Re: Part 1 (of 3): What are major aspects of evolutionary theory?

anon1@xxxxxxx wrote:

>>>So-far we have discussed duplicated segments of DNA, whose gradual
>>>drift away from each other after duplication shows the arrow of time,
>>>but that works only for very recent duplications where we have the DNA
>>>sequences of ancestors and/or several very near modern
>>>individuals/species which can be used to reconstruct the genomes at the
>>>presumed branch nodes.
>>
>>This is word salad, I'm afraid. "Reconstruct the genomes"? "DNA
>>sequences of ancestors"? And what you are saying (if I can understand
>>it) is not at all what I was talking about.

[snip simple, irrelevant explanation of parsimony]

> So what I envision from whole-ecosystem shotgun sequencing (see another
> message I posted previously) is that we find segments of DNA bases
> which fit entirely within single shotgun reads, and try to find which
> are so very similar they must be homologous, and we build unrooted
> trees for such, and use majority rule and variations thereof to predict
> the genomes of the inner nodes of our trees.

Whole-ecosystem shotgun sequencing? Where did that come from? (Please
don't explain whole-ecosystem shotgun sequencing to me. I'm trying to
figure out how all this fits into any sort of coherent argument, not
what the terms mean.)

>>>(Which is a circular argument if you're trying to root the tree in the
>>>first place.)
>>
>>Which nobody does.
>
> If nobody ever roots the tree in the first place, then you can't use
> the rooting somebody else already did, you're at square zero and can't
> get past there.

This is another case in which your deletion of all context removes the
meaning from a sentence. What nobody does is to root the tree using
character polarity and then use the same tree to assign character polarity.

>>>Um, slight problem: With only three domains of life, there is only one
>>>possible unrooted tree, and all three un[sic]rooted trees are satisfied by
>>>that one unrooted tree, so even if all three different unrooted trees
>>>applied to various homologous-gene-duplication groups, there'd be no
>>>way to check anything based on modern DNA evidence. Or am I wrong?
>>
>>Yes, you are wrong, but I have no idea where your confusion lies. You
>>are correct that there is only one unrooted tree. So where do you get
>>your three unrooted trees from?
>
> (That was a typo.)
>
>>Did you mean three rooted trees?
>
> (yes)
>
>>If so, the duplication would root the tree and distinguish among
>>them.
>
> (So we're in agreement that a duplication would show that the root
> occurred somewhere thataway in the original unrooted tree, on the side
> where there's only one copy, away from the side where there are two
> copies.)

No. Please look at my example again. In the example, all the taxa have
two copies. There is no "side where there's only one copy".

>>You have to know that X is a clade already before you can
>>use y to root it.
>
> So how do you ever know that X is a clade?

I have discussed this already, but I'll try it again. Classical
systematists formed lots of groups based on distinctness. Some of them
are monophyletic and some are paraphyletic. By provisionally assuming
monophyly, you can root a tree. You can test monophyly by provisionally
assuming monophyly of a more inclusive group. The only way this entire
framework could be wrong and yet be consistent is if the true root lies
very close to what we think is a tip, such as in my example where humans
are the root of the tree of life. If you consider that a serious
problem, we could appeal to molecular clocks.

>>But you want to know how to get into this system initially.
>
> Yes, that's what I keep asking you, and you keep evading.

It's not something I worry about much, because the initial system is a
historical question.

>>If you go far enough out, even with the worst molecular clock, you
>>reach a point at which it's not credible to root a tree at that much
>>distance from the midpoint, and thus you have outgroups.
>
> I agree with the basic idea, but I have no idea how to decide what is a
> reasonble threshold for such guesses.

Doesn't matter. However far you need to go, you can go there.

>>Some characters do seem to root themselves too, or at least that's
>>not a bad initial hypothesis. We might suppose that, for example, any
>>protist would make an outgroup to Metazoa, and that Metazoa is
>>monophyletic.
>
> I don't agree with that argument. There could be a whole bunch of
> protists, maybe even a phylum or two, which are just degraded metazoa.
> But even a single protist descended from metazoa would make metazoa
> non-monophyletic (not a clade, only 99.999% of a clade).

Irrelevant regarding the root of the metazoan tree, if you pick a large
sample of protists and there are at least some protists outside metazoa.

>>Now in fact some protists might conceivably be descended from
>>Metazoa. But if most of them are not, and we pick a bunch of different
>>examples, we would find a contradiction that let us see this. Only if
>>all protists are descended from Metazoa will we have a potential
>>problem.
>
> Let P represent major groups of protists, and let M denote major groups
> of metazoa, in the following condensed view of a large unrooted tree:
> PPP MM
> PP--MMMMM--PP
> |
> PPP
> PPPPP
> PPP
> We have no way to know which of the three groups of protist is
> ancestral and which are degraded metazoa. Accordingly we have three
> possible roots for metazoa and have no clue which is correct.

