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

anon1@xxxxxxx wrote:

>>>>Evidence for what, precisely? This seems like evidence for common descent.
>>>
>>>No no no, don't pre-judge what the evidence is "for". Just collect
>>>evidence like any good police detective, and then after you have
>>>all the evidence, see where it leads. For example, if you pre-judge
>>>that the husband probably killed his wife, you'll be so busy collecting
>>>evidence to "prove" he did it that you not bother to ask neighbors if
>>>they saw any one-armed men around the neighborhood, and as a result
>>>you'll achieve a false conviction Dr. Richard Kimble.
>>
>>Actually, no scientist does anything like this. You don't collect
>>evidence at random, and you don't collect evidence to prove any
>>particular hypothesis. What you do is collect evidence that bears on a
>>family of hypotheses and (we hope) distinguishes among them.
>
>
> You and I are in agreement, although before you have any hypothesis you
> just collect evidence regarding some particular kind of phenomen.

This is very seldom done. I can't really think of an example right off.
You almost always start with some kind of theory. That theory may be
based on evidence collected for some other purpose, by someone else. But
you never really start in a vacuum.

> After
> you have studied some evidence then maybe you are ready to guess
> various possible hypotheses, and from that point you may do as you say,
> collect evidence to distinguish the various possible hypotheses.
>
> To have the most convincing case against Creationists and IDiots,
> especially those leaning that way but not absolutely 100% certain that
> evolution is right and YEC or ID is correct, I want to take a
> historical approach. For example, first collect *all* evidence from the
> fossil record, without any single hypothesis or set of hypotheses in
> your mind. Then you look at it and see what it seems to show, namely
> some sort of descent (linear chains of related fossils), and then some
> sort of common descent (branchings whereby several modern species come
> from a single ancestral species).

Feel free to do whatever you want. But that's not how science is done.
Also, both of your chosen forms of evidence seem to rely on identifying
actual fossils as ancesors, which is just impossible. Instead you should
be thinking about nested hierarchy.

> So part 1 is just to look at the three kinds of evidence:
> characteristics of individuals that seem to be clumped into species
> that can mate together and clustering of such species at increasingly
> larger scales to form a nested hierarchy, fossils that seem to be lined
> up in time-consecutive lines with branchings from time to time, and
> cladograms from DNA sequences that actually can be computed and are to
> an unexpected degree stable unlike what you'd expect if the genomes
> were randomly generated.

It's probably not a good idea to think of these as separate. The first
and third especially seem like just two ways of saying the same thing.
The middle one is a garbled form of something that paleontologists
sometimes do, which is to test the stratigraphic fit of their
cladograms. It usually comes out as highly non-random despite the patchy
nature of the record.

> Then part 2 is to ask how to explain that
> evidence. At this point we have a prima facie case in favor of
> evolution and common descent, regardless of whatever mechanism might
> have caused it. I'm trying to show that if you just look at the
> evidence, with no pre-conceived idea what to expect, you come up with
> evolution and common descent.

That's pretty easy to do. Several people here and in the TO FAQs have
done just that. But you have a strange way to go about it. As I recall
we began this argument because you were trying to use the evidence above
as some kind of argument for natural selection, not just common descent.

> I leave for later the exhaustive testing,
> trying to find violations of the theories, failing to find any, and
> thereby getting more and more confident of the correctness of the
> theory. I'm just trying to collect a nice organization of the evidence,
> to ask who accepts it and who disputes it, and then for whose who agree
> to stipulate the evidence is correct, a nice organization of the
> theories of evolution and common descent, and ask them whether they
> accept those conclusions. For those who accept all the evidence, and
> evoution and common descent, then I ask them what theory, other than
> mutation + replication + selection/drift, they have to explain the
> cause of evolution and common descent.

You have a strange way of going about it.

> I really want a check-list of evidence, accept or deny, and then a
> check-list of theory, accept or deny, so we can pin down where exactly
> the others disagree with modern evolutionary fact or theory.

Good luck on that. I've had trouble getting creationists to admit what
they believe, especially in any consistent fashion, or to accept the
consequences of what they believe.

>>Anyway, I'm not pre-judging. I'm post-judging.
>
> Post-judging what? The topic of this thread (you dispute either the
> evidence or the theory to explain the evidence)? Or my style of
> debating the anti-evolutionists (you don't like my idea of asking them
> whether they dispute the evidence, and then if they accept the evidence
> askign whether they dispute the theory)? Or what??

