out of africa

Alan R. Rogers rogers at ANTHRO.UTAH.EDU
Mon Nov 18 12:16:46 EST 1991

My head is reeling with the parsimony analysis of migrations that has just
appeared on this interest group.  I want to point out what seems to me to be
a flaw in these analyses.  All are concerned with changes between the
character states "in Africa" and "out of Africa".  It is worth remembering
why it is that, with ordinary molecular characters, it is reasonable to talk
about the substitution of character states within populations.  There is a
mechanism---genetic drift---that tends to eliminate variation within
populations.  Because of genetic drift, each new neutral mutation has
probability 1/2N of eventually replacing all other alleles.  When this
happens, a substitution has occurred.  If we are studying evolution on time
scales that are long relative to the time it takes for a substitution to
occur, then we are justified in ignoring the process of substitution itself,
and speaking as though substitutions occurred instantly.  Thus, it makes
sense to talk about the nucleotide "A" being substituted for "T" within a

But what sort of sense does it make to talk about "in Africa" replacing "out
of Africa"?  Certainly the same mechanism---genetic drift---cannot be
responsible.  If, like other animals, humans have always tended to move away
from intense competition, then dispersal should always have tended to
maintain geographic dispersion, not reduce it.  It would have been
remarkable indeed had an entire population ever picked up and moved from
Asia to Africa.  More likely, dispersal has always involved the movement of
individuals and small groups over relatively short distances.  Some Asian
individuals may have moved towards Africa, and some of their descendants may
have ended up in Africa, but it seems incredible to posit that they and all
their relatives made that journey.  Let us therefore exclude as incredible
*all* the migrations that were proposed in the earlier parsimony analyses
and assume instead that movements have always involved small groups, and
that some relatives were always left behind.

Given this assumption, the "out of Africa" hypothesis implies that each of
the lineages that began in Africa would include some individuals who stayed
there.  Thus, the african population should include representatives of all
modern lineages, except perhaps for a few that have become extinct within
Africa.  On the other hand, some african lineages may have included no
emigrants, and Europe may have been colonized by lineages that were not part
of the movement into Asia.  Thus, the diversity of the African population
should exceed that of the other populations.  This, of course, is exactly
what is found.

I do not mean to imply that I think the molecular evidence for an african
origin of modern humans is overwhelming.  Only recently has it become clear
that the difference in diversity between african and non-african populations
is statistically significant.  This is interesting, but still does not
constitute a sound statistical test of the hypothesis of an african origin.
We need better statistical methods than we now have.

More information about the Mol-evol mailing list