An article on surprisingly troublesome wormy things

There have been two key differences so far between this course and my undergraduate. Compared to a typical autumn term back then, this has had:
1. Two vs. zero group presentations, which were painful but not quite as horrific as feared.
2. One vs. eight essays to write. But I'm probably spending the equivalent time on extra travel!

Anyhow, I present this term's finished essay. It's designed to be like Nature's News & Views articles, which talk about interesting new papers in a relatively friendly way, accessible to scientists from other disciplines and the interested amateur.

______________________________________________________________________________________

Xenoturbella and phylogenetic wanderlust

Phylogenetics, the evolutionary relationships of organisms, is crucial to understanding them fully. How else could we appreciate how, for example, vertebrate fins turned into limbs and back again? Finding the close relatives of an organism can also help us reconstruct its ancestors. For many areas in the animal tree of life, molecular analysis has complemented and clarified phylogenetic trees built on morphology alone, and we can be reasonably confident that the relationships we infer are real. Unfortunately, there are some problematic taxa whose placement, despite molecular analysis, is still a mystery. Xenoturbella, a genus of small benthic worms, is one. The debate over what type of animal Xenoturbella is has recently condensed to two hypotheses: does it belong at the base of Deuterostomia, potentially informing on the ancestor of vertebrates and our close relatives, or at the base of Bilateria, informing on the ancestor of Deuterostomes and a whole lot more? In their 2016 paper, Rouse et al.[6] raise the known number of Xenoturbella species to five, assess phylogenetic methods suggested for such deep timescales and present results that strongly suggest Xenoturbella belongs at the base of Bilateria... or maybe not. The puzzle is not quite solved yet.

Prior to their paper, Xenoturbella was represented by just two species, X. bocki and X. westbladi, both found only off the west coast of Sweden. As predicted by Nakano et al.[3], more Xenoturbella were waiting to be found in the deep sea, and Rouse et al. collected multiple specimens from three sites in the east Pacific. According to their mitochondrial genes, they represented four new species: X. monstrosa, X. profunda (figure 1), X. hollandorum and the delightfully named X. churro. The attentive reader may have noticed a mismatch in numbers; the mitochondrial genes of the original Xenoturbella species were similar enough that Rouse et al. treat them as synonyms, so X. westbladi has been cast out to taxonomic purgatory. Nevertheless, the diversity of the genus has been significantly increased. Having more species allows greater taxon sampling in phylogenetic analyses, which may improve results by reducing the impact of errors like long branch attraction (see below), though it has the potential to cause other problems[5]. However, more Xenoturbella species is probably not relevant to its position within the entire animal phylogeny; at such a deep scale the differences between species in the same genus are essentially zero.

Figure 1: X. profunda in a clam field near a hydrothermal vent, Mexico. Abbreviations: a, anterior; rf, ring furrow; p, polynoid scaleworm. From Rouse et al. (2016), figure 1c.

 Placing Xenoturbella using morphology has been dificult because they are very simple animals, lacking a centralised nervous system, coelom, excretory or reproductive organs. They glide across surfaces using ventral cilia. Quite how they find their way is unknown: an organ near the head-end may be a balance-sensing statocyst, and a sensory function has also been suggested for the furrows that encircle the body and extend down the sides, but neither are confirmed[3]. Rouse et al. pickled their new specimens shortly after collection; it would be interesting if return expeditions could perform behavioural studies on the new species.

Going back to phylogenetics, the review by Nakano et al. (2015)[3] summarises the many complicated relationships Xenoturbella has entertained across the animal tree (figure 2). It was first described in 1949 as a flatworm in the phylum Platyhelminthes [A]. Morphological studies (admittedly without much to work with) then went on to identify it as a basal metazoan [B], basal bilaterian [C], deuterostome [D], bryozoan [E] and even a bivalve mollusc [F]. When molecular techniques arrived in 1997, a study comparing three genes surprisingly confirmed this bivalve affinity. But soon afterwards, another study using the very same genes found Xenoturbella to be a deuterostome again. It turned out that Xenoturbella could be identified as a bivalve if DNA was taken from the whole animal, but as a deuterostome if its gut was removed. Instead of finding its relatives, the 1997 study had inadvertently found its food source. Studies of other genes continued to find Xenoturbella in Deuterostomia, but in another jump across the animal tree, the first phylogenomic study (comparing essentially the entire genome) argued that it was actually a sister group to Acoelomorpha, another clade of simple wormy animals, so belonged right at the base of Bilateria.

Figure 2: A simplified metazoan phylogeny showing the various placements of Xenoturbella. [A], in Platyhelminthes (Protostomia); [B], at the base of Metazoa; [C], at the base of Bilateria; [D], in Deuterostomia; [E], in Bryozoa and [F] in Mollusca (both Protostomia).

Despite subsequent work, basal-Deuterostomia and basal-Bilateria are still competing hypotheses. It has been suggested that the phylogenetic methods used are a source of conflict between them, which is quite reasonable. Because we cannot see back in time, it is usually impossible to know whether reconstructed phylogenetic relationships are true, making phylogenetics a field of better or worse hypotheses rather than facts. To address some potential methodological issues, Rouse et al. used several different methods to fit Xenoturbella into the animal tree and compared the results.

