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

> > Which of the taxa in the above tree is the outgroup? How would anyone know?
> > Picking one at random seems to be a circular argument.
> Nobody would know because you made them up.

That's right. With no data except the character matrix or DNA
sequences, you get an unrooted tree, but there's no way to know where
to root it. You need some *other* kind of data to root it.

> >>3) by presumed character polarity, including ontogenetic criteria;
> > All species are equally well adapted to the modern environment. None of
> > them seem to be "primitive" in the sense of having major lack of
> > development of adaptive features compared to the others.
> This is not what character polarity or ontogenetic criteria mean.

OK, I'll try again, this time looking on Google to try to find the
meaning of the jargon, instead of guessing what it means.

<http://www.blackwellpublishing.com/ridley/a-z/Character.asp>
Character polarity is the issue of the evolutionary history of a
character: given two character states, which we call a and a' , we
need to know whether a evolved from a' ' or the other way round.
Discerning character polarity is a fundamental task of phylogeny.
That begs the question. If you don't know where the tree is rooted, if
all branches are unknown-direction, how do you magically know which way
a trait evolved?

<http://pw1.netcom.com/~wsavary/larvae.html>
The outgroup comparison method for
establishing character polarity (Maddison, Donoghue and Maddison 1984)
clearly establishes transverse orientation as apomorhic.
That begs the question. If you already know an outgroup, then the
portion of interest is already rooted.

<http://www.sasb.org.au/TreeBuild/TreeBuilding2.html>
Having determined the homology of the character states, the key to
cladistic analysis is the distinction between derived character states
and ancestral character states (character polarity). It is important
to note that this is a local concept that applies only to a particular
set of taxa. By this I mean that a character state is only considered
to be derived relative to a specified ancestral state, and it may well
be the ancestral state for a further derived state. As an example, for
Again begging the question. You have no idea, within a portion of the
connection (unrooted) tree, which is ancestral and which is derived
until after already have found which direction (out which branch) the
root is located. For example, if you're looking at this portion of the
tree:
A1--+-------------+------B1
| |
(big clade 1)--+---+--A2 +--(big clade 2)
| |
A3 B2
where A/B indicates the two discrete values of the character, but we
don't know which of A,B is ancestral and which is derived, and we
haven't yet examined any members of the big clades to see which of A or
B they might have. We suspect the root might be somewhere within one of
the big clades, but we can't be sure which one. It might turn out that
if we study this character in the big clades, we'll discover that all
members of big clade 1 have character value A, while all members of big
clade 2 have character value B, but even if that's all confirmed it
still won't help us decide which of A or B is derived within this group
of five species. It may also be that one of the big clades is not
consistent in this character, for example:
A1--+-------------+------B1 B3 A4
| | | |
(big clade 1)--+---+--A2 +-----------+--+--+--+--(rest of big clade 2)
| | | |
A3 B2 B4 A5
We can't tell whether A was primitive on the left, evolved to B in
middle, then reverted back to B on right, or the exact opposite from
right to left.

<http://www.ucmp.berkeley.edu/education/events/carlson2.html>
Character Polarity
The direction of character change in evolution, from ancestral to
derived, is referred to as character polarity. It is not always
obvious which character states are derived and which ancestral.
I agree!
Some
criteria for determining the direction of character change in
evolution must be established.
Yeah, obvious, but what??
Ancestral and derived states are always
identified as such by comparison to a frame of reference (Figure 4,
Table 1). Of several possible character states (e.g., hair present,
hair absent), the ancestral state is the one present in a closely
related species outside the group of interest (outgroup criterion),
That doesn't work, or it begs the question. For example, if I'm
interested in various early dinosaurs, and I treat birds as an
outgroup, I've made the *wrong* decision in regard to establishing
direction of evolution of traits. But without any timestamps there's no
way to know that was a mistake.
or
the one appearing lower in the stratigraphic record (paleontological
criterion),
Aha, just like I said, timestamps are useful for rooting trees!!
I said it directly: You use timestamps on fossils to establish
direction among many of the branches. You said it indirectly, you use
"character polarity" to root trees, but then I see that "character
polarity" requires timestamps to decide which way they go.
or earlier in development (ontogenetic criterion).
Aha, early-development-fixed late-development-fluid, which I stated
within the past week but somebody said I was talking garbage, equated
that idea to late-development-add-on as in the "recapitulates" lie.

> >>4) by assuming monophyly of two divisions of the tree
> > That seems to be a circular argument.
> Only if you have nothing at all to go on, as, for example, if you had
> arbitrarily made up a tree with fake taxa.

What if I use real DNA sequence data, but I don't tell you the name of
the four species, so you can't just beg the question by already knowing
their location in the TOL. Well you can cheat by looking up the DNA
sequences in a database to find the species names, then beg the
question. So what if I delete all base pairs which are identical in all
four species, leaving only the SNPs, so you won't be able to simply
search for them to find the species names? You have to work from the
actual DNA data with no outside information? I bet with only the DNA
data, no outside information, you can't root the tree.

