Field of Science

Bird Evolution - Problems with Science

Hackett, S. J., R. T. Kimball, S. Reddy, R. C. K. Bowie, E. L. Braun, M. J. Braun, J. L. Chojnowski, W. A. Cox, K.-L. Han, J. Harshman, C. J. Huddleston, B. D. Marks, K. J. Miglia, W. S. Moore, F. H. Sheldon, D. W. Steadman, C. C. Witt & T. Yuri. 2008. A phylogenomic study of birds reveals their evolutionary history. Science 320: 1763-1768.

I'm afraid I'm going to be descending into cattiness for a moment later. I apologise in advance for any unwarranted snarkiness.

A paper (citation above) has appeared in today's edition of Science that adds to the ongoing debate on bird phylogeny. It is a fairly significant paper, giving the results of the largest molecular phylogenetic analysis to date for birds. As such, it largely supersedes the previous front-runner, the analysis of Ericson et al. (2006). However, most of the results of Hackett et al. (2008) are largely congruent with those from Ericson et al. (2006). So I'm a little bemused to read Chuck Hagner commenting that "What wasn’t expected was an apparent sister relationship between Passeriformes and Psittaciformes" and expressing surprise that Falconidae should cluster with that clade instead of with Accipitridae, when both these results had been reported in the 2006 paper. What is significant is that both these studies, conducted independently (no shared authors), found such similar results. Both studies (and the earlier Fain & Houde, 2004) found the same six major clades - Palaeognathae (ratites and tinamous), Galloanserae (gamebirds and waterfowl), Metaves (I'll explain in a minute), the "higher water-birds and allies" clade (including 'Ciconiiformes' and 'Pelecaniformes' intermixed), Charadriiformes and the "higher land-birds" (Passeriformes, Piciformes, Coraciiformes and allies).

Bird phylogeny as recovered by Hackett et al. (2008).

Metaves is one of the most controversial groupings of birds to have been proposed in recent years. It first made an appearance in a 2004 paper by Fain and Houde published in the journal Evolution. These authors coined the name Metaves for a clade containing nightjars, swifts and hummingbirds, pigeons and doves, sunbitterns, the kagu, mesites, tropicbirds and the hoatzin that was well-supported in an analysis of the β-fibrinogen gene. This clade was then completely unexpected - perviously, its members had been scattered among an assortment of other bird orders, and the only thing a number of them had previously had in common was that they had always looked a little out of place. In a message forwarded to the DML shortly after the publication of the 2004 paper, Peter Houde commented that the results had been so heterodox that it had been very difficult to get them published. The Charadriiformes, higher land-birds and higher water-birds together formed a clade that Fain and Houde dubbed "Coronaves". The Fain and Houde analysis did resolve the Charadriiformes and higher water-birds, but support was not great.

Ericson et al. (2006) increased the number of genes analysed to five, and again found the Metaves-Coronaves division of Fain & Houde (2004). They were also better able to resolve relationships within the major clades. However, the support for Metaves was completely reliant on the inclusion of the β-fibrinogen gene. If this gene was left out of the analysis, the clade collapsed.

Not too long after Ericson et al. (2006), a counter-sally from the morphological fort appeared in the form of the long-awaited Livezey & Zusi (2007) analysis. Using an awe-inspiring 2954 characters over 150 taxa, this morphological über-analysis bravely fought off the molecular novelties and called stridently for a return to more traditional relationships.

It is into this clash between the molecular data of Ericson et al. (2006) and the morphological data of Livezey & Zusi (2007) that Hackett et al. (2008) make their entrance. Hackett et al. increase the number of analysed genes to 19, and once again recover the much-maligned Metaves. Once again, though, the presence of this clade is dependent solely on the β-fibrinogen gene. The hoatzin abandons the Metaves and attaches itself to the base of the higher water-bird clade. I'm inclined to describe this as unsurprisingly surprising - once again, Opisthocomus is just being a prick. There seems to be a visible trail of respectability here - four years ago, Metaves had to fight its way for recognition in a respectable journal. With the publication of a paper supporting it in Science, it seems to have become a respectable hypothesis.