First, they analysed the 13 mitochondrial proteins from all five species. The first phylogeny used the maximum likelihood method. This assumes that lineages evolve independently of each other, so are not closely related, so it should be well suited to deep phylogenetic scales such as this. It found Xenoturbella to be a sister group to the acoelomorphs, forming Xenacoelomorpha, a result which also appeared in all of their other analyses. So, Rouse et al. show very strong support for Xenacoelomorpha. The analysis then put Xenacoelomorpha with deuterostomes, as had been found in other mitochondrial studies[4], but with only weak support.

They then used the same mitochondrial data with another method, PhyloBayes. PhyloBayes distinguishes itself by accounting for site-specific amino acid or nucleotide preferences, so should be among the most accurate models of evolution[2]. It too returned Xenacoelomorpha within deuterostomes, but again with weak support. Rouse et al. suspect the low support in the two mitochondrial analyses might have been caused by their additional data from the new Xenoturbella species, or by using all 13 proteins instead of a selection. As noted above, there is reasonable evidence that higher taxon and character sampling usually reduce uncertainty in a phylogeny[5], so Xenoturbella being a deuterostome is looking unlikely.

Rouse et al. then moved to phylogenomics, selecting just X. profunda and X. bocki to represent Xenoturbella. Taking just two species for a genus seems efficient for phylogenies of this scale, but mitochondrial data for all Xenoturbella species was used in those analyses, and genome sequencing is now relatively cheap. It is interesting that Rouse et al. did not sequence the other species' genomes. Perhaps after submarine expeditions the budget was just a little too tight.

The first phylogenomic analysis performed was another maximum likelihood analysis, and it returned Xenacoelomorpha as the sister group to Nephrozoa, right at the base of Bilateria. This was the other dominant hypothesis, and it came with much better support. Several other maximum likelihood phylogenies testing for potential pitfalls, such as genes evolving at different speeds, were produced and all came to the same conclusion with similar support. The basal position looked strong.

Rouse et al. then used PhyloBayes again, this time with specific settings recommended by Philippe et al. (2011)[4] to avoid the dangers of long branch attraction (LBA). In their study, Philippe et al. had also performed a variety of phylogenetic methods and concluded that the position of Acoelomorpha, and Xenoturbella when they were found together, was biased by LBA. LBA is the tendency for distantly related taxa that branch near the base of a tree to appear closely related when they are not; they evolve similarities to each other by chance, and their lack of true close relatives means these similarities have no context[5]. Convergences can be wrongly interpreted as signs of relatedness. Philippe et al. argued that coincidental similarities between xenacoelomorphs and truly basal animals pulled them to the base of the tree. With LBA minimised, they recovered Xenacoelomorpha as deuterostomes. However, when Rouse et al. tried that method with their data Xenacoelomorpha came out in yet another new position, next to the deuterostomes' sister group Protostomia, with moderate support. They do not offer an explanation. However, published in the same issue as Rouse et al., Cannon et al. (2016)[1] also examined whether LBA affected Xenacoelomorpha, and found that it did not. So in hindsight, this PhyloBayes model may be an inappropriate correction after all, making Xenoturbella as a protostome unlikely.

Rouse et al. have succeeded in elevating our understanding of a mysterious genus. They discredit the placement of Xenoturbella in Deuterostomia, giving most (though not unanimous) support to placement at the base of Bilateria. While investigating the different methodologies that may have led to Xenoturbella's phylogenetic wandering, they found problems with phylogenies based on mitochondrial analyses, a common alternative to using whole genomes. This suggests that mitochondrial data might not be appropriate at deep phylogenetic scales. Here, however, it was very useful in confirming four new species to an enigmatic genus. Nakano et al.[3] were correct in suspecting that more Xenoturbella species were waiting to be found in the deep sea, and it is likely that more are waiting still.



References
[1] Johanna Taylor Cannon, Bruno Cossermelli Vellutini, Julian Smith, Fredrik Ronquist, Ulf Jondelius, and Andreas Hejnol. Xenacoelomorpha is the sister group to Nephrozoa. Nature, 530(7588):89?93, February 2016.
[2] Nicolas Lartillot, Thomas Lepage, and Samuel Blanquart. PhyloBayes 3: a Bayesian software package for phylogenetic reconstruction and molecular dating. Bioinformatics, 25(17):2286?2288, September 2009.
[3] Hiroaki Nakano. What is Xenoturbella? Zoological Letters, 1:22, 2015.
[4] Hervé Philippe, Henner Brinkmann, Richard R. Copley, Leonid L. Moroz, Hiroaki Nakano, Albert J. Poustka, Andreas Wallberg, Kevin J. Peterson, and Maximilian J. Telford. Acoelomorph flatworms are deuterostomes related to Xenoturbella. Nature, 470(7333):255?258, February 2011.
[5] Steven Poe. Evaluation of the Strategy of Long-Branch Subdivision to Improve the Accuracy of Phylogenetic Methods. Systematic Biology, 52(3):423?428, 2003.
[6] Greg W. Rouse, Nerida G. Wilson, Jose I. Carvajal, and Robert C. Vrijenhoek. New deep-sea species of Xenoturbella and the position of Xenacoelomorpha. Nature, 530(7588):94?97, February 2016.