> You don't start in a vacuum here, as you do with your made-up tree. We
> do know something, or at least suspect something, about relationships
> before going in.

That isn't going to convince an anti-evolutionist that your entire
theory is based on circular argument. It also isn't going to correct
you if your original suspicion is mistaken.

> > When you use DNA evidence and/or character comparisons, without a
> > single reference to any dated fossil.
> You understand that this means we can't ever reconstruct the phylogeny
> of any group without a fossil record, or with one that's poor enough
> that there might as well be none? And yet we do that all the time.

For this purpose, I allow living bodies as "fossils" with zero million
years before current as timestamp. If you know of any species which
went extinct and never left any evidence that it ever existed, and it's
not simply an inferred species such as branch points on rooted trees,
how do you know there ever was such a species? If the only fossils are
so poor you can't tell what on Earth they might have been like, how do
you know they aren't just pieces of some other species you already know
about? How do you know they are an additional species not already in
the system? How do you "do it all the time"??

As for philogeny of inferred species, that's easy. You use a few
fossils you *do* have dated very accurately to root the tree, then you
have the whole tree converted to a philogeny, including those other
species that were only inferred.

> At bottom, it's a process of slowly assembling a foundation,
> testing that foundation with new data, and using that foundation as a
> basis for more hypotheses. We do know some groups, and know that other
> taxa are outside those groups. That's all the start you need.

How do you build that very first foundation, and be reasonably sure you
got it right? My answer: With a few dated fossils from long ago.
Fossils show that the fish group is not a clade unless the amphibians
are included too, because the first fish were before the first
amphibians, and even that's not a clade unless the reptiles are
included too, because the first amphibians were before the first
reptiles, and that's not a clade unless the mammals and dinosaurs are
included too, because the first mammals and dinosaurs were both before
the very first reptiles, and that's not a clade unless the birds are
included too, because the first dinosaurs were before the first birds,
but at that point you finally have a clade. Then working backwards you
get the dinosaur+bird clade, the reptile+mammal+dinosaur+bird clade,
and the tetrapod clade.

But without dated fossils to show that fish predates the first
amphibians, and yet lines of fish continued right in parallel with
amphibians, hence amphibians are a sub-group within fish, how do you
root the vertebrates? And how do you know vertebrates, without the rest
of the chordates, are indeed a clade, without fossils of chordate
non-vertebrates that pre-date the first vertebrates?

I'm to start from scratch in checking what evidence somebody accepts.
There is no framework of evolution already. The person doesn't believe
in evolution. The person says evolution is a joke, or a fraud, or a
lie, or a religion, and there's no evidence that there's any evolution
whatsoever. So without any framework a priori, how to establish the
framework of evolution, directly from evidence, without any circular
arguments, without any faith in evolution, etc.

If we try to root the tree of animals based on diversity, just as we
presumed humans came from Africa due to largest diversity there, then
with a vast majority of species being insects we must (falsely)
conclude that the root is within the insects, and figuring that about
half the tree is on each side of the root, we have something like this:
+--(half the insects)
root--+
| +--(other half the insects)
+--+
| +--(rest of arthropods)
+--+
| +--Mollusca
+--+
| +--(rest of Proterostomes)
+--+ +--Chaetognatha
| +--Deuterostomes------------+
+--+ | +--Chordata
| +--Pseudocoelomates +--+
| +--+ | +--Hemichordata
+--+ +--Accelomates +--+
| +--Echinodermata
| +--Radiata
+--+
| +--Placazoa
+--+
+--Porifera
So what evidence, other than dated fossils, would refute that cladogram?
(The correct root is between Porifera and the entire rest of the tree.)

> You can do this because you are not working in a vacuum.

Wrong. I *am* trying to start from the very beginning, no knowledge of
evolution whatsoever, only knowledge of the prerequisites such as
crystal formation and isotopes and genetics (including DNA) and plate
tectonics. What evidence is needed at the very start to show
conclusively that evolution is a good theory to seriously consider.

> Other people have learned a few things before you. One of those
> things is that amniotes are a clade, but fish aren't.

Presuming there's even the concept of a clade, much less anything
specific about fish vs. amniotes, already presumes the truth of
evolution as a fact, creating a circular argument if we are to use this
to show evidence for evolution, per the whole purpose of this thread.

> If they hadn't previously learned those things you would have to
> start somewhere else.

Yes, that's my purpose in this thread.

> But stratigraphic data really are not helpful in doing this.

What else gives clear evidence that something changed over time, that
things now aren't the same as things long ago, not just day to day
things like Kennedy born then assassinated, or century to century
things like Rome built then fallen, or even Babylon founded, but really
major things like a time when there were no humans whatsoever yet there
was other life already, not just two or three days of animals without
people, but hundreds of millions of years of animals without people,
and not even remotely the same kinds of animals during various eras of
that very long span of time?? Or ignore the absolute time span. Many
mega-generations of different forms of animals over whatever span of
time whether it be five years or five hundred million years,
trilobytes, then dinosaurs, then mammouths and primates. A whole series
of major changes in kinds of animals before the first human. Not just
all the non-human animals and plants virtually simultaneously, then
humans created specially apart from the rest. But all those major
different kinds of animals in temporal succession.