And that, really, is the source of my irritation. Nature and Science are widely regarded as the ultimate science journals, but it's difficult to escape the observation that many papers that appear in the two are, well, kind of crap. This is not the fault of the contributing authors, but results from the severe space restrictions on articles in these journals. At five very densely-written pages, Hackett et al. is a fairly long paper for Science, but the reader is left frustrated by the need to know about stuff that the authors were evidently forced to leave out. What happens when the analysis parameters change? If a given clade is collapsed, how does this affect the rest of the tree? Some of this is alluded to in the article, but there simply isn't the time for it to be explored properly. And was it lack of space that caused the authors to write clangers such as "flighted tinamous arose within the flightless Struthioniformes", which sounds to be suggesting that tinamous evolved or regained flight independently of other birds, rather than the far more likely scenario that flight was lost multiple times within the ratites? Nature and Science papers have been referred to as "extended abstracts" Sometimes, no matter how extended, an abstract just doesn't substitute for a paper.

Don't get me wrong, this is a very significant paper, and one that will provide a base-line for many future studies. It doesn't completely overthrow previous studies, but in the end that is exactly what is so fantastic about it - not that the results are completely unexpected, but that as more and more data is added, we can say more and more about the picture that has been developing over the past few years.


Ericson, P. G. P., C. L. Anderson, T. Britton, A. Elzanowski, U. S. Johansson, M. Källersjö, J. I. Ohlson, T. J. Parsons, D. Zuccon & G. Mayr. 2006. Diversification of Neoaves: integration of molecular sequence data and fossils. Biology Letters 2 (4): 543-547.

Fain, M. G., & P. Houde. 2004. Parallel radiations in the primary clades of birds. Evolution 58 (11): 2558-2573.

Livezey, B. C., & R. L. Zusi. 2007. Higher-order phylogeny of modern birds (Theropoda, Aves: Neornithes) based on comparative anatomy. II. Analysis and discussion. Zoological Journal of the Linnean Society 149 (1): 1-95.


  1. Using an awe-inspiring 2954 characters

    ...of which a few hundred were parsimony-informative.

    Also, the 150 taxa are impressive, but there are far too few fossils among them. We ain't seen nuthin yet.

  2. ARGH! A few hundred were parsimony-uninformative, I mean. The number of informative characters is still way over 2000, way more than in any previously published morphological cladistic analysis.

  3. John Harshman has a couple of things to say about the media presentation of the paper: here

  4. The media presentation of this is just all wrong. Look at this:

    It made me cringe. Horrible.

    While the paper is far too short, it was really interesting to see the several tables of analysis of their loci. I would've like to see the tree without B-fib

    ~ Nick

  5. Something I've been wrapping my head around is how a single gene can support such a different topology (as in metaves). Perhaps you have some light to shed on this subject?

    On another note, the next Linnaeus' Legacy is approaching. Time to start beating the bushes for links - I have none yet. Did you ever get a cool banner drawn for it?

    ~ Nick

  6. David: Yes, I agree that the Livezey & Zusi dataset needs something of a cleanup (do you know the reference for the critique Gerald Mayr wrote on it in which he noted a number of miscodings?) I'm also eagerly waiting for fossil taxa to be plugged into it - I'd vote for some fairly well-known but potentially significant taxa such as lithornithids, pseudasturids, phorusrhacoids or Limnofregata to start with. Still, Livezey & Zusi is the best large-scale analysis to date, and should be the base for future analyses.

    slybird: Ouch. That hurt. I whacked my head far too hard when I read in the second paragraph "The results of this five-year study are so broad that the scientific names of dozens of birds will have to be changed in biology textbooks and birdwatchers' field guides". Umm, no - Hackett et al. weren't looking at genus-level stuff, so the names will stay exactly the same. And I was reminded again of how unfortunate it is that there are two such different groups of birds known as "vultures".

