Discussion.—Although traditionally classed as hexactinellids, the relationships of many Cambrian-Ordovician hexactine-bearing sponges are now recognised to be unclear, and they may instead represent stem-group hexactinellids or stem-group siliceans. The common ancestor of Silicea + Calcarea is likely to have had hexactinellid-like spicules (Botting and Butterfield 2005), and so it is potentially difficult to distinguish true hexactinellids from basal and stemgroup sponges. A full discussion of this is beyond the scope of the paper, but we draw attention to the uncertainty as a matter of accuracy.

Remarks.—The higher-level classification of complex reticulosans has not yet been fully established. Although several superfamilies and families have been erected (see Finks and Rigby 2004b), these relate only to a few structurally distinct architectures, but even these as currently defined are inconsistent. Valospongia was classified by Rigby et al. (2010) as a hydnodictyid dierespongioid. The Hydnodictyidae were defined (Finks and Rigby 2004b) as possessing two nonaligned layers of hexactine-based spicules, each organised in parallel arrays. Their diagnosis of the Dierespongioidea expresses similar concepts, but excludes taxa bearing parietal gaps. Not only are such gaps present in the inner layer of Valospongia, but the prominent protuberances are not seen in other taxa in the family. It is also unclear whether the simple presence or absence of multiple skeletal layers with a degree of regularity is sufficient to define groupings on this level, and to what extent it is convergent. We therefore leave the genus in uncertain familial and superfamilial taxonomic position pending a better understanding of early complex reticulosans.

Type species: Valospongia gigantis Rigby, 1983, from the lower part of the middle Cambrian Marjum Formation of Utah, USA.

Species included. Valospongia gigantis Rigby, 1983; Valospongia bufo sp. nov.

Emended diagnosis.—Barrel-shaped to vasiform sponges with a complex skeleton composed of at least two layers; continuous inner layer or layers of semi-quadruled, semi-regular hexactines or their derivatives combining to form orthogonal and diagonal elements, and outer layer of similar spicules that is extended into prominent tubercles covered by a fine reticulation, which may or may not be spicular; prominent spicular strands developed in some species.

Remrks.—Valospongia is a complex sponge (see diagnosis of Finks and Rigby 2004b), with several apparently unique features, and the diagnosis has been reclarified here in light of the new material. It is unclear whether the two inner layers recognised by Finks and Rigby (2004b) are separate, or whether they form one layer with combined orthogonal and diagonal elements; the two series sometimes do appear distinct, however, and are treated as such here. The nature of the skeletal strands emphasized by Finks and Rigby (2004b) is unclear, but they may represent fusion or close overlapping of elongated spicule rays. The apparent parietal gaps are seen in the new material to reflect abrupt curvature of the wall into the mounds, which appear to be covered by a reticulate meshwork of small hexactine-based spicules in irregular arrangement, but it is possible instead that it represents soft tissues replaced by pyrite (or soft tissues supported by microscleres). The arrangement in these meshworks is not orthogonal, and the lines are irregular and perhaps curved, suggesting a non-spicular origin; it is possible that they were composed of similar material to that of Vauxia (e.g., Rigby 1986).

Despite superficial similarity in appearance of the sponge, there is certainly no dermal quadruled grid as seen in Hintzespongia and allies; however, neither is there any obvious reason to assign the genus to the Hydnodictyidae (cf. Finks and Rigby 2004b), representatives of which appear to have possessed neither the multiple layers of Valospongia, nor the meshwork-covered mounds. For these reasons, we prefer to leave the family-level assignment of Valospongia open. We include the new species within Valospongia, as although it does not appear to possess the skeletal strands of V. gigantis Rigby, 1983, the remaining structure conforms closely to the type material.

Stratigraphic and geographic range.—Currently confirmed only from the middle Cambrian of Utah,USA and the Early Ordovician of Wales, UK.

Etymology: From Latin bufo, toad, in appreciation of the pronounced wart-like mounds all over the dermal surface.