> If you have a gene duplication that happened
> before the time of origin of the group you are looking at, they will all
> have two copies. These two copies will make two trees (which should be
> identical), and the trees will be joined at some branch. That branch is
> where the root must lie.

That assumes enough time has elapsed *within* the group of interest
that every branch has "suffered" enough mutations to be sure of the
branching of those copies. Indeed I agree if the two unrooted trees run
parallel to each other, that's nice, and whichever pairs are most
distant show them to be furthest from the root. If there are two which
are definitely closest, then the root probably is between them. But if
there is one that is significantly closer than all neighbors, then it's
not clear which branch from that one node has the root. Still it does
narrow down the root to either one branch between two nodes or three
branches from one node to its various neighbors. Assuming all the
species are currently alive, and DNA is from living samples, All the
species should be approximately the same distance from the root, so I
would expected the unrooted tree to look something like this:
A--------+-----+-------------+----E
| | |
B------/ | \--D
\-----------C
Assuming equal rates of evolution, the root might be just to the right
of that central +, but we can't really be sure.

Now measuring the amount of divergence between copies of duplicated
gene at each of the present-day species, I suspect all five will be the
approximately equal, and that won't help us at all. On the other hand,
computing an unrooted tree of each duplicated gene by itself, and then
inferring what genome might be at the branch nodes, *then* we might see
the branch nodes having the two copies closer together than the leaf
nodes, so then the distance-between-copies might help locate the root.
Is that the method you've summarized?

> As for the same duplication happening twice, that's homoplasy, and it's
> something we have to worry about with all data of any sort.

Is there *ever* a case where exactly the same segment of DNA is
duplicated in two very different branches of life, with the copy
landing at exactly the same new location both times, so it looks like
the two branches are a lot closer than they really are? If so, what
mechanism might do that?

> Wrong meaning of terminal addition for our purposes. Irrelevant. The
> meaning relevant to Haeckel is the idea that new morphological features
> are added to the end of an ontogenetic series, i.e. late in development.
> Temporally terminal, not anatomically terminal. That's the necessary
> condition for recapitulation to work.

It's my understanding that the idea that a primitive species progresses
to some point and stops, but an advanced species progresses to that
same point but goes on another step, the "recaptulate" idea, what I
understand by the term "terminal addition", is wrong. But the idea that
two closely related species share steps 1 thru 8 but differ at step
number 9, and two species not quite so closely related share steps 1
thru 7 but start to differ from 8 onward, what I call "terminal
modification", is basically true. Feel free to correct me.

> Phylogenetic data is data we think might be useful in reconstructing
> phylogenies.

That's not a definition at all. It's a circular piece of crap if you're
trying to tell me which specific kinds of data you would consider
useful for reconstructing phylogenies.

Police cadet: What sort of evidence should I try to collect at the
crime scene?
Instructor: Criminal evidence.
Police cadet: What sort of evidence is criminal evidence?
Instructor: Anything that would be useful in diagnosing the crime.
Police cadet: And what specific kind of evidence would that be?
Instructor: Criminal evidence.

When I ask what sort of evidence might help root an evolutionary tree,
it doesn't help if you answer "Phylogenetic data", and when I ask you
what you mean, you answer "anything that would help root a tree" or
"anything that would help reconstruct a phylogeny" etc., because that's
what I was asking in the first place.

> Very few phylogenetic analyses make any a priori claims to know
> what's primitive or derived. These are instead conclusions gained from
> the trees.

But you have only unrooted trees until *after* you have decided where
the root lies. It's a circular argument to use an already rooted tree
to decide what's primitive and use that to decide how to root the tree.

> > DNA duplication events seem to be one way to
> > establish an arrow-of-time within unrooted trees to thereby narrow down
> > where the root really was.
> Not that reliable, at least in the way you have stated it, because
> duplicates can be lost as well as gained.

If a duplication event occurs, the two copies will be nearly identical.
After a period of evolution, they won't be the so similar any more. If
then one of them is deleted, the two that existed just before the
deletion won't be very similar at all. It's very easy to see the arrow
of time in an unrooted DNA-segment tree like this (horizontal
represents DNA base differences, vertical represents chain of
time-related species):
*
*
* *
* *
* *
#
*
(# represents two nearly identical sequences, i.e. ** on single grid point)
Can you see the arrow of time in that diagram? Does time go up or down?

> There are however some DNA characters that are close to being
> self-polarizing. Look up SINE insertions, for example.

Looked it up just now, some kind of spontaneous multiple repeats, can't
figure out what's the difference between these and "tandem repeats".
..

.