An exceptional Caradoc sponge fauna from the Llanfawr Quarries, Central Wales and phylogenetic implications

Abstract

An abundant sponge collection from basal Caradoc mudstones of the Llanfawr Quarries, Llandrindod, Powys is the most diverse fine‐sediment sponge fauna recorded from the British Palaeozoic. It is dominated by reticulosans, but includes some lyssakids and demosponges. Eight new genera and 11 species are erected and one new family proposed: Xylochos palindromica gen. et sp. nov., Pirania llanfawrensis sp. nov., Heminectere minima gen. et sp. nov., Asthenospongia cambria sp. nov. (Asthenospongiidae nov.), Acutipuerilis spinosus gen. et sp. nov., Granulispongia obscura gen. et sp. nov., Hemidiagoniella tenax gen. et sp. nov., H. caseus sp. nov., Solusrectus rosetta gen. et sp. nov., Cyathophycus loydelli sp. nov., and Dilatispongia tumidus gen. et sp. nov. Some specimens are preserved with organic or pyritic films representing soft tissue and most have high‐fidelity preservation of spicule external moulds, although iron oxide spicule replacement obscures details prior to cleaning. D. tumidus shows a thickened wall with specialised dermalia, while retaining some degree of quadruling and represents a lineage that may have given rise to the brachiospongioids. Others, such as S. rosetta, appear close to the ancestral state of the modern Lyssacinosida. The hex‐actinellid taxa are interpreted in a reasoned phylogenetic framework that stresses the importance of root tuft structure, the presence of zero‐order spicules and the number and nature of spicule layers. It also attempts to link extant and Lower Palaeozoic palaeontological classifications. The parallel evolutionary development of thick walls and strong attachment structures is interpreted as being due to an offshore‐onshore expansion, contrary to many other groups.

Full text

Journal of Systematic Palaeontology 2(1): 31–63 Issued 11 March 2004
DOI: 10.1017/S147720190300110X Printed in the United Kingdom C
The Natural History Museum
An exceptional Caradoc sponge
fauna from the Llanfawr Quarries,
central Wales and phylogenetic
implications
Joseph P. Botting
Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK
SYNOPSIS An abundant sponge collection from basal Caradoc mudstones of the Llanfawr Quarries,
Llandrindod, Powys is the most diverse fine-sediment sponge fauna recorded from the British Pa-
laeozoic. It is dominated by reticulosans, but includes some lyssakids and demosponges. Eight new
genera and 11 species are erected and one new family proposed: Xylochos palindromica gen. et sp.
nov., Pirania llanfawrensis sp. nov., Heminectere minima gen. et sp. nov., Asthenospongia cambria
sp. nov. (Asthenospongiidae nov.), Acutipuerilis spinosus gen. et sp. nov., Granulispongia obscura
gen. et sp. nov., Hemidiagoniella tenax gen. et sp. nov., H. caseus sp. nov., Solusrectus rosetta gen.
et sp. nov., Cyathophycus loydelli sp. nov., and Dilatispongia tumidus gen. et sp. nov. Some speci-
mens are preserved with organic or pyritic films representing soft tissue and most have high-fidelity
preservation of spicule external moulds, although iron oxide spicule replacement obscures details
prior to cleaning. D. tumidus shows a thickened wall with specialised dermalia, while retaining some
degree of quadruling and represents a lineage that may have given rise to the brachiospongioids.
Others, such as S. rosetta, appear close to the ancestral state of the modern Lyssacinosida. The hex-
actinellid taxa are interpreted in a reasoned phylogenetic framework that stresses the importance of
root tuft structure, the presence of zero-order spicules and the number and nature of spicule layers.
It also attempts to link extant and Lower Palaeozoic palaeontological classifications. The parallel
evolutionary development of thick walls and strong attachment structures is interpreted as being
due to an offshore–onshore expansion, contrary to many other groups.
KEY WORDS hexactinellid, rossellid, Reticulosa, demosponge, pyritisation, palaeoecology
Contents
Introduction 32
Locality information 33
Locality 1 33
Locality 2a 34
Locality 2b 34
Locality 2c 34
Locality 3 34
Environmental interpretation 34
Systematic palaeontology 35
Class DEMOSPONGIAE Sollas, 1885 35
Order UNCERTAIN 35
Family HAZELIIDAE de Laubenfels, 1955 35
Genus XYLOCHOS gen. nov. 35
Xylochos palindromica gen. et sp. nov. 35
Family PIRANIIDAE de Laubenfels, 1955 37
Genus PIRANIA Walcott, 1920 37
Pirania llanfawrensis sp. nov. 37
Class HEXACTINELLIDA Schmidt, 1870 38
Order RETICULOSA Reid, 1958 38
Superfamily PROTOSPONGIOIDEA Finks, 1960 38
Family PROTOSPONGIIDAE Hinde, 1887 38
Genus HEMINECTERE gen. nov. 38
Heminectere minima gen. et sp. nov. 38

