A chiton without a foot

Authors


Abstract

Abstract:  The palaeoloricate ‘polyplacophorans’ are an extinct paraphyletic group of basal chiton-like organisms known primarily from their fossilized valves. Their phylogenetic placement remains contentious, but they are likely to include both stem-group Polyplacophora and stem-group Aplacophora. Candidates for the latter position include ‘Helminthochitonthraivensis from the Ordovician of Scotland, which we redescribe here through a combined optical and micro-CT (XMT) restudy of the type material. The 11 specimens in the type series are all articulated, presenting partial or complete valve series as well as mouldic preservation of the girdle armature; they demonstrate a vermiform body plan. The valves are typically palaeoloricate in aspect, but differ in detail from all existing palaeoloricate genera; we hence erect Phthipodochiton gen. nov. to contain the species. The most notable feature of the fossils is the spicular girdle; this is impersistently preserved, but demonstrably wraps entirely around the ventral surface of the animal, implying that a ‘true’ (i.e. polyplacophoran like) foot was absent, although we do not exclude the possibility of a narrow solenogastre-like median pedal groove having been present. Phthipodochiton thraivensis presents an apparent mosaic of aplacophoran and polyplacophoran features and as such will inform our understanding of the relationship between these groups of extant molluscs. An inference may also be drawn that at least some other palaeoloricates possessed an ‘armoured aplacophoran’ body plan, in contrast to the ‘limpet-like’ body plan of extant Polyplacophora.

Mineralized shells comprise the overwhelming majority of the fossil record for taxa that bear them, but are unreliable indicators of total body morphology. In many groups, this may limit the ability of fossils to indicate both minor and major changes in body plan evolution (e.g. Briggs et al. 1983); reliance on shell characters has led to repeated misinterpretations of molluscan relationships even in living taxa (e.g. Schneider 1998; Bouchet and Rocroi 2005). The shell plates of polyplacophoran molluscs (chitons) represent an example of morphological conservatism through a long fossil record. However, certain exceptionally preserved taxa such as ‘Helminthochiton’ thraivensisReed, 1911, from the Ordovician of Scotland reveal an armoured vermiform morphology contrasting with the modern limpet-like chiton bauplan (Sigwart and Sutton 2007).

The classification of polyplacophoran molluscs (sensu lato) divides them into two subclasses: Neoloricata Bergenhayn, 1955, including the living species and fossil allies, and Paleoloricata Bergenhayn, 1955. The most important character dividing the two groups is the presence in neoloricates of a proximal shell layer (the articulamentum) and the concomitant development of extensive lateral insertion plates. Sirenko (1997) suggested that insertion plates may have evolved in more than one lineage; if true this could also imply that the Neoloricata may not be monophyletic. However, the Paleoloricata is an even less satisfactory taxon, defined primarily by the absence of derived characters; it is certainly paraphyletic, representing a grade of early valve-bearing aculiferans. Recent fossil discoveries (e.g. Sutton et al. 2001a; Vendrasco and Runnegar 2004) have emphasized the broad morphological range of ‘palaeoloricate chitons’. We contend (Sigwart and Sutton 2007) that many palaeoloricates, including ‘Helminthochiton’ thraivensis, represent stem taxa to modern aplacophoran groups (Caudofoveata, Solenogastres) rather than to modern chitons and that in these animals at least a vermiform body may be typical.

A thorough understanding of the evolution and phylogeny of basal ‘chitons’ is dependent on articulated and exceptionally preserved fossils such as the ‘H. thraivensis material. Here we present a detailed description of the total morphology of this species.

Materials and methods

Specimens of ‘H’. thraivensis were collected in the late 19th century from the upper Ordovician Lady Burn Starfish Beds of Girvan, Scotland. Specimens are preserved mouldically – all valve and much spicule materials have been dissolved away, leaving voids. The entire original series of the species (11 specimens) was re-examined. NHMUK G.47258 has been selected as the lectotype as it is the first illustration in the original publication (Reed 1911). Specimens with the best apparent preservation of valve and girdle elements were selected from this series for detailed examination by micro-CT scanning and form the basis of this study (Table 1). All specimens were studied optically using a stereo-microscope and photographed by the Natural History Museum photography unit.