Lost context is probably the reason for your confusion here. If you
hadn't snipped everything, you would be able to determine from context
just what I'm post-judging. As I recall, I was judging what the sorts of
evidence you were talking about were capable of telling us.

>>I don't think we can actually identify such fossil chains. What we have
>>are trees. There are various statistical tests you can do that will
>>determine the degree of "strain" between the tree and stratigraphic
>>order. When applied to the fossil record, these tests commonly show a
>>much better than random fit. That's about as close as actual science
>>gets to these chains of yours.
>
> Do you know of any Web page showing a complete worked out example for
> one "kind" of animal ("kind" in the naive sense, such as horses or dogs
> or cats etc.), whereby there's a description of each specific character
> being compared, with a criterion for assigning numeric values to each
> character (such as length of rear femur measured in centimeters, or
> curvature of spine between shoulders and rump measured somehow, or
> interpolation between claws and hoofs), and then a complete matrix
> showing the value (mean and standard deviation) for each character for
> each species? Then that matrix is fed to such-and-such computer program
> which generates the best-fit tree, if any, and shows a measure of
> fitness

Better call that "fit"; fitness has a different meaning.

> such as a P (confidence) value for tree-ness (and/or that
> specific tree model) preferred over the null hypothesis of random
> number input not belonging to any tree? It would be nice to
> have two such examples, one very small, such as apes, only five or ten
> species, such that the whole thing could be processed by hand or at
> least the final result verified by hand, and one much larger to show
> how really interesting trees can be demonstrated.

Only the one I did myself with mtDNA sequences. Did you read that post?
I've put it up many times. You will find something of the sort in
hundreds of scientific papers, but they never test a tree against
separate creation. However, we can suppose that any strong, nested
hierarchy is good evidence against the absence of a tree, and those are
plentiful in the scientific literature.

>>>When we fold together all the fossil chains, with the branch points
>>>that replace one ancestral fossil chain with two or more new fossil
>>>chains, we have a forest of trees. These trees are consistent with a
>>>single tree where lots of links are missing from our evidence so-far.
>>
>>That's not how it actually works. The trees we build all have real taxa
>>only at the tips of the branches. There are no linear "chains", because
>>we have no way to tell if one species is ancestral to another.
>
> If within a single geographic area there's an isolated species that
> lasts a few million years, then immediately afterward there's another
> species only slightly different from the previous but clearly showing a
> significant change from the previous, and it's in the same area as the
> first, we can reasonably guess that the second descends from the first.

No, we can't. We could if we could be confident that there had been no
movement between unsampled geographic areas, and that we had sampled all
species in that particular area, and that we could actually recognize
which individuals belonged to the same or different species. But we
really can't do any of those things.

> Are there no such examples whatsoever in the fossil record to show
> blatantly apparent descent?

Not to my knowledge. But feel free to propose one.

> Alternately, consider a marine species that
> is distributed worldwide, followed by a similar worldwide marine
> species, no examples whatsoever of that ever in the fossil record?

Not to my knowledge. There are certainly many with quite wide distribution.

> I'm
> saying if there are even a few such examples, they can establish the
> prima facie suspicion that lines of descent might be apparent, and then
> we can test that hypothesis to see whether there are other
> not-so-obvious lines of descent. But we need some really blatantly
> obvious lines of descent so that it makes it reasonble to even conceive
> such a hypothesis.

No, we don't. That good old nested hierarchy works fine. Also, a general
pattern of change in world biota (life of the Ordovician different from
Cambrian, Silurian different from Ordovician, etc.) argues for evolution
too, even without a hierarchy.

> Now Darwin's finches wouldn't be like that. He observed a whole bunch
> of species, which all seemed be be variants upon a common theme, and he
> guessed that they all descended from a common ancestor, but his "tree"
> had a single level wherein all modern species suddenly appeared in
> parallel, no successive splittings over time that he could discern
> without any fossils to look at.

In fact Darwin didn't even propose a tree for Darwin's finches, just
common ancestry for the lot. However, we don't need fossils to determine
this successive splitting. There are several papers on the phylogeny of
Darwin's finches, the most recent using DNA sequences.

> But surely among the actual fossils
> back on the mainland around the world, there might have been examples
> of apparent chains of descent?? Surely Lamarck knew of such when he
> formulated his linear evolution theory??