    As for how the analysis can be affected so strongly by a single gene, I don't know what are the exact factors are involved in this case, but here are some things to keep in mind - firstly, supporting characters for any tree topology are not distributed evenly through the dataset. Certain areas will strongly support one topology, others may support another, most will probably be random noise. However, just because one section of data strongly supports one result while the rest is noisy doesn't necessarily mean that one section is correct - if that one section has been subject to strong directional selection, for instance, convergence may have occurred. The supplementary data for the Ericson et al. (2006) paper did include a stack of alternative trees under different parameters (such as no FGB), but of course Ericson et al. did have a much smaller data set so I don't know how much their results would have resembled Hackett et al.'s.

  7. Oh yes, and I'll put up a call for Linnaeus' Legacy links over the next few days and starting trawling for some (I was about to say 'trolling', but then I remembered that that means something very different in blogging from in fishing). No banner as yet, I'm afraid - the person that was designing it has been having a hard time of things lately, I believe, so I assume she has more important things to deal with.

  8. I hope for a combined analysis of molecular and morphological evidence...

  9. Gerald Mayr's response paper to Livezey & Zusi is...

    Mayr, G. 2008. Avian higher-level phylogeny: well-supported clades and what we can learn from a phylogenetic analysis of 2954 morphological characters. Journal of Zoological Systematics and Evolutionary Research, 46, 63-72.

    Because some of the reporting of Hackett et al. (particularly in the blogosphere) has been produced by people who probably don't follow the phylogenetic literature, they are perhaps unaware that many of the relationships posited here have been proposed before. So, as you note, getting these conclusions into Science has done wonders for bringing them to wider attention.

  10. ...many papers that appear in the two are, well, kind of crap. This is not the fault of the contributing authors, but results from the severe space restrictions on articles in these journals.

    I disagree, actually. I think it's more that they're competing for, in a sense, "shock value", and so they publish articles that have striking results (in this case, loads of data + supporting previously-published but unconventional hypothesis), even when they really should be more skeptical of them. The article a couple of years ago on phasmids repeatedly re-evolving flight is another example. The lack of space just gives less detail for people to criticize.

  11. The June 2008 bird genetics study of Chicago’s Field Museum of Natural History (the Early Bird Assembling the Tree-of-Life Research Project) re-writes the avian evolutionary tree and is stunning in its impact. The genetics study shows that, though they look alike, for example, falcons are not closely related to other birds of prey, hawks or eagles. Several birds that look very different, including woodpeckers, hawks, owls and hornbills, are all closely related to perching birds. Flamingos, tropicbirds and grebes, all of which are closely related, did not evolve from water birds. Contrary to conventional wisdom, daytime hummingbirds evolved from drab nocturnal nightjars. Tropicbirds (white, swift-flying ocean birds) are not closely related to pelicans or other waterbirds. But perching birds, on the one hand, and parrots and falcons on the other, which do not look all that much alike, in fact descend from a recent common ancestor. Shorebirds are not a basal evolutionary group, which refutes the established view that all modern birds evolved from shorebirds. It is an understatement that appearances are deceiving. Birds that look or act similar are not necessarily related. Modern birds evolved relatively rapidly within a few million years during an explosive radiation, sometime between 65 million and 100 million years ago. The same is true of flower evolution. One day there are no flowers in the fossil record, and the next ‘day’ there are flowers, which troubled Darwin. Likewise, flowers that look alike are frequently not closely related. So how does this genome-scale phylogenetic evidence on avian evolution possibly jive with Darwinian gradualism, incremental changes over millions of years? It does not and cannot. Something else would appear to be going on to explain this planet’s remarkable biodiversity. The best hope to answer these questions is more and better science, not reliance on 19th century conjecture. Whatever the causative mechanism for bird or other biological evolution, avian evolution is not the result of natural selection. Phylogeny is more complex than we thought back in May. Good luck, guys.
    John Umana, Wash, DC

  12. If John Umana had actually read some of the comments I made in my post and not just cut and pasted a large block of text from the Field Museum press release, he might have come out of things looking a bit better off. "Phylogeny is more complex than we thought back in May"? The entire point of my posting was that the findings of the Hackett et al. paper largely corroborate what has been found before elsewhere. Anywho...