32 J. P. Botting
Family ASTHENOSPONGIIDAE fam. nov. 40
Genus ASTHENOSPONGIA Rigby et. al., 1981 40
Asthenospongia cambria sp. nov. 40
Genus ACUTIPUERILIS gen. nov. 42
Acutipuerilis spinosus gen. et sp. nov. 42
Family ? ASTHENOSPONGIIDAE fam. nov. 42
Genus GRANULISPONGIA gen. nov. 42
Granulispongia obscura gen. et sp. nov. 42
Family UNCERTAIN 44
Indeterminate protospongioid sp. A 44
Indeterminate protospongioid sp. B 46
Indeterminate protospongioid sp. C 46
Indeterminate protospongioid sp. D 46
Order UNCERTAIN 47
Family uncertain 47
Genus HEMIDIAGONIELLA gen. nov. 47
Hemidiagoniella tenax gen. et sp. nov. 47
Hemidiagoniella caseus gen. et sp. nov. 49
Genus SOLUSRECTUS gen. nov. 49
Solusrectus rosetta gen. et sp. nov. 50
Family ? DICTYOSPONGIIDAE Hall, 1884 51
Genus CYATHOPHYCUS Walcott, 1879 51
Cyathophycus loydelli sp. nov. 52
Order LYSSAKIDA Zittel, 1877 52
Genus DILATISPONGIA gen. nov. 52
Dilatispongia tumidus gen. et sp. nov. 54
Root tufts 56
Phylogenetic considerations 57
Palaeoecology 61
Acknowledgements 61
References 61
Introduction
Non-lithistid siliceous sponges are becoming recognised as
important or dominant members of many Ordovician si-
liciclastic ecosystems. Their diversity and abundance in
Cambrian deposits has been well described (e.g. Walcott
1920; Rigby 1978, 1986; Chen, Hou & Lu 1989; Chen,
Hou & Li 1990), and several diverse assemblages are known
from the Ordovician and Silurian of North America and else-
where (Rigby 1971; Rigby, Hannun & Frest 1979; Rigby &
Webby 1988; Mehl, Rigby & Holmes 1993). However, most
occurrences consist of very distinct, largely endemic faunas
(e.g. Rigby & Chatterton 1994), reflecting patchy preser-
vation and, probably, life distribution. It is possible that
sponges were much more abundant than is presently recog-
nised, since almost all non-cemented sponge skeletons are
disarticulated and inconspicuous as fossils. Siliceous spic-
ules are also highly prone to dissolution in shallow marine
environments (Land 1976). In spite of this, many Lower
Palaeozoic sequences contain horizons or intervals where
spicules comprise a significant proportion of the sediment
and they are often extremely abundant on individual bedding
planes. In general, however, these are of no taxonomic use
due to the similarity of basic morphotypes in many diverse
taxa, particularly in fine sediment facies, and the impossibil-
ity of isolation from the matrix. Preservation is usually either
mouldic or as iron oxides (after pyrite), with occasional re-
crystallisation as silica, or replacement by calcite. Although
fresh pyritised material is, in principle, extractable by hy-
drofluoric acid dissolution of the siliciclastic matrix, this is
rarely, if ever, achieved and unpublished studies have shown
that pyrite replacement is only partial within at least some
examples (A. Roberts, pers. comm., 2001).
Described sponge faunas from the Welsh Basin are
limited to isolated occurrences of disarticulated spicule as-
semblages (Pulfrey 1933; Woodland 1939; Lewis 1940) and
rare articulated material (Hinde 1887; Owens et al. 1982;
Botting in press a).Shallow-water deposits often contain
large elongate monaxons assigned to root tufts of the genus
Pyritonema M’Coy, discussed by Hinde (1887), Finks (1960)
and Reid (1967). Botting (in press a) reinterpreted this as a
derived hexactinellid, based on silicified Llanvirn material
from Llandegley Rocks, Powys.
Isolated hexactinellid spicules are very widespread in
siltstone facies of the Builth–Llandrindod Inlier, especially
the Llanfawr Mudstone Formation (sensu Furnes 1978;
Davies et al. 1997). Heteractinid spicules are, however, very
rare and no articulated material has yet been recovered from

Caradoc sponge fauna from Central Wales 33
spiculite
chitinozoan beds
trinucleid beds
trilobite beds
Dicellograptus
beds
Cyathophycus beds
sponge beds and
shales?
diplograptid beds
diplograptid beds
Dolerite
Dolerite + unknown
Cyathophycus loydelli
Pirania llanfawrensis
Heminectere minima
Hemidiagoniella tenax
Hemidiagoniella caseus
Acutipuerilis spinosus
Asthenospongia cambria
Dilatispongia tumidus
Granulispongia obscura
Solusrectus rosetta
Xylochos palindromica
d
D
d
C
ch
s
tr
t
Bristol
Cardiff
St. David’s
Aberystwyth
Caernarvon
Hereford
R. Severn
R. Wye
ENGLANDWALES
LLANDRINDOD
50 km
A
C
Lake
12a
2b
2c
3
t
ch
s
tr
d
D
C
d
ch
100 m
N
to Llandrindod B
total sediment thickness
approx. 3 m
fine sediments; the few isolated spicules are simple hex-
aradiates and pentaradiates that may have been transported
from shallow-water assemblages, or represent the remains
of rare individuals after reworking. Sponges were recorded
from this formation by Hughes (1969) and Sheldon (1987),
but no descriptions were included. Where discussed at all, or
labelled in museum collections, they are invariably referred
to tentatively (and incorrectly) as species of Protospongia.
The only non-lithistid genera, in addition to Protospongia,
that have been adequately described from the Lower
Palaeozoic fine sediments of the British Isles are Phor-
mosella ovata Hinde, 1888, from the Ludlow of Shropshire,
Amphispongia oblonga Salter, 1861, Dictyophyton danbyi
(M’Coy, 1852) and Plectoderma scitulum Hinde, 1883,
from the Silurian of Scotland and northern England. The
attribution of individual monaxons to Recent demosponges
such as Reniera by some authors (e.g. Lewis 1940) was
essentially groundless, due to the indistinctiveness of indi-
vidual spicules.
Locality information
The four Llanfawr Quarries are situated about 500 m east
of the main road in Llandrindod, Central Wales (SO 065
617) and were excavated for dolerite building stones in the
19th and early 20th centuries. The three excavated bodies are
laccolithic (Jones & Pugh 1948) or loppolithic (J. Davies,
pers. comm., 2001) and were intruded into fossiliferous mud-
stones, which are exposed at the quarry margins. Mudstones
occur in three quarries, the lower, main and upper (Fig. 1) and
each has produced articulated sponges. Precise thicknesses
and correlations are complicated by limited, largely strike-
parallel, exposures, deformation and thermal metamorphism
from the dolerite intrusions as well as local faulting. Never-
theless, several distinct beds are recognised in the succession,
on the basis of sedimentology and faunal content (Fig. 1).
The utility of sponges in defining such horizons is presently
unclear, since the consistency of the distributions cannot be
established. Sponge populations are normally patchy in mod-
ern deep shelf environments, with localised regions of very
high density, where their accumulated spicules can influence
the distribution of other organisms (Bett & Rice 1992). Five
specific sponge-bearing localities are distinguished.
Locality 1
Lower quarry: small remnant outcrops adjacent to the lane,
overlain by dolerite. Approximately 0.7 m of highly trilobitic
blue mudstones, including Ogygiacarella (large), Cnemido-
pyge,Barrandia,Trinucleus, orthoconic nautiloids, inartic-
ulate brachiopods and conulariids, with locally abundant
graptolites, particularly Climacograptus brevis brevis.The
Figure 1 (A) Regional location of Llandrindod, with the
Builth–Llandrindod Inlier shaded. (B) Locality map and local
stratigraphy of the Llanfawr Quarries complex on the eastern edge of
Llandrindod town centre; numbers in bold refer to localities,
abbreviations refer to informal stratigraphic divisions. (C) Simplified
stratigraphy of the Llanfawr Quarries succession, showing the
distribution of the named taxa described in this paper.