Table 1.   Summary of Phthipodochiton thraivensis specimen material.
Museum specimen numberValve preservationGirdle preservation
  1. Measurement data from four specimens noted (*) are presented in detail in Table 2.

Specimens with XMT scans
 NHMUK G.47258* (lectotype)I–VIII presentPresent
 NHMUK G.47246* (paralectotype)I–VI present; V–VI missing, VII–VIII presentPresent
 NHMUK G.47250 (paralectotype)I–II missing; III(?)-V(?) present; VI–VIII missingPresent
 NHMUK G.47251* (paralectotype)I missing; II–VIII presentPresent
 NHMUK G.47253 (paralectotype)I(?)-II(?) present; III–VIII missingPresent
 NHMUK G.47254*I–VIII presentPresent
 NHMUK G.22679*I–VII present; VIII missingPresent
Additional material
 NHMUK G.47247 (paralectotype)I–VI present; VII–VIII missing (remains of a potential second specimen in block)Present
 NHMUK G.47260indet. intermediate valvesPresent
 NHMUK G.47261indet. intermediate valvesPresent
 NHMUK G.47265indet. intermediate valvesPresent

Seven specimens were micro-CT (XMT) scanned using a Metris X-Tek HMX-ST scanner housed in the NHM (London). A tungsten reflection target, 3142 projections and no filter were used for the scans, performed at (190 μA/225 kV) using a 2000 × 2000 detector. Resolution (voxel size) varied with specimen size (28 μm for the lectotype, NHMUK G.47253). Where possible, specimens were scanned with part and counterpart held together in original position, so valves present on both could be reconstructed in entirety. Computer models of the void space (i.e. ‘negative’ models or virtual casts) were created using the custom SPIERS software suite, implementing the methods of Sutton et al. (2001b, 2002); reconstructed images for Figures 1 and 2 were generated as screen captures from this software. Several valves contained entirely within either part or counterpart (i.e. not visible on any surface of the specimen) were imaged in this way; valve VIII of NHMUK G.47254 (Fig. 2P–R) is an example. Data sets were manually interpreted using SPIERS on a slice-by-slice basis to isolate valves and spicular masses or sheets from cracks, debris and other extraneous material. Figure 2Z, EE shows representative tomograms to illustrate the data on which these interpretations are based. In many cases, valves were initially isolated as arbitrary spherical regions of interest for convenience, prior to more detailed noise removal (see Fig. 1I, K, I, N–O, Y).

Figure 1.

 Reconstructions (virtual casts) of valves of Phthipodochiton thraivensis (Reed, 1911). All ×2. V I–V VIII; Valves 1 (head, V I) through 8 (tail, V VIII). Shaded bars link equivalent valves from different specimens, and white vertical lines separate specimens. A–L, NHMUK G.47258 (lectotype). A, V I (head valve), dorsal view. B, V I (head valve), ventral view. C, V I (head valve), anterior view. D, V II (subcomplete), dorsal view. E, V III (partial), dorsal view. F, V IV (partial), dorsal view. G, V V, ventral view. H, V V, anterior view. I, V V VI, ventral view, J, V VII, ventral view. K, V VIII (tail valve), ventral view. L, V VIII (tail valve), anterior view. M–X, NHMUK G.47246. M, V I (head valve), ventral view. N, V II (partial), dorsal view. O, V II (partial), ventral view. P, V VI, dorsal view. Q, V VI, ventral view. R, V VI, posterior view. S, V VII, dorsal view. T, V VII, ventral view. U, V VII, posterior view. V, V VIII (tail valve), dorsal view. W, V VIII (tail valve), ventral view. X, V VIII (tail valve), posterior view. Y-JJ, NHMUK G.47251. Y, V II, ventral view. Z, V III (partial), posterior view. AA, V III (partial), ventral view. BB, V III (partial), dorsal view. CC, V IV (partial), dorsal view. DD, V V (partial), ventral view. EE, V VI (partial), ventral view. FF, V VII (partial), ventro-posterior view. GG, V VII (partial), ventral view. HH, V VIII (tail valve), ventral view. II, V VIII (tail valve), dorsal view. JJ V VIII (tail valve), posterior view.

Figure 2.

