An OpenLetter to the OpenWings Project

According to their website:
“The goal of the OpenWings Project ( is to understand the evolutionary history of and evolutionary relationships among birds.”

“One of our goals in this project is to collect genomic data from DNA samples that have an associated voucher specimen in a research collection.”

Figure 2. Newell's shearwater (Puffinus newelli) in vivo.

Figure 1 Newell’s shearwater (Puffinus newelli) in vivo. We’ll examine the skeleton of this bird in an upcoming blogpost. 

“The fundamental and missing piece of this otherwise powerful comparative biology toolkit is an accurate and complete avian phylogeny. The overarching goal of the OpenWings Project is to fill this gap by producing: a complete phylogeny for all 10,560 bird species that will provide a unifying framework for understanding the origins and maintenance of avian diversity… as well as serving as a case study of the benefits and challenges of sampling all species in a major clade.”

Suggestion #1:
Start with a dozen diverse birds. IF the genotypes and phenotypes produce identical tree topologies, double that number and test two dozen birds. IF those produce identical tree topologies, test four dozen birds. Etc. Etc. You’ll get to 10,560 at that rate in ten steps with confidence that your results have been validated at every step.

If at any point the genotypes and phenotypes don’t produce identical tree topologies, review your paradigms and hypothesis. There is something wrong if they don’t match. In my testing, birds genes do not recover the same tree topologies as bird traits. Don’t waste your time and money testing 10,560 bird genes only to find they don’t deliver the same tree topology as bird traits.

Suggestion #2:
Since no large scale genomic and phenomic studies of birds have ever matched, just study phenotypes. Then you can include fossils and you won’t have to exclude taxa critical to understanding the phylogeny of birds. Click here for a starter list of taxa.

I would have contacted the OpenWings Project directly,
but (at present) they don’t provide access except through apps (like Twitter) I don’t have or want.

We looked at another DNA analysis of birds
by Prum et al. 2015 here and here and found it matched dissimilar taxa while separating similar taxa. So beware of DNA. It can only be validated with trait analysis and too often it produces odd results needing odd explanations.


13 thoughts on “An OpenLetter to the OpenWings Project

  1. “Since no large scale genomic and phenomic studies of birds have ever matched…”

    There’s only been one large scale phenomic phylogenetic analysis of birds (Livezey and Zusi, 2007), and it was flawed by inaccurate codings (Mayr, 2007).

    You realize your insistence that molecular analyses cannot resolve suprageneric relationships means you are calling what thousands of biologists do for a living a pseudoscience? Enough biologists to justify the existence of entire journals like the Journal of Molecular Phylogeny. Do you honestly think your single analysis of a few hundred characters (or anyone’s analysis of a few hundred characters) is good enough to dismiss an entire field of biology?

  2. Yes.
    Look at the DNA analyses and tell me which one
    1. delivers a gradual accumulation of derived traits for every series of taxa (see prior posts on this subject for examples of bad pairs)
    2. employs fossil taxa

    Will we ever find turtles nesting with archosaurs based on traits? We have not, so far.

    I don’t know why DNA does not work in phylogenetic analysis. But it doesn’t. My open letter request is a reasonable test that starts with a dozen taxa, so it is not time/money/talent expensive. Ultimately a trait test is needed in any case to validate a DNA test.

    • “delivers a gradual accumulation of derived traits for every series of taxa”

      Every single one does, for the molecular traits the study analyzes (which are orders of magnitude more numerous than your few hundred traits). In turn, your LRT does not show a gradual accumulation of molecular traits. So by your own criteria, you should reject the LRT. It’s the DNA and RNA that are actually evolving after all, and which cause the phenotype to change. They’re looking at the basic units of change, you’re looking at the messy developmental outcome.

      “I don’t know why DNA does not work in phylogenetic analysis. But it doesn’t.”

      Another problem- you have no mechanism for explaining why molecular analyses are bad, but there are plenty of reasons morphological analyses could be bad- miscoding, convergence to a similar niche, poorly formed characters, etc..

      Do you realize how bad it looks that you, with no background in molecular biology, just flat out says you know better than an entire field of science? That you uniquely have the solution (score every taxon for 228 characters) and that those thousands of other biologists are just wasting their careers.

      • It’s not that I ‘know better’. It’s that I raise a hand to say, ‘DNA doesn’t always confirm trait analysis.’ Others have also noted that discrepancy, as you know, but are shy about discussing. It’s too soon to have a mechanism for explaining the problems of DNA coding over large phylogenetic differences. First it has to be recognized as a problem. At present it’s enough to say, ‘let’s take another look at our results, because sometimes they are verified by trait topologies, but other times they are not.’ Currently DNA trumps trait analysis in the bird community. As a result we get ‘odd bedfellows’ that cannot be verified by trait analysis, like flamingos and grebes.

        The LRT provides a topology that provides a gradual accumulation of physical traits. It was designed to do so. Traits can be used to tell the story of evolution because they enable organisms to survive, reproduce, etc. On the other hand, DNA is invisible. Moreover, sometimes DNA is expressed in biological traits. Sometimes it is not, as you know. We can only hope and pray (have faith) that DNA tree topologies reveal some sort of order that tells the story of evolution. Too often, unfortunately, something is amiss.

        Like other scientists that came before me, I don’t care how bad it looks for me. Your reaction is historically typical. The results can be replicated by any sufficiently large character list, so long as the taxon lists are similar.

      • Just from your last paragraph- “Your reaction is historically typical.”

