Re: Date for Last Common Ancestor?

John Hawks (
Mon, 12 Aug 1996 15:02:05 -0400

Susan and Stephen (and whoever else),

Hi, I'm John Hawks, and I study with Dr. Wolpoff at the University of Michigan. I want to let you know that
you're both right about "last common ancestors." However, the topic is one that readily invites confusion,
so I'm glad to see this thread appear as it promises to be a very useful discussion.

The concept of a "last common ancestor" is of course perfectly clear to all of us. We would readily admit
that a person and his or her first cousin share at least one grandparent as their "last common ancestor." We
also may claim that A. africanus is the "last common ancestor" of A. boisei and A. robustus. Some of us may
even go so far as to suggest that all living humans share a single female "last common ancestor" that we
might call "Eve."

The point that many people miss (including the authors of far too many papers) is that the situations that I
just listed relate to each other only by analogy, not because of any feature of their sets that would cause
them to share a function called "last common ancestor." Indeed, our concept of "last common ancestor" as we
usually discuss it is an intuitive one. Rarely does anyone define it rigorously.

What you have been talking about up to this point is not a confusion over the concept of "last common
ancestor," but instead a confusion about how that concept maps upon two very different types of sets: the
set of species and the set of individuals.

> > Susan S. Chin wrote:
> >
> > > It sounds to me like we have a disagreement over the unit of evolution
> > > here. I'm referring to the LCA as a species, or a population of the
> > > species which gave rise to modern man. Once this LCA appears, any later
> > > purported LCA's are really artificially recognized units of evolution.
> > > . . .
> > > The emphasis should be on the Common Ancestor part
> > > of LCA, not on Last, since that might create some confusion.

Susan is referring to a speciation event here. The founder population of the new species is in a way a
"common ancestor population" for that species. Depending on the subsequent population history of the
species, it might even be the "last common ancestor population." As Susan notes, all members of a species,
living and dead, must be descended from *some* of the members of this population. However, it might be more
useful to refer to this population simply as the "founder population," since (1) not all of the members of
the founder population are necessarily ancestors of later individuals, and (2) the status of this population
as a "common ancestor" or even "last common ancestor" of later individuals depends on the subsequent
population history of the species. (Note: when Cann et al. and Vigilant et al. refer to the population in
which their "Eve" lived, they specifically mean it to be a hypothetical founder population of H. sapiens

> Stephen Barnard wrote:
> > 2. The sense in which Mitochondrial Eve was our last female common ancestor
> > has nothing to do directly with the beginning of our species. Her
> > immediate ancestors were just as human as all of her descendants.

Stephen is, and has been, talking about individuals, not species. And he is perfectly right about the way in
which the "last common ancestor" can change as a consequence of demographic history (though the logic
solution in his earlier post is flawed). However, the sense in which "mitochondrial Eve (mtEve)" is our
"last female common ancestor" is not the way that he describes. In truth, we might even have a "last female
common ancestor" who is much more recent than mtEve. The reason for this is because of the types of sets
that we are talking about.

When we talk about "last common ancestors" in biology, we would like them to conform to Susan's concept of

> > Susan S. Chin wrote:
> > >
> > > The "last" part of LCA is crucial in identifying when the split between
> > > lineages occurred. At that point of splitting, branching, whatever you
> > > want to call it, there is an ancestral species known as the LCA. If there
> > > is further splitting or speciation which later occurs, that is totally
> > > irrelevant to the original LCA which has already been identified wrt Homo
> > > sapiens sapiens or any other species. Further splitting produces LCA's for
> > > OTHER species which result from the split, but has nothing to do with the
> > > original LCA. So if anything, LCA's are relative to what 2 organisms you
> > > are comparing. The ancestral relationship to their LCA though is absolute.

This is true of species. Two species of animals *must* share a single "last common ancestor," which is
itself a species. This is because the set of animal species (and the set of all species of organisms if we
assume a single origin of life) exhibits a property called coalescence. Coalescence is a very confusing
concept, but simply put it is this: If we start at the present and look backward in time, then the diversity
of elements of our set will coalesce to a single element. Coalescence describes sets in which the only way
that new elements are created through time is by divergence. New species are created only by the splitting
of an old species. Each species has one and only one parent species. New species are not the product of a
mixture of two old species.

The set of animal species is a coalescent set. The set of sexual animal individuals is not. If you look back
at the history of individuals, they do not coalesce to a single individual at some time in the past.
Instead, they branch out into a family tree, increasing exponentially into the past. This is because the
process of sexual reproduction is not one of divergence. Instead, the genes of two individuals are mixed
with the creation of each new individual. Even if some of the branches of a person's family tree lead to the
same individual, even if most of the branches should happen to lead to some small number of individuals, the
"last common ancestor" of individuals in the geneological sense says nothing about the pattern of splitting
or branching of the individuals of a species. Indeed individuals cannot be said to split or branch in the
same way as species.

In a similar sense, the "last common ancestor" of populations cannot say anything about the branching
history of the set of populations of a species except in certain special cases (more discussion on this
point may continue at a later time), and so my preference of "founder population" for the idea Susan is
writing about.

Individuals make lousy subjects for study because they are the product of a mixture of their parents.
Specifically, they have alleles from each of their parents, and from each of their grandparents, and so on.

Alleles, on the other hand, are great for studying splitting and branching because they (outside of the
action of recombination mutations) *do* feature coalescence. The mitochondrial DNA is, as you all know, just
one big allele, not subject to recombination and *mostly* maternally inherited. When Cann et al. and
Vigilant et al. refer to mitochondrial Eve, what they are talking about is a female who contained the
coalescent "last common ancestor" of human mitochondrial DNA. Note that this is not necessarily the "last
female common ancestor" of living humans, and it is almost certainly not the "last female common ancestor"
of all H. sapiens sapiens--even if mtEve lived in the *hypothetical* founder population of H. sapiens
sapiens. Note also that this does not even require that mtEve herself be a member of H. sapiens sapiens. All
living humans *must* have a common mtDNA ancestor because the set of human mtDNA alleles is a coalescent
set. The date and species associated with this ancestor are thus matters for scientific investigation. (I
would add that anyone who wants to start a thread on mtEve would find an eager participant here.)

I will summarize:

"Last common ancestors" do not have the meaning we would like when talking about sets of biological entities
that are not coalescent sets. These include sexually reproducing individuals and populations (though certain
subsets of populations may be coalescent). In these sets, the "last common ancestor" of two elements takes
on a strictly genealogical meaning, one that is obtained only by analogy.

Coalecent sets include sets of species and sets of alleles (though in certain circumstances, new alleles may
be produced through recombination). In these sets, the "last common ancestor" of two elements refers
specifially to the shared element immediately preceding in time the split between the lineages leading to
the two elements.

Looking forward to further discussion, and
Apologetic for the length of the post,

John Hawks