Phthipodochiton thraivensis (Reed, 1911). A–R, reconstructions (virtual casts) of valves. S, U–Y, photographs of specimens. T, DD, reconstructions of entire specimens. AA–CC, virtual sections (see materials and methods), Z, EE; XMT tomograms. V I–V VIII, valves 1 (head, V I) through 8 (tail, V VIII); V?, uncertain valve; SS, spicular sheet; M, matrix; V, void (cracks or valves); DE, ‘dark edge’ (see material and methods). A–I, NHMUK G.47253, ×3. A, V I (head valve), dorsal view. B, V I (head valve), anterior view. C, V I (head valve), ventral view. D, V II, ventral view. E, V II, posterior view. F, V III, anterior view. G, V III, ventral view. H, V IV (partial), ventral view. I, V IV (partial), anterior view. J–R, NHMUK G.47254, ×3. J, V II (partial), dorsal view. K, V II (partial), posterior view. L, V III (partial), dorsal view. M, V III (partial), posterior view. N, V VII (partial), dorsal view. O, V V II (partial), posterior view. P, V VIII (tail valve, near complete), ventral view. Q, V VIII (tail valve, near complete), dorsal view. R, V VIII (tail valve, near complete), posterior view. S–T, NHMUK G.47258 (lectotype), ×1. U, NHMUK G.47254, ×1. V, NHMUK G.47251, ×1. W, NHMUK G.47246, ×1. X, NHMUK G.47253. ×1. Y, NHMUK G.47250, subdorsal view, ×2. Z–BB, NHMUK G.47253. Z, unprocessed CT slice showing preservation of spicular sheet and solid voids inside matrix, ×3. AA, BB, virtual sections (see materials and methods). AA, subtransverse section though middle of V V, ×3. BB, subtransverse section through middle posterior of V II, ×5. CC–EE NHMUK G.47250. CC, virtual subtransverse section (see materials and methods) through middle of left-hand valve visible in 25, ×4. DD, ventral view of reconstructed specimen with matrix removed showing impersistently continuous ventral spicular sheet (darker shade) and ‘dark edge’ material (see material and methods, lighter shade), ×2. EE, unprocessed CT slice (subexsagittal section) showing preservation of spicular sheet inside matrix, ×3.

Raw computed tomograms (the ‘slice’ images resulting from a CT scan) were orientated essentially randomly with respect to the specimens under study; they are hence difficult to interpret individually and do not provide transverse sections. The transverse (or subtransverse) sections of Figure 2AA–CC were generated indirectly by virtually sectioning the three-dimensional models, whose generation from the CT data is discussed earlier. This process generates outlines of the two-dimensional regions modelled; these were manually filed for clarity, a patterned fill being chosen to represent the impersistent ‘speckly’ voids of the spicular sheet (see Fig. 2Z, EE). The solid shading in Figure 2AA–BB represents other clear voids in the rock, valves and cracks. For specimen NHMUK G.47254 (Fig. 2Y, CC–EE), a slightly different reconstruction methodology was used. Here, pixels of a mid-level of grey in the raw CT data were picked out for modelling; these represent material either assignable to the spicular sheet or to the ‘edge’ of the rock matrix, where it grades rapidly from the light ‘normal’ matrix shade to the dark outside the specimen (e.g. at label DE, Fig. 2EE). This latter material is referred to herein as ‘dark edge’ material; one edge of this material models the true surface of the specimen (representing the internal cast of valves and/or the rock surface away from the valves), while the other is essentially arbitrary.

Stereo images of individual valves were measured on-screen to determine valve dimensions and the angles of valve apices using measurement tools in Adobe Illustrator; all dimensions presented are an average of a minimum of three successive recorded measurements and averaged across multiple images of the same element (Table 2).

Table 2.   Biometric data for valves of four specimens of Phthipodochiton thraivensis (Reed, 1911).
 Valve length (mm)Valve width (mm)Mixoperipheral fold (mm)Apical angle (degrees)Dorsal angle (degrees)
  1. The four specimens represent the lectotype (NHMUK G.47258) and three paralectotypes. Measurements represent a calibrated average recorded from screen images of computer tomographic models generated from CT scans of embedded fossils.