        And historically, how typical is it for a lone researcher to claim to have the solution to every vexing problem in a scientific field, using methods no professional in that field agrees are sufficient (if a phylogeneticist said your methods were good, you would have announced it by now) which are universally rejected by peer-review, and turn out to be correct in the end? And to correctly discredit an entire field they have no training in in the process?

        That doesn’t mean your ideas are wrong, but the kind of savant who accomplishes this basically doesn’t exist in history. Even if you stretch the parameters, it’d be a short list like Darwin/Wallace, Galileo and Newton/Laplace which surely you don’t see yourself as.

        “The results can be replicated by any sufficiently large character list, so long as the taxon lists are similar.”

        I’ve actually disproven this twice. First when I corrected all your LRT scores for dinosaurs back on my blog where I did not get the topology you had at the time. Second when I scored five times as many Mesozoic theropods for three times as many characters in my Lori analysis and got very different results from you. It’s honestly just not that easy, or else everyone else would be doing what you do and publishing their results.

  3. I agree with your suggestion #1. Hope they take it up, the results could be very useful and interesting.
    As for #2, doesn’t this really amount to pre-judging the outcome of #1?

  4. Please send your theropod topologies (cladograms and .nex files). I don’t recall having seen them. Then we’ll discuss them.

    And since when are phylogenetic analyses universally rejected? Those are the standard tools by which paleontologists operate.

    If you’re talking about my published analyses, see Peters 2000 where I added taxa to prior analyses. If other workers prefer to ignore that peer-reviewed publication in their ?search? for the origin of pterosaurs, that’s an issue you need to bring up with them. I find it funny/strange.

    re: Lone researcher… I have reported every instance that other workers have confirmed my earlier phylogenetic findings: Eoraptor, Chilesaurus, Lagerpeton, Diandongosuchus.

    I implore you, if the results are wrong, please report the sisters that should not be sisters in the LRT. Step up to the plate and swing (baseball analogy).

    • The Lori paper is in the process of peer review, so I’m not going to send you the character list or matrix, but that’s fair of you object since the data is not publicly available. I sent you the matrix and character corrections for my first example though, which I don’t think you’ve ever commented on here. I wouldn’t be surprised if your basal dinosaur phylogeny changed between then and now, but it’s the codings that are in dispute there, so feel free to say why you agree or don’t with my corrections for characters 1-3-

      “1. Shuvosaurus (Lehane, 2005; Long and Murry, 1995), Agilisaurus (Peng, 1992- table 1, pg. 6) and Hexinlusaurus (He and Cai, 1984- pg. 9, tab. 2) have a skull shorter than half the presacral length. No postcrania are preserved for Massospondylus kaalae (Barrett, 2009), so it cannot be coded.

      2. Shuvosaurus (Lehane, 2005; Long and Murry, 1995), Herrerasaurus (Sereno and Novas, 1993- tables 1 and 2), Agilisaurus (Peng, 1992- table 1, pg. 6) and Hexinlusaurus (He and Cai, 1984- pg. 9, tab. 2, 3) have a skull shorter than the cervical series. As the cervical series of Gracilisuchus was a composite from several specimens, the very close cranial and cervical lengths in Romer’s (1972- fig. 1) reconstruction suggest it should be coded unknown. Marasuchus (Sereno and Arcucci, 1994) does not preserve enough cranial elements to confirm the skull was shorter than the cervicals. Daemonosaurus (Sues et al., 2011), Pampadromaeus (Cabreira et al., 2011- fig. 1) and Massospondylus kaalae (Barrett, 2009) do not preserve enough cervicals to know the skull is shorter.

      3. Arizonasaurus (Nesbitt, 2005- fig. 2) has a skull wider than 1.2 times its height. Effigia (Nesbitt, 2007; after angling quadrate ventrally), Silesaurus (Dzik, 2003- fig. 6), Daemonosaurus (Sues et al., 2011- fig. 1A; based on frontal width), Pampadromaeus (Cabreira et al., 2011; based on parietal width), Pantydraco (Yates, 2003- fig. 2), Heterodontosaurus (Norman et al., 2011- fig. 7), Scelidosaurus (Norman et al., 2004- fig. 15.1) and Hexinlusaurus (He and Cai, 1984- fig. 3) have a skull narrower than 1.2 times its height. The width of Marasuchus’ skull is unreported, besides not being flat like proterochampsids (Bonaparte, 1975). Trialestes’ (Reig, 1963), Tawa’s (Nesbitt et al., 2009) and Saturnalia’s skull width is unreported. The skull remains are too fragmented in Scutellosaurus (Colbert, 1981; Rosenbaum and Padian, 2000) for an estimate. Panphagia (Martinez and Alcober, 2009) lacks preserved temporal bones to estimate this value. “

    • You’re always asking what characters are miscoded (even though that’s not actually your main problem), and I list almost thirty incorrect entries here, so why don’t you respond to them? We don’t have to wait for the Lori theropod differences to have the conversation, since your problems reconstructing/scoring basal dinosaurs are the same kinds of issues.

  5. Over the last few days I’ve been under the spell of a virus while adding and reviewing turtles while not in bed. Finally on the mend. Mickey, I think you’re working with old data. The score changes you suggested have been made who knows how long ago — when appropriate. The skull of Marasuchus can be reconstructed shorter than the cervical series, given known elements. Massospondylus is here a generic chimaera. Best to score skull width based on palatal elements, since cranial elements can be narrower.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.