I don't know whether Lamarck used fossil evidence of descent. But I
don't know of any "chains", just trees. You can turn trees into chains
by ignoring some of the branches, if you really want a chain. But the
assumptions are unwarranted and unnecessary.

>>>We can state strongly that there are fossil chains, and branch points,
>>>yielding lots of small common-ancestry trees. We can guess with less
>>>assurance that they all belong to one universal common-ancestry tree.
>>
>>If what you're trying to say is that not all nodes are resolved, then
>>yes. But we don't in fact end up with lots of separate trees. We end up
>>with one tree some of whose nodes are unresolved.
>
> I believe such trees would show low levels of confidence for some
> branchings, expecially pre-Cambrian, and if the program is forced to
> exclude all low-confidence parts of the tree, it would yield a bunch of
> separate trees instead of just a single tree. (Some programs would
> simply refuse to give any result at all.

Not true. I only know of one such program, and I doubt sincerely that
you have ever heard of it.

> I'm suggesting a program that
> would break apart low-confidence trees to achieve a forest of
> high-confidence trees instead.)

That's not what we would get. The node of all eukaryotes, for example,
is a very high confidence one. So whatever lack of resolution there may
be above that node, everything must come together there.

> By the way, by "unresolved", you mean three or more subs in a rooted
> tree, or four or more branches in an unrooted tree, can't be resolved
> into a binary-rooted-tree or ternary-unrooted tree respectively, right?

Yes. Or to put it more simply, a node at which 4 or more branches come
together. All fully resolved nodes have exactly 3 branches. (In the
rooted tree, that's one ancestral branch and two descendants.)

> Note I'm assuming that if you have only a bunch of current-day species,
> with no information about past species, such as when making DNA
> cladograms, or when including embryiological data which you have only
> for present-day organisms, you generate unrooted trees, but if you have
> time-dated species, such as when dealing with the fossil record, you
> can then generate rooted trees, right?

No. Trees can be rooted in many ways, and we seldom if ever, even with
fossils, use stratigraphic position to root a tree. The most common
method is by outgroup. This of course requires the assumption that some
particular species is outside the clade of interest.

> Here's an interesting data-analysis experiment: Use all the fossil
> evidence, but discard all time and place data, include *only* the
> characters of the various species, and use that data to generate an
> unrooted cladogram. Then check if the that unrooted tree matches the
> rooted tree that you got when you included time-and-place data from the
> outset. If they indeed do match, that's refutation of ID, because a
> supernatural Designer wasn't restricted to placing designed species in
> a natural-looking chronological sequence, He could have designed a
> Jurassic species, kept the design in his Mind, then designed five
> different variations on it for the Cambrian, then manufactured those
> five when the Cambrian time came, then waited for the Jurassic and
> manufactured that ancestor species at that time, no need for ancestral
> designs to be manufactured before descendent designs.

You have conflated ID, which at least claims to be a theory about
mechanism, with separate creation. This would be evidence against
separate creation, not against ID.

Something like what you have proposed here (though quite different in
detail) is actually done by paleontologists. They assess the strain
between trees constructed without stratigraphic data and the
stratigraphic data themselves. There are a number of different tests.

There are also a couple of methods that allow incorporation of
stratigraphic data into tree-building, but they are rarely used. If
you're interested, look up stratocladistics.

> Even computer
> software designers sometimes do that: Design a primitive program, but
> not release it to the public, then design variations on it, and release
> those variations, then later finally release the prequel for those who
> have been asking for it. Some TV programs or movies are like that too,
> releasing the main movie before the prequel (although in those cases
> the prequel was designed after the main movie, but they were
> retro-evolved to make them look like ancestral movies).
>
>
>>>We can state with intermediate assurance that we have a single
>>>common-ancestry tree for each phylum, and sometimes for several phyla
>>>together, from the Cambrian explosion to the present. The case for
>>>joining all those trees into one universal tree before the Cambrian
>>>explosion is not firmly established by fossil evidence alone.
>>
>>True. Were we limiting ourself to fossil evidence? I was unaware.
>
> What, you're recanting your claim that we *always* get a single tree?

If we limit ourselves to fossils, that single tree is not so clear or so
strongly supported. I'm not really sure what characters you would even
use that you could score across all metazoan fossils.

