A Blog by

The Changing Science of Just-About-Birds and Not-Quite-Birds

When people say that scientists are always changing their minds, it’s usually meant as a slight. How can anyone trust conclusions that are so prone to revision? But the fluctuating nature of science is a feature not a bug. It means that our knowledge of the world is constantly being updated in the face of new evidence.

For example, scientists from the University of Bristol recently showed that the evolutionary relationships between different dinosaurs are continuously changing in the light of new fossils. It’s a bit of an etch-a-sketch science—no sooner are family trees drawn before they’re re-drawn again. Even well-known transitions are prone to big shake-ups.

Consider the origin of birds. There is now overwhelming evidence that birds evolved from small predatory dinosaurs. Hundreds of stunning fossils illustrate the transitions from dino-fuzz to flight feathers and from grasping arms to flapping wings. The avialans (all birds, living and extinct) fit within a group of dinosaurs called the Paraves, which also includes dromaeosaurids (sickle-clawed predators like Velociraptor and Deinonychus) and troodontids (large-brained predators like Troodon).

But which of these creatures were the first birds, and which specific group of paravians were their closest relatives? That’s still the subject of heavy debate.

The famous Archaeopteryx, with its winged arms, clawed hands, toothed jaws, and long bony tail, was one of the first fossils to suggest a link between birds and other dinosaurs. Since its discovery in 1861, it has been widely heralded as one of the earliest birds (avialans). But two years ago, Chinese palaeontologist Xing Xu cast its pivotal position into doubt.

Xu, who has discovered more feathered dinosaurs than anyone else, had just found a new species called Xiaotingia. By comparing this creature with Archaeopteryx and other related species, Xu created a family tree that put Archaeopteryx outside the avialans (see diagram below). Instead, it sat next to the dromaeosaurids and troodontids, together with relatives like Xiaotingia and Anchiornis. The first bird was no bird at all.

If Xu was right, the implications would have been profound. For a start, Archaeopteryx clearly had wings with what looked like flight feathers. If it was a bird, then flapping flight probably evolved once in the lineage leading to modern birds. If it wasn’t a bird, then flight evolved twice—once in Archaeopteryx’s group and again in modern birds. (Or, alternatively, the dromaeosaurids and troodontids all lost the ability by reducing their wings.)

But was Xu right? He himself said that his revised family tree only had “tentative statistical support”. Later in 2011, Mike Lee from the South Australian Museum showed that a different tree, built with different methods, reinstated Archaeopteryx as a bird.

Pascal Godefroit from the Royal Belgian Institute of Natural Science entered the debate this January, with a new dinosaur called Eosinopteryx. There were many echoes of Xu’s study: Godefroit also placed the paravians on a family tree, also concluded that Archaeopteryx was no bird, and also said that the result was statistically weak.

Now, his team is back with yet another maybe-a-bird fossil and yet another revised paravian family tree… and this one contradicts his own earlier conclusions (see diagram below). It once again restores Archaeopteryx as an early bird, while settling related species on different perches.

Simplified Paraves family trees over time.
Simplified Paraves family trees over time.

The fossil in question is called Aurornis xui, and it lived 160 million years ago in northeastern China. Aurornis is Latin for “dawn bird” but the animal’s species name honours Xing Xu. For Godefroit, the decision was an easy one. “Xu has completely revolutionised our vision of dinosaur biology and evolution,” he says. “Although he is still very young and extremely modest, he is probably the most important living vertebrate palaeontologist.” Modest is right—in typical form, Xu says that he’s treating the “great honour” as recognition of the contributions that Chinese palaeontologists have made as a group.

Aurornis is beautifully preserved and has much the same features as Xiaotingia, Anchiornis, Eosinopteryx, and Archaeopteryx. Godefroit’s team believe that it’s a distinct species based on a few characteristic features, although Steven Brusatte from the University of Edinburgh says, “I can’t shake a nagging suspicion that some of these may be juveniles and adults of the same [species]. The only way to know for sure will be to check the internal structures of the specimens’ bones.

Aurornis skeleton. Credit: Thierry Hubin/IRSNB
Aurornis skeleton. Credit: Thierry Hubin/IRSNB

Regardless, it’s clear that these animals all lived at roughly the same time, in the same place—the Tiaojishan Formation in northeastern China. This treasure trove of fossils was once home to an entire flock of not-quite-birds and just-about-birds that all and looked rather similar and lived near each other. It’s no wonder that their evolutionary relationships are difficult to entangle.