NHMUK G.47258
 Valve I5.789.231.13121.12
 Valve II8.40
 Valve III8.741.46
 Valve IV
 Valve V11.9413.762.0291.15
 Valve VI
 Valve VII9.9813.64
 Valve VIII
NHMUK G.47246
 Valve I4.655.700.98120.96
 Valve II5.806.79
 Valve III
 Valve IV
 Valve V
 Valve VI11.078.692.2195.53108.69
 Valve VII10.469.381.89132.41105.38
 Valve VIII13.148.591.70119.5798.14
NHMUK G.47251
 Valve I
 Valve II11.4310.333.16119.69
 Valve III13.063.60104.38114.03
 Valve IV13.23
 Valve V10.872.99
 Valve VI12.082.24111.84
 Valve VII13.403.01
 Valve VIII12.202.94
NHMUK G.47254
 Valve I
 Valve II100.22
 Valve III6.706.0189.44
 Valve IV
 Valve V
 Valve VI
 Valve VII5.335.6286.38
 Valve VIII6.486.3887.80
NHMUK G.22769
 Valve I4.234.761.4078.81128.63
 Valve II11.507.814.0798.2391.92
 Valve III13.1411.924.00120.90109.40
 Valve IV11.6695.87
 Valve V9.53
 Valve VI
 Valve VII10.51
 Valve VIII

Systematic palaeontology

Class POLYPLACOPHORA Gray, 1821
Order PALEOLORICATA Bergenhayn, 1955
Genus PHTHIPODOCHITON gen. nov.

Derivation of name.  From the greek ‘Phthi-’ (‘waned’, or ‘faded away’) and ‘pod’ (foot), as the first fossil chiton demonstrably lacking a true foot.

Type species. Helminthochiton thraivensisReed, 1911, from the Lady Burn Starfish Bed, Threave (Thraive) Glen, Girvan, Scotland (Upper Ordovician).

Diagnosis.  Valves not articulated in life position, lacking articulamentum, overall shaped as a half-ellipse, folded on the shorter axis with the rounded edge at the anterior. Posterior margins pinched at the midline fold, forming a pronounced point but not a significant projecting beak. Apical angle weakly obtuse, jugal angle weakly acute. Head and tail valves slightly smaller than intermediate valves. All valves mixoperipheral with significant (ventral) apical area extending to 30 per cent of valve length. Ornament of growth lines only.

Phthipodochiton thraivensis (Reed, 1911)
Figures 1, 2

  • 1911 Helminthochiton thraivensis Reed, pp. 337–339, pl. 15.

  • 1932 Helminthochiton thraivensis Reed; Quenstedt, p. 85 [annotation].

  • 1939 Helminthochiton thraivensis Reed; Berry, p. 207.

  • 1960 Helminthochiton thraivensis Reed; Smith, p. I48, fig. 33, 2a–c.

  • 1967 Helminthochiton thraivensis Reed; Curry and Morris, p. 423.

  • 1977 Gotlandochiton thraivensis (Reed); Sirenko and Starobogatov, p. 31.

  • 1981 Septemchiton thraivensis (Reed); Rolfe, p. 677.

  • 1987 Septemchiton? thraivensis (Reed); Smith and Hoare, p. 55.

  • 2004 Helminthochiton thraivensis Reed; Cherns, p. 452.

  • 2004 Septemchiton? thraivensis (Reed); Vendrasco and Runnegar, pp. 679, 686.

  • 2006 Helminthochiton thraivensis Reed; Hoare and Pojeta, p. 15.

  • 2007 ‘Helminthochiton’ thraivensis Reed; Sigwart and Sutton, pp. 2413–2419 and electronic supplement, fig. 2.

  • 2010 ‘Helminthochiton’ thraivensis Reed; Donovan et al., pp. 935–937, figs 2, 3.

Lectotype.  Here designated as NHMUK G.47258 (Reed 1911, pl. XV, fig. 1).

Material.  Specimens collected by Mrs Gray, Natural History Museum (London); Paralectotypes examined; NHMUK G.47251 (Reed 1911, pl. XV, fig. 2), NHMUK G.47247 (Reed 1911, pl. XV, fig. 4), NHMUK G.47246 (Reed 1911, pl. XV, fig. 7), G.47250, G.47253.

Additional material examined; NHMUK G.22679, NHMUK G.47260; NHMUK G.47261; NHMUK G.47265; NHMUK G.47254.