> Yes, I'm treating the three lines of evidence (modern taxonomy,
> historical fossil, and DNA cladograms) as three independent lines of
> evidence at first. For fossils, we get a rooted tree or forest of trees
> of which the largest start just before the Cambrian. For each of the
> other two we get an unrooted tree with a bunch of unresolved nodes or
> even a forest of disconnected unrooted trees.

No, not true. You are confused about rooting. There is also no reason to
separate any of these. Are you proposing that analyses of morphology
should include fossils only, or living species only? Are you proposing
that no combined analyses of morphology and molecules should be done?
And your (apparent) distinction between morphological and molecular data
is artificial. There is no "twin nested hierarchy". And independently
assembled data set is testable against any other: one gene vs. another
gene, one part of the body vs. another part, etc.

> The fact that we can get
> most of the data fitted into high-confidence trees is powerful evidence
> already. *then* we compare those three trees (or forests of trees) to
> see whether they are approximately mutually consistent), and the fact
> that they are indeed is really powerful evidence.

> The two sets of
> unrooted trees merely show there's hierarchial clustering of
> characteristics, a Platonic sort of thing,

Not so. In fact (and ignoring your idea of rooting), the nested
hierarchy itself is the most powerful evidence of common descent. Nobody
has ever been able to come up with an alternative explanation. Try it
yourself. Attempt to flesh out your "Platonic sort of thing" into an
explanation that holds up.

> but the one set of rooted
> fossil trees shows some actual chronological pattern of descent with
> modification (either actual descent of living species, whereby species
> begats species, or descent of Intelligently-Designed productions where
> design begats design and each design is released into production
> separately, like the evolution of MicroSoft Windows software designs).

No, it doesn't. Like I said, we can't actually see the begats. We can
only infer from the nested hierarchy that they have happened. However,
stratigraphic fit is a separate criterion that we can use to evaluate
the evidence, so you're not entirely off base. There's a germ of a real
idea there, though you state it confusingly.

> So indeed for the one line of evidence from fossils, indeed at first
> I'm limiting the evidence *only* to fossils, nothing else. That's why
> the trees are disconnected before the Cambrian.

> When we later try to merge the disconnected rooted trees (from fossil
> evidence) with DNA evidence, we then hope to fit those various rooted
> fossil-trees onto branches of the DNA unrooted trees, and then use the
> central parts of the DNA trees to link the fossil trees together into a
> single unrooted tree.

Again, this is not the way it's done, or the way it should be done, or
even a way it could be done.

>>>This is where we need the other two lines of evidence (comparative
>>>embryology, and DNA cladograms) to piece together the separate trees
>>>into a single universal tree. And last I heard the best we could do is
>>>establish three trees for the three major domains, with unknown linkage
>>>earlier than that.
>
> (Let me clarify that slightly: When drawing whole-cell-species trees,
> at each endosymbiosis event I expect to start a new tree, since it
> doesn't make sense to fit a combination of separate symbionts into a
> single ancestor. If we ignore plastids however, we may be able to get
> by with just two trees, one for prokaryotes and one for eukaryotes. Or
> if we ignore the endo-horizontal gene flow from mitochondria to
> nucleus, pretend like the DNA just magically appeared out of nowhere
> when it first got into the nucleus, and only the nuclear DNA is
> calculated in our cladograms, then we might get a single unrooted tree
> for both prokaryotes and eukaryotes. I've seen published reports where
> they seem to have done that last thing, but failed to make that clear,
> so their report seemed poor-quality to me.)

!

>>The problem is in defining the root.
>
> For unrooted trees, such as cladograms of modern DNA or cladograms of
> modern characters including embryology, there's never a root, so who
> cares?

Like I have said already, this is just plain wrong.

> When we fit rooted trees from fossils in with unrooted trees
> from cladograms (both kinds), we lose the roots. We have directed links
> where pairs of dated fossils show the direction, but non-directed links
> where there are no dated fossils, so the central part of the unrooted
> tree consists of non-directed links while the rest is all directed
> links.

And this isn't even wrong.

