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

anon1@xxxxxxx wrote:

>>you are again assuming that stratigraphic data root trees, and that
>>in the absence of stratigraphic data a tree is unrooted. As I have
>>explained, this is not true.
>
>
> OK, here's an unrooted cladogram, where branch lengths are
> proportional to count of DNA base differences:
> Portlandphyta Renophyta
> | |
> Salemphyta---+-------+----------------------+---+-Vegasphyta
> | |
> Chicagophyta---+----------Bostonphyta Hendersonphyta

> Now without any timestamped nodes, and suspecting that different
> branches evolve at different rates so you can't use branch length to
> assign time durations, where do you place the root?

No idea. But we don't work in a vacuum like this. There are many ways of
rooting trees, and I will mention some of them: 1) by outgroup; 2) by
midpoint (bad idea if the evolutionary rates are as different as you
suggest here); 3) by presumed character polarity, including ontogenetic
criteria; 4) by assuming monophyly of two divisions of the tree -- that
one needs explanation, so here's an example: if you are looking at a
sample of galliforms and anseriforms, you can assume that each of these
is a clade and that the root lies between them.

> My claim is that the only way you can definitily place the root is if
> you have a prior information which of the above is an outgroup compared
> to the rest, in which case you can place the root along the branch
> leading directly to the outgroup.

That's not the only way, though it's by far the most common.

> But if you just have the cladogram,
> no other information whatsoever, you have no way to guess which might
> be an outgroup, or even whether there is one. Instead of all but one
> taxa being monophyletic, you might be dealing with some other
> partition, i.e. two or more distinct non-trivial (>1 taxon in each)
> clades with possible outgroup(s) in addition.

True. There is no way to root a naked tree with no additional
information, except midpoint rooting, which assumes a clock. So? When do
we ever lack all additional information?

> Now in the case above, you might be able to guess from the names of the
> taxa where the root is. But that's cheating.

And it's silly too.

>>There are indeed programs and/or methods that take stratigraphic data
>>into account. They are not much used, for reasons we don't have to get
>>into here.
>
> If stratigraphic data are available, is the usual method to ignore it
> when constructing the unrooted tree, then fold in the stratigraphic
> data and use it to root the tree and also to test stress?

No. The usual method is to ignore it when constructing and rooting the
tree, which is commonly rooted by outgroup. Stratigraphy may be compared
to the tree for various reasons, one of which is to test the agreement
between them.

>>>No, when I say "ID" I really mean "Intelligent Design", not intelligent
>>>tinkering.
>>
>>Then use a different term to avoid confusion. How about "separate
>>creation" which means what you seem to want here?
>
> What other term would you propose me to use when I really mean
> Intelligent Design (not tinkering)?

Didn't I just suggest "separate creation", in the sentence immediately
preceding your question?

> I'm not aware that the IDiots have
> ever stated their definitions clearly.

Yes. They are vague on purpose to promote their "big tent" strategy.

> I'm not aware they've even
> separated the design process from the manufacturing per an
> already-conceived design. Does the alleged design process include
> feedback from earlier designs, or are all designs conceived in vacuuo
> in the mind of the Designer with no awareness of the results except
> mental predictions? Does the manufacturing process create each
> different instance of life from inorganic materials, or by recycling
> low-level biochemicals such as sugars, or by tinkering with a genome to
> breed a variation from the parent(s)? If you know of a Web site that
> provides standard IDiot jargon for each variation possible, please cite
> the URL. In particular, is there a cladogram for the design process and
> another cladogram for the manufacturing process?

There is no such thing. Two reasons: first, as I've already mentioned,
detail is the enemy of the big tent. Second, you should be aware by now
that IDiots aren't interested in science, or in hypothesis testing, or
any of that. ID is a political movement whose basic tenets have nothing
to do with science, so they have no interest in making their models
detailed.