Horizon.  ‘Lady Burn Starfish Bed’, Threave Glen, Girvan district, Ayrshire (Strathclyde), southwest Scotland; NGR NS 250 038. Drummuck Subgroup, South Threave Formation, Farden Member (Harper 1982a, b); Upper Ordovician, Katian; Ashgill Series, Rawtheyan Stage (= Richmondian, Cincinnatian in terms of the North American succession); see Ingham inFortey et al., 2000, fig. 24.

Diagnosis.  As for genus.

Description.  Animals preserved in variably curved posture (Fig. 2S–T), interpreted as analogous to the post-mortem contortion typical of extant chitons and aplacophorans, curled in toward the pedal groove. Body shape long and narrow, from seven to ten times long as wide; total body length 40–99 mm, width 5–10 mm. The valves in series are preserved in presumed life position, in almost all cases separated by gaps. Within these gaps between valves, the tissue is covered in spicules; although these are rarely resolvable as individual elements, they form a distinct spicular sheet (e.g. Fig. 2Y). Spacing between valves is variable between and within individuals. Valves do not articulate but do overlap in more curved specimens (compare Fig. 2S–U); this difference is interpreted as variation following body posture.

Valves are subcarinate, keeled at posterior and rounded at anterior. Elevation (height/width) ratio in intermediate valves varies from c. 0.35 to c. 0.65; ratio does not appear to vary consistently along series, although middle valves (IV–VI) do not include any examples of low elevation. Valves slightly beaked. Valves thin, maximum observed values <0.5 mm (e.g. Fig. 2BB); thinner at dorsal median.

Head valve (Figs 1A–C, M–O, 2A–C) generally similar in outline to intermediate valves but shorter. Head valve of same width as tail valve, slightly narrower than intermediate valves. The valve is also lower, with an elevation ratio of below 0.1 in some cases (e.g. Fig. 2B). Head valve forms somewhat variable in apical angle and expression of median jugal embayment.

Intermediate valves subrectangular, lateral areas differentiated by a variably expressed diagonal depression normally more visible toward posterior apex (see e.g. Fig. 1D). Posterior margin with pronounced apex (mean apical angle 110 degrees); apical angle does not vary appreciably along valve series (including tail valve). Anterior margin with shallow median jugal embayment, median valve length at midline more than 90 per cent of maximum total valve length. Side slopes slightly convex, rounded. Shape of valve II not differentiated from other intermediate valves (in contrast to Recent Chitonida). Tail valve (Figs 1K–L, V–X, HH–JJ, 2P–R) shaped as intermediate valves, with mixoperipheral growth and without dorsal mucro. Planar valve width is slightly longer than total valve length, shape is as a ellipse folded on shorter axis on the jugal ridge. Side slopes straight. Anterior margin convex, evenly rounded. Posterior margin convex, pinched in middle, strongly subcarinate. Lateral areas separated by shallow diagonal depression, more pronounced at posterior margin.

Dorsal valve surface without sculpture other than growth lines, which are strong in many cases (e.g. Fig. 1D, S). Growth lines occasionally reflected on ventral surface (e.g. Fig. 2H) as well as on the apical area. Valves preserve no evidence for aesthetes.

Ventral valve surface smooth with posterior mixoperipheral fold. The mixoperipheral margin (ventral apical area) is shortest in the head valve; in other valves, it occupies a larger portion of more anterior valves to a maximum of 35 per cent of the length in valve II compared to a minimum of 13 per cent of the length of valve VIII. There is general decrease through the valve series, with more posterior valves having a smaller mixoperipheral area; however, there are not sufficient specimens to determine whether this is a robust trend. Valves in two of the specimens examined by XMT have evidence for a ventral posterior tunnel (NHMUK G.47251 valve VII, Fig. 1FF; NHMUK G.47254 valve VIII, Fig. 2P); these are not consistently preserved, and the two tunnels observed may represent different structural artefacts.

