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

anon1@xxxxxxx wrote:

> Regarding Venter's whole-ecosystem shotgun sequencing project:
>
>>The benefit is that it lets you get genomes for species you can't see
>>and can't culture. The problem is that you don't know what you have
>>genomes for. And at any rate it doesn't speed anything up.
>
>
> The biggest benefit may be that you quickly get a nearly unbiassed
> sample of *all* the sequences of DNA present in the environment,
> weighted according to how commonly they occur, which therefore gives
> you an idea which sequences occur often enough that they are lkely to
> be accidently picked up by a virus and dropped in some random place
> such as one of the genomes you have been studying.

I disagree. I don't think there's likely to be any correlation.

[snip]

> I personally feel such an exhaustive neither-over-nor-under-sampled
> survey of variation of genome would be a useful piece of nature-study.
> If that is a goal, then Venter's method seems to be a considerable
> speed-up compared to old methods of trying to collect a sample of cells
> and trying to figure out ways to grow cultures of them and only then
> finally reading the sequence of their genomes.
>
> Furthermore if your goal is to get complete gneomes of
> commonly-occurring micro-organisms, then Venter's method would show
> blatant gaps in coverage, where we need to find a way to culture some
> previously uncultured but very common micro-organism, which we would
> completley overlook by the tradition means of looking for our keys only
> under the light where culturing the cells is easy. Venter's method is
> like pressing the remote button and hearing beeps from all the many
> sets of keys in various dark locations and thereby knowing where to
> look for missing keys, giving us a good chance of finding *our*
> particular keys somewhere among those many beeping sets of keys that
> various people lost, and meanwhile collecting a large number of sets of
> keys that may each be interesting in their own right.

The survey is interesting. It may have value. I don't think it's at all
useful for finding where random chunks of DNA come from.

>>>Semantics: just a single node-to-node thingy whole part of tree cut off
>>>My terms: link or edge branch
>>>Your terms: branch what??
>>>Example: +--+ (just the -- part, not the +s) --+--+--Luke
>>> | |
>>> Noah Joseph
>>
>>That right-hand term would be "subtree", or, if rooted, "clade".
>
> I omitted guessing that because it's the math term and you said the
> jargon here is not like mathematics and I should divorce my language
> from math when dealing in biological systematics.

Not precisely. I said the terminologies are not the same. That doesn't
mean they have no correspondences.

> I guess this is a
> case where you (and presumably standard jargon) would stick with the
> math term. That's fine with me. I'll use sub-tree instead of
> bipartition in the unrooted case, and either sub-tree or clade in the
> rooted case depending on what my emphasis is (the diagram structure
> which are a sub-tree visually/mathematically, or the set of taxa which
> are a clade evolutionarily).
>
> So the all-but-one theorem of unrooted trees is that if you divide the
> unrooted tree (which is asumed to be the shadow of an unknown rooted
> evolutionary tree) into disjoint sub-trees plus a backbone that
> connects those sub-trees together to make the whole tree, at most one
> of those sub-trees is not a clade, i.e. all but one, or all, are in
> fact shadows of clades in that hidden rooted tree.
>
> In any large unrooted tree, there are many different ways to decompose
> it into sub-trees plus backbone, and in each case you can apply the
> theorem to say at most one of *that* set of sub-trees is not a clade.

Why are we doing this again? And there's a simpler statement of your
theorem: for any bipartition of the tree, one of the subtrees is a clade
and the other is not, unless the root falls exactly on that bipartition,
in which case both of the subtrees are clades.

>>the reason we know the polarity is that we are mapping the whole
>>thing on an already-rooted tree. This doesn't root the tree for you.
>
> Are you playing a game of who's on first with me?
> I'm asking the question, how do you know the "official" rooted tree is
> correct in the first place. Do you always just bow to authority?
> Bowing to authority, with only circular arguments to explain why you
> bow, isn't going to be acceptable evidence of evolution.

I have no idea. You snipped so much context that I don't even know what
I was replying to there.

>>>I would grant that, but for reasons you would reject, so even if we
>>>agree as to conclusion we don't agree as to reason why to believe that
>>>conclusion.
>>
>>What are your reasons?

Again, at this point, I have no idea what the subject is. Hopefully it
will become clearer from your answer. But have you considered leaving
enough context to make the question comprehensible to begin with?