Fortunately, this dilemma does not arise. There would be ways of
resolving it, but they are not necessary.

> On the other hand, if after sequencing corresponding parts of *every*
> species of eukaryotes whatsoever, the unrooted tree looks like this:
> PPPPPP MMMMMMMMM
> PPPPPPPPPPPP--MMMMMMM
> PPPPPPPPP MMMMMMMMM
> P PPP P M M
> I.e. just a single clump of each, then I would guess maybe the root is
> somewhere in that one protist half-tree, and so we know the root of the
> metazoa tree.

Bingo.

> Of course it's going to be amusing if it comes out like this:
> PPPP MM
> PPPPPPPPPPP--MMMMM M MMMMMMMMM
> PPP PPPP MMMM MMMM MMMMMMMMMM
> P P PP--MMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMM
> MMMMMMMMMMMMMMM MMM MMMMMMMMMMM
> MMMM MMMMMMMMMMMMMMMMMM
> M M M MMMMMMMMMMM
> MMMMMMMM MMMMM M
> whereby most of metazoa is one clade descended from one kind of protist
> but a small group of metazoa is a different clade descended from a
> different kind of protist. It would be especially amusing if it turns
> out that convergent evolution made the two clades look nearly the same
> in the way they develop toward adulthood despite using very different
> mechanisms to evoke such similar development.

That turns out not to be the case.

[snip]

> Do you know of any online Web page with an unrooted tree that covers a
> good fraction of all known protist and metazoa phyla, everything
> sequenced to date, as a preliminary view of this? (And if it also
> includes plants and fungi, so much the better.)

No. There probably is one. Oh, wait, I do. Just a minute:
http://www.zo.utexas.edu/faculty/antisense/tree.pdf

It's based on SSuRNA, which isn't that great, but in gross features it
matches other trees.

>>You remember that we had many human sequences, not just "Human" as a
>>terminal taxon. The question was whether "Human" was a clade, and
>>whether chimps were an outgroup.
>
> Are you talking about nuclear DNA, or mitochondrial DNA?

Doesn't matter. Same question, same outgroup.

> Mitochondrial
> DNA is only a very tiny portion of the whole human genome. But nuclear
> DNA is involved in meiosis so there's no such thing as a "clade" within
> a single species. Each species must be considered an "individual", a
> single node, when drawing either rooted or unrooted trees with nuclear
> DNA, so the question you ask is meaningless.

You are very naive. First off, if populations were panmictic you would
be quite correct, but most populations are structured in some way; there
are partial barriers to gene flow that do produce something quite like
trees that can be usefully analyzed. Second, even if there is
reticulation within a species it can usefully be called a clade if all
the genotypes are more closely related to each other than to any
genotype outside the species. Third, there is no magic point at which
you can always delimit one species from another genetically. Depending
on the species concept you use, individuals can be divided in several
different ways.

> I'm going to assume you're restricting this question to mitochondrial
> DNA, where it makes any sense. By the way, is the 3.1 billion figure
> for human genome the total of nuclear and mitochondrial, or is that the
> subtotal for nuclear only? How large is the mitochondrial portion?

Very tiny, only about 16,000 bases. So it doesn't affect the total at all.

> <http://www.actionbioscience.org/evolution/ingman.html>
> * Mitochondria have their own genome of about 16,500 bp that exists
> outside of the cell nucleus. Each contains 13 protein coding
> genes, 22 tRNAs and 2 rRNAs.
> Is that correct, only 16.5k bp total for human mitochondria?

Yes.

> I guess the 3.1 billion figure to that many significant digits would
> apply to either the sub-total or the grand total equally well, because
> the mitochondria part is essentially zero at that scale.

Yep.

> So how many humans, and how many chimps, have gotten their
> mitochondrial DNA totally sequenced, so that all of them could be put
> into a single unrooted tree?

Why would you need the whole mitochondrion for this? Why not just a
part? I don't know the answer exactly, but it would be "lots" for each.

> What does that tree look like?
> One clump of chimp connected via a single link to one clump of human?
> If, after drawing the tree, we compare different branches of it, do we
> discover any lateral gene flow, or is it consistent with none at all?

One clump, no lateral gene flow.

> Anyway, back to questions of clades: Suppose we get lots of
> mitochondrial genomes of all five species, and the unrooted tree looks
> like this: H H H H H
> BBBBBBBBBBB-----------+-------+--------+--HHHHHHHHHHHHHHHHHHHHH
> B B B | | | H H H
> BB BB B OOOO GGGGGG CCC-+-CCCCC HHHHHHHHHH
> B B BBB O G G C C C H H H
> CC HHH H
> Then we can say for sure at most one of those species is not a clade.
> (Treating all species of gibbon as if a single species here, but
> showing each species of chimp separately.)

