Is everything OK with Olfactores? A call to properly assess morphological implications

When molecular phylogenies do not coincide with the morphological phylogenies, this is a serious problem. Experience tells us that its not a matter of assuming the morphological data is equivocal; specially if it is completely uncontroversial within (morphological) phylogenetic systematics. Many times the conflict thereafter disappears, specially upon better taxon sampling of molecular data (Note: increasing the number of adequate taxa seems to be of much greater consequence than increasing the number of genes). We all remember that "sharks are teleosts" thing. This is why it is good to know the exact "morphological cost", or extra morphological transformations, that are implied by molecular hypotheses that openly conflict with morphological phylogenies. If the conflict shows no resolution by further studies, it is not a matter of assuming the molecular data must be "the correct one"; or at least, if you are going to make that assumption, KNOW the implications for the evolution of morphology!

Is something smelling fishy about the phylogeny of the chordates? Or is everything OK? At record speed, the evo-devo community has accepted the results of the latest molecular phylogenies, that vertebrates are closer to urochordates than to cephalochordates. Olfactores = Urochordates + Vertebrates. This term was born in the context of a marginal theory of Jefferies (1981) within his interpretation that the fossil Homalozoa ("calcichordates") are the ancestors of chordates (other paleontologists consider Homalozoa to be basal echinoderms).

The recent publication of the entire genome of Amphioxus has further repeated this result for the comparison of an astounding 1090 genes in a phylogenetic analysis (Putnam et al 2008) that included important groups that are frequently left out, such as an acorn-worm, and a larvacean urochordate.
Given that the Olfactores are becoming accepted as some "new truth", it is interesting to review why never before had this notion attained popularity, the most common assumption being that cephalochordates and vertebrates were closest, conforming a clade Euchordata (also called "Myomerozoa" for the presence of somites). Actually, just a glance at amphioxus, a very fish-looking creature, should make us immediately suspect that the olfactores is probably not consistent with the most parsimonious morphological history; that is, that we may have the proverbial case of a clash of molecular vs morphological phylogenies.

Everybody agrees that urochordates have secondarily lost traits. This is obvious when the traits absent in urochordates are present even in hemichordates , such as coelomic cavities, and several (not just one) pairs of branchial openings. However, the list of things lost in urochordates increases substantially if we consider Olfactores to be real: for instance, the loss of somites, and several vertebrate-like gene expression patterns in the developing neural tube.

In this sense, it is important to point out that the exact morphological cost of the olfactores, in terms of assuming extra steps beyond parsimony, has not being adequately discussed or investigated yet. An important conflict with morphology and gene expression is looming, but people are failing to see it.

For instance, a recent comment (Swalla and Smith 2006) says "an extensive cladistic reanalysis of morphological data found strong support for Olfactores ( Ruppert 2005)". However, the cited work of Ruppert is not a cladistic analysis at all, but a "homology analysis", the mere mapping of proposed events assuming that Olfactores is real, for a limited set of traits. Further, the morphological cost of the olfactores is hardly rescued by Ruppert 2005, which cites only a few of the most notorious losses we must assume occurred in the urochordates (for instance, Ruppert fails to mention any of the similarities of gene expression between the neural tube of cephalochordates and vertebrates)

In good faith, we must assume that Swalla and Smith were thinking about another morphological analysis that is cited as support for the olfactores, namely the cladistic analysis of all metazoa made by Zrzávy et al (1998). While indeed Zrzávy et al. is an extensive analysis (238 traits), it does not specifically address the question of the phylogeny of chordates , but of metazoa in general. So actually only a minority of these traits are bound to be relevant to the question of chordate phylogeny. Further, despite the non-traditional placement of the urochordates retrieved by this analysis, Zrzávy et al did not make any mention of this result in their discussion, concentrating on other aspects of animal phylogeny. No list of "Olfactorian" synapomorphies was discussed. The analysis by Zrzávy has been criticized at length by Jenner 2001 on various grounds, such as the assumptions made in the definition and polarization of traits.

With no explicit cladistic analysis of chordates for an alternative phylogeny , the review by Rowe (2004) of chordate phylogeny does not even mention the work by Zrzávy or the possibility of the olfactores. Indeed, the best references for a cladistic analysis specifically focused on the chordates are earlier works (Maisey 1986, and Schaeffer 1987), which support the euchordata, and are largely accepted within the community of phylogenetic systematics. However, these studies will fail to collect numerous newly described traits shared by cephalochordates and vertebrates, from gene expression patterns to fine structure of the nervous system. So, in fact, a new updated cladistic phylogenetic analysis of the cephalochordates is needed to establish the precise consequences of the Olfactoria for the morphological history of the chordates.

It is certainly possible that this new analysis may show the history implied by olfactores to be too absurd, implying too many reversals or convergences. If this is the case, it is not just a matter of going with the molecular phylogeny over the morphological; rather, the possibility of an artifact in the molecular studies must be taken into account (yes, even with 1090 genes! )


This is the tree of the 1090 genes (Putnam et al 2008). Some observations:

1) As usual, the longest branches of the chordates are the Urochordates. Also, notice that this effect is not mitigated by the large amount of genes; that is, abnormally high substitution rates is a genome-wide phenomenon. Is it possible that high substitution rates may distort the phylogenetic signal of entire genomes? Also, notice the low 76.4 % bootstrap value support for the monophyly of chordates. Traditionally, molecular evidence has had some problems retrieving this node, which is, from a morphological perspective, a very straight-forward conclusion

2) Hagfish, morphologically the most basal vertebrates , were not included in this analysis. In fact, the position of hagfishes haunts this entire issue, since it is a case in which a clear conflict of molecules vs morphology still lives on

3) Despite the general trend of nuclear genes to support the olfactoria, the comparison of entire mitochondrial genomes supports the classic hypothesis of euchordata (Bourlat et al. 2006). Further, this is also the case when discussing the relationships of Hagfishes, supporting them as basalmost vertebrates (Yu et al. 2008). Why this diference ? Is it possible that the mitochondrial genomes have been spared from some source of artifact affecting the phylogenetic signal of nuclear genes?

References.

Bourlat et al 2006 Nature 444:85-8

Jefferies 1981 Zool. J. of Linnean Soc. 73, 351-396

Jenner 2001 Syst. Biol. 50(5):730-742

Putnam et al 2008 Nature 453: 1064-1070 doi:10.1038/nature06967

Rowe 2004 In: Cracraft & Donoghue, Ed. Assembling the tree of life. Oxford. pp 384-409

Ruppert 2005 Can. J. Zool. 83: 8–23

Swalla & Smith 2006 Phil. Trans. R. Soc. B doi:10.1098/rstb.2007.2246

Yu et al. 2008 J. Genet. Genomics 35: 285-290

Zrzávy et al. 1998 Cladistics 14, 249 -285