Background

Molecular data [1, 2] have radically reconfigured many aspects of metazoan phylogeny, and in doing so have effected a renaissance in this area of evolutionary biology that has uncovered many hitherto unsuspected relationships. An unavoidable drawback, however, is that typically the extant forms are highly derived and disparate with the result that the natures of common ancestors are often matters for unconstrained speculation. In principle, however, by providing direct access to members of stem groups of particular phyla (or other major clades), the fossil record can provide insights into the early divergence of major groups and hence the construction of their respective bodyplans [3]. In reality the available fossil record remains problematic, because although in many cases stem-group representatives have proved of great value in documenting the nature of transitional forms, in the Cambrian there is a plethora of seemingly enigmatic taxa. Interpretations of such taxa can be so disparate that it is sometimes questioned whether what is known of these fossils makes them capable of serving any useful phylogenetic role [4, 5]. A key area of dissent is the recognition of homologous structures. This, combined with often apparently bizarre combinations of character states, has generated controversy rather than consensus. Despite being widespread and often showing excellent soft-bodied preservation, the enigmatic vetulicolians [6–13] exemplify this difficulty.

This group is united by a distinctive bodyplan. This consists of a prominent anterior unit, ranging from oval to quadrate in shape, and possessing a wide oral opening. Attached to this unit is a segmented tail with a terminal anus. Despite a strikingly arthropod-like arrangement, none of the known taxa (including Vetulicola [6], Ooedigera [13], Pomatrum [14], ** dynamics

The demonstration of unequivocal lateral perforations (Figures 1, 2, 3) is consistent with water entering the animal via its oral opening and subsequently being expelled along either side of the animal (Figures 2G and 3D). Irrespective of whether, as we argue below, the apertures are homologous with the gill slits of deuterostomes, the inferred water flow invites a consideration as to whether the mechanism was effectively passive, relying on currents induced by ciliary action, or forced by active pum**. It is the case that a wide range of suspension-feeding deuterostomes, notably enteropneusts, most urochordates (apart from salps; see below), amphioxus and by implication extinct gill-bearing echinoderms, employed ciliary mechanisms, and this would accord with a consensus that among the deuterostomes the primitive mode of feeding was passive. While this cannot be ruled out for Vetulicola, we suggest that several features of the body of this animal are consistent with a more active mode of water flow. In our view, the most compelling observation revolves around the arrangement of the anterior unit and especially the longitudinal grooves, which we propose played a key role. While evidently having the same composition as the rest of the anterior body wall, the grooves appear to have been more flexible [28]. This is particularly evident from the range of configurations that vary from specimens where the groove is effectively closed (Figure 4A) to others where either margin is widely separated (Additional file 1). In the latter the floor of the groove displays a narrow median depression that anteriorly diverges into a Y shape (Additional file 1); this we presume would have been used to accommodate any extension. Construction of the anterior part of the vetulicolian body as four semi-independent plates with junctions along the dorsal and ventral midlines, combined with dilation of the grooves, would permit expansion of the internal cavity and so allow ingress of seawater. Correspondingly, contraction of the anterior unit would serve to expel the water through the gill slits. Dorsal and ventral sutures [16], that defined the meeting points of the plates, may also have been zones of relatively weakness and by rotating would have contributed to the expansion and contraction of the anterior unit. We have considered whether any other function for these grooves is more plausible. If, for example, vetulicolians ingested large prey then such dilation might assist capture, but overall the evidence for suspension feeding seems stronger.

Figure 3
figure 3

Gill structures of Didazoon haoae. (A) Laterally compressed specimen ELI-2010A (external view); internal openings (inhalant apertures) are revealed (red arrowheads) in left gills 1 to 3 by removal of gill cowls; intact left gill slits 4 and 5 show cowls and exhalant apertures (yellow arrowheads). (B) Laterally compressed specimen ELI-JS1001A, viewed from the interior. (C) Close-up image of right gill slit 3 in (B) (boxed area); note the oval inhalant aperture (arrowhead) surrounded by plate-like structures and radiating strands. (D) Schematic horizontal section through the lateral midline of Didazoon, with inferred ventilation of feeding-respiratory currents (dotted blue lines). Abbreviations: Co, cowl-like structure; LG1-5, left gills 1 to 5; M, mouth; Phx, pharynx; Po, gill pouch; RG1-4, right gills 1 to 4; Wf, water flow. Scale bars: 1 cm in (A), (B); 1 mm in (C).