34 J. P. Botting
sponges include an almost monospecific mass bedding-plane
assemblage of Cyathophycus loydelli sp. nov. and rare Xy-
lochos palindromica gen. et sp. nov., with occasional well-
preserved material of the same species at several other levels.
Two specimens of Pirania llanfawrensis sp. nov., a single
specimen each of Granulispongia obscura gen. et sp. nov.
and Hemidiagoniella sp. and additional unidentified frag-
ments, have also been recovered. These exposures are strati-
graphically the highest at Llanfawr. Specimens are preserved
by spicular replacement by iron oxides, presumed after pyr-
ite, but usually insufficiently hardened to provide detailed
spicule moulds.
Locality 2a
Main quarry: extensive scree with thin outcrop at margins,
in the quarry corner immediately above the pool. Thin (5–
10 cm) ferruginous spiculite containing X. palindromica,
Acutipuerilis spinosus gen. et sp. nov., Heminectere min-
ima gen. et sp. nov., Hemidiagoniella tenax gen. et sp. nov.,
G. obscura and C.loydelli and overlying blue mudstones
with Asthenospongia cambria sp. nov., abundant Hemidiag-
oniella tenax,H. caseus sp. nov., G. obscura,H. minima and
indeterminate protospongiids. Associated fauna consists of
trilobites (often articulated), graptolites (sometimes three di-
mensionally pyritised), chitinozoans (including helical as-
semblages), inarticulate brachiopods and rare orthoconic
nautiloids, gastropods and conulariids. Sponges are pre-
served with spicules being replaced by weathered iron oxides,
presumably after pyrite, but retaining relief; gentle brush-
cleaning yields high-fidelity external moulds. In some spe-
cimens, the soft tissue is outlined by an iron-rich film that
may represent decay by sulphate-reducing bacteria, as in the
Hunsr¨
uckschiefer (Raiswell, Bartels & Briggs 2001). Only
sponges appear to have been preserved in this way, perhaps
because of the nucleation sites provided by spicule replace-
ment. Small-scale, high-angle cross-bedding is visible in sec-
tion, suggesting that burial was by abrupt sediment input.
These are stratigraphically the lowest exposures at Llanfawr.
Locality 2b
Main quarry: slope-parallel beds exposed in a narrow cutting
above the northeast corner of the quarry. Thin (10 cm) dark,
baked mudstone containing sparse chitinozoan–graptolite
fauna with bivalves, nautiloids, dendroid graptolites and
conulariids; rare soft-bodied organisms have been recovered,
including a possible bivalved arthropod. Sponges are repre-
sented by frequent specimens of C. loydelli and X.palindro-
mica (often monospecifically clustered) and rare G. obscura.
Minimal relief is preserved, spicules are mouldic without fer-
ruginous replacement, but soft tissue outlines can be shown
by dark, presumed organic, staining, occasionally with some
preserved silica. Although sedimentologically homogen-
eous, examples of sponges (SM X.41150) and other organ-
isms with Leptograptus wrapped around them indicate signi-
ficant current activity, here and in the trinucleid beds (Fig. 1),
from which only disarticulated spicules have been recovered.
Locality 2c
Main quarry: north wall. Rock-fall material (dark mudstones)
with abundant graptoloids, chitinozoans, occasional trilob-
ites and brachiopods (primarily inarticulate). Sponge fauna
comprises occasional specimens of C. loydelli and X. palin-
dromica; the preservation is similar to locality 1, but gener-
ally with less relief.
Locality 3
Upper quarry: slope-parallel beds incised by scalloped cut-
tings. Blue mudstones immediately overlying black grap-
tolitic shales, with trilobites, graptolites and brachiopods.
Single slab with sponges, preserving three large specimens
of C.loydelli in poor relief, lacking spicule replacement.
Additional spicules and partially articulated fragments
occur at several levels, most notably among the celebrated
Trinucleus-dominated beds of the main quarry, from which a
single pentaradiate, presumed heteractinid, spicule was also
recorded. Root tufts are limited to rare specimens from loc-
alities 2a and 2b and also from fallen blocks around the
northeastern face between localities 2b and 2c. The lat-
ter have not yielded any other sponge material, except for
rare hexactinellid spicules and correlation between these and
the sponge bodies is difficult (but see section on root tufts,
below).
The quarries constitute a Site of Special Scientific In-
terest and permission to collect must be sought from the
Countryside Commission for Wales.
Environmental interpretation
Establishing the water depth of organic-rich siltstones is no-
toriously difficult and estimates for the Llanfawr Mudstone
Formation range from around 700 m (Sheldon 1987) to 100–
150 m (Botting 2000). The trilobite fauna is very similar to
the Arenig Raphiophorid Community of Fortey & Owens
(1978), which they attributed to the middle or outer shelf.
Pelagic taxa (including agnostids) are conspicuously rare,
but this could be attributed to offshore, fault-generated el-
evations enclosing a relatively shallow basin, although this
would not explain the ‘Raphiophorid Community’ fauna.
No other faunal element is particularly useful, although bi-
valves are frequent in some beds, such as locality 2b; Ordovi-
cian bivalve faunas are rarely found below the inner shelf.
Orthoconic nautiloids such as Alumettoceras (Evans 1994)
have been examined for possible use in estimating water
depth from septal implosion (Stridsberg 1990), but most are
decalcified and compressed. One specimen was recovered
retaining calcitic infill of the chambers and some mouldic
specimens retaining partial details of septae, but these yiel-
ded only a tentative maximum depth limit of 600 m, having
no imploded chambers.
The sedimentary environment became progressively
deeper following the Builth Volcanic Formation, deposited
during the murchisoni Biozone, a sequence that represents
both erosion of the quiescent volcanic cone and increas-
ing eustatic sea level. Sediment grain size at Llanfawr is
finer than in most localities in the underlying teretiusculus
Biozone, where there is a higher proportion of mica flakes,
but this may reflect proximity to shoreline more than abso-
lute water depth. Sedimentary structures are limited to planar
or gently inclined lamination in most beds, although cross-
bedding at up to 20◦has been observed in some sponge-
bearing horizons; the laminations are defined by slight grain