The girdle is preserved as a large mass or more typically as a sheet of mouldic spicules. Preservation is not sufficient to make any detailed description of the morphology of girdle elements or to determine whether there are multiple types of spicules present. Girdle extending ventrally around whole body and intruding between valves. The extent of the spicule preservation varies in different specimens, but in most, it appears as a contorted, folded sheet, impersistently preserved but in places clearly continuous ventrally. NHMUK G.47250 illustrates spicule distribution most clearly in three dimensions (Fig. 2Y, CC); in other specimens variously the contortions of specimens, impersistent preservation and presence of cracks render satisfactory three-dimensional visualization of the spicule sheet impossible. Virtual cross-sections (Fig. 2AA–CC), however, demonstrate ventral continuity in NHMUK G.47253 as well as in NHMUK G. 47250; note that the light-coloured object descending from the left of the valve in Figure 2AA represents a major crack, which we interpret as following the line of weakness left by the sheet of mouldic spicules. The spicular sheet includes erratically placed ventral disturbances; because of the asymmetric and inconsistent nature of these disturbances, we interpret them as taphonomic artefacts, where the spicules have been disturbed in process of preservation, rather than indicating the substantial gaps in the girdle armature that would be required to accommodate a polyplacophoran-like foot. The spicules appear to have contact with the ventral part of the valves (Fig. 2AA–CC), perhaps indicating direct attachment.

There is no evidence for preservation of a radula nor any softer tissues in any of the specimens examined.

The lectotype, NHMUK G.47258, preserves a series of associated shelly debris extending from valve IV to the posterior end of the tail valve (Fig. 2T). Items are arranged in a discontinuous string ventral to the midline of shell series, in a position that corresponds to the presumed placement of the gut in the living animal. Where identifiable, debris includes pieces of calcified subcircular rings with variable-sized hole and small fragments with crenulated structures; we interpret these to represent crinoids and trilobites. The crinoid material in these gut contents has been described by Donovan et al. (2010, 2011).

Remarks. Reed (1911) originally assigned Phthipodochiton thraivensis to the genus Helminthochiton essentially on the basis of its overall elongate shape. The Silurian type species of this genus, H. griffithi, was recently redescribed by one of us (Sigwart 2007). H. griffithi remains poorly characterized (known only from the holotype), but differs from the material described here in the much closer articulation of its valves, the possession of radial ornament, in many details of valve outline (such as the lack of an anterior embayment, lack of a sharp apex and lower length/width ratio). The differences are particularly apparent in the head valve, which in H. griffithi is relatively large and more elongate than the trunk valves. We hence consider the assignment of thraivensis to this genus to be untenable and assign it to the new genus Phthipodochiton. The systematics of paleoloricate ‘polyplacophorans’ is in need of further general revision, and in particular, we lack confidence in the monophyly of certain familial and higher-level taxa in use for these fossils. To establish the generic distinctiveness of Phthipodochiton, therefore, we have exhaustively compared Phthipodochiton with all other ‘chiton’ genera reported from Ordovician rocks; these are dealt with alphabetically below. None of these are sufficiently similar to warrant placement of Phthipodochiton thraivensis within any established genus, but we tentatively note that it seems morphologically most closely allied to Alastega, Robustum and Septemchiton.

AlastegaCherns, 1998a, was erected to house a Silurian-aged species; Hoare and Pojeta (2006) have since assigned other Ordovician fossils to the genus. Alastega intermediate valves differ from those of Phthipodochiton in their acute apical angle and low length/width ratio, and head valves are not closely comparable (those of Alastega are elongate and lack a clear apical area). AmblytochitonHoare and Pojeta, 2006, was placed within the same family as Alastega; it is relatively poorly characterized, but differs from Phthipodochiton at least in its head valves (elongate and subrectangular) and in the acute apical angle of its intermediate valves. CalceochitonFlower, 1968, is also poorly known but similar to Matthevia. ChelodesDavidson and King, 1874, has become something of a ‘bucket genus’ to which a variety of Ordovician species have been assigned. The Silurian type species C. bergmani was, however, redescribed by Cherns (1998a); while only one intermediate valve is known, it clearly differs from Phthipodochiton in its relatively long apical area, rounded posterior margin, lack of anterior embayment, higher length/width ratio, rounded jugum and other details besides. EochelodesMarek, 1962, is Chelodes like in possessing a long apical area and straight anterior margins, and hence not closely comparable to Phthipodochiton. GotlandochitonBergenhayn, 1955, is another genus with a Silurian type species to which Ordovician fossils have been assigned (by Smith inSmith and Toomey, 1964). The type species G. interplicatus was redescribed by Cherns (1998b), but is known only from the dorsal surface on a single intermediate valve; this differs from Phthipodochiton valves in its lower apical angle, convex anterior margin, concave posterolateral margins and lower length/width ratio. As discussed by Vendrasco and Runnegar (2004), HelminthecellaStinchcomb and Darrough, 1995, is very similar to Matthevia.