Full size image
Figure 4
figure 4

Gill anatomy and body-wall musculature of Vetulicola rectangulata. (A) Laterally compressed specimen ELI-SJ0605A, with preserved muscular fibers beneath the plate-like unit; note also the tightness of the lateral groove. (B) Internal view of right gill slit 4 in (A); note the internal opening (red arrowhead) surrounded by striations. (C,D) Enlargement of boxed areas in (A) at left and right side, respectively, showing fibrous structures interpreted as longitudinal muscles, defined by diagenetic iron minerals; note their location beneath the plates and arrangement as continuous bundles (approximately six fibers per bundle) extending almost the entire length of the anterior section; note also the equidistant expansions (arrowheads), interpreted here as corresponding to a column of horizontal muscles. (E) Specimen ELI-0000306, showing inferred muscular imprints represented by longitudinal structures in the anterior section (notably the focused area). (F) Reconstructed horizontal section (membrane in blue, plates in grey, musculature in red), equivalent to section A-A' in (A), demonstrating inferred dynamics of pharyngeal dilation, governed by the collective action of the longitudinal muscle fibers and their horizontal derivatives. Abbreviations: Co, cowl-like structure; Fp, fin-like process; LG, left gill slit; M, mouth; Phx, pharynx; RG, right gill slit; T, tail; Wf, water flow. Scale bars: 1 cm in (A), (E); 1 mm in (B) to (D).

Full size image

A second reason to support the idea of a more dynamic mode of suspension feeding in Vetulicola concerns its overall morphology, which is indicative of relatively active swimming. In principle this too would have assisted with intake of water. The final point, which is less secure, concerns the relatively small size of the apertures as against the inferred capaciousness of the interior cavity (as judged by the sediment infill). While this involves a number of imponderables, the ratio of aperture diameter to interior volume suggests that ciliary flow alone may not have been sufficient to remove the water efficiently.

We suggest that swallowing and the subsequent expulsion of water most likely would have been coordinated, with control of both the size of the oral opening and lateral apertures. If it is accepted that pum** of water was indeed an active process, then the alternating dilation and contraction most likely would have been mediated by a musculature. Exceptionally preserved material reveals a series of longitudinal structures connected by vertical units (Figure 4A,C-E; see also [22]). Similar, albeit less organized, soft tissues have been tentatively compared to either internal appendages or muscles [22]. The structures described here are provisionally identified as body-wall musculature. We make here the assumption that originally they consisted of horizontal and longitudinal components so that their contractions would determine the volume of the anterior cavity (Figure 4F). Alternative possibilities, not least these structures might represent some part of the filtration unit, certainly merit attention. Nevertheless, the evidence already mentioned in support of a dilatory capacity by the grooves would imply the existence of some sort of antagonistic system. Our reconstruction of this muscular activity, which was possibly assisted by an inherent elasticity of the body wall, provides a working model.

Gill slits in other vetulicolian taxa

Taxa such as Didazoon, Pomatrum, Ooedigera, ** might invite comparison to advanced members of this clade, but because of the distinctive bodyplan of the vetulicolians we suggest that this functional feature re-evolved in groups such as the gnathosomes with which, apart from the pharyngeal openings, they share no significant features. We prefer to argue that the vetulicolians are members of the stem-group deuterostomes [13, 16, 17, 24]. Crucial in this regard is the interpretation of the laterally located openings, which as pharyngeal gill slits are taken to be diagnostic of the deuterostomes [36]. Other than what at best seem to be remote analogies in the form of the minute anterior pores in the meiofaunal gastrotrichs [37] no phylum appears to possess comparable structures. On the assumption, therefore, that these openings are not convergent, then the deuterostome status of the vetulicolians means that the spacious anterior would be homologous with the pharynx and the lateral perforations are genuine gill slits.

This conclusion invites brief consideration of three inter-related questions: (a) what implications might the proposed homologies of the gill slits have on other features of the vetulicolians in the context of a deuterostome affinity? (b) What bearing do vetulicolians have on our understanding of deuterostome evolution, not least with respect to the chordates, ambulacrarians and xenoturbellids? And (c) what light might other Cambrian deuterostomes throw on the early evolution of this group?