Caradoc sponge fauna from Central Wales 35
size variations and organic carbon and oxidised iron content,
probably related to organic decay. There is little volcanic
input, although a single thin ash bed occurs in the main
quarry (Huff et al. 1993). Benthic oxygenation appears to
have been variable, based on faunal colonisation. Trilobites
are generally ubiquitous, although absent from an unfos-
siliferous pyrite nodule horizon low in the trinucleid beds
and shales (Fig. 1), while they are rare in the chitinozoan-
rich, sponge-bearing bed (locality 2b). Limited oxygenation
might be inferred for the latter, supported by the high carbon
content, although sponges, dendroid graptolites, inarticulate
brachiopods, gastropods, bivalves and the chitinozoan pro-
ducers were evidently able to tolerate the conditions and
other parameters may have been more critical (see section
on palaeoecology, below). Articulate brachiopods (Dalman-
ella) are generally rare, occurring locally as disarticulated
individuals or very small clusters. Echinoderms have not
been certainly recovered from this locality and bryozoans
occur primarily as encrustations on orthocone nautiloids,
although some of this was post-mortem, occurring on in-
ternal and external surfaces. Occasional isolated bryozoan
colonies may also represent in situ benthic colonisation, or
encrustation onto unmineralised organisms such as float-
ing seaweeds or arthropods. At some horizons, numerous
subspherical pyrite nodules were developed; although this
may be partly related to the dolerite intrusion, pyritisation of
sponge soft tissues implies very early mineralisation in some
beds.
Overall, the environment appears to have been of middle
to outer shelf depth, with typically quiet conditions separ-
ated by episodes of sudden sediment input or remobilisa-
tion. Oxygenation varied from moderately aerobic to an-
oxic, sometimes sulphidic. Although consisting entirely of
organic-rich silt and mud, the sedimentology and faunal vari-
ation indicates a significant range of depositional rates and
conditions.
Systematic palaeontology
REPOSITORY. All specimens, unless otherwise stated, are
housed in the Sedgwick Museum, University of Cambridge.
Class DEMOSPONGIAE Sollas, 1885
Order UNCERTAIN
Family HAZELIIDAE de Laubenfels, 1955
Genus XYLOCHOS gen. nov.
TYPE SPECIES.Xylochos palindromica gen. et sp. nov.
DIAGNOSIS. Globular demosponges with thin walls consist-
ing of vertical and horizontal fine monaxon tracts, parallel
to wall. Extensive fringe of erect, slightly divergent mar-
ginalia and similar broad root tuft, slightly convergent, both
of coarser monaxons.
ETYMOLOGY. Greek for copse, thicket, referring to the
clustered distribution. Gender f.
OCCURRENCE. Known only from the basal Caradoc Llanfawr
Quarries, Llandrindod, Powys.
REMARKS. The new genus is easily separated from Hazelia
by the extensive marginalia and basalia and by the body wall
structure. The latter is similar to Hazelia, comprising fine
bundled monaxons in a cross-hatch arrangement, but oriented
only tangentially and not radially. Because of the basic sim-
ilarity, however, a close relationship is assumed. Of the other
genera of Hazeliidae, Crumillospongia Rigby, 1986, has re-
duced horizontal tracts, whereas Falospongia Rigby, 1986,
appears to represent a form that is convergent with the rhizo-
morine lithistids (Rigby 1986). Both are more distinct from
Xylochos than Hazelia is. A similar arrangement is seen in
the Leptomitidae, which form elongate cylindrical bodies of
long monaxons, but have spaced, rather than pervasive, tracts.
Although normally related to the haplosclerid demosponges,
de Laubenfels (1955) placed the Leptomitidae tentatively
among the protospongioid hexactinellids. Rigby (1986) and
Walcott (1920) preferred to classify them as demosponges,
but Rigby (1986), in particular suggested a position close
to the demosponge–hexactinellid transition. However, the
Leptomitidae possess more complex characteristics in the
wall itself, such as internal spiral threads, while lacking sig-
nificant marginalia or basalia; the same is true of the re-
lated, but more irregular, Halichondritidae. Xylochos there-
fore appears to be an extremely primitive demosponge, with a
globose morphology that may reflect the ancestral state of the
Hazeliidae.
The similarity of this genus to the internal layer of
Cyathophycus is striking and many reticulosans possessed
pronounced marginalia and basalia. Reduction and ultimate
loss of the hexactine-based dermal layer could have resulted
in the appearance of a form similar to X. palindromica,so
that under phylogenetic taxonomy it would be classed as a
hexactinellid. This putative relationship is discussed at length
by Botting (in press b).
Xylochos palindromica gen. et sp. nov. (Figs 2A–D, 3)
TYPES. Holotype: SM X.39452; obliquely compressed spe-
cimen from locality 2c, lacking basalia. Paratypes: SM
X.39393, 39353–61, 39363–5, 41078–81.
OTHER MATERIAL. SM X.39362, 39366–77; further poor par-
tial specimens from localities 2b and 2c.
DIAGNOSIS. As for genus.
ETYMOLOGY. From Greek palin and dromos, yielding ‘pal-
indrome;’ after the difficulty in recognising the correct ori-
entation.
OCCURRENCE. Recorded from localities 1, 2a, 2b and 2c;
particularly abundant at 2b and 2c.
DESCRIPTION. Small globular sponges with flattened length:
width ratio of 1.2–1.9 in known specimens, typically around
1.5. Maximum observed length =11.2 mm, obliquely flat-
tened. Maximum width occurs at approximately half height,
with width symmetrical across the equatorial plane, until os-
culum. Main body comprises cross-hatch arrangement of
densely-packed fine monaxons, in most cases with indi-
vidual spicules being obscure. Interweaved structure pre-
vents estimates of spicule length; thickness approximately
0.01–0.02 mm. Cross-hatched on variable scale, but clearest
divisions approximately 0.1–0.2 mm per edge of square in
adults (for change of skeletal parameters with growth, see
Botting 2003). There is no evidence of wall thickness beyond

36 J. P. Botting
Figure 2 Xylochos palindromica gen. et sp. nov. (A)–(D)andPirania llanfawrensis sp.nov.(E), (F). (A)X. palindromica SM X.39452,
holotype, ×6; (B) SM X.39464, paratype, with basalia, ×11; (C) SM.X.39465, paratype, ×10; (D) SM X.39459–60, paratypes, ×3.
(E)P. llanfawrensis SM X.39478, holotype, under water, ×20; (F) SM X41011, paratype,×10.
that of the spicule mesh, and no other spicules are present,
excluding marginalia and basalia. Allowing for the unknown
length of monaxons, the skeletal wall thickness is estimated
at several to 10 spicule thicknesses (i.e. up to 0.2 mm). There
are no obvious ostia or parietal gaps and inhalant canals are
assumed to have been 0.2 mm in diameter at most. Osculum
revealed by marginalia and was approximately one-half of
maximum sponge width. In some specimens (e.g. Figs 2B &
C), the oscular rim is visible as an impression on the spic-
ule mesh, possibly indicating reinforcement of the margin.
Orientation of cross-hatching is approximately orthogonal
to the sponge axis, leading to a radial pattern in obliquely
compressed specimens. All known specimens are completely
articulated, excepting basalia, implying a high degree of co-
herence. There is no evidence for distinct gastralia, which
are therefore presumed absent.

