On 9/14/21 3:36 PM, Peter Nyikos wrote:
> On Tuesday, September 14, 2021 at 11:37:11 AM UTC-4, John Harshman wrote:
>> On 9/14/21 7:50 AM, Peter Nyikos wrote:
>>> On Tuesday, September 14, 2021 at 9:29:13 AM UTC-4, John Harshman wrote:
>>>> On 9/14/21 5:40 AM, Peter Nyikos wrote:
>>>>> On Tuesday, September 14, 2021 at 12:41:28 AM UTC-4, John Harshman wrote:
>>>>>> On 9/13/21 8:09 PM, Peter Nyikos wrote:
>>>>>>> On Monday, September 13, 2021 at 3:04:02 PM UTC-4, John Harshman wrote:
>>>>>>>> On 9/13/21 11:39 AM, Peter Nyikos wrote:
>
>>>>>>>>> Back to the specimen. On a thread about a Cambrian invertebrate [IIRC a stem arthropod]
>>>>>>>>> you were disappointed by the lack of a phylogenetic analysis. The following should put you
>>>>>>>>> in hog heaven where this "shark-tooth" is concerned:
>>>>>>>>>
>>>>>>>>> "Two phylogenetic analyses, each including the new taxon Ulughbegsaurus, were conducted with the
>>>>>>>>> software TNT v. 1.5 [41] (electronic supplementary material, text S1: figures S3 and S4). Our first
>>>>>>>>> analysis was performed using the data matrix proposed by Carrano et al. [42] then modified by
>>>>>>>>> Hendrickx & Mateus [10] where Eoraptor represents the outgroup; the characters are treated as
>>>>>>>>> unordered. Our second analysis was conducted using the matrix originally proposed by Porfiri et al.
>>>>>>>>> [43] then modified by Chokchaloemwong et al. [40] where Ceratosaurus represents the outgroup;
>>>>>>>>> characters 2, 4, 6, 13, 15, 17, 27, 69, 106, 148, 155, 158, 160, 167, 169, 171, 179, 181, 194, 195, 205, 208,
>>>>>>>>> 217, 233, 241, 259, 267, 271 were treated as ordered. A traditional heuristic search was done with 1000
>>>>>>>>> replicates of Wagner trees using random addition sequences, followed by the tree bisection and
>>>>>>>>> reconnection branch swapping that holds 10 trees per replicate. Consistency index and retention index
>>>>>>>>> were calculated with PAUP 4.0a [44]."
>>>>>>>>> https://royalsocietypublishing.org/doi/pdf/10.1098/rsos.210923 [Section 4, paragraph 1]
>>>>>>>>>
>>>>>>>>> What does "were treated as ordered" mean?
>>>>>>>
>>>>>>>> A multistate character can be either ordered or unordered. If unordered,
>>>>>>>> going from one state to any other takes one step. If ordered, going from
>>>>>>>> state 1 to state 3 requires two steps, etc.
>>>>>>>
>>>>>>> Ah, like state 1 feathers (dinofuzz, optimistically called protofeathers), state 2 feathers,
>>>>>>> state 3 feathers [barbs but no barbules, as in kiwis], state 4 feathers, and state 5 asymmetric flight remiges?
>>>>>
>>>>>> That depends on how you choose to code them. What you have there is a
>>>>>> multi-state character, but you have provided no information on whether
>>>>>> it's to be considered ordered or unordered in analysis.
>>>>>
>>>>> What, don't you remember Prum's hypothetical stages in the evolution of feathers?
>>>>> You seemed to be quite a fan of his about half a dozen years ago.
>>>
>>>> Again, it all depends on how you choose to code it. Prum's (and Brush's)
>>>> stages could be considered as an ordered or unordered character.
>>>
>>> It looks like a no-brainer: count the steps, like you said. It's plain to see in which direction
>>> the characters changed.
>
>> Yes, you could definitely code that as an ordered character. There are
>> potential complications that I won't go into.
>
>>>>>>> But you don't include any of these in your phylogenetic analyses, despite a "consensus" that the majority of
>>>>>>> coelurosaurs had at least state 1, eh?
>>>>>
>>>>>> Yes. The consensus is based on the phylogenetic distribution of known
>>>>>> feathers and optimization on a tree constructed from other data. The
>>>>>> sparse data for feathers just don't contribute much to supporting a tree
>>>>>> even if you include them.
>>>>>
>>>>> So you think that, unlike the case of mammalian hair, loss of feathers in non-avian dinosaurs
>>>>> was a great rarity?
>>>
>>>> No. Where are you getting all these notions?
>>>
>>> Isn't it obvious? absence of feathers in fossil after fossil of the same species
>>> could mean actual loss of feathers.
>
>
>> How would you distinguish loss of feathers from non-preservation?
>
> It's not a certainty, but a hypothesis. As more and more good fossils
> accumulate, the hypothesis is supported. You know, ye olde "scientific method".

I don't think that works. You have to take taphonomy into account. The
conditions for the preservation of feathers or skin are extremely rare,
and you can't just count up fossils with or without them.

>> Usually we infer loss (when coding characters) only when there is
>> preserved skin. And even there it tends to be a patch here and a patch
>> there, so you don't know if there were feathers on some other part of
>> the body.
>
> So you seem to want certainty here, but not in telling whether X is a sister group of Y.

No. I'm saying that absence of evidence (of feathers) is not evidence of
absence.