Correct.

> I think in general, when we get an unrooted tree like that, the best we
> can say is that all but one of the clumps is a clade, and the remaining
> clump may or may not also be a clade, and we have no idea which clump
> is that one maybe-clade. Maybe it's best to draw only unrooted trees
> and let the reader guess which if any of the clumps is not a clade.

That would be true in a vacuum. However, even with no further
information it's better than that. Consider, in the case you have here,
that all the species form tight clusters with large distances between
the base of each species and any other node. Even if we assume that the
root might be anywhere with equal probability, the statistical
probability is very high that it's not within any species. If we make
any sort of stronger clock hypothesis, the root ends up between gibbons
and others. And of course we could always introduce outgroups till the
cows came home.

>>>>If humans still form a single group relative to the chimp when the
>>>>tree is rooted on a gorilla
>>>
>>>Huh? The tree is rooted on a branch between two clades, not on a
>>>single-species clade.
>>
>>Don't be pedantic. This is the commonly used terminology. Nobody is
>>thereby claiming that gorillas are ancestral to humans. You could say
>>"rooted on the branch leading to a gorilla", but that's needlessly long.
>
> Oh, thanks for explaining the slightly illogical jargon which confused me.

Welcome to human language. We hope to make your visit to this species as
comfortable as possible.

>>The question at hand is whether Homo sapiens is a clade, and whether
>>a chimp is a safe outgroup
>
> Which makes sense only if you're restricting discussion to mitochondria.

Not true.

>>As long as there is a branch separating the "Human" part
>>of the tree from all others, our conjecture that humans are a clade is
>>unfalsified.
>
> So long as you restrict discussion to asexual reproduction, such as
> mitochondria, that makes sense. With link between two parts of unrooted
> tree, at most one part can contain the true root, so at most one is not
> a clade.
>
> But if we start talking about nuclear DNA, then each species is a
> single node, it makes no sense to talk about clade or tree when dealing
> with meiotic-crossing DNA. The only interesting questions are whether
> traditional/natural groups of more than one species, such as "placental
> mammals" or "birds" or "bats" or "monotremes" or "tetrapods" or
> "chordates" or "metazoa", are clades (as originally defined, without
> any "adjustments" of moving some small groups into or out of the
> natural group to fix its cladeness).

This is just plain false.

>>Are you saying that species can't be paraphyletic?
>
> If it's a sexually reproducing species, in the sense that every member
> can potentially mate with every other member (either directly if of
> different sex, or with one intermediary if of same sex), and if in fact
> there is travel between different local populations to avoid clades in
> the local-population sense, then it makes no sense to even ask the
> question in regard to nuclear DNA.

If the population is panmictic, we can still ask the question. You
confuse structure within the species with monophyly of the species as a
whole.

> Each species is a single clade because it's a single individual for the
> sake of computing trees or cladograms with nuclear DNA.

I thought you claimed that species couldn't be clades. Now they are
clades by definition? No, you are entirely wrong about all this. There
are in fact known paraphyletic species. Here's a buzzphrase to chew on:
interbreeding is plesiomorphic.

.

Relevant Pages

  • Re: Part 1 (of 3): What are major aspects of evolutionary theory?
    ... >> but that works only for very recent duplications where we have the DNA ... Suppose we have an unrooted tree with three terminal nodes whose genome ... A before the inner node M, and both descendents K and L inherited ... So how do you ever know that X is a clade? ...
    (talk.origins)
  • Re: Part 1 (of 3): What are major aspects of evolutionary theory?
    ... > bipartition in the unrooted case, and either sub-tree or clade in the ... > are a clade evolutionarily). ... > unrooted tree (which is asumed to be the shadow of an unknown rooted ... real node somewhere on that stem, such that some real species falls into ...
    (talk.origins)
  • Re: Part 1 (of 3): What are major aspects of evolutionary theory?
    ... But if I have an unrooted tree, ... If somehow I can determine that the branching node adjacent to ... Harrietphyta-and-four-others clade), so the root must lie somewhere on ... extremely long-lived species which is ancestral to the branch point ...
    (talk.origins)
  • Re: Part 1 (of 3): What are major aspects of evolutionary theory?
    ... and either sub-tree or clade in the ... are a clade evolutionarily). ... unrooted tree (which is asumed to be the shadow of an unknown rooted ... Any living species could be the common ancestor ...
    (talk.origins)
  • Re: Archaeopteryx
    ... >>>into that clade as part of its definition. ... Neornithes includes more extinct species (most of them ... > events in the part of the tree I'm browsing. ... just now I tried "Neornithes" in the search form ...
    (talk.origins)