Caradoc sponge fauna from Central Wales 37
Figure 3 Cut-away reconstruction of adult Xylochos palindromica
gen. et sp. nov.; scale bar =2 mm.
All specimens have a broad fringe of marginalia, dia-
meter approximately three times height. Marginalia are fine
monaxons, but much coarser than body spicules; length up
to 2.8 mm and thickness 0.03 mm in large specimens. Distal
terminations oxeate, but proximal terminations are embed-
ded in the body wall and not clearly seen. Marginalia densely
packed, but individual monaxons are separable in good speci-
mens, particularly at margins. The fringe was slightly diver-
gent, but not held rigidly; several specimens show convergent
vertical flattening, where the sponge body has sunk into the
sediment on its side and the marginalia were pushed up by
drag against the sediment.
Basal tuft comprises 15–20 monaxons, slightly broader
and shorter than marginalia: length 1.5 mm, thickness
0.04 mm. Again, no proximal terminations seen, but distal
terminations oxeate. Basalia orientation more variable than
body spicules or marginalia, although in general slightly con-
vergent. In some cases, perhaps where damaged, the tuft may
be divergent, when it closely resembles the marginal fringe.
Overall size of tuft is approximately equal to marginal fringe.
Smallest known example is 3 mm long and morpholo-
gically very similar to adults. The marginalia are slightly
more pronounced and basalia absent in this specimen (SM
X. 39458).
REMARKS. The species occurs among a range of sub-facies
at Llanfawr, usually associated with the dictyospongioid
Cyathophycus loydelli sp. nov., with which it shares several
morphological features, although the outer hexactine layer
is absent. However, there is no likelihood of confusing the
species with any previously described taxon. Where found,
Figure 4 Cut-away reconstruction of Pirania llanfawrensis gen. et
sp. nov.; scale bar =2 mm.
X. palindromica tends to be locally clustered, with a very
patchy distribution, but is sometimes common. The dense
mesh inhibits recognition of the spicule arrangement in poor
specimens.
Family PIRANIIDAE de Laubenfels, 1955
Genus PIRANIA Walcott, 1920
TYPE SPECIES.Pirania muricata Walcott, 1920, from the
Middle Cambrian Burgess Shale, Canada.
OCCURRENCE. Known from the Middle Cambrian of North
America and the Middle Ordovician of Wales.
Pirania llanfawrensis sp. nov. (Figs 2E, F, 4)
TYPES. Holotype: SM X.39478; lacking apex. Paratype: SM
X.41082; complete but damaged, with less detail preserved
than the holotype.
DIAGNOSIS.SmallPirania in which the fine dermal monax-
ons are restricted to the basal region.
ETYMOLOGY. After the Llanfawr Quarries, Llandrindod,
Powys; the type locality.
OCCURRENCE. Locality 1 only.

38 J. P. Botting
DESCRIPTION. The holotype is a well-preserved proximal re-
gion, lacking apex and showing possible broken base of a
second branch. The paratype is a complete, larger speci-
men. Basal region consists of dense, fine monaxons radiat-
ing from base of sponge body in all orientations, wavy in
places and without individually projecting basalia. Spicules
are too densely clustered to specify proximal terminations,
but appear to be oxeate distally.
Dermal skeleton comprises projecting styles or sub-
tylostyles – some spicules show slight basal thickening. Spic-
ules are primarily straight, directed upwards at 30–60◦to the
axis, some curving towards the axis distally, through no more
than 10◦. Spicules are up to 2.4 mm long, with basal diameter
of 0.03 mm, tapering slowly until near the termination. Style
density is approximately 15/vertical mm proximally, chan-
ging to 8–10/mm distally. There is no evidence of fine mon-
axons between the styles except in the basal region, extending
for only 1.6 mm distally. Spicule bases are concentrated in a
region 0.16 mm thick, presumably representing the soft tis-
sue thickness. The sponge body then consists of a simple
tube, 0.6 mm wide at the base, widening slightly to 0.8 mm
at the observed length of 7.1 mm. In the paratype, which is
approximately twice as large in all measurements, the in-
terior is infilled with microcrystalline quartz, showing that
even the basal region was largely hollow. The apex is present
in the paratype, but shows no distinctive features; prostalia
curve upwards and towards the axis in the described manner.
The fine monaxons at the basal region are 0.7 mm long,
with basal ray diameter of 0.01 mm, very densely clustered
in radiating tufts. A possible branch base in the root region of
the holotype is shown by a blunt projection of similar width
to the proximal sponge body, but there is no clear evidence.
Fine monaxons in this area terminate abruptly, but due to
the variable orientation of basalia tufts, undamaged speci-
mens would be needed to confirm branching. The paratype
is certainly unbranched.
REMARKS. The present species differs from P. muricata
Walcott, 1920, redescribed by Rigby (1986), in that it lacks
the dense fine monaxons between the styles over the major-
ity of the sponge body. Although it is also much larger than
the present species, only two specimens of P. llanfawrensis
are known and discussion of size differences, either of body
or spicules, is inappropriate. The spicules are otherwise very
similar in the two species, as is the overall morphology, relat-
ive wall thickness and basal morphology. Although the distal
spiculation is clearly distinct from the type species, there is
little question that P. llanfawrensis belongs to Pirania.
De Laubenfels (1955) gave the range of Pirania as
Cambrian to Silurian, but did not give references. No other
record of Pirania from above the Middle Cambrian has been
found and the present species thus indicates a significant
expansion of its stratigraphic range.
Class HEXACTINELLIDA Schmidt, 1870
Order RETICULOSA Reid, 1958
Superfamily PROTOSPONGIOIDEA Finks, 1960
Family PROTOSPONGIIDAE Hinde, 1887
Genus HEMINECTERE gen. nov.
DIAGNOSIS. Protospongiids in which orthogonal quadruling
has been partially, but incompletely, replaced by tractose
development through extension of spicule rays until extensive
overlap occurs; mesh comprises several orders or stauracts
and/or hexacts, but without development of more complex
morphologies.
TYPE SPECIES.Heminectere minima gen. et sp. nov. from the
Caradoc of central Wales.
OTHER SPECIES,REASSIGNED.Protospongia conica (Rigby
& Harris, 1979).
ETYMOLOGY. Latin hemi,halfandnectere, twine, describing
the partial development of spicule tracts. Gender f.
OCCURRENCE. Widespread in the Silurian and Devonian of
North America and the Caradoc of Wales.
REMARKS. There has been much confusion relating to the
assignment of protospongioids that clearly differ from Pro-
tospongia, while retaining some degree of quadruling and
lacking particularly distinctive features. The general diffi-
culty in separating hexactines from stauracts in flattened
material, typically with coarse iron oxide replacement, has
also led to numerous species being included within Proto-
spongia, despite not conforming to the generic diagnosis of
Rigby (1986), who emphasised the regularity of arrange-
ment, or of Hinde (1888), who restricted the genus to those
whose spicules were entirely stauracts. Rigby (1966) sugges-
ted that the genus Protospongia could be subdivided on the
basis of spicule arrangement, particularly of the first-order,
but this suggestion has so far been barely adopted. There
are presently no other genera available for such species, un-
til the differences become sufficient for inclusion in forms
such as Cyathophycus Walcott, 1881, Asthenospongia Rigby,
King & Gunther, 1981, Testiispongia Rigby, 1983, or
Ratcliffespongia Rigby, 1969. Botting (2003) showed that
the variation in tract development could be partly ontogen-
etic rather than taxonomic. However, in this case, there are
sufficient specimens to be assured that the species is not
merely an early growth stage of a Protospongia species. Al-
though clearly related to that genus, it differs in the spic-
ule morphology, extensive first-order ray overlap and less
regular subdivision by low-order spicules. Heminectere dif-
fers from the Middle Cambrian Testiispongia in morphology,
lesser regularity and absence of root tuft, although anchor-
ing spicules were recorded in H. conica (Mehl et al. 1993).
It is considered likely that several further sub-divisions of
this group of sponges will be adopted in future, perhaps
on the basis of quantitative characteristics of the spicule
mesh; this is not warranted by the present material, although
related but qualitatively different genera are also erected
below.
Heminectere minima gen. et sp. nov. (Figs 5, 6)
TYPES. Holotype: SM X.41161. Paratypes: SM X.41162–
41185; various complete and partial specimens.
DIAGNOSIS. Very small, stauract-dominated spindle-shaped
Heminectere, with vertical spicule rays longer than hori-
zontal and at least three orders of spicules, including some
clinopentactines.
ETYMOLOGY. Latin, small.
OCCURRENCE. Locality 2a only, from both spiculite and blue
mudstone.