Type material: Holotype: NIGP154629 a near-complete specimen showing both skeletal layers. Paratypes: 8 specimens, NIGP154630 to NIGP154637. NIGP154630 a partial, three-dimensional specimen showing outer layer and poorly preserved spiculation from inner layer in some areas; from loose but local block at Amnodd Wen; NIGP154631, a near-complete but weathered specimen found as a loose block beside the track near Amnodd Wen, 100 m up the track towards Ceunant-y-garreg-ddu; NIGP 154632, a partial specimen from Ceunant-y-garreg-ddu approximately 39 m stratigraphically up in the section from where the stream crosses the path; NIGP154633, a small mesh fragment mostly preserving parietal outer layer, and weak spicule preservation from Ceunant-y-garreg-ddu, approximately 18 m stratigraphically up in the section from where the stream crosses the path; NIGP154634, three partial specimens on a loose block from Ceunanty-garreg-ddu, including profile of margin with protruding mounds; NIGP154635, loose block with two partial specimens, from Ceunant-y-garreg-ddu; NIGP154636, loose block with three partial specimens, from Ceunant-y-garreg-ddu; NIGP154637, near-complete specimen with three-dimensional preservation of the skeletal mesh but weak spicule preservation, from loose but local block at Amnodd Wen; age for all specimens as for holotype.

Type locality: Ceunant-y-garreg-ddu stream section, found as a loose block below outcrops 40 m stratigraphically above the point where it crossed the track (SH 82103600).

Type horizon: Late Tremadocian (early Migneintian).

Diagnosis.—Relatively small Valospongia without obvious skeletal strands; walls between parietal gaps in outer layer narrow (less than half diameter of gaps); inner layer(s) of orthogonal and diagonal reticulation densely arranged and sub-regularly quadruled; spicules small relative to parietal gap size.

Description.—Rounded, roughly barrel-shaped sponge body (Fig. 4A₁, B, C₁). Holotype (Fig. 4C) 31 mm wide and over 40 mm tall (estimated height c. 45 mm), with other specimens of comparable size. Base broadly rounded. Oscular margin is flat, nearly as wide as widest point of sponge, and slightly irregular. Appearently no distinct marginalia, but very short (0.3–0.4 mm) spicule rays project subvertically from the edge. Weakly-preserved, slightly longer (less than 1 mm) spicule rays also project laterally from the body wall, but do not constitute discrete prostalia.

Body wall consists of a tuberculate primary (outer, probably dermal) layer, and a thin inner (probably gastral) spicule layer. Primary skeletal wall layer with acute rounded mounds (Fig. 4A₁ B, D₁) that penetrate into the matrix, giving the appearance of numerous subcircular spaces, typically 2 mm in diameter (clearest in Fig. 4A₁, A₂) but proportionally smaller in small specimens (around 1 mm in the 18-mm tall NIGP154634; Fig. 4D₁). Some smaller spaces are interspersed irregularly with the larger ones, and occasionally a large gap is subdivided by an additional curving wall (Fig. 4C₄, at right edge), typically leaving one half smaller and with a concave side. At the upper margin of the sponge, these gaps are less clearly defined and smaller, around 1 mm in diameter. Mounds form a disordered array, roughly closepacked, and with skeletal areas between gaps normally 0.5 mm wide at thinnest, widening slightly in triple-junctions but in general maintaining constant thickness (although dependent on degree of compression of specimens); this produces slightly polygonal, rounded spaces rather than strictly circular apertures. The inter-mound areas may be proportionally wider at earlier growth stages, but too few small specimens are available for confirmation.

Mounds, where visible in profile (Fig. 4B, D₁) at the sponge margin, can be up to 2 mm tall in large specimens, and normally around 2 mm wide; this accords with the diameter of spaces described above, indicating that the skeletal wall of the mounds was of negligible thickness. Mounds in early growth stages are proportionally smaller, and perhaps disproportionately shorter relative to diameter (Fig. 4D₁. Mounds are covered with a similar irregular mesh (Fig. 4D₂) to that seen in the body wall near the oscular margin, although nowhere preserved clearly enough to confirm the presence of discrete spicules. Where broken through the domes, the internal section is often preserved with a thin reflective film (Fig. 4D₁), presumed to be composed of aluminosilicates, but fainter than in the skeletal wall. In many of the larger gaps, this surface is slightly domed, with subradial and concentric wrinkles (Fig. 4C₄). This indicates the presence of a thin soft-tissue membrane covering the interior of mounds in life, creating an enclosed space within the mounds.

Spiculation of the outer wall is normally obscured by substantial pyritisation, particularly at the margins of parietal gaps, but individual spicules are visible in places (Fig. 4C₄). Where pyritisation is less extensive, the skeleton is covered by a thin reflective sheet of probable aluminosilicates. Spicules appear to be simple stauractines, although additional gastral or dermal rays cannot be excluded. Rays are straight and simple.

