Rectifying Evolution and Geneaology

[Richard Pyle]

> The point of course is to allow for a coherent view of a single
> core aspect of evolution, the geneaology.

Unfortunately, when you approach the level of species, the task of mapping
the geneaology and the task of mapping the evolutionary history (in the
sense of representing evolution as a "tree" of bifurcating lineages) start
to get at odds with each other - at least for sexually-reproducing organisms
(or organisms that can share, swap or blend genes with individual organisms
other than direct progeny).  The reason the two tasks come at odds boils
down to the fact that in an evolutionary tree, you have a repeating pattern
of one entity being split to produce two (or more); whereas at
individual-level geneaology, you have a repeating pattern of two entities
being blended together to produce one (or more).  Somewhere around the point
of the "species threshold", the distinction becomes critical, and probably
represents the "Grand Unified Explanation" for why a number of seemingly
"unsolvable" debates in biology (e.g., the "species definition problem")
continue to persist.

A pattern of individual-to-individual geneaology within a given population
looks like a criss-crossing network, representing the pattern of gene flow
within that population over historical time. It's not a perfect "mesh" of
criss-crosses, but rather a complex pattern with some areas that are
"thinner" (indicating relatively resticted gene-flow between two
sub-populations), and some areas that are "thicker" (indicating relatively
higher prevalence of in-breeding). Keeping the image of this un-even
criss-crossing mesh of geneaology in mind, take a broader view (in time
scale), and you see that some of the "thinner" areas continue to get thinner
and thinner until eventually one side hardly ever exchanges genes with the
other side.  In some cases (e.g., as a consequence of certain vircariant
events), this can occur essentially instantaneously (within one generation).
In other cases (e.g., assortative mating, gradual increases of barriers to
dispersal over time, "pulse" dispersal), this can occur over over long
geological time periods (many generations).  Regardless of how suddenly or
gradually they begin, some of the "thinner" areas of the criss-crossing
individual genaeology mesh will eventually re-connect over time (leaving
what amounts to a "hole" in the mesh).

A good example of "holes" could probably be seen in the case of humans.
Until relatively recently, the major human races undoubtedly had
considerably less gene-flow between them, than within them (hence the
substantial morphological divergencies).  Civilization and technology
dramatically reduced barriers to dispersal, and without other barriers to
reproduction (biochemical, behavioral), gene flow among the different human
races has been re-established to a degree not likely seen since soon after
the time when the races first began to diverge.  I suspect that, absent some
catastrophic introduction of new barriers, sufficient gene-flow will
continue among the races of our species such that we will preserve (and
perhaps even strengthen) our conspecific status until such time as we cease
to exist (go extinct).  Thus, the descendant populations on either side of
the "thinner" areas in the human genaeology mesh that existed at the time
when the races were isolated and diverging from each other, ended up (or
will likely end up) re-establishing gene-flow, thereby leaving a hole in the
historical mesh, rather than a point of divergence.

It's always dangerous to use the example of humans for representing patterns
of evolution; but I think that it doesn't require much imagination to
acknowledge that similar patterns exist (dominate?) in most species of life
on earth.  All that's required is imperfect gene-flow within a population
(not hard to imagine), and fluctuations in patterns of gene flow over
geological time (also not hard to imagine).

So....suppose we draw a line through the global mesh of geneaology of all
life on Earth at one particular moment in time. That line will cross
enormous chasims in the mesh (e.g., reflecting the span of time that has
passed since the most recent common ancestor between myself and, say, the
plam tree in my front yard); chasims that are merely "large" (most recent
common ancestor between myself and the porcupinefish in the lagoon off my
back porch); somewhat smaller chasims (between me and the rats that engage
in all sorts of noisy activity on said porch late at night); very small
chasims (between me [european ancestry] and my neighbor [asian ancenstry]);
even smaller chasims (between me and my wife [european ancestry]); and no
chasims at all (between my daughter and her brother).  The vast majority of
chasims will be large and obvious enough that we'll be able to safely
predict that the gene-flow across the chasim will remain effectively zero
into perpetuity.  However, there will be many chasims ranging in scale
roughly from the one between me and the rats on my porch (high confidence
that neither us nor our recent ancestors will ever have a common
descendant), to the one between me and my neighbor (reasonably good
confidence that some of our recent ancestors will eventually share a common
descendant, or perhaps even he and I will eventually have a common
descendant).  This particular range in scales of chasim more or less
represents the "species threshold", and it's the genesis of many of the
aforementioned "unsolvable" debates in biology.