Caradoc sponge fauna from Central Wales 39
Figure 5 Heminectere minima gen.etsp.nov.(A) SM X.41166, paratype,×20; (B) SM X.41163, paratype, ×15; (C) detail of SM X.42263, ×60;
(D) SM X.41178–81, four paratypes in dense association, ×15; (E) SM X.41168, paratype, ×19; (F) SM X.41161, holotype, ×13. All photographs
taken under water.
DESCRIPTION. Fusiform sponges with maximum height
around 6 mm in observed material and 1.5–2.5 times as high
as wide (close to twice in the best-preserved specimens).
Maximum flattened width is at around two-fifths of sponge
height and equal to half of sponge height. The oscular width is
approximately half of the maximum width. Outline smoothly
curved, with maximum convexity in the lower third. Spicules
parallel to sides, almost exclusively closer to orthogonal than
diagonal; fewer than 5% are strongly inclined.
Spicules form a single layer except for overlapping
rays, with certainly three, perhaps four orders. All spicules
with at least four rays, and some first-order, are certainly
clinopentactinal. No definite hexactines observed, although
small orders are unclear; no large hexactines at margins.
Spicule rays are straight, or slightly curved proximally in
clinopentactines; there are no spines, branches, expansions
or other visible modifications. Rays taper evenly, from basal
diameter of one-fifteenth of ray length. There are no zero-
order spicules (sensu Botting 2003). Larger spicules have
approximately 30–50% longer vertical rays than horizontal,
corresponding to a vertically elongated rectangular first-
order grid. First-order spicule rays overlap by at least 50%,