Our aim here is as much to pose these questions as it is to arrive at unequivocal conclusions. Central to any argument is the recognition of homology. Patterson [38, 39] crystallized the discussion by proposing that homology might be recognized in terms of similarity, conjunction and most importantly congruence. Similarity stands on the shakiest ground given that while clearly these are lateral perforations located on the anterior, these proposed gill slits find no exact counterpart in any other deuterostome. However, pharyngeal openings show a remarkable diversity within the deuterostomes, not least among some of the extinct echinoderms [40]. Their possession is, however, universally regarded as a synapomorphy of the deuterostomes [36]. In addition, developmental data are consistent with the earliest deuterostome gill slits being derived from endodermal pockets [36] and in principle originally these might have been little more than pores. In the case of Patterson's [38, 39] criterion of what he termed conjunction we can at least argue that (a) vetulicolians do not possess any other structures that are more plausibly identified as gill slits, and (b) that their location on the anterior section is consistent with this identification. By congruence Patterson [38] meant a test for homology based 'on the equivalence of homology and synapomorphy' so that the proposed homology is consistent with the other homologies that are employed to define a monophyletic group. First, we can observe that beyond the deuterostomes no other group of metazoans has an array of lateral openings and as noted despite their undoubted homology [36] they still show an enormous range of anatomical configurations. What then of the relationship of this homology to others that might pin down more precisely where vetulicolians fall in the deuterostomes? The range of phylogenetic conclusions that have already been proposed [22] underlie the difficulty in finding other homologies that one might employ in the use of Patterson's criterion of congruence. In the absence of molecular data and the relative unlikelihood of identifying vetulicolian embryos indicating the fate of the blastopore, very few characters serve to unite the deuterostomes as a whole [41]. Elsewhere we have suggested that vetulicolians had an endostyle and a mesodermal endoskeleton [16], but the evidence is somewhat circumstantial. It is, however, also worth drawing attention to Romer's prescient suggestion [42] that an ancestral form had a bipartite form of a 'somatic' unit with musculature and a 'visceral' component with gill slits. Romer applied this concept to the chordates, but we have argued that his 'dual animal' hypothesis is just as consistent when applied to the vetulicolian bodyplan [16].

If one accepts vetulicolians as deuterostomes then this leads to the question of their possible relationships to the other major groups within this superclade. As already noted we argue that a position within the stem-group deuterostomes is the best phylogenetic solution, but as is often the case in such problematic animals from the Cambrian the correct identification of homologies is crucial. In this context the case of the putative cephalopod Nectoccaris [31, 32, 34, 35] provides a useful parallel. Thus any attempt to link the vetulicolians to one or other group within either the ambulcrarians or chordates is largely frustrated not only by the paucity of obvious homologies, but in addition evidence for radical reorganization of some bodyplans (notably echinoderms [43]) as well as both molecular [44] and fossil [45] data that suggest other groups (notably amphioxus) are more or less degenerate. Nowhere is this more evident than in the xenoturbellids (and acoelomorph flatworms) which now appear to be massively simplified deuterostomes (now comprising the Xenocoelomorpha) and the sister group to the ambulacrarians [46]. As such they are uninformative as to either the appearance of the most primitive deuterostomes (let alone bilaterians [47]) or by implication the putative position of the vetulicolians.

Finally, we can ask as to whether, apart from bona fide chordates and the like [24, 45], are there other Cambrian animals that might be informative with respect to early deuterostome evolution? Foremost in this regard are the coeval yunnanozoans, which elsewhere we have argued [16, 24, 48] provide a link not only between the vetulicolians but possibly also with representatives of the stem-group chordates [45]. As with the vetulicolians their distinctive bodyplan has led to a diversity of opinions as to their exact placement, but a general consensus places them in the deuterostomes [16, 20, 24, 48]. While a key piece of evidence is the series of external filamentous gills, especially in Haikouella [48], unequivocal evidence of pharyngeal openings in any yunnanozoan appears to be wanting.

The results we report here have a bearing on the yunnanozoans in as much as just as the vetulicolians have been compared to arthropods, so too Bergström [19] agrees with earlier observations [24, 49] that yunnanozoans have a segmented cuticular surface, at odds with their interpretation as myomeres [48]. This feature, however, need not equate with either an arthropodan or ecdysozoan relationship; nor do we accept Bergström's [19] interpretation of a specimen described by us earlier [16] as evidence for molting. The evidence for a cuticular surface suggests that at least some members of the stem-group deuterostomes possessed a hardened, albeit non-mineralized, exterior. Elsewhere, a remnant of this feature has been tentatively identified in the stem-group chordate Pikaia [45].

To conclude, in our view the demonstration that the anterior lateral structures in Vetulicola not only have a complex configuration but also unequivocally include perforations is consistent with this animal and related taxa belonging to the deuterostomes. So too the identification of a capacious pharyngeal cavity lined with ventral and dorsal food grooves is consistent with the deuterostome bodyplan and the likely origin of the lateral openings so as to allow exit of seawater.