>>Some of the IDers have said this clearly. The problem is that some of
>>them have come down for separate creation and some have come down for
>>theistic evolution (e.g. Michael Behe).
>
> So which of them is using the jargon in the correct way, the way you
> approve of, the way you want me to use it, and which are mis-using the
> jargon as you've accused me of doing?

All of them use it in the correct way; after all, it's their term, so
they can define it. They use it to mean that God -- oops, the unknown
designer -- had something unspecified to do with the origin of life's
diversity. Whether that unspecified thing was tweaking, or poof
creation, or whatever, is, well, unspecified.

>>>The Great Tinkerer produces results indistinguishable from Darwin's
>>>Descent with Modification. But the Great Designer could produce
>>>anything whatsoever, which is refuted by evidence in the sense that the
>>>Great Designer theory doesn't explain *why* there are no cross-time
>>>cross-space chimeras, or even why the nested hierarchy of geneology
>>>follows a strict time sequence.
>>
>>Not sure what "follows a strict time sequence" means here, or what data
>>you think you have to show it.
>
> A Great Tinkerer, i.e. a supernatural or alien being or society which
> modified already-existing species to create new slightly-different
> species, building up variation toward a longterm goal, would show this
> evidence: Each new species in the fossil record is either a slight
> modification of some previous species or a chimera of different
> previous species. Depending on the physical abilities of the
> tinkerer(s), the multiple species going into a chimera, and the
> locations of the species that resulted from single or chimeric
> heritage, may or may not be restricted as to geography.
>
> A Great Designer, who thought everything in His head and threw them out
> into the world any time he felt like doing it, could violate all
> temporal and spatial limits.
>
> Strong temporal correlation but spatial randomness even for very local
> species would argue against natural evolution and in favor of Great
> Tinkerer that was equipped with a worldwide transport system.
>
> Strong temporal and spatial correlation (what we actually observe)
> argues in favor of either natural evolution or Great Tinkerer with
> limited means to move things around the world.

I don't think you have thought this out very well. Nor do I think you
are familiar enough with the paleontolical data to have a feel for what
it can or can't distinguish.

You are assuming that the time ranges between tinkerings, or whatever,
are sufficient to be distinguished in the fossil record. And you are
assuming that our geographical and temporal sampling are good enough to
distinguish among hypotheses. And you are assuming that there is some
objective criterion by which you can recognize what features of
organisms are taken from what other organisms.

> Both temporal and spatial randomness would argue against natural
> evolution or either kind of Tinkerer, in favor of supernatural special
> creation at the whim of a maniac such as described in the Bible. The
> supernatural Creator could either move back and forth through time (not
> allowed by natural law) to deposit species in random times, or could
> create everything during a six-day period (as claimed by Genesis).
>
> Great Omphalos Deceiver of course could mimic any of the various
> models, so remains a philosophical possibility regardless of evidence.
> But a supernatural being or space aliens not deliberately trying to
> fake the evidence would leave evidence that by statistics gave a clue
> to the underlying mechanism (such as supernatural being or space aliens
> who can or cannot travel through time, who can or cannot preserve
> ancient life for hundreds of millions of years before release, etc.).
>
>
>>Because the general opinion is that stratigraphic data are not relevant
>>to phylogenetic estimation, given the poor quality of the record. While
>>there is a tendency for the early members of a group to be primitive,
>>it's not reliable enough to base an analysis on. It's reliable enough to
>>test after the fact, though. And these methods may be useful in cases
>>where the fossil record is unusually good.
>
> That seems to be a well written summary of state-of-art in that area
> where I made a reasonble guess based on years of reading _Science_ but
> somehow botched that aspect of it. Thanks.
>
> In the example I concocted an hour or so ago (the one with
> ElvisImpersonator re-interpreted as LazarusStillAlive), did I do a good
> job of making the cladogram *only* on the basis of untimed data, then
> aligning it with known times (both completed before the first diagram I
> posted), and finally fixing one branch where it really did look like
> one set of fossils descended from the other rather than being on sister
> branches (plus ghost taxa to yield the second posted diagram)?

