Some animals are so familiar that we can scarcely believe we know little about their origins. Take earthworms, for example, common to everyone's garden, which belong to a large phylum of invertebrates — the annelids (ringed worms). The hitherto shrouded evolutionary history of annelids is now illuminated by Struck et al. on page 95 of this issue1.

Annelids are global players in terrestrial and freshwater environments, and in marine ecosystems, where they live in and on the sea floor. But the identity of their nearest relatives (maybe molluscs, maybe flatworms), and even their affinities within the phylum, has remained a puzzle. This lack of understanding has another edge, given that various annelid species serve as model organisms for the investigation of basic biological processes. Embryologists and neuroscientists have studied leeches for decades. And the ragworm Platynereis has recently emerged as a valuable model for studying development, evolution and neurobiology2, along with Capitella, Hydroides and other marine species3. Just imagine working on vertebrate models such as fish, mice and frogs without knowing their evolutionary interrelationships.

Struck et al.1 have made a significant advance in the reconstruction of annelid phylogeny, having resolved their internal affinities. Using molecular techniques, they have studied the relationships between various annelid families and orders, and have obtained a surprisingly clear result. It turns out that annelids are deeply subdivided into two main groups, the Errantia (to which the ragworm belongs) and the Sedentaria (to which the leech, earthworm, Capitella and Hydroides belong).

As the names suggest, the subdivision of annelids into the Errantia and Sedentaria matches their overall lifestyles (Fig. 1). Members of the Errantia are free to move about, and crawl, swim or burrow. Many are predators or feed on macroalgae. By contrast, representatives of the Sedentaria are hemi-sessile burrowers or tube dwellers (apart from the highly specialized, parasitic leeches). They eat sediment or surface deposits, or filter the surrounding water with their tentacle crowns.

Figure 1: An illustration of the Errantia and Sedentaria by Ernst Häckel, dated 1904.
figure 1

MARIXVERLAG

To take just two examples, top right is Eunice magnifica (Grube, 1866), Eunicidae, an errantian; top left is Sabellastarte spectabilis (Grube, 1878), Sabellidae, a sedentarian. Errantians often have especially prominent lateral appendages with bristles, for undulatory crawling. Many sedentarians exhibit beautiful tentacle crowns for filtering plankton and other food particles from the water.

Full size image

In 1865, a similar grouping, but excluding the earthworms and leeches, was put forward by Jean Louis Armand de Quatrefages de Bréau. However, it was dismissed by later authors, who considered the similarities in lifestyle to be convergent adaptations due to similar ecological constraints. Ironically, Struck and colleagues' study1 reveals that the older classification was closer to the truth, thus 'revising the revision'. The authors' phylogeny demonstrates that broad features of lifestyle and morphology, even if sometimes challenging to quantify, can be at least as informative as ultrastructural or fine morphological characteristics, and are not necessarily much more prone to the complication of convergent evolution.

The work is another success story in the young discipline of phylogenomics — which attempts to resolve evolutionary history by genomic comparisons — and is one of the first aimed at probing deep within a phylum. Struck and co-workers have sequenced about a thousand expressed-sequence tags (complementary DNA library clones) from 17 members of annelid families and complemented these collections with existing data, yielding a total representation of 34 annelid species. They selected 231 genes common to at least one-third of the total taxa, and aligned and concatenated them into a supermatrix of 47,953 amino acids.

Next, they reconstructed a phylogenetic tree using refined methods capable of handling the diversity of amino-acid substitution processes in such a supermatrix. Many of the nodes in their tree, especially that separating the Errantia from the Sedentaria, had remarkably high support values (contrasting with those of previous annelid phylogenies based on single genes4), making it highly likely that this grouping is definitive.

The case of the annelids exemplifies both the beauty and the pitfalls of phylogeny reconstruction when applying the principle of parsimony, which settles on the tree minimizing gain or loss of particular characteristics. At the molecular level, this approach has proved very powerful, and it has been further enhanced by the advent of phylogenomics. But it is becoming increasingly obvious that, on the basis of morphological characteristics alone, there is a serious problem: the apparent ease with which such characteristics are lost.

This point is illustrated by a morphological parsimony analysis of annelid phylogeny5 that established a group, the Palpata, whose members were defined by the presence of specific head appendages (palpae) — the implication being that other groups without palpae had never had them. The new annelid phylogeny instead indicates independent loss of palpae in errantian and sedentarian groups, as was previously suggested6,7 by two of the co-authors of the current paper1. This example corroborates the general idea of frequent and independent loss of traits during animal evolution.

Finally, the new work1 nicely illustrates how, once the (molecular) phylogeny has been solved, matrices of morphological characteristics can be used to reconstruct the common ancestors of the respective groups. If a given characteristic is found in both branches resulting from a node, it must have been present in the common ancestor. In this way, we can infer a lot about the 'urannelid', the last common ancestor of all annelids. Most significantly, it was an animal richly equipped with sensory organs. It had chemosensory nuchal organs and palpae; a pair of two-celled larval eyes for phototaxis2; and possibly a pair of more elaborate, multicellular adult eyes with an alternating arrangement of rhabdomeric photoreceptors and shading pigment cells. The latter combination is found in extant errantians8 and in the sipunculans9, which represent an outgroup to both errantians and sedentarians (see Fig. 1 of the paper1).

The urannelid was segmented, a detail that is clear from the nested position of two unsegmented taxa, the echiurans and sipunculans, within segmented groups10. This ancestor probably lived on the sea floor, using its relatively complex lateral appendages for undulatory crawling (as seen in today's Errantia and for example in the Spionidae, which lie in the basal part of the Sedentaria branch of the tree). Given the power of phylogenomics, we might soon know what the urannelid mollusc- or flatworm-like relatives looked like in the ancient oceans.