A paper on a pocket-sized winged “dinosaur” is being retracted after new unpublished findings cast doubt on the authors’ characterizations of their discovery.
The study, “Hummingbird-sized dinosaur from the Cretaceous period of Myanmar,” was published in Nature on March 11, 2020. Many news outlets, including the New York Times, Newsweek and National Geographic, picked up on the findings.
Then on March 18, Zhiheng Li of the Chinese Academy of Sciences, along with co-authors, posted a comment on bioRxiv about the study, casting doubt on whether the amber-encased specimen was in fact a dinosaur or avian species.
Nature updated the study with an expression of concern on May 29, which said:
Readers are alerted that doubts have been expressed about the phylogenetic placement of the fossil described in this paper. We are investigating and appropriate editorial action will be taken once this matter is resolved.
The study’s corresponding author, Jingmai O’Connor, also of the Chinese Academy of Sciences, posted a reply to Li’s comment with her collaborators on June 14. It said that there wasn’t adequate evidence for Li’s claim that the fossil was a type of lizard.
The study is being retracted today. The retraction notice reads:
We, the authors, are retracting this Article to prevent inaccurate information from remaining in the literature. Although the description of Oculudentavis khaungraae remains accurate, a new unpublished specimen casts doubts upon our hypothesis regarding the phylogenetic position of HPG-15-3.
This is one of two articles being retracted from Nature today. The other, “Exploring the quantum speed limit with computer games,” was published in 2016 and claimed to show that optimization problem games played by human participants could outperform the “most prominent established numerical methods.” Other readers, however, discovered “an error in our computer code that means the quantitative results reported are not valid,” according to the retraction notice.
Lead author doesn’t agree with retraction
O’Connor, who has made other recent high-profile discoveries and contributed to 16 publications in 2019, according to her personal biography, wrote to Retraction Watch:
I don’t agree with the retraction but there is no point in fighting it, so we all signed it.
I cannot say why Nature chose to retract, I cannot hypothesize on their inner machinations. I am fairly certain we were wrong in our identification but as we have demonstrated in a Matter’s Arising reply…the specimen cannot be unequivocally identified as either a bird or a squamate without more material (which has come to light but is as yet unpublished and effectively does not exist to science yet).
The author went on:
It is also not that unusual for paleontologists to misidentify things and for new information to correct previous hypotheses. However, Nature chose not to publish the Matter’s Arising and instead retracts our paper – they must have their reasons. It’s unfortunate because this way science can’t simply correct itself (as it is supposed to do) and on top of that, according to the [International Code of Zoological Nomenclature] ICZN the nomenclatural acts are valid whether retracted or not creating a complex grey area. The paper is retracted, yet will continue to be cited. So science will correct itself and cite the paper even though it is retracted making the retraction pointless.
Magda Skipper, the editor in chief of Nature, told Retraction Watch through a spokesperson:
We are committed to maintaining the integrity of the scientific record and carefully investigate all concerns that are raised with us in conjunction with the authors and independent reviewers as appropriate. These two retractions [the dinosaur paper and the computing article retracted the same day] reflect the scientific process in action: a process of constant appraisal by the scientific community which can incorporate the latest evidence as it becomes available.
O’Connor determined the classification of the bird-like creature, according to the work’s contributions section. She was also quoted in a New York Times article published the same day as the study, which explained that some researchers were choosing to boycott amber fossils — such as the one described in the Nature paper — from Myanmar due to the country’s genocide of the Rohingya Muslim minority. O’Connor opposed the boycott.
O’Connor added in a follow up email:
I agree we were wrong and an unpublished specimen will eventually prove it, but I disagree that a retraction was the best way to handle the situation.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
I agree with O’Connor. A retraction of an article is not the best way to handle this. Retractions should be for fraud, plagiarism, or an error in analysis, not a scientific work superseded by further evidence.
You really believe that Nature wanted to trtracted this paper because of a new unpublished specimen? There are quite a few independent studies re-studying this original specimen, which all indicated that the specimen was nothing but a lepidosaur!
From what I gather, the original interpretation was challenged but the authors nonetheless stood by their work until new evidence appeared that demonstrated to their satisfaction that they were wrong, at which point they acquiesced.
To me, at least, this sounds like the regular process of scientific study and debate, so I still don’t understand the reason for the retraction.
I’m reminded of one of my professors who had worked on the excavation of Hasanlu in Iran. They originally misinterpreted an Urartian wall surrounding the site because they did not recognize the architectural style, and they had published the misinterpretation in several articles. Later publications corrected the mistake, but the previous articles were not retracted, nor does it seem to me that they should have been.
I do strongly agree with your comment. Only from error we can learn – it sparks further discussion and eventually leads to the modification of approaches, theories and paradigms. If all scientific works and theories which have been proven wrong in the history of science would have simply vanished shortly after they were challenged, how could progress ever have happened? How short would the list of ancient scholars be we know today? I think that the notion that science should produce unassailable facts is dangerous to the scientific system.