Godefroit’s team didn’t want to create another weak family tree, so they started from scratch. Italian scientist Andrea Cau scoured the literature and compiled data on 101 species of dinosaurs and birds, scoring each skeleton according to almost 1,000 characteristics. “It’s very impressive,” says Lee. “They considered more than twice as much anatomical information as even the best previous analyses.”

The results put Archaeopteryx back in its traditional roost as an avialan, but no longer as the earliest one. That honour goes to Aurornis itself. It’s now the most primitive bird, followed by Anchiornis, Archaeopteryx and Xiaotingia in that order. (Eosinopteryx, perhaps surprisingly, emerges as a very early paravian that preceded the groups I’ve already mentioned.)

“If Aurornis is the most primitive bird, then it is a huge discovery,” says Brusatte, “but I am not convinced that this paper resolves the early history of birds.” Xu agrees. He says the results deserved to be taken seriously, but adds that several parts of the family tree are inconsistent with earlier work. He’s not just talking about Archaeopteryx. For example, among the other dinosaurs, the troodontids emerge as the group that’s closest to the birds. And the biggest surprise isn’t even mentioned in the paper! Here it is:

From Godefroit et al, 2013. Nature.
From Godefroit et al, 2013. Nature.

Balaur is a Romanian dinosaur that was discovered in 2010. Its discoverers (including Brusatte) concluded that it was a close relative of Velociraptor; the two dinosaurs look very similar, although Balaur is stockier of build and has two sickle-claws on each foot rather than one. But Godefroit’s family tree has it perching firmly within the birds! It’s not alone; other supposedly non-bird species like Rahonavis and Shenzouraptor have been similarly displaced.

That’s a huge shake-up, and puzzling one given these animals’ appearances. “I am suspicious,” says Brusatte. “It is certainly possible that Balaur is a bird but I would be surprised if this is the case.” Godefroit says, “It was also a big surprise for us!” One of their team—Gareth Dyke—even went to check the original specimen to make sure that they hadn’t made any obvious mistakes.

Godefroit plans to publish a separate paper to address this discrepancy. For the moment, it raises some scepticism that has spread to other parts of the tree. “It suggests that some other results, such as the status of Archaeopteryx, Anchiornis, Aurornis and Xiaotingia need further evaluation,” says Xu.

“I am confident that specimens from the Tiaojishan Formation will end up solving this debate somewhere down the line,” says Brusatte, “and I predict that something that all workers agree is a true bird will eventually be found there. Maybe Aurornis is that bird; maybe not.”

As more evidence comes in, the etch-a-sketch will almost certainly shake again, and new versions of the paravian family tree will emerge. “I hope so,” says Godefroit. “Otherwise, palaeontology will become as dull as dishwater!”

Reference: Godefroit, Cau, Yu, Escuillie, Wenhao & Dyke. 2013. A Jurassic avialan dinosaur from China resolves the early phylogenetic history of birds. Nature http://dx.doi.org/10.1038/nature1216

22 thoughts on “The Changing Science of Just-About-Birds and Not-Quite-Birds

  1. I hadn’t considered that some of these animals might be ontogenetic morphs of each other, but I suppose it’s a real possibility, and the fact that so many dromaeosaurs end up in Avialae (especially Balaur) is cause for concern. Still, I always like reading about new dinobirds!

    And isn’t it usually “dull as DISHwater?”

  2. The placement of Rahonavis outside Unenlagiids and more derived than Anchiornis and Xiaotingia looks similar to the one recovered in Agnolin and Novas’ Avian Ancestors (2013). If Shenzouraptor (Shenzhouraptor?) is synonym of Jeholornis, this result will also agree (more or less) with Agnolin and Novas (2013). The placement of Balaur, on the other hand, is not evaluated in their work, and I’m not sure if it was recovered in a similar position before. I don’t have this new paper, thus a final question: are other unenlagians included in this analysis (Unenlagia, Buitreraptor, Austroraptor)? because they were placed in different places of the maniraptorans tree in different studies, and would be interesting if they were here included…. great post, as always!!!! (but you should ask for a bigger “comment box”, it is very small IMHO).