No idea, since I can't see any data. I think you just made it up. It
does look rather like a tree that incorporated stratigraphic ranges
would look, with one exception that I explained. Trees like this are
fairly common in the literature, though I didn't find any in a quick
look through the web.

I ignored the part about one species being descended from another, but
you just can't be confident of any such judgment. I think you decided
that a reconstructed branch length of zero between species A and its
common ancestor with species B means that A is the ancestor of B. And
you can do that if you like -- certainly no data falsify that position.
But if you make that assignment, what have you gained, and how confident
can you be? My claim would be nothing, and not very.

> By the way, I notice at least three different formats of cladograms:
> (1) Free-form two-dimensional unrooted tree (like my Portlandphyta
> diagram above), where it's virtually impossible to show names attached
> to any of the branch nodes;
> (2) Symmetric-rooted-tree (like my Elvis tree in my previous posting),
> where names on branch nodes are slightly messy but still usable, but
> usually all branch nodes are left nameless in published diagrams;
> (3) Parent-first rooted-tree (like most of the cladograms I copied from
> the net, such as from
> <http://palaeos.com/Vertebrates/Units/340Theropoda/340.200.html>>,
> which look best if every ghost node has a name, making it blatantly
> obvious whenever such a node is unnamed.

I'm afraid that the term "ghost node" is inappropriate, and I'm sorry if
I misled you earlier. I'm not clear on what the differences are among
your three styles, because your non-standard terminology doesn't summon
up any image to me. It may be that one of the differences is whether or
not internal nodes are given names, which is purely a matter of taste.
Note that giving them names (as in the last example) does not imply that
any real organisms are located at those nodes. There are many more than
three styles, anyway.

I think that by "branch node" you mean "internal node" and exclude
"terminal node". Internal nodes (on a fully resolved tree) are the
meetings of three branches; terminal nodes connect to just one branch,
i.e. are the tips of the branches. But I'm not sure how you mean "branch
node" to differ from "ghost node". I also have no idea what
"symmetric-rooted" means.

> I've seen mostly formats 1,2 in published articles, but format 3 in
> online cladograms. Other than the density of unnamed branch nodes, and
> rootedness, and capabilities of software used by the author, are there
> any other factors in choosing one or another format?

Sure. Space and esthetic value. What are you trying to show? What
information do you want to stand out? Do you want branch lengths to mean
anything, or not?

>>Again you make the bizarre assumption that fossils are the only data
>>that you can use to construct a phylogenetic tree.
>
> No you're misunderstanding me. I merely claim that dated fossils are
> the only way to root a tree with high confidence you have the root at
> the correct place.

Why would you think that?

> For example, if you use traditional taxonomy to root
> the tree, you have birds and reptiles and dinosaurs and mammals as
> separate clades (or perhaps dinosaurs and mammals and birds inside
> reptiles, but surely not birds inside dinosaurs).

That's not rooting. That's constructing a tree by traditional taxonomy.
Rooting is a separate thing entirely. Once you have the tree topology
defined, you pick a place to pull down into a root.

> But with fossils you
> can show the evolution from reptiles to dinosaurs occurred before the
> evolution from whatever to birds, which then allows locating birds
> within dinosaurs, breaking away from the traditional direct
> reptile-to-bird hypothesis. (Actually with Biblical taxonomy, Whales
> are fishes not mammals. But dated fossils show whales clearly evolved
> *after* mammals had been going along for hundreds of millions of years,
> allowing rooting whales within mammals.)

Again, you confuse rooting with tree construction. No fossils are needed
to root the tree. If, for example, you had a tree of all the amniotes,
all you would need to root it is some non-amniote. And this requires
onlyh the assumption that amniotes are monophyletic. Having done that,
all without stratigraphy, birds come out as a subset of theropods,
themselves a subset of dinosaurs, archosaurs, diapsids, sauropsids, etc.

