Arthropods and annelids have their body divided into a series of repeated units that bear pairs of appendages in most cases. These body units and their appendages have long been regarded as homologous structures and have been the basis for uniting annelids and arthropods as sister taxa in the taxon Articulata 1. The so-called "new animal phylogeny", however, does not support this close relationship between annelids and arthropods and rather places them in two different branches of protostome phylogeny termed Lophotrochozoa and Ecdysozoa, respectively 2. This suggests that body segments and appendages in annelids and arthropods might have originated separately and are therefore not homologous. On the other hand recent results suggest that at least segmentation might have an ancient origin that predates or coincides with the origin of the Bilateria 3.

In arthropods the body units are first specified in a parasegmental register 4 and later these parasegments are transformed into segments by re-segmentation during embryonic development 5. Intriguingly, recent work using the expression of the segmentation genes engrailed (en) and wingless (wg) has demonstrated that re-segmentation does not occur in annelids and the body units in annelids thus remain parasegments 67. This also makes annelid appendages (parapodia) parasegmental rather than segmental structures and this is further evidenced by the expression of the appendage marker Distal-less (Dll/Dlx) 7. This calls into question the homology between annelid parapodia and the appendages of arthropods (arthropodia). Here I show that in an arthropod species, the brine shrimp Artemia franciscana, the early body organisation including the limb primordia is identical to the annelid condition. This demonstrates that arthropodia like parapodia are initially parasegmental organs and suggests that both share a common evolutionary origin.

Artemia franciscana development includes a larval stage (nauplius) (Figure 1A). The nauplius consists of an anterior part, comprising the ocular region, labrum, first and second antenna and mandibles, and a posterior part, which is a more or less undifferentiated trunk. In nauplii at stage III (staging after 8) this trunk region develops two bulges (boxed in red in Figure 1A) followed by a smooth region, the "growth cone". The reason for the two bulges is that there are already the mesodermal blocks forming beneath the ectoderm 8. The large bulge contains, still in a single block, the mesoderm for several future segments (first and second maxillary, and first thoracic segment), the second smaller bulge contains the mesoderm for the second thoracic segment. These two bulges are thus the first morphological sign of subdivision in the trunk region. I used the expression of en [GenBank:X70939] and wg [EMBL:AM774593] to identify the location of parasegment boundaries in the trunk. en is expressed in the anterior portion of the smaller bulge and in the anterior part of the growth cone following the smaller bulge (Figure 1B). wg is expressed at the posterior border of the large bulge and in the posterior portion of the smaller bulge (Figure 1C). In some specimens, that are further into stage III, the wg stripe in the posterior part of the large bulge is separated from the morphological groove by about two cell diameters (Figure 1D). The significance of this is presently unclear. It could be a sign of the beginning resegmentation at this location or might be correlated with patterning mechanisms specific to the large bulge which is a complex structure comprising several future segments. Double-label in situ hybridizations of wg with en could clarify this, but have been technically impossible in Artemia so far. Based on the opposition of wg and en expression across the morphological grooves between large bulge and small bulge (at least in early stage III) and between small bulge and growth cone (throughout stage III) the morphological units at stage III are still parasegments and the morphologically visible indentations (grooves) between them coincide with the parasegment boundaries. I then used the expression of the appendage marker Dll [EMBL:AM774594] to identify the appendage primordia. Dll expression shows that the first appendage primordia are already specified at stage III (Figure 1E). Surprisingly, Dll is expressed in groups of cells anteriorly adjacent to the grooves and thus in front of the parasegment boundary. Thus, the limbs in Artemia initially are parasegmental structures, identical to the parapodia of the annelids (see 7).

Hejnol and Scholtz 9 have studied limb primordium formation (as marked by Dll protein expression) in the crustaceans Orchestia cavimana and Porcellio scaber. Because in these species the cell lineage is known, these authors were able to demonstrate that the limb primordium starts as a single cell and expands by activation of Dll in adjacent cells. I propose that arthropods use this expansion mechanism to transform their initially parasegmental appendage primordia into segmental appendages after re-segmentation (Figure 1F, upper row). By contrast, annelids do not re-segment and the clonal boundaries of the parasegments are retained as the definite morphological borders between the body units. I suggest that because of this, the restrictive influence of the morphological borders between the parasegments cannot be overcome by the primordia of the parapodia, resulting in a lack of expansion of the primordia (Figure 1F, lower row). This model thus derives the adult arthropod condition from a common body organisation in immature arthropods and annelids by the arthropod-specific processes of re-segmentation and subsequent expansion of the appendage primordium across the parasegment boundary.

In the adult animals arthropodia and parapodia have different positions on the body (segmental versus parasegmental). This argues against their homology. The present data, however, show that the primordia of these appendages have identical positions in the immature animals and thus arguably derive from a common ancestral structure (i.e. they are homologous). This strongly suggests that Urprotostomia, the last common ancestor of annelids and arthropods, was not only already parasegmented 67, but also had parasegmental appendages. This also has consequences for the discussion of animal evolution 1011 because it suggests, that the last common ancestor of the Lophotrochozoa (Urlophotrochozoon) and of the Ecdysozoa (Urecdysozoon) had appendages and the appendage-less forms in both Lophotrochozoa and Ecdysozoa have lost their appendages secondarily (Fig. 2). Data from the Onychophora show that in older embryonic stages en is expressed in the posterior portion of the appendages and thus in a segmental fashion like in arthropods 12. This suggests that re-segmentation was already present in the Urecdysozoon (see arrow in Fig 2).

The author declares he has no competing interests.