> For example, there was a revolutionary pair of papers which upset the {theropod, sauropod, ornithischian}
> sister group relationship.
>
> Baron et al. 2017, "A new hypothesis of dinosaur relationships and early dinosaur evolution",
> https://www.semanticscholar.org/paper/A-new-hypothesis-of-dinosaur-relationships-and-Baron-Norman/f03f6a706a883b18303e61b4cca6a56d942bf637
>
> Langer et al. 2017, "Untangling the dinosaur family tree":
> https://www.semanticscholar.org/paper/Untangling-the-dinosaur-family-tree-Langer-Ezcurra/8c8ff01b0526f76d1a98c88c01d1978c9de75614
>
> A couple of years ago, you said that these papers had been superseded and we are
> back to the traditional division. But I don't recall what evidence you provided.
> Can you cite a paper or two that put this Ornithoscelida hypothesis to rest, in your opinion?

I'm not sure it's put to rest. It's still a matter of contention,
though. Several dinosaur paleontologists criticized the original paper,
if I recall. In fact, you will note that the second paper you reference
finds the traditional topology, not the Ornithoscelida topology.

> Temporarily, at least. I expect you to think that a stake has been driven through
> its heart, but I would like to see the opinion of some well known authority.

I would hope that data analysis would be a better guide than authority.

>>> That's why I think it is a good idea to include feathers
>>> [in the broad sense that seems to be the consensus these days]
>>> in a phylogenetic analysis in which the taxa have enough preserved
>>> of the places where one might expect feathers to have been attached in real life.
>>
>>> Really, what's the harm in doing that?
>
>> No harm if you code most taxa as missing data. Just no benefit either.
>
> Except that you might discover that there is no correlation between having
> hairlike growths optimistically called "protofeathers" and having feathers
> with barbs and barbules. Then the whole "feathered dinosaur" hypothesis could
> be split in two: dinofuzz and feathers.

What exactly do you mean by that? There could only be a correlation if
you scored them as separate characters. But why would you expect a given
fossil to have both, if one evolved from the other? I'm not sure you
have thought this through.

> [Trivia: General Winfield Scott, Mexican War hero, was called "Old Fuss and Feathers."]
>
>
>>>>>>>>> The section continued with:
>>>>>>>>>
>>>>>>>>> "The first phylogenetic analysis produced 6320 most parsimonious treeswith a strict consensus tree placing
>>>>>>>>> Ulughbegsaurus within a poorly resolved Neovenatoridae (Aerosteon, Australovenator, Chilantaisaurus,
>>>>>>>>> Fukuiraptor, Megaraptor and Neovenator), a clade within Carcharodontosauria (figure 3a).
>>>>>>>>>
>>>>>>>>> How can there be so many "most parsimonous trees"? Is there a quantitative measure of parsimony,
>>>>>>>>> or is it just an order ranking, with 6320 maximal trees, none of which is comparable to any of the others?
>>>>>>>
>>>>>>>> The quantitative measure is the number of steps (changes) required to
>>>>>>>> fit the data to the tree. Not sure what you mean by "maximal" here;
>>>>>>>
>>>>>>> Not all orderings are linear. Take the proper subsets of the set {1, 2, 3} with the order "is a subset of";
>>>>>>> by proper subset I mean a subset missing at least one element. The maximal proper subsets
>>>>>>> are {1,2}, {2,3} and {1,3}. They are maximal, because there is no proper subset that contains any
>>>>>>> of them without being that subset itself.
>>>>>
>>>>>> This has nothing at all to do with the meaning of "most parsimonious
>>>>>> trees". Nor does "maximal".
>>>>>
>>>>> Thanks for clearing up that much. But it seems that biologists sometimes work with an
>>>>> impoverished model where you try too hard to shoehorn things into linear order,
>>>>> and even to assign numbers to them. [Mohs scale of hardness was a lot like that,
>>>>> until some physics-savvy person fixed it up.]
>>>>>
>>>>> On the other hand, in a phylogenetic tree, EVERY SPECIES is incomparable to every
>>>>> other species. ALL species are at maximal points, at the tips of all the branches.
>>>>> I could kick myself for not thinking of this example right off the bat.
>>>
>>>> I really have no idea what you're talking about here,
>>>
>>> Explaining the difference between "maximal" and "maximum = above everything else"
>
>> Still have no idea what you mean.
>
> Oh, really? you are still unable to comprehend the difference, are you?

I still don't know, based on your example, what the difference it. The
tips of the trees are indeed above everything else.

> >What does this have to do with
>> parsimonious trees?
>
> We could rank them in a nonlinear order, "X is more parsimonious than Y", and
> then have a bunch of maximal trees. The benefit might be that some maximal
> ones may be found to be parsimonious for weightier reasons than some others.

Still no idea what you're saying. What does "maximal" mean here? If it
means "most parsimonious", just use that word. The most parsimonious
trees are those that require the least number of steps to explain the
data. In the current study, there are 6320 different trees, all of the
same length, that have the least number of steps.

I'm not sure you mean the same thing by "parsimonious" as I do, or what
that has to do with "maximal".

>>>> but it seems as
>>>> far as I can discern to have nothing at all to do with phylogenetics.
>>>
>>> It has everything to do with phylogenetic classification, which is based on
>>> the prohibition against putting taxa at the nodes of phylogenetic trees.
>
>> So nothing to do with phylogenetics, then.
>
> Now you know why I make a distinction between cladistics and cladistic
> classification,

Everyone makes that distinction, not just you. But I don't understand
your reason, though it seems to have something to do with not liking
cladistic classification. This is all irrelevant to what we were discussing.

> and don't allow luster to get transferred from one to the other.
> The classification adds no information to that obtainable from the phylogenetic tree
> that is the basis for it in the first place. And most people have a much
> easier time understanding the trees one sees on the internet than the classifications.

All true. But what does this have to do with what we were talking about?
Is this you wandering off down a rabbit hole?