> As I said earlier, you need some a priori assumption in order to guess
> where the root might be, and that assumption must be based either on
> religious myth or dated fossils, and you presumably know which I prefer.

Strangely, systematists manage to root trees without any of that. I
can't figure out how you imagine they do it.

>>Fossils can't connect all life, or even all metazoans, into a single
>>tree with any great confidence.
>
> I agree. But that's a moot point at present because no method
> whatsoever has yet connected all life into a single *rooted* tree,
> which is what we're debating here.

Actually, you are wrong. The root is debated, and there is some argument
that there is no single root at all (just a set of basal reticulations)
but there are methods that have been used to root the tree of all living
species. I will mention two: 1) midpoint, assuming a molecular clock; 2)
rooting on a gene duplication that happened before the basal split in
extant life; one copy roots the other copy.

> (For unrooted trees, dated fossils
> are totally unnecessary, fossils with dates disgarded are just fine to
> mix in with other evidence.) The best we are even close to is rooting
> each of three domains separately and then connecting their three roots
> in an unrooted master tree, with perhaps some guess where the root
> exist based on comparative degrees of diversity,

Incorrect.

> just as we root human
> origins in Africa due to greater diversity there.

Actually, an outgroup root is commonly used here too.

> If you believe we have a *rooted* tree of all life already, please cite.

We have several rooted trees, but they don't agree. At any rate, what is
the point of talking only about the tree of all life? I will agree that
it's harder to root than less inclusive trees are, but that has zip to
do with the availability of fossils.

> The Creationists complain that science can't answer all questions, so
> it must be worthless. The flaw is that their religion can't answer all
> questions either. You are arguing fossils can't answer all questions,
> i.e. create a single rooted tree of life, but I'm pointing that the
> other evidence can't (yet) create a single rooted tree of life either.
> So just as the Creationist argument against science is invalid, your
> argument against my desire to use timed fossil data is invalid.

No, you misunderstand my argument entirely. I'm saying that fossils
don't root a tree, and that you have no clear idea how they would. I'm
saying that we have several methods of rooting trees, of which outgroup
rooting is by far the most common, and that this makes no use of
stratigraphy. Nobody uses stratigraphy to root trees.

>>Morphology of extant organisms is somewhat better, because there in
>>fact *are* characters scorable across all metazoans;
>
> You've switched topics mid-sentence there.
> It is not valid to conclude, since metazoans work well, that all extant
> organisms work well, as you are mis-concluding there.

I was unaware that we were talking exclusively about the tree of all life.

>>Molecular data, though, are better.
>
> Both Morphology (including development of embryo or whatever,
> life-cycle, etc.) and molecular data are great for producing unrooted
> trees. I've agreed with that all along. If you believe they are
> sufficient for rooting a tree, see my Vegasphyta unrooted tree earlier
> and tell me where the root is supposed to be.

I will agree that placing a root on an arbitrary and fictional tree,
lacking any information about the taxa, is pointless. Why you think that
means anything is beyond me.

Do I have to resort to all caps here? Nobody uses stratigraphic position
to root trees. Trees constructed with or without fossils use (mostly)
outgroups to root them. Most trees you will find in the literature are
rooted, not unrooted, and they are rooted by outgroup. Whatever you
think to the contrary is your personal fantasy based on I'm not sure what.

Now, we can ask whether the methods used by systematists to root trees
are in fact valid, but I would think that nobody uses your claimed "only
valid method" suggests that you should at least think about the
possibility that I'm right here. I root trees all the time. I root them
using outgroups. What's the problem?

> If all evolution were climbing the tree of complexity, never loss of
> any feature formerly evolved, then you could score each taxon per
> complexity and thereby deduce a direction to each branch of the tree,
> and thereby root the tree. But you know that ain't true!!

Sure do. I wonder why you even brought it up.