The largest spicules in the primary skeletal layer are centred in the centres of wall areas between parietal gaps, with the largest spicule identified having ray length 1.2 mm (basal ray diameter approx. 0.1 mm, possibly taphonomically enlarged); orientation is irregular. Smaller spicules are distributed largely irregularly except at parietal gap margins, where they are often arranged with rays parallel to the margin, such that one ray projects towards the centre of the gap (Fig. 4C₄). Typical ray length for smaller spicules is c. 0.2 mm. Near oscular margin, the spicules are arranged to form an irregular polygonal meshwork, oriented roughly diagonal to the sponge axis with spaces approximately 0.1 mm across (Fig. 4C₃).

Inner layer (possibly layers) composed of locally dense felt of small, fine hexactines or their derivatives in parallel, perhaps quadruled array. Traces of discrete orthogonal and diagonal spicule series, superimposed onto each other, and visible in the holotype (Fig. 4C₂); the paratype NIGP154631 (Fig. 4B) also shows a small area at the upper right corner with the orthogonal grid visible. The best-preserved regions in the holotype show an apparently regular grid with square spaces consistently around 0.1 mm per side.

No evidence for basalia is seen in any specimen.

Discussion.—If the structure of the sponge is well preserved, as reconstructed in Fig. 5, recognition of the species is straightforward. Separating the species from other known parietal gap-bearing taxa is difficult if the preservation does not show clear spiculation of the inner layer(s) or the skeletal structure of the mounds, but several features are useful. The parietal gap size appears to be tied consistently to body size, and is larger than that in co-occurring taxa. The near-circular, slightly polygonal form of the gaps, arrangement into locally clear transverse rows, and the occasional division of a large gap by a curving wall, are also characteristic. The narrowness of the walls between the gaps produces a relatively high proportion of space in this layer. Where spiculation is visible, the irregularity of the orthogonal and diagonal layers contrasts with that in Hintzespongia Rigby and Gutschick, 1976, in which the regular layer is external to the layer with parietal gaps (supported by Rigby 1983). In specimens from Amnodd Wen, the walls are typically preserved with mounds in three dimensions, allowing easy separation from Hintzespongia-like forms; the same distinction is visible at the margin of flattened, complete specimens, where mounds are visible in profile.

The new species is distinctly separated from Valospongia gigantis Rigby, 1983 by the lack of spicular strands, the denser packing of the parietal gaps, and perhaps also by the expression of the dermal spiculation over areas outside the mound surfaces; the holotype of the type species shows the fine-scale reticulation only over the mound surface rather than also within the inter-pariety wall region as seen in V. bufo sp. nov. The type species is also an order of magnitude larger than any specimen in the present collection, and with disproportionately larger spicules, although it is possible that the proportions of a juvenile V. gigantis would be similar to those of V. bufo. Other parietal gap-bearing taxa from the Burgess Shale-type faunas include Stephenospongia Rigby, 1986, which is known from fragments only; this species is assigned to the Hintzespongiidae (Finks and Rigby 2004b), but may also show a reticulating grid over the parietal gaps, as in Valospongia. It can nonetheless be distinguished by its elongate parietal gaps, by the large size of the spicules relative to the parietal gaps, and by the lack of a regularly ordered grid. The allied Ratcliffespongia Rigby, 1969 is a difficult genus; the type species, R. perforata Rigby, 1969, appears to have possessed only a single wall of spicules in an irregular array, and with transversely-elongated parietal gaps. Neither this nor the Chinese species R. multiforamina Wu, Yang, Janussen, Steiner, and Zhu, 2005 shows the reticulation-covered mounds or the regularly-oriented inner layer(s) of Valospongia. Ratcliffespongia wheeleri Rigby and Church, 1990 is discussed below, where we reassign it to Hintzespongia.

Stratigraphic and geographic range.—Known only from several sites within the upper part of the Dol-cyn-Afon Formation (Conophrys salopiensis Biozone; early Migneintian, Late Tremadocian) of the Arenig Fawr area, near Bala, Wales, UK.

