Re: molecular clocks: (was Re: Letter from Reasons to Believe)

On 5/25/06, Dawsonzhu@aol.com <Dawsonzhu@aol.com> wrote:
>
> David Campbell wrote:
>
>
> I have yet to encounter convincing evidence that molecular clocks can be
> reliable enough to give precision like that, despite the persistant use of
> such calculations by molecular biologists.
>
>
>

Here's an extreme case:

Based on a sequence of a protein from a snail, a chiton, a clam (which was
highly divergent), and an arthropod, a paper focusing mainly on the protein
biochemistry threw in a molecular clock calculation that claimed that the
clam diverged from the other mollusks at about 440 million and the chiton
from the snail at 220 million. This was based on some other molecular paper
that claimed the arthropod-mollusk split could be put at 545 million. All
of those dates are way too young (bivalve fossils are known from 540; snails
and chitons are a little trickier to identify by shell alone but appear by
500, and the lophotrochozoan [inc. mollusk]-ecdysozoan [inc.
arthropod] split is minimally into the Edicacaran), and the chiton diverged
from snails before clams did based on all available evidence.
Widespread problems of molecular clocks:
Inadequate calibration of dates and not considering uncertainties in the
dates as part of the error in calculations. In the example above, a date
was pulled out of another molecular paper; sometimes a rate is pulled out of
another paper without any time calibration internal to the new paper. Many
papers have used a single calibration point, which gives no statistical
confidence.
No more than superficial knowledge of the geology and paleontology.
Checking if a date is correct and invocation of possible factors at the time
of the posited divergnece event are affected by this. A prominent molecular
clock paper about ten years ago claimed that the divergence times for
mammals better fit times of continental breakup during the Mesozoic than a
radiation immediately after the K/T boundary. In addition to having only a
single calibration point for the molecular dates, the purported correlation
to continental breakup was entirely uncalibrated-there was a graph
purporting to show divergence times along with continental breakup times,
but the divergence times didn't line up so well with the breakups even on
their little figure. My current work on freshwater snails has led to
finding out an error in the widely used The Fossil Record II. In the late
1800's, Walcott reported in Science three groups of freshwater snails from
Carboniferus rocks in western North America, a big jump in age. These are
cited as the oldest record for the groups in various sources; however, they
stick out both from all other records and from related groups. Geological
reinvestigation many decades ago demonstrated that Walcott failed to notice
subtle evidence that shows he was into a bit of Cretaceous rock instead.
Relying solely on a database like that, without double-checking (especially
any record that stands out), will cause problems.

Clear evidence that variations in mutation rates exist. This includes known
changes in extrinsic factors affecting mutation (e.g., decrease in
radioactivity decay and increase in ozone produces appreciably less mutation
than the early Precambrian, but the increase in oxygen produces more
mutation) and variation between species (or higher groups, e.g. eukaryotes
generally having better error checks than prokaryotes and vertebrates having
additional error checks) as well as simple observations that different
organisms or different genes yield different results. For example, I
checked if I could get a molecular clock out of 18S sequence data for
bivalves. Because of a couple of groups with very high variation at
relatively low taxonomic level, I actually got a negative correlation
between time and mutation.

There exist some better papers, with much more complex models of genetic
change over time and obvious involvement of paleontologists. I remember one
talk where a paleontologist teamed up with a geneticist to look at possible
cnidarian affinities of Precambrian organisms. Some of the modern taxa
proposed as possible relatives of the Precambrian (>545 million) forms had
molecular divergence dates on the order of 500-600 million years, and so
they thought those might actually have been around in the Precambrian.
Others had divergence dates around 200 million or less, and they thought
they were fairly safe in assuming that those probably were not around in the
Precambrian.

An obvious test of the claim that some genes show more recent exchange
between humans and chimps than the initial split between the species would
be to compare the same genes to other apes. If these genes purportedly
transferred by hybridization also seem to have a younger divergence from
gorillas than other genes, this looks like the rate estimates are wrong.

Stereotypes are not universally true, but there is generally just enough
truth for them to stick. The general sterotype for molecular biologists is
that the only thing biological about them is the source of the molecules
they study. The team across the road from me with many honors for work on
C. elegans has no interest in nematodes as such; it's just a handy model
organism for investigation of genes with medical interest. They are doing
valuable work, but I don't think they would pretend to be able to comment on
nematode evolution.

-- 
Dr. David Campbell
425 Scientific Collections
University of Alabama
"I think of my happy condition, surrounded by acres of clams"