The type species of Ivoechiton Smith inSmith and Toomey, 1964, is imperfectly known, but intermediate valves possess a much lower length/width ratio, higher jugal angle and a higher apical angle than those of Phthipodochiton, which they do not closely resemble in outline. KindbladochitonVan Belle, 1975 (= Eochiton Smith inSmith and Toomey, 1964), is also imperfectly known; its intermediate valves are broader than those of Phthipodochiton and lack a sharp apex and an anterior embayment. ListrochitonHoare and Pojeta, 2006, differs from Phthipodochiton in its lower apical angle and distinctive heart-shaped valve outline in all valves, lower jugal angle and in its much longer apical area, extending to near mid-length. Valves of LitochitonHoare and Pojeta, 2006, lack a sharp jugum and straight sides of Phthipodochiton, possess parallel lateral margins and lack a true apical area. They also possess strong concentric ornament. MattheviaWalcott, 1885, is a distinctive form that occurs in both Cambrian and Ordovician rocks. It differs from Phthipodochiton, and indeed from most paleoloricates, in the extreme elongation and low apical angle of its intermediate and tail valves, in the extremely long apical area not fully flattened onto the valve surface (forming the lower part of a compressed cone shape) and in its flattened holoperipheral head valves. OrthriochitonVendrasco and Runnegar, 2004, was erected for Cambrian material, but an Ordovician genus was assigned to it by Hoare and Pojeta (2006). The genus differs significantly from Phthipodochiton in its valve outline, notably in its acute apical angle, and its deep and straight-sided anterior embayment. The putative head valve figured by Vendrasco and Runnegar (2004, fig. 12.26) is subcircular, and not like that of Phthipodochiton.

Paleochiton Smith inSmith and Toomey, 1964, is poorly characterized; the holotype and three paratypes are all fragmentary intermediate valves. These differ from those of Phthipodochiton in their outline, which is subrectangular with straight and parallel lateral margins and an apical angle near 180 degrees; they also possess shorter apical areas. PreacanthochitonBergenhayn, 1960, is similar to Phthipodochiton in valve outline although with a lower apical angle; however, Phthipodochiton valves lack the distinctive pustulose ornament of the type species Preacanthochiton cooperi and possess a much sharper jugum. PriscochitonDall, 1882, is similar to Matthevia. RobustumStinchcomb and Darrough, 1995, possesses elongate subcylindrical intermediate valves similar to those of Septemchiton. Vendrasco and Runnegar (2004, p. 686) drew attention to the apparent saddle-like form of an internal mould of Phthipodochiton thraivensis figured by Reed (1911, pl. XV, fig. 3), comparing it to similar saddle shapes in Robustum; we do not consider this observation to be indicative of any particularly close relationship between the genera. SarkachitonDzik, 1994, is poorly characterized, but differs from Phthipodochiton in possessing apparent spiny ornament, subrectangular tail valves and an acute apical angle. SeptemchitonBergenhayn, 1955, possesses elongate subcylindrical intermediate valves, with complete overlap when articulated; it also differs from Phthipodochiton in possessing a very small head valve (see Rolfe 1981). The head valve in some specimens of Phthipodochiton is reduced but not to the dramatic extent seen in Septemchiton. While some authors (e.g. Vendrasco and Runnegar 2004) have tentatively assigned Phthipodochiton thraivensis to the genus Septemchiton, we consider the differences in intermediate valve morphology to be sufficient for differential diagnosis. Note also that the genus Solenocaris is considered a synonym of Septemchiton (see Rolfe 1981). Lastly, SpiculchelodesCherns, 1998a, was erected to house a Silurian-aged species; Hoare and Pojeta (2006) have since assigned Ordovician fossils to the genus. The type species S. pilatis differs from Phthipodochiton in its distinctive pustulose ornament, low jugal angle and rounded jugum, long apical area and high length/width ratio.