>>But don't confuse the lack of resolution with a lack of a tree. The
>>three domains are clearly united by a host of characters. The questions
>>are which two are more closely related, and if a branching tree is in
>>fact the best model.
>
> Yes, we agree about this. In particular I actually am leaning against a
> whole-cell (or whole-genome) tree-system of heredity for the very early
> eras when the three domains originally came to be. I think horizontal
> gene flow and/or frequent merging of genomes to yield new composite
> genomes (formation of endosymbiosis by merging of formerly separate
> genomes) may have been very common way back then. Two strains of
> RNA-world pre-prokaryote might have separately evolved different
> aspects of biochemistry, such as one of them evolving a way to make DNA
> bases and the other evolving an efficient form of sulfur respiration,
> or one of them producing formate and the other consuming it, and
> initially they survive as exo-symbionts via external exchange of
> chemical products, and later the two merge to form a single strain that
> performs both types of chemosynthesis within a single cell. It's
> possible that all such genomic merging occurred before the first
> DNA-world true-prokaryote happened, but it may also be possible that
> exo-symbiotic biochemistry webs persisted right into the age of DNA
> prokaryotes and only later a bunch of additional merging of genomes
> happened, yielding five or ten different domains, of which three were
> so much more fit/successful than the others that they drove the others
> to extinction.

I think you have confused all manner of early events here. RNA world
comes way before any of the prospective symbioses you speculate about
here. It's also very unlikely that any two prokaryotes actually merged.
Endosymbiosis would require that one cell incorporate the other, and
that wouldn't happen until the invention of endocytosis by
proto-eukaryotes. What you can have is exhange of genes by plasmid or
viral transmission.

>>>Accordingly *any* theory of evolution (to fit the evidence on Earth)
>>>*must* involve some kind of selection, hence any *scientific* theory of
>>>evolution must include stochastic selection.
>>
>>I don't think so. For one thing, "stochastic selection" is a terrible
>>term, conflating drift and selection.
>
> But as somebody else pointed out, drift and selection *are* already
> conflated in fact.

No, they aren't. They are entangled, if you like, which is quite a
different thing. Using terms that confuse two different mechanisms is
just, well, confusing.

> They are caused by the same mechanism: Differential
> survival rates among different allelles,

That's not a mechanism. It's a result. Selection and drift are
mechanisms. Or if you want to think of it another way, being eaten by a
frog is a mechanism. Any affect your phenotype has on the chances that
this does or does not happen is selection. Other effects (being in the
wrong place at the wrong time, say) are drift.

> which *may* be partly caused
> by some difference in fitness or may be caused entirely by random
> events unrelated to any difference of fitness. (Or as you point out
> below, with supernatural intervention, all random elements whatsoever
> might be eliminated, and the differential survival rate might be
> *entirely* caused by differential fitness in the Mind of the
> Intervener.) In most cases, both difference in fitness (not necesarily
> in the allele under consideration, but possibly in nearby genes which
> are linked to the allelle under consideration) and randomness are
> involved in the differential survival, not just statistically, but for
> each and every life/death decision which occurs. An individual might
> die tonight because its cells lacked the ability to synthesize citric
> acid, and by chance it wasn't able to consume any food containing
> citric acid for a while, and its stockpile ran out, and Kreb's cycle
> slowed down so much that the cell exhausted its ATP supply and could no
> longer perform basic functions of life. How can you say that death was
> *entirely* 100% due to the faulty citric-acid pathway, or it was
> *entirely* 100% due to the chance lack of citric-acid-containing food?
> I say that both the differential fitness and the bad luck in the food
> department caused the death.

That's a bad example. It's selection if some other individual can
synthesize citric acid (really you mean ascorbic acid, i.e. vitamin C,
by the way) and, encountering the same circumstances, doesn't die.
Anyway, this has been covered at length already. That's not the point at
all. The point is that "stochastic selection" is a bad description of
any of this. The features that are selection are not stochastic, and the
features that are stochastic are not selection, regardless of whether or
not they are happening at the same time. Use the right word, not its
third cousin.

[snip]

>>>and the DNA cladograms can be shown live in the
>>>laboratory so only a fool would dismiss them.
>>
>>I don't know. But it's all just the nested hierarchy under another name,
>>and I expect the same responses: it exists but it's explained by
>>goddidit, not common descent, or it doesn't really exist.
>
> That's why I want to clearly organize the various kinds of evidence,
> and the various elements of the theoretical explanation for the
> evidence, and ask each individual person which items he/she accepts or
> disputes. Once they are pinned down as to where they disagree with our
> chain of evidence and logic, we can work on discussing thet particular
> breakdown-point with them. Maybe they'll learn they were overlooking
> something, and they convert to our logic. Or maybe we'll discover
> there's a flaw in our logic and our case isn't yet proven.
> .

I'm not sure you know what you want to do, or have any idea how to go
about it.

.

Quantcast