>>>For convincing newcomers that "evolution happens", it would suffice to
>>>show that even in one small clade there's ample evidence of evolution
>>>happening there. That's like the toe into the tent, the first domino to
>>>fall, the first crack in the ice, the first invasion on the shore of
>>>Normandy, the first *** in their armor, just getting the
>>>anti-evolutionists to admit that evolution *ever* happens even in one
>>>small clade.
>>
>>I would recommend using primates for any examples, since the
>>non-relationships of humans and is the essential position of any YEC,
>>and also because relationships within primates are such a slam dunk
>>using just about any sort of data.
>
> Continuing the Normandy metaphor: Normandy was a good place to invade
> because Hitler believed we were planning to invade somewhere else, so
> he didn't have his best defenses at Normandy. Likewise, Creationists
> have their best defenses in the primates and apes, so that's not a good
> place for us to try to establish our first *** in their
> anti-evolution armor. Best to make the *** somewhere less defended,
> and *then* move our forces to confront them at primates/apes after
> theyir defenses are already breached. Horses seem one likely place to
> start if we can figure out a small clade that is easy to present in a
> single attention-span of a television couch-potato sound-bite zombie.

The problem with horses is that all but one genus is long extinct. I
know you seem bent on ignoring all but fossil data, but is that really a
good idea?

>>Do you remember from post to post what you were saying in the last one?
>
> No. I just respond to what others say, except occasionally when I
> present a brand-new topic such as demoting Fungi from kingdom status.

It would help to have some kind of coherent position, or some coherent
goal, and to keep your story straight from one moment to the next.

>>There are not three kinds of data. That's my point. There are different
>>collections of data, and different ways to collect them. No more.
>
> I disagree. I say that reading of DNA sequences, and studies of fossils
> (bones and shells and hardly nothing else), and studies of living forms
> (whole life cycle in many cases, including development of embryo,
> mating studies such as ring species and artificial insemination), are
> three different kinds of data.

OK, go ahead and defend that position.

> There's some overlap between fossils and
> currently living species, but not much. 99% of all species are extinct
> now, so the intersection covers only 1% of all species. At least 90% of
> kinds of information we can get from living species are unobtainable
> from fossils, so the intersection covers at most 10% of all characters
> of any individual. I think we'll have to agree to disagree here.

All irrelevant, unless you think that, for example, characters of the
head are a different kind from characters of the pelvis, since there is
no overlap.

>>>- Do you accept that we can read out the genomes of various species?
>>>- Do you accept that we have already read out the genomes of ten species?
>>
>>This is silly. The bulk of molecular evidence doesn't come from complete
>>genomes, but from homologous fragments of many species.
>
> There you go attacking a strawman again. I never asked:
> - Do you agree that the bulk of DNA sequences we currently know and
> compare are from the twenty or so complete genomes we've sequenced?

Then why go into the "genomes of ten species"?

>>>- Do you accept that the stages from blastula to gut-forming fall into
>>>several major groups, such as proterostome and deuterostome?
>>
>>That's "protostome".
>
> Sorry, I was in a hurry and didn't spend the extra 20 minutes to look it up.
>
>>This sounds suspiciously like a plea for Haeckelian terminal
>>addition, too.
>
> Absolutely not. You've made up another strawman, although at least in
> this case you're not original, you're copying Haeckel's big lie as the
> strawman. I said quite clearly that Haeckel was wrong about that:
> <http://groups.google.com/group/talk.origins/msg/cb88d16e52e76899>
> = Message-ID: <d4a5c$43a633c1$c690c02a$29551@xxxxxxxxx>
> Search for "Haeckel was wrong, flat out wrong"
> See right after that where I say where Haeckel was correct about the
> non-strawman point I used here where you tried to turn it to straw.
> (Hey, that gives me an idea for another metaphor: Anti-Rumplestiltskin,
> turning gold into straw. I'll nickname you Aunti Rump for short.)

That would require the further assumption that you are consistent from
moment to moment, and that hasn't been my experience here. Of course
when you snipped out all the context, that made it difficult to
reconstruct what I was talking about.