In summary, none of the genera examined appear especially closely comparable to Phthipodochiton thraivensis; while many are inadequately known, in all cases, there are differences in multiple characters. It should also be noted that the valves of P. thraivensis are relatively thin. While valve thickness has rarely been described for other taxa, this may serve as another distinguishing character. The erection of the new genus Phthipodochiton is hence taxonomically warranted although we prefer not to assign this to any higher taxa beneath the level of the Paleoloricata until further revisionary work has been completed.

Discussion

The preservational style of the P. thraivensis specimens can fairly be described as challenging. First, the mouldic nature of the fossils necessitates a reliance on casts; while virtual casting via XMT is clearly a more practical solution than physical casting, specimens are obscured by the part/counterpart boundary crack, many secondary cracks, the spicular material and the curvature of the animal itself, and are difficult to reconstruct and image in their entirety. See Figure 2T and Donovan et al. (2010, fig. 3) for the most complete available views (of the lectotype), and Figure 2DD for the best model of the disposition of spicules. Additionally, the spicular material itself is represented as a distorted and in some cases partially broken sheet, complicating interpretations. Nonetheless, it is clear that for much of the length of the animal, this sheet was ventrally complete or near complete. Spicules are associated solely with cuticle in aculiferans. We hence infer that P. thraivensis had a ventrally complete or subcomplete cuticle and must have lacked a ventral foot of the type found in all extant Polyplacophora, although we do not exclude the possibility of a median pedal groove analogous with that of the neomeniomorphan aplacophorans (Solenogastres).

Spicule preservation around the posterior of our specimens is equivocal, and we lack direct evidence as to the continuity of the cuticle in the region of the tail valve. Despite this, we infer that a solenogastre-like posterior respiratory cavity must have existed, unless the taxon had a respiratory mode quite unlike that of any extant mollusc. Its mode of locomotion must have been different to that of extant chitons, which creep on their foot; we suggest again that a solenogastre model may be appropriate here. Note that the other group of extant aplacophorans, the chaetoderms (= caudofoveates), are infaunal and likely derived and highly modified (Scheltema 1993; Giribet et al. 2006); we consider them to be inappropriate as a model of plesiomorphic character states.

Valve-bearing aplacophoran-like fossils are already known in the form of the Silurian Acaenoplax (Sutton et al. 2001b, 2004), albeit bearing valves less obviously polyplacophoran like than those in the current study. Our phylogenetic analysis of the Aculifera (Sigwart and Sutton 2007) placed both Phthipodochiton and Acaenoplax in the aplacophoran stem lineage; if Acaenoplax provides clear evidence of the presence of valves on an aplacophoran-like organism, Phthipodochiton demonstrates that the aplacophoran-like body plan extended to animals bearing more typically chiton-like valves.

Conclusions

The taxon Phthipodochiton thraivensis is important for understanding molluscan evolution. Its valves are different in detail from any previously described taxon, but they fit easily nonetheless within the morphological range of palaeoloricate ‘chitons’, for which its footless bauplan may well have been typical. It also demonstrates the importance of exceptionally preserved fossils for the understanding of molluscan evolution and the power of XMT to resolve details of ‘difficult’ mouldic fossils.

The next major goal in the study of fossil polyplacophorans, and hence for understanding the evolution of the Aculifera, should be a concerted effort to resolve the phylogeny of ‘paleoloricate’ chitons; we contend that this group includes stem-group Polyplacophora, stem-group Aplacophora and likely also stem-group Aculifera. This work must be realistic about the limitations of nonarticulated shell material, which may be misleading for, or impossible to apply to, the inference of real phylogenetic relationships (Sigwart et al. 2007). To expand on the sparse analytical work to date (Cherns 2004; Sigwart and Sutton 2007; Pojeta et al. 2010), shell morphology should be combined with stratigraphic and geographic distribution information and supplemented by the careful analysis of articulated material, especially where novel imaging techniques can throw new light on old fossils.

Acknowledgments

Acknowledgements.  We thank the Natural History Museum, London, for permitting access to the specimens and XMT scanner, and Lesley Cherns and Bruce Runnegar for constructive reviews. Mrs Catherine Bothwell (Saintfield, Northern Ireland) graciously provided expert advice for the etymology.

Editor. Dr Phil Lane

Ancillary