What's my point?  I guess it would be that using cladograms to represent
evolutionary "geneaologies" is conceptually "tidy" when the chasims within
the true genealogical mesh are unambiguous (i.e., higher taxa), but when you
zoom in closely on the actual individual-level genaologies in the vicinity
of cladogram nodes, things can start to get blurry.  For historical nodes we
can ignore the blurryness, because we have the benefit of hindsight (i.e.,
to see that what began as "thinner" portions of the mesh, never re-closed to
form "holes").  But when you get to the terminal nodes (present-day
species), the blurryness often *does* matter. When we see evidence of a gap
between two evolutionarily "sister" populations, how can we predict whether
that gap represents the genesis of a perpetual chasim, versus whether that
gap will eventually close itself up, becoming a "hole"?

For instance, the aforementioned porcupinefish that lives off my back porch,
which would be classified as Diodon hystrix
(http://images.google.com/images?q=Diodon+hystrix) shares the same lagoon
with other porcupinefish that we would classify as Diodon holacanthus
(http://images.google.com/images?q=Diodon+holacanthus). If you look at the
pictures, you'll see why these two species are regarded as close relatives
of each other.  Despite their similarities, we still confidently regard them
as different species, because we are confident that the gap in their
respective individual-level geneaologies will never close up to form a hole,
and thus represent the genesis of a chasim.  We get this confidence from the
fact that the two species are broadly sympatric, and no hybrid between them
has ever been reported.  Even if we did occassionaly find a hybrid, we could
safely regard these as anomalies (or in the context of the geneaological
mesh, the last few tendrils that span an otherwise growing chasim between
two populations).  But there are many, many cases where we don't have that
sort of confidence -- either because the populations in question are
allopatric, or because they are parapatric with some level of hybridization
taking place at the boundary.  In such cases, how can we predict whether the
"thinner" section of the geneaology represents the genesis of a true
perpetual chasim, or simply a hole that will be sealed up over time (e.g.,
when climate or other environmental parameters change to reduce historical
barriers to gene flow). In some cases (e.g., plants and corals), the
potential for rampant hybridization across large chasims of evolutionary
geneaology is so vast that we've even given it a name ("reticulate
evolution").  I won't even go into the cases where hybridization can itself
lead to new species (which is represented in the geneaological mesh as a
trifurcation with a hole at its base).

Goodness...I spent a LOT more time on that than I had originally intended,
and I sort of ended up losing my direction along the way.  I guess the
take-home message is that, although it's very convenient to represent
evolution as a clean bifurcating tree of descent, and it's logical to think
of evolutionary history as a "geneaology"; one shouldn't forget that the
"true" geneaology of individual organisms works in a reticulate pattern, not
a bifurcating one.  Actually, the "true" GENEaology isn't even about
individual organisms, it's about individual genes (as Dawkins summarized so
well in his classic book); and when you get to this level the reticulation
becomes even messier (especially when you get into bacteria containing genes
that evolved in humans, and vice-versa). On a brighter note, maybe this
gives credibility to the idea that cladograms represent the evolution of
characters, rather than populations of organisms -- given that "characters"
and how they're used to construct cladograms represent the genes, rather
than the individuals.  But I won't go there....

Aloha,
Rich

Richard L. Pyle
Ichthyology, Bishop Museum
1525 Bernice St., Honolulu, HI 96817
Ph: (808)848-4115, Fax: (808)847-8252
email: deepreef at bishopmuseum.org
http://www.bishopmuseum.org/bishop/HBS/pylerichard.html
"The opinions expressed are those of the sender, and not necessarily those
of Bishop Museum."