  3. Yes, this is not settled by any means. It is an interesting new bit of information, but this is hardly a settled matter.

    And that is perfectly fine! What we are finding is that animals like Archaeopteryx and Anchiornis and Aurornis are close to the common ancestry of dromaeosaurids, troodontids, and birds, regardless of the particular topology.

  4. And to show how unsettled this is, you did not include the Agnolin & Novas (2013) phylogeny, which had troodontids, classic dromaeosaurids, microraptors (including Sinornithosaurus), unenlagiids, Anchiornis, Archaeopteryx, and Rahonavis as successively closer groups to unquestioned birds: that is, dromaeosaurids as typically recognized were paraphyletic, but closer to birds than to troodontids.


  5. Very nice job on this story, Ed. I covered it for Science online but wasn’t able to go into as much detail! Some of my sources did raise issues about the authenticity of fossils from dealers and collectors, however.

    [Thanks, Michael! And since you’re too modest to link to your piece, I’m going to because the provenance point is important and well-told. http://news.sciencemag.org/sciencenow/2013/05/earliest-bird-claim-ruffles-feat.html – Ed]

  6. “…he is probably the most important living vertebrate palaeontologist…”

    Aren’t all palaeontologists vertebrates?

  7. “It’s not alone; other supposedly non-bird species like Rahonavis and Shenzouraptor have been similarly displaced.”

    I thought Shenzouraptor was a synonym of Jeholornis, which has always been considered a bird?

  8. I agree with Mr. Balter.,so many “new” fossil finds from one area–many proven to have been tampered with.Who then tries to qualify the dealers claims? Others from the same area and two of the people that wrote the paper. I cry “foul!”

  9. Very cool and interesting new find! I’m really surprised that so many of the articles and the paper itself keep saying that this “settles” things, or in this case that it “resolves” the phylogenetic history. No doubt people will disagree, and the next find will change it again!
    An aside: the link to the paper didn’t work for me, not sure if anyone else has had this problem?

  10. OK, so there’s a huge grey area, but what are the features that people are now looking at to distinguish putative avialans from the pack?

  11. That (its being a bird and all) would explain how Balaur came to be living on a remote Romanian island otherwise populated by relics from the Jurassic.

  12. You know what this means in a big-perspective, meta way?

    They can’t resolve the paraves tree with good statistical results because all the more primitive ones are too similar to each other and Troodontids, Dromeosaurids and Avialans all descended from a cluster of small species that had four wings, teeth, sickle claws, long tails and were glieders and could unskillfully fly. Bigger dromeosaurids and troodontids lost their ability for unskillful flight and later Avialans lost their sickle claws, leg-wings and teeth (in that order). Every new species switches the relationships back and forth because the methods of phylogenetic analysis by morphology of fossils is insufficient to resolve the true relationship: All those small primitive species metioned above, but also things like microraptor, are more closely related to each other and more morphological similar to each other than any of them is to the later forms (Avialans, Dromeosaurids, Troodontids, Balaur…). Current phylogenetic methods won’t recover that real relationship because they are biased towards recovering phylogenetic trees with successive branching of paired clades. With that in mind, of course 1) things like Balaur will end up all over the tree, because they are so weird and different from all other input species. 2) Everytime new characters and species are added, the branching pattern at the base of the tree shifts (again, because the method is biased from resolving the true basal relationship.)

    To reject my hypothesis, you must:
    1) build a tree by a method that is not biased towards resolving successive pairwise branching at the base;
    2) Statistically test the resulting tree against (all) trees that resolve basal pairwise successive branching.

    Have fun!

  13. Hi all,
    thanks for the comments on our Aurornis paper.
    Most of you raised important questions, and I hope future iterations of the analysis will solve some of the controversial points.
    Some clarifications to a few questions:
    The main aim of the analysis was to include as much characters as possible among those mentioned by theropod workers as relevant in bird origins. Since bird origins is linked to coelurosaur relationships, it was necessary to include also a large amount of coelurosaur features. Note that many other authors did this in the recent past: I’ve simply tried to fuse all the different data sets into one: IF my analysis has resulted useful in some way it is by standing on the shoulders of Giant theropod palaeontologists that produced most of the data I used! If it results wrong, all the errors are mine.
    The complete set of species in the analysis (listed in the supplementary material of the paper: the complete topology of the tree is more detailed than the one showed in the main paper) includes all unenlagiines.
    “Shenzouraptor” was a typo of “Shenzhouraptor”, the name we consider has priority over Jeholornis (it’s documented that the former was published two days before the latter: the important thing is that we are referring to the same animal as Jeholornis).
    The position of Balaur, although unexpected by most, is congruent with the many bird-like features of its skeleton, and we also checked them directly on the specimen: a study discussing this in detail is going to be completed very soon.