> For one, fossils of primitive fish occurred long before any fossils
> of amphibians or mammals or dinosaurs or turtles or lizards or birds
> or crocodiles. If the unrooted tree shows:
> AllBirds--+-+-+-----+--------------------+---AllLizards
> | | | | |
> [Dinosaurs] Crocodiles AllMammals--*--AllTurtles
> |
> +----LivingAmphibians
> |
> SomePrimitiveFish------+------ModernFish
> |
> MorePrimitiveFish
> then from the fossil evidence we can be confident the root is somewhere
> within SomePrimitiveFish or within MorePrimitiveFish or along the
> branches/links connecting those two. We canfidently predict that any
> new taxa we add to the tree totally dissimilar to any of those modern
> clades, or dissimilar to *anything* in that tree including even the
> [S/M]PrimitiveFish, such as seastars, would connect to this tree
> somewhere in/near [S/M]PrimitiveFish, even if it's very recent. We can
> already say that the smallest sub-tree containing LivingAmphibians and
> AllTurtles is a true clade.
>
> Note that the oldest ModernFish fossil is so short a time after the
> first PrimitiveFish fossils that we can't say for sure that
> [S/M]PrimitiveFish came strictly before ModernFish, so that's why the
> claimed clade doesn't also include ModernFish.

There are many problems here. Most importantly, you are dating the
minimum ages of groups only. Maximum ages are not determined. What you
are doing is assuming that the order of minimum ages in the known fossil
record is the same as the order of true ages. That requires claims about
the quality of the fossil record, such that a time gap of greater than
length x is so unlikely to be due to sampling error that we can ignore
that possibility. This method is not completely useless, especially if a
large number of data points are used and error bars are taken into
account, but considering it superior to the methods that are more
commonly used seems unwarranted. As it turns out, we tend to get the
same results from this method as from other methods. That would tend to
validate your method, but then again it also validates the other methods
you consider flawed.

>>>Is there a hemogloben event which can be polarized
>>>from yeast to tetrapods, thereby assuring that the minimal whatever
>>>(branch) containing tetrapods are a true clade?
>>
>>Only if you assume a root for the tree.
>
> So is there *any* way to be sure tetrapods are a clade other than an
> appeal to authority?

If you reject the methods that people have used, yes, there are no other
methods in waiting.

>>You can suppose that the root is more likely to lie near than far
>>from the midpoint.
>
> You accept that, but I consider it too vague. If there are
> approximately 800 million years of evolution between two modern taxa,
> 400 million years along each branch, and we compare molecular distances
> in selected genes (the only ones stable over such long spans of time),
> do we claim the root is within the middle 30:40:30 span of genetic
> differences, or more narrowly within the middle 40:20:40 span, or do we
> allow anywhere within the middle 10:80:10 span? If the narrow
> middle-of-40:20:40 rule is used with all known sequences that have been
> aligned across widely different kinds of eukaryotes (animals, plants,
> and several very different phyla of protists), where are the possible
> places for the root?

I notice you demand rigor from other methods but not from your own "time
stamp" method. Modern fish were placed in a polytomy because their
fossils were too near in time to Primitive fish fossils. But how near is
too near? You don't say. All animals meet in a node that is far enough
from the midpoint of the tree of life that we really don't have to
quibble about whether they're a clade.

>>You can suppose that the polarities of some characters are more
>>likely to go one way than another.
>
> You accept that, but it's not clear which particular kinds of
> characters you consider polarized in this way.

Not important here. We're listing conceptual methods.

>>You can use gene duplications that happened before the common
>>ancestor of all included taxa (and which are retained in all included
>>taxa) to root each other.
>
> I assume you're referring to DUP-then-diverge situations? You say this
> isn't used, as if it weren't reliable. Why do you disdain it, or why
> don't professionals use it? It sounds to me like this is the most
> reliable method of knowing the polarity of a link/branch, whenever it
> can be actually applied.

When did I say it isn't used? It isn't used to test clades we already
know reliably, but it is used.

>>Yeast is a clade if it shows up as monophyletic on the tree.
>
> Given an unrooted tree, plus polarity of some links/branches known within
> that tree, what is your defintion of "shows up as monophyletic on the tree"?
> The only definition I would accept is that it is a sub-tree and that
> the trunk of that sub-tree is on the after-side of any polarized
> branch, which is exactly what I showed in that diagram:
> | Yeast--+---<<<---+-----...
> | | |
> | Other1 ...
> Given the unrooted tree showing Yeast as a single sub-tree, and given
> that single polarized link shown there, Yeast is assured of being a
> clade. (Other1 is also an assured clade. Yeast--+--Other1 is also an
> assured clade. But those matters are moot to the dispute over whether
> Yeast is or is not a clade.) Do you disagree?

Yeast is shown as a terminal node. That assumes monophyly. Perhaps you
didn't mean to make that assumption. I don't really know what we're
arguing about here because once again you removed all context.