“I agree we were wrong and an unpublished specimen will eventually prove it, but I disagree that a retraction was the best way to handle the situation.”
It seems like Jingmai O’Connor, the corresponding author, wants to have her cake (keep the Nature paper) and eat it too (acknowledging the phylogenetic error).
If new data show that the title-defining main conclusion of the paper is incorrect, the paper should be retracted. Authors should republish including their new data.
I addressed your comment further down, but should have replied directly.
As O’Connor says, misidentifications happen all the time. Those papers aren’t retracted, even if it turns out your fancy pterosaur crest is actually a bit of turtle shell. You (or another team) publish a new paper (or a correction) pointing out the error and so it goes on.
If we knew several months ago what we know today, this paper would not have been accepted by Nature. Misidentifications happen, but it was the claim of “hummingbird-sized dinasour” that made this a Nature paper. Once this claim is disproven, the rest of the descriptions, naming etc. are not of interest to non-specialists, so they should be republished in a specialty journal.
I disagree with the idea that the papers like this, in which there is no obvious fraud, should not be retracted. If an extraordinary claim is proven to be an artefact of a simple verifiable error such as sample mix-up, computational errors, lack of proper controls, or gross misinterpretation, such claims should also be retracted from peer-reviewed literature. Let’s remember the saga of xenotropic murine leukemia virus-related virus in chronic fatigue syndrome. This association was spurious because of contamination (not fraud, to my best knowledge), and retracting the original study was the right move.
“Once this claim is disproven, the rest of the descriptions, naming etc. are not of interest to non-specialists, so they should be republished in a specialty journal.”
That’s not how acts of nomenclature work though. It doesn’t matter that non-specialists won’t care about the section of the paper that names the specimen and provides the diagnosis- which is accurate, and doesn’t (to the best of my knowledge) contain mistakes, for specialists that’s the part that will forever be cited whenever people discuss this taxon (and quote the name). The first paper is the important one. Its why there are so many arguments about which names have priority, and what citation to use, it gets very complicated at times.
There are plenty of papers published with misidentifications in specialty journals too. They aren’t retracted. Plenty of papers are published with bad phylogenetic analyses. They aren’t retracted either. The proper course of action in this case is to publish a corrigendum or “matters arising” communication pointing out the error. Nature papers don’t get to have special rules just because non-specialists might read them. There’s nothing wrong in this little area of science with publishing a note saying “please ignore everything after page 2 of this paper”, either as a letter to the original journal, or as part of a followup article in the specialist literature.
The question of whether it should have passed peer review, or been submitted in the state it was (there are rumours that the authors’ had been told that it was a lizard before the paper appeared) is another matter however. But once its been published it doesn’t get to be disappeared.
>If we knew several months ago what we know today, this paper would not have been accepted by Nature.
The majority of scientific papers wouldn’t get published in nature, knowing what we do today. Scientific knowledge is always expanding.
As I reported to O’Connor et al., they need to expand their taxon list to include Cosesaurus, a lepidosaur that nests at the base of the Pterosauria. In an online analysis that includes over 1700 taxa, Oculudentavis nested with Cosesaurus, which was originally and mistakenly considered a Middle Triassic bird ancestor. That was corrected twenty years ago. Details here: https://pterosaurheresies.wordpress.com/2020/06/17/oculudentavis-reply-bird-lizard-or-option-3/
Oracle:
“If new data show that the title-defining main conclusion of the paper is incorrect, the paper should be retracted. Authors should republish including their new data.”
That’s not how systematics works. For specialists the important part is the erection of the name and the reasoning behind it, not necessarily the phylogenetic analysis. I can think of plenty of papers in the last few years where something was described but misidentified. Any phylogenetic analysis in the paper isn’t useful, but the description is, and that’s the part that will be cited. (And that’s before we get into issues of synonymy. You don’t retract a paper because someone already described your animal either).
What this means for the name is uncertain too- according to Henry Gee the publication no longer exists, so this thing doesn’t have a name and the name could be reused for something else (normally this is not allowed), but I suspect it’s more complicated than that, and will probably need the ICZN to weigh in.
You cannot retract a validly published name and the description meets the criteria as specified in Art 8 of the code. Its wrongful phylogenetic placement does not come into play as that is a completely different hypothesis: a) is it a new species? b) belongs to this clade. Even if it was a synonym, once published in accordance to the code, the name exists and can only be synonymized via a pub and not via retraction, because synonymization is also a hypothesis that can change in the future. Only if the article contained a disclaimer (Art 8.2) to the effect of defining the publication as not issued for permanent scientific record (i.e. temporary names like Aus “green Sp”) then the name wouldn’t exist. But as this was not the case the name exists and is valid. But now, thanks to Nature’s profound ignorance of taxonomy, we have a mess where the “public” (pretty much everybody who thinks you can retract and be done with it, including both CAS teams) assumes the name has disappeared and that the fossil is unnamed and can be rebaptized. Both assertions are wrong.