Thursday, February 23, 2006

Old icons will die hard

It would be grossly inappropriate for a blog called The Lancelet to avoid discussing some of the interesting findings reported in today's issue of Nature. The new study, puts an interesting twist on our deep ancestry. In what is almost prototypical for science, it seriously challenges the common sense perceptions that have led us to our traditional classification. That is, science is almost at odds with how our subjective impressions tell us things ought to be. In this case, the authors have challenged a prevailing idea of how vertebrates are related to our neares invertebrate relatives.

Lancelets, or Branchiostoma, or amphioxus, (seen in the header of this site) are almost iconic in their status as the protovertebrate. There is even a song about our kinship with the lancelet. Lancelets have a notochord (a firm rod of tissue that underlies the dorsal hollow nerve chord -- one of the unique features that unites chordates, the group that includes tunicates, lancelets, and vertebrates), an elongate row of gill slits, and segmented muscle blocks.

A debate has raged among biologists as to who is more closely related to vertebrates: lancelets or tunicates. Tunicates (or sea squirts) are sessile animals that live kind of like a sponge, rather than like a fish. As was noticed by the early 19th C. embryologist, Karl Ernst von Baer, they have a free-swimming larval stage that has gill slits and a notochord and looks a lot like a lancelet. The ongoing debate (especially in recent years with the advent of molecular phylogenies) has been who is more closely related to whom: are lancelets more closely related to vertebrates? Are tunicates? Or are tunicates and lancelets closer to each other than either is to vertebrates?

The new study tested tested the interrelationships of the deuterostomes: the branch of the animal family tree that includes vertebrates, the various chordates, and echinoderms (sea stars & co.). The researchers used 146 nuclear genes of 36 different animals and two fungi. Their results were rather unorthodox. They recovered a tree that placed the tunicates as the sister group of the vertebrates. But what was more striking was the placement of lancelets with an echinoderm lineage.



The results, if correct, imply that the last common ancestor of vertebrates and echinoderms was an animal very much like the lancelet and that sessile filter feeders are more closely related to us than is a particularly vertebrate-like swimmer. A common, but fallacious, interpretation of such a phylogeny is that it implies we evolved from a sessile filter-feeder or, in this case, a tunicate. But it makes no such implication, since all the tunicates converge to a single unique node in the tree (the red branches marked "Tunicata"). What the tree actually implies is that their condition evolved along the "trunk" of that cluster of branches. This in itself partly explains why tunicates have a free-swimming, lancelet-like larval stage.

Common sense would make us want to lump lancelets with vertebrates. After all, lancelets look most like fishes. However, the question raised by this analysis is more along the lines of: what if tunicates split from the tree after lancelets split from vertebrates? This is an entirely possible scenario, but one that is not accounted for by a classification that focuses on overall similarity. The way evolution works is not necessarily reflective of the way we think it should work.

The reality is that looking back at evolutionary history from the present can be as biased as observing the solar system from here on earth. Appearances can be deceptive. The reason why we share so much in common with lancelets may not be because we have a special common ancestor with them. Instead, it may in fact be because lancelets haven't changed much since the common ancestor of vertebrates and echinoderms. Instead, it may be that tunicates and echinoderms are the ones that have diverged the greatest. This is certainly what the new result implies.

However, this phylogeny is far from being the last word on the interrelationships of our deepest ancestors. The authors themselves call for the need to test the grouping of lancelets with echinoderms. One of the potential problems here may be that each is represented by a single taxon. In the case of lancelets, we haven't got much choice. However, more echinoderms will be needed. A single, ancient lineage in an analysis can cause problems due to a phenomenon that systematic biologists call "long branch attraction". In simple terms, this just means that the longer and more rapidly evolving your lineage is, the greater the chances that parts of your genetic sequence will happen to match that of another ancient lineage, just by chance. When this happens, two unrelated branches can "snap together" implying recent common ancestry, when there is in fact none.

One of the strengths of this study is that they tried to circumvent this and other potential problems as much as possible. They concentrated on what were considered to be slowly evolving lineages. One of the problems that happens when a gene evolves too quickly is that it changes too much and, consequently, looks a lot less like the ancestral gene would have looked like. The result is that fast-evolving genese amount to a lot of "noisy" data over very long periods of time. The old information gets overprinted by new information and confuses the analysis and also confuses the analysis process.

There will certainly will be a number of researchers who will dispute this analysis. People who know their molecular evolution far better than I do will probably have much to say about it. In the meantime, it serves as a welcome challenge to the prevailing "dogma". This is not the first analysis to cast doubt on the status of lancelets, and certainly won't be the last.

EDIT: Predictably, Carl Zimmer has written another summary with an interesting twist. Check it out.



Delsuc, F., Brinkmann, H., Chourrout, D., and Philippe, H. 2006. Tunicates and not cephalochordates are the closest living relatives of vertebrates. Nature 439: 965-968. link.

Gee, H. 2006. Careful with that amphioxus. Nature 439: 923-924. link.

3 comments:

RPM said...

Yeah, those long branches make the relationships between tunicates, echinoderms, cephalochordates, and vertebrates pretty unconvincing. It's not just the long external branches (which, by the way, aren't because of rapid evolution of the sequences, but because of undersampling in a few taxa), but the short internal branches that lead to long branch attraction.

Also, portraying tunicates as sessile, sponge-like organisms is a bit misleading. Tunicates have two body openings (an incurrent and excurrent siphon), like all other chordates. The juvinile stage is a free swimming chordate.

Martin Brazeau said...

Hi RPM,

Thanks for the comments.

Yes, I agree. I'm not terribly convinced by the result, but I wasn't (and will not be) disputing it in the form of a blog post. The grouping of Branchiostoma with that one lonely echinoderm sets off the alarm bells. I tried to make clear at the end of my post that this was far from being the last word.

"It's not just the long external branches (which, by the way, aren't because of rapid evolution of the sequences, but because of undersampling in a few taxa), but the short internal branches that lead to long branch attraction."

I didn't think that I implied any differently: "A single, ancient lineage in an analysis can cause problems due to a phenomenon that systematic biologists call "long branch attraction". Rapid evolution will also cause saturation with a similar statistical effect, no?

"Also, portraying tunicates as sessile, sponge-like organisms is a bit misleading. Tunicates have two body openings (an incurrent and excurrent siphon), like all other chordates. The juvenile stage is a free swimming chordate."

You seem to have missed the part where I wrote: "they have a free-swimming larval stage that has gill slits and a notochord and looks a lot like a lancelet." Tunicates do indeed have a sessile adult phase and this is quite different from the business of a lancelet. Sponges also have incurrent (ostia) and excurrent openings (oscula), though they are not homologous. I wouldn't go much further than that, but it helps a layperson envisage the dramatic differences in lifestyle and body form that exist between an adult tunicate and a lancelet or, indeed, a vertebrate.

Caio de Gaia said...

I've been reading some papers about this subject lately and it's funny how fast the relationships tend to change. I thought that hagfishes were outside Vertebrata and lampreys where in, but then I saw a paper from 2002 about cyclostome monophyly. The cladogram you present also puts hagfishes and lampreys in the same node, and I wonder if this is the consensus view these days.

Now the lancelet appears closer to echinoderms. What a mess. Probably even groups with extinct forms only, and no chance of genetic studies, are also moving up and down. I wonder if Euconodonta are still considered vertebrates. Time to read a good recent review paper I suppose.