40 J. P. Botting
Figure 6 Cut-away reconstruction of Heminectere minima gen. et
sp. nov.; scale bar =1 mm.
sometimes entirely, with adjacent spicules. The smaller size
orders have all rays of subequal length, but are difficult to
distinguish clearly; rays are generally non-overlapping.
The smallest well-preserved specimen is 3.0 mm high,
with six or seven first-order squares in the flattened half-
diameter. In the largest specimens, there are 13–16 squares,
indicating interstitial addition of one lower size order into the
first-order grid between these growth stages, in the manner
described by Botting (2003).
No basalia are preserved in any specimen. The proximal
first-order spicules have ray apices overlapping or adjacent,
forming a sharp point, of approximately 90◦when flattened,
which may have been the only attachment structure. The
oscular margin is simple, without distinct marginalia.
REMARKS. The morphology and structure of H. minima most
closely resembles the larger H. conica (Rigby & Harris 1979),
also described by Mehl et al. (1993), but differs in several
features. The smaller size is regarded as a reliable feature
due to the substantial number of specimens. The nature
of the spicules is seen here to be, in at least some cases,
clinopentactinal, whereas H. conica is reported to possess
only stauracts or hexactines, although imperfect preservation
may have resulted in misdiagnosis of spicule morphology.
The spicule arrangement, however, is similar; H, minima is
slightly less bundled, with a smaller ray overlap in adult spe-
cimens. The number of spicule orders is probably the same,
although the holotype possesses only two (Rigby & Harris
1979), but H. conica also possessed anchorage spicules
(Mehl et al. 1993).
Family ASTHENOSPONGIIDAE fam. nov.
DIAGNOSIS. Protospongioids with partially retained quad-
ruling and containing an enlarged hexactine morphotype
(simple or modified), which acts as prostalia. Spongocoel
partially enclosed by sharp break in body wall from sub-
conical to horizontal upper surface, significantly reducing
oscular diameter.
INCLUDED TAXA.Asthenospongia Rigby et al., 1981, Acu-
tipuerilis gen. nov. Also ‘Protospongia’ spina Mehl et al.,
1993 and, questionably, Granulispongia obscura gen. et sp.
nov.
REMARKS. Finks (1983) included only the Protospongiidae
within the Protospongioidea and no further families have
since been added. The morphology and the distinctive char-
acteristic of well-developed prostalia derived from modified
first-order spicules are here considered sufficient to separate
Asthenospongia and Acutipuerilis gen. nov. from other pro-
tospongiids. They appear to represent a coherent group with
no obvious descendants. The phylogenetic significance of
zero-order spicules is considered in the concluding section.
Genus ASTHENOSPONGIA Rigby et al., 1981
TYPE SPECIES.Asthenospongia acantha Rigby et al., 1981,
from the Arenig of Idaho.
OTHER SPECIES.A. cambria sp. nov and, questionably, Pro-
tospongia spina Rigby & Mehl, 1994.
DIAGNOSIS. (Emended after Rigby et al. 1981.) Thin-walled,
conical to cylindrical, with moderate to regular spicule ori-
entation, primarily orthogonal; rough quadrangles formed by
first-order spicules. Large zero-order hexactines with non-
orthogonal but sub-equal length rays are dispersed around
the body wall, producing substantial prostalia; these are
concentrated around the dorsal wall flexure, such that non-
orthogonal rays are parallel to the wall.
OCCURRENCE.Asthenospongia is known from the Middle
Ordovician of Wales (Caradoc) and the Lower Ordovician of
Idaho. ‘P.’ spina is from the ? Lower Devonian of Nevada.
REMARKS. Not previously known with a sufficient degree
of articulation for a full description of overall morphology.
‘P.’ spina Mehl et al., 1993, is almost certainly a member
of this family, but differences in low-order spicule arrange-
ment and body morphology prevent certain assignment to
Asthenospongia itself.
Asthenospongia cambria sp. nov. (Figs 7, 8)
TYPES. Holotype: SM X.39479. Paratypes: SM X. 39397,
39480–82, 39484, 39486–9, 39491.
OTHER MATERIAL. SM X. 39483, 39485, 39490. Fair speci-
mens, disarticulated, but not certainly of A. cambria.
DIAGNOSIS. High conical Asthenospongia with osculum oc-
cupying approximately one-half to two-thirds of dorsal dia-
meter; body wall retains quadruled mesh with local regular-
ity maintained over 1–2 first-order squares. Large hexactines
with ray length 2–3 times first-order ray length.

Caradoc sponge fauna from Central Wales 41
Figure 7 Asthenospongia cambria sp. nov. (A) SM X.39479, holotype, overall view, ×3.5; (B) SM X.39479 detail of oscular margin and spicule
ray orientation, ×10. (C) SM X.39486, paratype, disarticulated, ×5. All photographs taken under water.
OCCURRENCE. Locality 2b only.
ETYMOLOGY. Latin Cambria, Wales.
DESCRIPTION. Details of basal region unknown, but body is
high conical to sub-cylindrical with slightly convex sides.
The only articulated specimen is 18.3 mm wide and more
than 23 mm high, with the maximum width around 5 mm be-
low the upper surface. Osculum approximately 10 mm dia-
meter, in the centre of the upper surface, following inflexion
to horizontal. The oscular rim is pyritised in the holotype,
preventing detailed description of spicule arrangement in that
region.
Distal inflexion is marked by zero-order hexactines, of
which most rays are straight; a few curve smoothly, with
radius of curvature around twice ray length. Rays are of sub-
equal length within any spicule, distorted in orientation to
conform to body wall, with distal ray prostalial. Zero-order
spicules also present on lateral walls, but less frequent; these
also have irregular ray orientations, although generally with
coplanar rays parallel to wall.
Smaller size orders are regular hexactines and perhaps
some pentactines and/or stauracts, in very loose quadruling.
At least five orders of spicules present. All smaller spic-
ules with near-orthogonal, straight rays, tapering smoothly.
First-order spicules are dominantly sub-orthogonal to sides of
sponge, but with substantial variation. Second-order spicules
are of varied orientation, but smaller orders are dominantly
diagonal. There is no tract development, or systematic ray
overlap. As a result, the wall is very fragile, as also described
Figure 8 Cut-away reconstruction of Asthenospongia cambria sp.
nov.; scale bar =5 mm.
for A. acantha Rigby et al., 1981. Although much of the mesh
is open space, there is no evidence of distinct parietal gaps.
REMARKS.A. cambria can be distinguished from A. acantha
Rigby et al., 1981, by the more regular arrangement of