  14. “The avialans (all birds, living and extinct)”

    Oh. Another strange change of meaning. “Birds” has been always equal to Aves, but now several scientists and science writers are equating “birds” with the bigger clade Avialae. Why? Now, in addition to non-avian dinosaurs we have non-avian birds (and avian birds). Unnecessary and confusing.

    1. This is a very good point. Paleontologists are using the term Avialae in two different ways, and don’t seem to agree on its definition. It has been suggested that researchers use the definition that is most convenient to the conclusions they want to reach, but that might not be fair.

  15. When he erected Avialae, Gauthier (1986) stated: “Because of feathers and the presumed ability to fly. Archaeopteryx has always been considered a bird. This informal usage has been maintained above, and use of the informal term “bird” for this taxon will be continued in the following discussion. In a formal sense, however, “birds” and Aves will not be synonymous.” Gauthier restricted Aves to the modern birds only, and used Avialae for modern birds + all bird-like theropods more closely related to modern birds than dromaeosaurids and troodontids. Thus, Avialae equates to “birds” since its original definition, and the equation “Avialae” = “all birds, living and extinct” is correct.

    1. “Thus, Avialae equates to “birds” since its original definition”

      Why? Sorry, I don’t get it. Avialae includes Aves (undoubtedly birds), Archaeopteryx (traditionally a bird), and also bird-like theropods, more basal than Archaeopteryx, possibly non volant, and not necessarily “birds”.

  16. You seem interested in what the vernacular term ‘bird’ means, but this is not much important for taxonomy, what matters are the clade names. ‘Bird’ is a pre-darwinian vernacular term, a ‘typological’ name anchored on the birds we see every day. Such concept is not useful in evolutionary palaeontology, where ‘kinds’ evolve. Gauthier coined ‘branch-based’ clades, names anchored on evolutionary relationships, not on typologies, and Avialae is the branch-based clades referring to all theropods closer to modern birds than deinonychosaurs. We to consider that clade the taxonomic equivalent of the vernacular ‘bird’, but the two names are not referring to the same concept. You can restrict ‘bird’ to a less inclusive clade, if you wish, but originally, Avialae was used for the clade of birds as known in 1986 (modern birds and a few Mesozoic fossils). Nevertheless, Archaeopteryx is not relevant in Avialae definition, since Gauthier anchored it on modern birds and used deinonychosaurs as external specifier: everything closer to modern birds than deinonychosaurs results thus an avialan.
    This is the original and simplest meaning of Avialae. Animals as Anchiornis and Aurornis, once found to be closer to modern birds than deinonychosaurs, are avialans by definition. Unfortunately, after the original definition (‘branch-based’) of Avialae, other definitions have been proposed (for example avialans as theropods with ‘wings’ or ‘the Archaeopteryx node’), and this created ambiguity on the meaning of ‘avialans’. I have no problem in restrincting ‘avialans’ to a less inclusive clade, but – if possible – prefer to follow the original meaning. As far as I known, no palaeontologist has ever claimed to consider ‘avialans’ distinct from ‘birds’, although, it must be remembered, the former is a taxon-name, the latter a vernacular name, and thus they pertain to different – partially overlapping – concepts. I don’t care much on discussion on what ‘bird’ means, since it’s a linguistic and not taxonomic issue: ‘Avialae’ is a taxonomic term, and this matters in palaeontology. Hope we could reach a consensus on the use of ‘Avialae’, I agree there is a bit of confusion in its application and definition. We just followed a definition as closer as possible to the original meaning of the name.

    1. “You seem interested in what the vernacular term ‘bird’ means, but this is not much important for taxonomy”

      Of course! It’s not important for taxonomy. It’s important for the popularization, teaching and understanding of science.
      It’s also important for translation to other languages.