>>>Other3+Arthropoda (hence Arthropoda is a clade)
>>>Other5+Chordata (hence Chordata is a clade, and within it Tetrapoda, and "birds")
>>
>>All true if in fact the parts of the topology you don't show are
>>consistent with that.
>
> I agree, none of our evidence is absolute, so a contradiction within
> our data simply means we have made a mistake in estimation or
> measurement, or horizontal gene flow has broken the assumption of
> common descent across *entire* genomes, or that evolution is not true
> in the first place, not that we've found a contradiction in nature
> which will make the whole Universe fall into a black hole tomorrow.

Nothing to do with what I said, as far as I can tell from the little
remaining context.

[snip]

> But is that standard jargon that each of the following is called a "branch":
> ---
> +---
> ---+
> +---+
> and that each of the following is called a "subtree":
> +---AfricanApes
> ---AfricanApes
> AfricanApes
> (What I mean by AfricanApes there is the entire diagram showing Hum +
> Chi + Bon + Gor with the usual connection between them and one external
> link connected to the rest of the tree not part of the "subtree".)

Mathematicians may be interested in the distinction between closed and
open segments, but systematists aren't. Even the interesting difference
between even node- and stem-based taxa emerges only if there is some
real node somewhere on that stem, such that some real species falls into
one group but not the other.

>>>If you're dealing with *current* taxa at the terminal nodes, then
>>>there's no way any of them could be the universal common ancestor,
>>>which would imply current life evolving not from past life but from
>>>other current life.
>>
>>Yes. But unless you require some limits on evolutionary rates, you must
>>accept that possibility. Any living species could be the common ancestor
>>of all life, absent a rooted tree or some limits on rates.
>
> No, I don't accept that possiblity, given that we have fossils of
> living species going back only a few tens of millions of years, most of
> them going back much shorter times, yet fossils of non-modern species
> way back more than a billion years ago, and none of them lasting more
> than a few tens of millions of years per our sampling, except for very
> primitive species that are so vague in shape we could be seeing a
> succession of hundreds of different species that all look the same.
> There's just no way a current species could be unsampled over such a
> long span, with today's population perfectly capable of interbreeding
> with the ancestral 1.2-billion-year-ago population, while so many other
> species come and go in our sampling over that span.

Much easier just to accept that evolutionary rates can't remain zero for
billions of years. By your reasoning, you would not be able to reject
the notion that all metazoans were descended from Lingula.

>>>Hmm, I think I figured out what the analagous stem-based definition
>>>must be in an unrooted tree. Instead of largest clade containing one
>>>taxon but not another, it would be largest mumble (what I call
>>>"branch") containing one taxon but not either of two others.
>>
>>Subtree. Yes, that would work, if anyone had any reason to define
>>stem-based groups for unrooted trees. Generally, we try to avoid
>>defining groups unless they are monophyletic.
>
> OK, then for either stem-based or node-based subtree, we can refer to
> it as such locally within a report where we need to talk about some
> particular part of the unrooted tree that has been drawn. For example,
> we might define subtree #2 = max(Chordata, not Other4, not Other2),
> and then note that the root of that sub-tree is polarized *toward* that
> subtree, which shows it to be a true clade, hence the node-based
> sub-tree just inside it is also a true clade, etc. Elsewhere we might
> define the larger subtree #3 = max(Chordata, not Other1, not Other2),
> and note that we don't have any polarized branch leading into it, so we
> don't at present know whether it's a clade or not.

Yes, if we cared to.

>>Feel free to call it whatever you like that isn't already taken by
>>something else. The general term would be subtree. So you probably
>>want <modifier> subtree as your term.
>
>
> Yes, either:
> - stem-based subtree: max(Chordata, not Other4, not Other2)
> or
> - node-based subtree: min(Chordata, Other5, not Other4)
>
> I have a general question, for anyone who knows: Of all the various
> individual genes or other DNA segments which have "sequenced" across
> many species, which particular such has been sequenced across the very
> most species? (This question for eukaryotes only at the moment, but I
> don't care whether it's a nuclear or mitochondrial DNA segment.) I'm
> guessing it's something I've heard about a lot, such as Cytochome C, or
> DNA Replicase, or ribosomal RNA, etc.

By bet would be 12S rRNA. Other candidates: cytochrome b, cytochrome c,
cytochrome oxidase II, other rRNAs. If you're looking for any short
fragment, then the winner will soon be (if it isn't already), the little
piece of cytochrome oxidase I chosen by the DNA barcoding project.

.