles Seribner's primates. In Molecular Anthropology, Goodman M, Tashian
RE (eds), pp 141-170
l(an~~y y wDozyp 215AM,Trecartin R, Todd D (1970) Identification Southern EM (1975) Detection of specific sequences among
of a nondeletion defect in a-thalassemia. N Engl J Med DNA fragments separated by gel electrophoresis. J Mol Biol
297:1080-1084 98:503-517
k~inaura M (1968) Evolutionary rate at the molecular level. Wallace DC (1981) Assignment of the ehloramphenicol resis-
Nature 217:624-626 tance gene to mitochondrial deoxyribonucleic acid and
Layrisse Z (1980) The HLA system in the Warao, In Demo- analysis of its expression in cultured human cells. Mol
graphic and Biological Studies o f the Warao lndians. Wilbert Cell Biol 1:697-710
J, Layrisse M (eds) 'UCLA Latin American Center Publica- Wallace DC, Surti U, Adams CW, Szulman AE (1982) Complete
tions, Univ Calif, Los Angeles moles have paternal chromosomes but maternal mitochon-
Layrisse Z, Layrisse M, Heinen HD, Wilbert J (1976) The histo- drial DNA. Am J Hum Genet 61:145-147
s~ system in the Warao Indians of Venezuela. Watterson GA (1978) The homozygosity test of neutrality. Ge-
cienee 194"1135-1138 ob k . . . . netics 88:405-417
MOUrant AE, Kopec AC, Domaniewska-S cza K ~etts) (1978) Watterson GA, Guess HA (1977) Is the most frequent allele
Blood Groups and D&eases, Oxford Univ Press, Oxford the oldest? Theoretical Population Biology 11 : 141-160
Nei 1~I(1982) Evolution of human races at the gene level. Human Wilbert J, Layrisse M (eds) (1980) Demographic and biological
Genetics, Part A: The Unfolding Genome, Alan R. Liss, studies of the Warao Indians. UCLA Latin American Center
Ine, New York, pp 167~181 Publications, Univ of Calif, Los Angeles, p 252
Nei M, Li W-H (1979) Mathematical model for studying genetic
variation in terms of restriction endonucleases. Proc Natl
N .Aead Sci USA 76:5269-5273
el M, Roychoudhury AK (1974) Genetic variation within and
between the three major races of man, Caucasoids, Negroids,
and Mongoloids. Am 1 Hum Genet 26:421-443 Received August 10, 1982/Accepted February 2, 1983