Emended diagnosis.—Thin-walled, obconical, ovate or vasiform Reticulosa in which an inner, gastral layer of parallel, slender-rayed hexactines or derivatives, as in Protospongioidea, is overlain by a layer of slender-rayed hexactines or derivatives in nonparallel orientation that surround closely spaced, circular gaps or aporhyses.

Discussion.—The only material of Hintzespongia consists of the holotype of H. bilamina Rigby and Gutschick, 1976, and additional fragments of Hintzespongia from the Wheeler and Marjum formations described by Rigby et al. (2010). Another potential specimen is the holotype of Ratcliffespongia wheeleri Rigby and Church, 1990, if that species should be placed in the genus. A fragment of a similar sponge from the Burgess Shale was described by Rigby and Collins (2004), and assigned to H. bilamina, but re-examination (JPB, personal observations) shows this specimen to be a different, undescribed sponge with a single spicule layer. The only non-Laurentian sponge recorded as Hintzespongia (Yang et al. 2010) appears to show no evidence of a secondary spicule layer, and is likely to represent instead Ratcliffespongia or a related genus.

The relative positioning of the two spicule layers in H. bilamina could not be stated with certainty by Rigby and Gutschick (1976) due to the fragmentary nature of the specimen, and none of the subsequent discoveries provides a near-complete specimen in which the two layers are clearly distinct. As a result, the dermal and gastral placement of the two spicule layers has never been confirmed.

Type species: Hintzespongia bilamina Rigby and Gutschick, 1976, from the middle Cambrian Marjum Formation of Utah, USA.

Species included. Hintzespongia bilamina Rigby and Gutschick, 1976, Hintzespongia wheeleri (Rigby and Church, 1990).

Diagnosis (emended after Finks and Rigby 2004b).—Sponge thin-walled, conical to barrel-shaped, with a probably gastral layer of parallel, slender-rayed stauractines of at least four orders of size, underlying a layer of slender-rayed stauractines and hexactines in nonparallel orientation.

Discussion.—Definition of this monospecific genus is complex, and requires a re-evaluation of similar and related taxa. Finks and Rigby (2004b) list Hintzespongia, Ratcliffespongia, Cyathophycus, and Stephenosphongia as being assigned to the family, but some of these assignments are perhaps unsound. Cyathophycus is a complex, widely distributed genus that is in need of revision (and probably subdivision), but the inner layer of at least some species appears to be composed of monaxons (discussed by Botting 2003, 2004). Stephenospongia is a poorly known genus that may not even have an outer spicule layer, although there are traces of reticulation in some parts (Rigby 1986) that might resemble the reticulating outer mesh of Valospongia, discussed above. None of these genera are particularly similar to Hintzespongia bilamina, but there are definite similarities between it and Ratcliffespongia wheeleri (Rigby and Church 1990).

The taxonomy of Ratcliffespongia is complicated, despite being based on only two specimens. Rigby (1969) described R. perforata based on a unique, semi-complete specimen from the Wheeler or Marjum formations, of a sponge with prominent, somewhat transversely elongate parietal gaps in a thin wall that was constructed from large, thin-rayed but clearly defined stauractines. The spicule rays do not generally demark the margins of the gaps, which were visible as a soft tissue impression. The sponge apparently consisted of a single layer of spicules, and was assigned to the Teganiidae. Another unique specimen from the upper Wheeler Formation was described by Rigby and Church (1990) as R. wheeleri, but although showing prominent parietal gaps, this species differs substantially. The gaps are more circular and irregularly arranged, and the spiculation of the wall is rather obscure, with small hexactines or their derivatives often marking very clearly the margins of the gaps. Most fundamentally, R. wheeleri appears to have possessed two layers, with a homogeneous, sub-reticulate grid of small spicules overlying(?) the pariety-bearing layer. The diagnosis of Ratcliffespongia was emended by Rigby and Church (1990) to include the double-layered wall, and both species were thereby assigned to the Hintzespongiidae. The similarity was accepted only through the assumption that the holotype of R. perforata represents a decorticated inner layer of a hintzespongiid (Finks 1983; Rigby and Church 1990). However, the holotype is apparently well preserved, near-complete (the only damage appears to be due to erosion and distal breakage of the loose block containing the specimen), and fully articulated, even showing traces of the soft tissues marking the margins of the parietal gaps. Decortication seems very unlikely in this case, especially as the fragment forming the holotype of R. wheeleri is much less intact and appears to be less well preserved in some areas than others.

