Re: DNA sequencing (sort of long)

Chris Frazier (
Mon, 2 Oct 1995 22:03:47 -0600

>A phylogenetic reconstruction exclusively based on
>sequence comparison or molecular data is in my opinion pure waste of time &

I am a big proponent of morphological analyses, but to dismiss the
power of molecular phylogenetic techniques out of hand is like, well, like
dismissing lightbulbs since candles work so well.

>The phenotype *must* be considered in the first line, other data can
>be used for confirmation, never as the basis of a phylogeny.

Actually the phenotype, or how things look, may be what selection
is acting on, but that has precious little to do with the reconstructing
the historical fact of how lineages diverged, that is, the pattern of
evolution. Neutral markers make the best indicators of this pattern. In
other words, to reconstruct phylogeny, it is best to use characters that
change as a function of time without being influenced by selection.
Molecular data work particularly well in this regard.
There are of course several caveats to this. First, one can never
tell if a gene or sequence is free from the influences of selection.
Second, an investigator may not be strictly interested in the pattern of
divergence, i. e., how long in mutation time lineages have been genetically
isolated from each other. One may weight certain morphological characters
as being a priori important for distinguishing species even though those
character change rather readily (say, hypothetically, pitcher shape). One
might use this character to split up two closely related species that
differ in this character, while leaving two taxa that have actually
diverged genetically a great deal but which have the same pitcher shape. I
guess the point is that even if you know the pattern of how lineages are
related it still doesn't tell you what a species is. Figuring out what a
species is is an art, especially in groups where species are widely
interfertile like Nepenthes, orchids, oaks, etc.

>I am not as enthusiastic as you about "modern molecular methods" (in fact,
>I cannot support any "monotheistic" approach to systematics).

I don't know what you meant by "monotheistic approaches to
systematics" (what God has to do with systematics is a whole other line of
discussion), but if you mean roughly "don't put all your eggs in one
basket" I agree. Sequencing of one gene gives you the divergence pattern
of that gene which isn't necessarily the same as that for the organisms
involved. As with all science, accumulating independent lines of evidence
is the best approach for discerning "true" patterns of nature. I might add
that sequencing, even of different genes, is just one molecular
phylogenetic technique. There is protein electrophoresis and RFLPs just to
name two. Depending on what questions one is asking, molecular techniques
may be an extremely important, if not the only way, to get appropriate
data. For example, understanding the effect of hybridization on
co-occurring interfertile species (like Nepenthes) can only be studied so
far without using molecular techniques to get at the underlying structure
of the population.

>This is rather exactly the kind of short-sighted implications which are the
>greatest danger of "modern molecular methods".

The gene tree in the Albert et al. 1992 that Fernando was writing
about may have its limitations, but this dataset yields as valid an
estimate of phylogeny as any other character based analysis (and better
than many). I think the main point of the paper is still fascinating:
some carnivorous plants are closely related even though they use different
capture mechanism (e. g. Nepenthes, Drosera and Dionaea), whereas other
groups of carnivorous plants that use the same capture mechanism (e. g. the
pitcherplants Cephalotus, Nepenthes and Sarracenia) are, nevertheless,
distantly related. Perhaps you will say that was patently obvious before
this dataset was presented, but the sequence data is still a strong
confirmation of this surprising fact. Nothing "short-sighted" about it.

>This is not new but probably wrong in this oversimplified statement (well,
>they are of course both dicotyledons...). Furthermore, the most interesting
>families in this respect (viz. Dioncophyllaceae and Ancistrocladaceae) have
>not been considered.

Actually, Kenneth M. Cameron from University of North Carolina gave a
rather interesting talk at the American Institute of Biological Sciences
Meetings (co-authored by Mark Chase and Susan Swenson) this August entitled
"Molecular evidence for the relationships of Triphophyllum
(Dioncophyllaceae) and Ancistrocladus (Ancistrocladaceae)." It sounds like
they have this area pretty well worked out at least as well as the rbcL
sequence is concerned.

Last note: This has probably been posted before, but just in case the
following is is an important follow-up paper from this same investigative
group for anyone wh cares to see more of this kind of work: Williams,
Albert and Chase, 1994, Relationships of Droseraceae: A cladistic analysis
of rbcL sequence and morphological data (American Journal of Botany


Chris Frazier
Dept. of Biology, Castetter Hall, UNM
Albuquerque, NM, USA 87131

Wk: (505) 277-0683
Home: (505) 255-2176