Deuterostome disparity, both between and within the xenoturbellids, ambulacrarians and chordates, is considerable. Despite now largely secure molecular phylogenies a stumbling block has been how best to envisage the common ancestor. There seems little evidence to suggest that the vetulicolians can be interpreted as the sister group of any of the major groups of deuterostome. There is, for instance, no evidence that the gill slits emptied into a common atrium or otherwise collected the water and expelled it via a single opening. In this and other respects vetulicolians differ from the tunicates [25]. There is, however, one point of comparison that may be relevant. While nearly all tunicates [50] are suspension feeders, among the pelagic representatives the salps are exceptional in employing a muscular pum** [51]. While this is associated with a form of jet propulsion [52], the antagonistic system that involves the tunic and the possibility that this pum** system may have originated in the context of filtration [53] invites analogies with Vetulicola. This analogy is, of course, with respect to the dynamics of pum** and given the large and evidently propulsive tail there is no suggestion Vetulicola employed any sort of jet propulsion.

We suggest that despite uncertainties the anatomical evidence presented is consistent with Vetulicola and its allies as being deuterostomes. While aware that the disparity among the crown-group deuterostomes makes comparisons with the vetulicolians problematic, we are unable to identify any character that would secure the place of this Cambrian group within either the ambulcrarians or chordates. We suggest that the vetulicolians are more likely to be members of the stem-group deuterostomes. If our interpretation of the vetulicolians as being among the earliest deuterostomes is correct then this has a number of important implications. First, it is consistent with the suggestion that such ancestral deuterostomes were not sessile, but free-living animals [13, 16, 54, 55] with an expanded pharyngeal cavity. We take this feature, along with five pairs of gill slits and dorsal and ventral feeding gutters, to represent the primitive condition, and is consistent with the supposition that a filter-feeding pharynx had evolved in the last common ancestor of all deuterostomes [36, 56, 57]. In addition, we conclude that at least in Vetulicola active pharyngeal pum** had evolved, and presumably it arose independently of the system seen in the salps and also more advanced gnathostomes.

Even making allowances for loss and simplification, as well as once popular comparisons (such as the hemichordate stomochord to chordate notochord [58]) very few anatomical homologies serve to unite all the deuterostomes. In this latter respect the pharyngeal openings are a key character. For example, in his proposal that suspension feeding was a feature to emerge very early in the history of deuterostomes, Cameron [57] remarked how 'Clearly the evolution of slits in the pharynx is one of the most important events in the evolution of the deuterostomes'. In our view their identification in the vetulicolians supports the idea that they helped to open the doors to deuterostome diversification, first in terms of effectiveness of feeding and subsequently efficiency of respiratory exchange [17]. Elaborations in the form of the gill slits in both the ambulacrarians and chordates are marked by convergences [41], but in their various configurations they presumably helped to contribute to the success of this phylum. So too we see evolutionary innovation in terms of the disposal of the water, such as the subsequent development of an atrium, and arrangement of food grooves to a single ciliated tract either ventrally (enteropneusts) or dorsally (amphioxus and tunicates). Much of subsequent deuterostome evolution appears to have been driven by ecologies that either favored sessility, with the reduction (as in the pterobranchs) or ultimate loss (as in the echinoderms) of gill slits, or increased motility. The latter may have ultimately led to the transformation of the dorsoposterior segmentation seen in the yunnanozoans [24, 48] into a myotomal arrangement with the corresponding evolution of the notochord in the chordates [45]. In conclusion, calls that more data be collected to help secure phylogenetic arguments [4, 5] can now be met on the basis of exceptional material. On this basis we argue that the vetulicolians provide an important benchmark in the documentation of the emergence of the deuterostomes.

Methods

A total of 480 specimens of various vetulicolian taxa (10 species referred to 7 genera), recovered from five localities (Erjie, Sanjie, Jianshan, Tanglipo, and Maotianshan sections) at more or less coeval horizons in the Heilinpu Formation (Lower Cambrian, approximately 520 million years ago), Yunnan, southern China, were prepared and analyzed. About 50 three-dimensionally preserved specimens were 'dissected' to reveal internal anatomy. Details were analyzed using a Zeiss Stemi-2000C microscope (Jena, Germany).

Specimens were reposited in the Early Life Institute (ELI), Northwest University, **'an, the Early Life Evolution Laboratory (ELEL), China University of Geosciences, Bei**g, China, and the Chengjiang Fauna Museum, Chengjiang Fauna National Geopark, Kunming (CFM), China.