  17. I always look forward to new dino-bird discoveries in general & finding out what they can tell us about dino-bird evolution in particular. However, I have to side w/Chiappe on this 1 (Quoting Ehrenberg: “Still, the fossil’s forelimb is much shorter than that of true birds, Chiappe says. That detail, he says, along with the fossil’s longish reptilian tail and several skull features, suggest it is a dinosaur”). I’m also concerned about what Brusatte & Xing said (as you mentioned in this article).

  18. The reason people think of “scientists are always changing their minds” as a slight isn’t due to areas of constant refinement, nor even areas of great uncertainty with constant attempts at discerning the truth; rather, it’s from seeing too many things stated as if they were facts, that there would not need to be another change.

    Case in point here: “There is now overwhelming evidence that birds evolved from small predatory dinosaurs. Hundreds of stunning fossils illustrate the transitions…” The fossil record has often been called “spotty,” and that of birds has been notoriously spotty. Much of the current view depends vitally on several fossils found in China and, as noted, have a bit of a cloud over them. If a farmer in Arkansas produced a fossil that overturned a popular theory, but hadn’t documented its removal, wouldn’t the mere lack of solid provenance be enough for it to be dismissed? Likewise, when fossils start popping up from the sticks of China that seem tailored to meet the “if only’s” of the previous one, it might not be wise to give them places of honor.

    I’m glad to see that different possibilities within the favored view are given air, but similar possibilities could easily show that modern birds have an origin that doesn’t include these dinosaurian birds. Here we go: “…Archaeopteryx clearly had wings with what looked like flight feathers. If it was a bird, then flapping flight probably evolved once in the lineage leading to modern birds. If it wasn’t a bird, then flight evolved twice—once in Archaeopteryx’s group and again in modern birds. (Or, alternatively, the dromaeosaurids and troodontids all lost the ability by reducing their wings.)” Then add this: “…it’s clear that these animals all lived at roughly the same time, in the same place… an entire flock of not-quite-birds and just-about-birds that all and looked rather similar…” Archie, of course, was found in Germany but is dated about the same time and is quite similar. So feathered flight might have evolved twice, AND a number of dino-to-bird transitional might be (re-)assigned to a group of dinosaurs that had been feathered but gradually lost them.

    Now consider that these phylogenies are constructed with techniques that are meant to eliminate subjective judgments — but in so doing, can give equal weight to features which may not be significant, or are likely to be convergent or parallel, such as the hypertrophied “killer claw” which may have a totally different function on small feathered creatures and those on large, running carnivores. They also don’t take into consideration the dates assigned to the fossils, or the pattern of appearance of features especially related to flight. (BTW, The difficulty with Rahonavis may be due to the possibility that its wings actually belonged to another critter, as they weren’t joined to the rest and lower parts of a bird were also found nearby. That’s the only specimen, so the question can’t be resolved.) So some “basal” or “ancestral” fossils (such as Velociraptor) come long after this four-winged flock, and have to be considered descendants of a population that didn’t get fossilized. With “absence of evidence” so built into a construct that includes possible parallel development of feathered flight, what surprises might yet change what is now presented as a sure thing?

    Consider also that footprints almost identical to those of modern birds have been found in Triassic rocks, over 55 million years older than the Chinese flutterers. The record of later birds is still spotty and features oddballs like the pseudontorniths and is heavy on the sea and shorebirds, until the end of the Mesozoic, and then the stars of the show are huge flightless “terror birds” until suddenly recognizable modern forms appear, either rapidly or from unfossilized ancestors going back into the Cretaceous: “Mass survivals” by Penny and Phillips, _Nature_ v. 446, 29 March 2007: “…an evolutionary tree of more than 4,500 mammals, and conclude that more than 40 lineages … survived from the Cretaceous, some 100 million to 85 million years ago… This is paralleled by Brown and colleagues’ analyses for birds, just published in _Biology Letters_: they claim that more than 40 avian lineages have likewise survived from before the extinctions at the Cretaceous/Tertiary (K/T) boundary 65 Myr ago. …”

    BTW, cute typo: “It’s no wonder that their evolutionary relationships are difficult to entangle.” I’m pretty sure that last word was meant to be “untangle,” as the various trees show how easily the relationships can be tangled.

  19. OH… full disclosure: I’m a creationist. Hope you don’t mind my talking evolution from such a different perspective; I try to keep my evolutionary thinking cap on in these cases.

Leave a Reply

Your email address will not be published. Required fields are marked *