    The same is true of flower evolution. One day there are no flowers in the fossil record, and the next ‘day’ there are flowers, which troubled Darwin.

    Not actually true (apart perhaps from the bit about Darwin, who lived in a time when people knew a lot less about the fossil record than we do today, but that's largely irrelevant). The fossil Bennettitales, which are strong contenders as stem-angiosperms, had "flowers" with the reproductive structures surrounded by large bracts (specialised leaves) that may have been colourful in life and attracted insect pollinators. Many modern flowering plants (such as poinsettias and bougainvilleas) have similar structures, while the modern gnetopsid Welwitschia produces nectar to attract insect pollinators. Genetic studies show that petals and sepals of flowers are homologous to leaves, and it seems that a relatively small number of genes are involved in the changes in development ( There is no need to postulate any unknown process in the evolution of flowers.

  13. Oh yes, and:

    Modern birds evolved relatively rapidly within a few million years during an explosive radiation, sometime between 65 million and 100 million years ago.

    Accepting an average age to maturity of two or three years would, I think, be a conservative estimate for birds. Even that allows for a million generations over the course of a few million years. That's a lot of potential for natural selection, and a lot of time for change.

    If (for whatever reason) average human height were to increase by 0.1 of a millimetre with every generation, from a starting average height of 170 cm, then at the end of a million generations the average person would be over 100 metres tall.

  14. Bird sequence data flew the coop?
    First to commiserate with Christopher. These big biology projects produce many data types and thousands of data points and yet readers are only tantalized with the subjective highlights noted by the authors in the allocated succinct journal space, what about all the other novel knowledge gems hidden in the raw data?
    I find myself in a familiar position having digested the paper and downloaded and reviewed the supplemental on line material yet still wondering how to get a hold of the raw sequence data. The paper itself has provided few accession numbers for example reference 33 in the main paper indicates Sequence accession numbers in GenBank are EU737149 to EU740386, EF521416 to EF521576, and EU302706 to EU302748. I was able to pull sequences in the last two EU ranges, but NCBI 's Genbank has no entires for the first range of ids shown. All accnos queried in this range indicate Id/Accession does not exist. The only other accno reference in the SOM was the Treebase accession number S2079 and yet a detailed search of Treebase revealed zero alignment hits for this study( queried by accno, taxon and author) the page indicates data was updated 9 days ago so maybe it will still be deposited?(
    Several bioinformatic findings noted in the paper are unusual and thus I sought to rerun the alignment and tree generation using differing algorithms and sub-setting the data. Is this not the goal of peer reviewed publications to report high quality, REPRODUCIBLE findings? How is it possible to reproduce findings when the raw data is not available through the cited source? Makes me wonder if Science and Nature even test the links to the data pointers prior to publication?
    This is a common experience in my field of genomics, huge data sets distilled down to a 4 page Science article a sizable S.O.M. and zero ability to access the underlying data. In the past I have written to authors to ask for clarification and yet disparity in cross field scientific prestige means no response is the norm and since no web portal (even an FTP would suffice) is indicated for the early bird project where would one go from here to even address potential bioinformatic faux pas and start data analysis mining for other novel findings ?????
    Anjanette D Watson PhD (I would appreciate anyone with contacts in the bird world putting pressure on these authors or other field experts to provide a correction to the deposited accnos or provide a valid source for the raw data—is this too much to ask?)

  15. This comment has been removed by a blog administrator.

  16. Are you thinking of writing about zebrafish as well? A phylogeny tree for zebrafish could be so handy these days! Looking forward to it :)

  17. i know this is probablay the wrong site to ask, but if a flamingo doesnt get enough iodine what kind of bird is he most resembling?

  18. if a flamingo does not geta proper mount of iodine what bird does he become


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS