New reconstruction of the Wiwaxia scleritome, with data from Chengjiang juveniles

Wiwaxiids are a problematic group of scale-covered lophotrochozoans known from Cambrian Stages 3–5. Their imbricating dorsal scleritome of leaf-like scales has prompted comparison with various annelids and molluscs, and has been used as a template to reconstruct the articulation pattern of isolated Small Shelly Fossils. The first articulated specimens of Wiwaxia from the Cambrian Stage 3 Chengjiang Konservat-Lagerstätte show that the Wiwaxia scleritome comprised nine equivalent transverse rows associated with outgrowths of soft tissue, but did not possess a separate zone of anterior sclerites. This serial construction is fundamentally incompatible with the circumferential disposition of sclerites in early molluscs, but does closely resemble the armature of certain annelids. A deep homology with the annelid scleritome must be reconciled with Wiwaxia’s mollusc-like mouthparts and foot; together these point to a deep phylogenetic position, close to the common ancestor of annelids and molluscs.


Material and Methods
Five new Wiwaxia specimens, each comprising part and counterpart, have been collected from Chengjiang by the Early Life Institute working team, and deposited in the Early Life Institute and Department of Geology, Northwest University, Xi'an, China (Prefix: ELI). These complete, articulated specimens represent organisms in various states of enrolment, preserved at various orientations to the plane of splitting. Sclerites and mouthparts are represented by regions with a deep purple to black colouration, and in the best cases correspond to an intact layer of carbon. Regions corresponding to soft tissue are coated with rust-coloured framboids, apparently arising through the oxidation of pyrite.
Burgess Shale specimens of W. corrugata are deposited in the Smithsonian Institution National Museum of Natural History (NMNH), Washington DC, and the Royal Ontario Museum (ROM), Toronto, and represent unweathered carbonaceous compression fossils associated with diagenetic aluminosilicate films 15 . Systematic Palaeontology. This published work and the nomenclatural acts it contains have been registered in Zoobank: http://zoobank.org/Referencesurn:lsid:zoobank. org:pub:1B5E0AE5-2FB2-4EFF-B35B-8293D919DEE8 Family Wiwaxiidae Walcott, 1911(nom. corr. Howell, 1962 Genus Wiwaxia Walcott, 1911 Emended diagnosis. Ovoid body bearing nine transverse rows of ribbed carbonaceous sclerites, arranged in bundles and directed towards the posterior. Anteriormost sclerite row terminal. Sclerites comprising narrow root and wide blade, and incorporating narrow internal longitudinal chambers. Sclerite morphology varying consistently across each transverse row; medial sclerites rounded, ventro-lateral sclerites elongate and curved, usually with pointed tip. Dorsal surface of adults often with elongate spinose sclerites. Ventral surface comprising unarmoured 'foot' . Toothed feeding apparatus comprising two to three rows of curved carbonaceous teeth arranged on tongue-like supporting apparatus. Holotype. ELI-W001 ( Fig. 1a-b), an almost complete dorsoventral specimen preserving mouthparts and soft tissue.
Stratigraphic setting. Specimens were collected from the yellowish-green to greyish-green mudstones of the Chengjiang Lagerstätte at the Jianshan Section in Haikou, Kunming. Other taxa recovered from this site include the early agnathan Haikouichthys 16 and the echinoderm-like vetulocystids 17 .
Remarks. The examined material resembles juvenile specimens of Wiwaxia corrugata in terms of its overall body size, the form of its mouthparts, the relatively large size of dorsal sclerites, the broad yet short ventrolateral sclerites, and the absence of dorsal spines. Adult specimens are conceivably represented by the larger isolated sclerites that have also been reported from Chengjiang 18 (Fig. 3), though this material is difficult to exclude from other Wiwaxia species. W. papilio sp. nov. is distinguished from W. corrugata based on the low number ribs on its sclerites; detailed comparison with other species is hampered by the shortage of comparative material 1,5 .
Description. The articulated specimens of W. papilio sp. nov. (  . Two specimens preserve mouthparts with two to three rows of carbonaceous teeth (Fig. 2a,b). As the morphology of individual teeth is indistinct, it is not clear whether or not small lateral teeth are present; otherwise, no substantive differences from W. corrugata (Fig. 2c,d) are evident. The anterior row of sclerites is made up of two bilaterally-paired rosettes, across which the morphology of sclerites varies in the same fashion as it does elsewhere on the body: the most ventral sclerites are siculate, whereas the more dorsal sclerites are rounded and occasionally asymmetric (Figs 1c and 4b).
Siculate sclerites also form part of the anterior row of Burgess Shale specimens of Wiwaxia corrugata (e.g. Fig. 5b,d), occurring at the lateral edges of the scleritome but not skirting the front of the animal (contra ref. 4). As such, the anterior sclerites do not form a distinct zone of the scleritome, but represent a (ninth) transverse row of sclerites.
The ventral surface of the fossils is represented by an iron-rich region that we interpret as soft tissue in a position dorsal to the foot. Bundles of sclerites insert into lateral projections of this iron-rich region (Fig. 1a-c). Sclerite bundles are anchored in equivalent projections in W. corrugata (Fig. 5a-g; see ref. 12), where equivalent projections are connected by transverse bands of connective tissue (Fig. 5h); these themselves are embedded in the soft tissue of the organism (Fig. 5h).  12 . This reflects a more general scarcity of adult Wiwaxia specimens in shallow-water settings. Bedding-surface fossils from the Buchava, Hongjingshao and Kaili formations exclusively correspond to juvenile size ranges and morphologies 2,3,19 . (These localities, like Chengjiang, preserve shallow-water communities; in the case of Kaili, shallow-water taxa were washed into deeper waters before burial [19][20][21][22][23] ).
In contrast, deep-water settings are replete with adult Wiwaxia. The deep water Tsinghsutung (= Qingxudong) Formation contains disarticulated sclerites that correspond to the size range of sclerites in adult Wiwaxia corrugata, and includes elongate sclerites that conceivably represent spines 3,24 . The Spence Shale and Sinsk Biota, which were deposited below storm wave base 25,26 , contains articulated and disarticulated sclerites belonging to Wiwaxia adults 1,6,27 . And in the Burgess Shale, adult Wiwaxia are present in great abundance at the deeper water localities on Fossil Ridge and Mount Stephen 4,12,28,29 but have not yet been found in the shallow-water Marble Canyon locality 30 .
Wiwaxia juveniles occur in almost all geographic and ecological settings 31,32 , perhaps reflecting planktonic larval dispersal 33 . The rarity of adult specimens in shallow waters may therefore represent failure to reach maturity in these environments -whether through active migration to deeper water, or through accentuated predation pressure on adult organisms.

Phylogenetic implications of scleritome constitution.
Since the discovery of the first articulated specimens 10 , sclerite disposition has played a central role in determining Wiwaxia's biological affinity.
One obvious analogue to Wiwaxia sclerites are the conspicuous dorsal scales (elytra) of aphroditid and polynoid annelids 10,34 -but these fleshy outgrowths are not secreted by microvilli, so cannot be equivalent to Wiwaxia sclerites 4,11 . The modified paleal chaetae of chrysopetalid annelids represent a more promising analogue 11 ; as with Wiwaxia sclerites, chrysopetalid paleae occur in a series of bundles or fans across a transverse rows 35,36 , and indeed sclerite morphology even varies from siculate lateral sclerites to more symmetrical dorsal sclerites 37 . This correspondence also rings true on the level of sclerite construction: chrysopetalid paleae, like Wiwaxia sclerites, comprise a proximal root and a broad distal blade, and on a more superficial level may exhibit ribs, a granular ornament, and a distal prong (cf. ref 32).
Despite this compelling similarity, there is a fundamental objection to a chrysopetalid affinity: chrysopetalids are fundamentally derived crown-group annelids [38][39][40][41][42] , whereas Wiwaxia lacks key synapomorphies such as biramous parapodia, palps and aciculae and thus belongs outside the annelid crown group 7,[42][43][44] . Equally problematic is the location of the ventral mouthparts in Wiwaxia beneath the second or third sclerite row: this is difficult to reconcile with the anterior position of the annelid prostomium. As such, the detailed similarity between Wiwaxia sclerites and those of chrysopetalids must be attributed to convergent evolution.
Could the molluscs provide a more reasonable analogue for the Wiwaxia scleritome? Of the extant molluscs, only aculiferans (= Polyplacophora + Aplacophora) bear comparable sclerites. Polyplacophoran sclerites exhibit crystalline cores that are surrounded by a thin cuticular layer with a microvillar texture 45,46 , which is conceivably homologous (at a deep level) to the sclerites of Wiwaxia 12 -but polyplacophoran sclerites are arranged in concentric zones rather than transverse rows, and exhibit a broadly quincuncial disposition 47 rather than occurring in bundles. Aplacophoran molluscs do exhibit transverse rows of dorsal sclerites at early developmental stages 13,48 -but stem-group aplacophorans resemble polyplacophorans 49,50 , meaning that Wiwaxia would have to represent a surprisingly early and extremely derived aplacophoran that retained larval features to adulthood and developed a precise sclerite organization unseen in modern representatives 12 . As such, no living mollusc provides a convincing analogue for Wiwaxia's scleritome.
Although the detailed construction of the Wiwaxia scleritome has no precise equivalent in modern or fossil groups, this is not to say that it does not share homologies at a deeper level. Carbonaceous sclerites are secreted by microvilli in bryozoans, brachiopods, molluscs and annelids 46,51,52 , and are likely homologous across Lophotrochozoa 53 . The iterated arrangement of Wiwaxia sclerites is paralleled by basal annelids, and possibly brachiopods [54][55][56] . Finally, the bundling of sclerites in Wiwaxia could conceivably foreshadow the neuropodial and notopodial bundles observed in crown-group annelids. Whilst it is conceivable that the iterated nature of the Wiwaxia scleritome arose convergently, we prefer to assume homology in the absence of clear evidence to the contrary.
The broadly annelid-like serial construction of the Wiwaxia scleritome must be balanced against the conspicuously mollusc-like nature of its mouthparts and foot 12,14,57,58 . An equivalent paradox is represented in Hallucigenia, which bears onychophoran-like claws alongside cycloneuralian-like mouthparts; this is resolved if cycloneuralian-like mouthparts occurred in the common ancestor of onychophorans and cycloneuralians 59 . As the affinity of Wiwaxia is less well established, there is more than one way to reconcile its molluscan and annelidan features. Either annelid-like sclerite rows or molluscan mouthparts may have been present in the common ancestor of annelids and molluscs and been retained for some time in the stem lineages of each phylum.
If Wiwaxia is a stem-group mollusc (Fig. 6a), a scleritome of iterated rows was conceivably ancestral to annelids and molluscs, and was later rearranged into the circumferential format of aculiferans. If Wiwaxia is a stem-annelid (Fig. 6b), a muscular foot and radula-like mouthparts are ancestral to molluscs and annelids, with a serially arranged scleritome unique to the annelid stem and ultimately leading to metamerism and full segmentation in the crown group. Under this arrangement, annelids replaced their ancestrally mollusc-like mouthparts with an independently-derived and non-homologous 14 jaw, just as onychophorans replaced their cycloneuralian-like mouthparts with independently-derived jaws 59 . The third possibility is that Wiwaxia falls in the stem lineage of Mollusca + Annelida. Under this scenario, both phyla exhibit a loss or overprinting of primitive features: the foot and mouthparts in annelids, the transverse sclerite arrangement in molluscs.
Ultimately, a more complete record of early Lophotrochozoan evolution is needed before the polarity of these distinctive characters can be resolved. But whatever the exact phylogenetic position of Wiwaxia, it clearly diverged before the modern phyla had attained their distinctive body plans, and thus represents a valuable proxy for the common ancestor of molluscs and annelids.