>>>- Do you accept that each single phylum has very similar development
>>>even beyond the gut-forming level, well into the organ-development stage?
>>>- Do you accept that similarities up to some point in development but
>>>then major differences past that point yield a hierarchial organization
>>>of the various "kinds" of animals?
>>
>>I certainly don't. This is "terminal addition" at its worst. If you
>>don't know what terminal addition means, look it up, perhaps in
>>connection with "biogenetic law".
>
> No that's not terminal addition at all, it's terminal mutation. If you
> mutate something that affects early development, all the later stages
> that depend on it are screwed up, and the animal dies promptly except
> under unusual circumstances, and even then it doesn't have anybody to
> mate with due to gross anatomical differences. But if you mutate
> something that happens only late in development, there's not much
> dependent on it, so you can easily create a viable new species or
> variation within an existing species that can interbreed with original
> genomes until the new variant becomes common. So most early variation
> already happened pre-Cambrian and won't ever happen again. Most
> variation since then, i.e. within any single phylum, is either
> late-development or biochemical pathways that don't affect develoment.

That's an argument suggesting a mechanism for terminal addition. Since
terminal addition is not a correct model, it's irrelevant. I'll ask
again: how much do you think a chicken blastula looks like a frog blastula?

>>>- Do you accept that it's possible to compare different
>>>trees-of-descent for the same set of "kinds" of animals, to see whether
>>>the trees are basically in agreement with each other?
>>
>>Presumably you mean trees based on different data. Note that for this
>>purpose, two unlinked loci are just as good as morphological vs.
>>molecular data, or whatever pairs you had in mind.
>
> Agreed.
>
> What I posted was just an off-the-top-of-my-head example of the format
> of sequential questions to ask in the poll, not even a first draft of
> proposed actual questions, although I actually like some of those
> questions and would rather adapt then instead of discard them and start
> from square one again. If I do that, I'd probably insert additional
> questions, such as whether they agree it's possible to find similar
> genes with similar functions in different species, and compute a
> philogenic tree, and whether they agree the different trees gotten from
> different such sets of similar-genes turn out to agree with each other
> much more than would be accounted for by randomly-generated trees.
>
>
>>>- Do you accept that in fact the trees-of-descent generated from
>>>fossels, and from embryology, and from genomes, have been compared with
>>>each other, and the result was that they are indeed mutually
>>>consistent?
>>
>>Poorly stated. There really aren't many trees constructed solely from
>>embryological data, so there's not much to compare there.
>
> Agreed. Like I said, this wasn't even a first-draft attempt. I was just
> trying to show the style of my proposed survey, how I'd build up to the
> final question. By the time I got to that final question, I was already
> exhausted and couldn't word it well. I was rushing it and making it as
> terse as possible, violating the second clause of Einstein's criterion.
> (Make it as simple as possible, but no more simple.)
> When I said "fossils", I really meant "time-stamped location-stamped fossils".

Your problem is that nobody constructs trees using that sort of data --
that is the time and locality are known, but are not used on
constructing the tree. And yes, not in rooting the tree either.

> When I said "embryology", I really meant "studies of living forms,
> including development of embryos, structure of adults, biochemical
> pathways, etc., but not including DNA sequences which are a relatively
> recent capability".

As Mark Twain said, try to use the right word, not its second cousin. At
least if you want anyone to understand you. "Embryology" has a clear
definition, and if you use it to mean "everything about living forms
except their DNA", you will get into trouble.

> When I said "genome", I really meant "DNA sequences, including both
> fully-read genomes and partial reading of similar-function genes
> compared across multiple taxa".
> Can you think of a succinct way to ask that question correctly? I can't.

Hard to say, since it's not clear what you think you mean. I would ask
something like this: "Do you agree that a single, consistent, nested
hierarchy results from analysis of disparate sets of phylogenetic data?"

And if you wanted clarification, I would add "such as different unlinked
DNA sequences, morphological characters from different anatomical
regions, protein sequences, behavioral characters, etc."

.

Quantcast