The skeletal architecture is fundamentally different in the two Ratcliffespongia species, and they have little in common except for being reticulosan hexactinellids with parietal gaps. We therefore do not recognise these two species as belonging to the same genus, and restrict the use of Ratcliffespongia to the type species R. perforata. Instead, the fundamental structure of R. wheeleri is identical to that of Hintzespongia, and we therefore reassign it to that species. The separation from Hintzespongia by Rigby and Church (1990) was on the basis of differences in the regularity of spicule organisation in the reticulate layer, and parietal gap dimensions; these are superficial differences that do not obscure a close relationship. We therefore include both R. wheeleri and the new (open nomenclature) sponge described here in Hintzespongia.

Stratigraphic and geographic range.—The confirmed range for Hintzespongia is restricted to the middle Cambrian Wheeler and Marjum formations of Utah, USA, together with the probable new record from the Tremadocian of Wales, UK.

Material.—NIGP154638, fragment of the body wall preserving detail of both spicule layers, from the Late Tremadocian (early Migneintian), loose block in Ceunant-y-garreg-ddu stream section.

Description.—Fragments of a partial specimen patchily exposed over 24 mm of roughly broken surface, and with spicules often partly replaced by iron oxides. Spicules are distributed through a thickness of matrix, rather than splitting precisely along a plane, producing intermittent overlap of the skeletal mesh from inner and outer layers. The best preserved region displays a dominant diagonal fabric superimposed on a somewhat irregular orthogonal array with major spicule rays visible. Entire individual spicules are difficult to observe, due to the irregular surface and replacement of spicules by iron minerals and reflective films (probably aluminosilicate).

Largest spicules observed with ray length exceeding 3 mm, but no complete rays of largest spicules seen; maximum ray diameter (probably near-basal) approximately 0.20–0.25 mm, but preservation makes exact width difficult to assess. At least three (probably four or five) size orders visible, but few complete spicules. One of the smallest has ray length 0.6 mm and basal diameter around 0.1 mm in broadest ray (around 0.05 mm in narrowest), with rays evenly tapered. These proportions suggest that spicule rays became more elongate as they grew, but better specimens would be required to confirm this. Spicules of orthogonal layer less clearly preserved, but appear to be consistently narrower for given spicule length, with some rays in excess of 1 mm long, but only 0.05 mm in diameter, and not obviously tapering over exposed length. Fewer small spicules are visible in the inconsistently exposed orthogonal array, which may be entirely due to preferential preservation, but might indicate the presence of obscure parietal gaps in that layer. One area (Fig. 6A, pg) shows what may be the pyritised margin of a parietal gap, around 1 mm in diameter.

Discussion.—The limited material available for this species prevents a categorical assignment even at genus level, but the structure appears to be diagnostic of Hintzespongia. Although normally described with the two planes of spicules clearly distinct (e.g., Rigby and Gutschick 1976), specimens are known in which the layers are less clearly defined and spicules are superimposed (Rigby et al. 2010). This is a preservational effect that complicates recognition of the sponge, but does not in itself necessitate reconsideration; the specimen was probably originally preserved in a similar manner to the type material, but authigenic aluminosilicate growth during low-level metamorphism is likely to have caused disintegration of particular sedimentary laminations. The preservation is entirely consistent with a regular diagonal layer of spicules overlying a somewhat less regular, sub-orthogonal layer. The only other significant difference from the previously described material is the rarity of obvious parietal gaps in the inner layer, but at least one probable example is visible, and the spicules interpreted as representing that layer in our material are much more sparsely preserved.

The skeletal structure is somewhat similar to that of Valospongia bufo sp. nov., but specimens are in general easily separated by preservational differences; in V. bufo, the most prominently preserved feature is the array of apparent gaps, the margins usually heavily pyritised. In side view these are clearly seen to be mounds in V. bufo, but profile views are not often available due to fragmentation of the sponge skeletons.

This difference is not dependent on exact bed; a fragment of V. bufo, preserved in the typical way, is present on the edge of the Hintzespongia? sp. slab. If only a spicule array is present, there are differences in the spicule dimensions, with the largest spicules in Hintzespongia? sp. having rays several times longer than the largest seen in V. bufo. Growth stage must be taken into account, but it also appears that parietal gaps in Hintzespongia? sp. are substantially smaller than the apparent gaps in V. bufo, even when the spicules are substantially larger.