42 J. P. Botting
zero-order spicules in the latter and the relative sizes of first-
and zero-order spicules. There is also a suggestion that A.
acantha was much shorter, in that some specimens were ap-
parently compressed vertically in the type material. The large
hexactine morphology is, at present, sufficient to distinguish
the species from all others, including Acutipuerilis spinosus
gen. et sp. nov.
Genus ACUTIPUERILIS gen. nov.
TYPE SPECIES.Acutipuerilis spinosus gen. et sp. nov.
DIAGNOSIS. Asthenospongiids in which enlarged hexactines
have only one extended, prostalial ray, which is highly de-
veloped in the juvenile phase.
OCCURRENCE. Only known from the Caradoc of central
Wales.
ETYMOLOGY. From the Latin acutus, sharp, violent and pu-
erilis, juvenile, for the extreme spinosity of the early growth
stages.
REMARKS. This genus is clearly separated from all other
reticulosans by the remarkable elongation of prostalia in an
otherwise simple protospongiid body wall. It is separated
from Asthenospongia by the extreme development of one
ray of the large hexactines, rather than equal growth of all
rays to large size.
Acutipuerilis spinosus gen. et sp. nov. (Figs 9, 10)
TYPES. Holotype: SM X.41083; complete. Paratypes: SM
X.39396, 41084–5; two apical specimens and one complete
juvenile.
OTHER MATERIAL. SM X.41086. Disarticulated, but appears
to be this species.
DIAGNOSIS. As for genus.
ETYMOLOGY. Latin spinosus,spiky.
OCCURRENCE. Locality 2a only.
DESCRIPTION. High conical sponge with divergence of wall
from axis of 20–30◦, generally straight-sided but walls some-
times reducing in angle distally to near-parallel. Ventral sur-
face marked by change in angle to sub-horizontal, forming
partial lid over approximately two-thirds of diameter; flex-
ure is rounded, but with small radius of curvature. Flexure
is marked by zero-order spicules with hypertrophied distal
rays, curving upwards towards axis. The hypertrophied ray
is up to 10 mm long in largest specimens, with basal ray dia-
meter of 0.05 mm. The remaining rays are greatly reduced.
Zero-order spicules also over body wall surface, also with
distal ray curving upwards towards axis, and form isolated
basalia near the base, there curving downwards.
Spicules of body wall dominated by small, simple hex-
actines of at least four orders, possibly some stauracts, but
not confirmed. Mesh sparse, with little coherence; first-order
spicules separated, arranged irregularly, but each generally
associated with locally orthogonal spicules of smaller or-
ders. Approximately one-fifth of smaller orders are diagonal
to the associated first-order spicule, the rest more-or-less or-
thogonal.
Holotype is 10.8 mm high, with maximum width
6.0 mm. Zero-order spicules extend at least 7.6 mm from
margins of upper surface and 4.7 mm from base. First-order
spicules have ray lengths of up to 0.6 mm, with basal ray dia-
meter of 0.04 mm. The partial paratype X.41085 has max-
imum width 5.8 mm and zero-order spicules extending at
least 9.3 mm distally. First-order spicules have ray lengths
up to 0.7 mm, suggesting that although the body width is
slightly less than that of the holotype, it represents a later
growth stage.
Paratype X.41084 has body length 3.8 mm and max-
imum width 1.8 mm. Zero-order spicules extend over 4.1 mm
distally and 2.5 mm basally, the former probably incomplete.
Short, 0.4 mm lateral prostalia also occur. Spicule rays in the
disarticulated specimen X.41086 do not exceed 10 mm, des-
pite the much larger array of spicules, which implies a larger
body size. This suggests that the importance of the prostalia
was emphasised in the early growth stages and growth rate
reduced through ontogeny.
Family ? ASTHENOSPONGIIDAE fam. nov.
REMARKS. It is unclear whether the large spicules of Gran-
ulispongia should be regarded as first-order or zero-order,
but they do not appear to be prostalial. However, in all other
respects, Granulispongia is similar to the asthenospongiids
and may best be questionably included, at least temporarily,
within the family. If eventually rejected from the Astheno-
spongiidae, it should fall instead in the Protospongiidae,
pending further division.
Genus GRANULISPONGIA gen. nov.
TYPE SPECIES.Granulispongia obscura gen. et sp. nov.
DIAGNOSIS. Thin-walled protospongiids with six or seven
orders of dominantly hexactine spicules; the largest are of
appropriate size to be first-order, but are sparsely distributed
and with a greatly reduced radial axis; second-order spic-
ules rare. Spicules are arranged in a coarse grid, but with
little regularity at any scale and poor organisation within the
approximate squares. Includes spicules with all rays curved
in one orientation and others with the two primary axes ob-
lique. Skeleton with numerous small granules, concentrated
near the apex.
ETYMOLOGY. Referring to the small mineralised granules
that particularly characterise this species, but are lost in more
derived protospongioids and their descendents.
REMARKS. Since the genus is currently monospecific, Gran-
ulispongia can be identified from the diagnostic spicule mor-
phologies and arrangement. The granules are not unique to
this genus, but have so far received little attention, due in
part to the difficulty of distinguishing them from embedded
spicule rays or entirely diagenetic features.
Granulispongia obscura gen. et sp. nov. (Figs 11, 12)
TYPES. Holotype: SM X.41087. Paratypes: SM X.39492,
41001–8, 41010. Primarily disarticulated, but containing the
diagnostic spicules.

Caradoc sponge fauna from Central Wales 43
Figure 9 Acutipuerilis spinosus gen.etsp.nov.(A) SM X.41083, holotype, ×6, taken under ‘dry water’ polarisers (Crabb 2001); (B)SM
X.41085, paratype, under water, ×5; (C) SM X.39396, paratype, under water, ×12; (D) SM X.41084, paratype (juvenile), under water, ×9.
Figure 9(c) reproduced with permission from Botting (2003).
DIAGNOSIS. As for genus.
ETYMOLOGY. Latin, obscura, indistinct; referring to the un-
certain body morphology.
OCCURRENCE. Localities 1, 2a and 2b.
DESCRIPTION. The available specimens are poorly articulated
or consist of associations of spicules only, with the morpho-
logy partly visible only on the holotype. Isolated spicules
are also known, suggesting that the sponge was not rare, but
relatively fragile. Sponge body broadly globose with oscu-
lum approximately half of maximum width. Basal region not
seen. Diameter of compressed body approximately 20 mm,
although with some disarticulation at margins.
Spicules comprise single layer of hexactines with re-
duced distal rays, organised in very weak orthogonal quad-
ruling, only clearly visible around the osculum of the holo-
type. Only hexactines definitely present, although some with
distal ray apparently reduced to a small node. Rays are un-
usually slender, with ray length to basal diameter ratio being
approximately 20–25:1 in observed material. Ray diameter
is constant for a large part of the length, then tapers rap-
idly. Maximum observed length 3.3 mm, corresponding to
a width of 0.12 mm. Lower size orders reduce by one-third

44 J. P. Botting
Figure 10 Cut-away reconstruction of adult Acutipuerilis spinosus
gen. et sp. nov., including position of osculum; scale bar =5 mm.
at each order, with at least six, probably seven, size orders
present in the holotype. The largest size order appears to be
slightly hypertrophied, but no complete spicules are visible;
however, one suggests a ray length of at least 4.8 mm, and
the distal ray is often strongly reduced in these spicules. It is
unclear whether it should count as first-order or zero-order,
but other characteristics also suggest affinity with the as-
thenospongiids. Two common modifications of spicules are
those with non-orthogonal tangential rays, typically altered
to 70–80 and 100–110◦, and smooth curvature of all rays in
one direction. Each modification can occur in any size order.
REMARKS. The modified spicules are diagnostic at present,
although the curved morphotype is similar to the hyper-
trophied spicules of ‘Protospongia’spina Mehl et al., 1993,
which differs in possessing a much more regular grid. De-
tails of spicule growth parameters may be useful for dis-
tinguishing closely related taxa, should they be discovered.
The only species currently described that could be confused
with it are Asthenospongia acantha and A. cambria sp. nov.,
both of which possess modified zero-order spicules with un-
predictable ray orientations and they lack both the regular
modifications and the abundant granules seen here.
Family UNCERTAIN
Indeterminate protospongioid sp. A (Fig. 13A)
MATERIAL. SM X.41097; single, near-complete; base dam-
aged.
Figure 11 Granulispongia obscura gen. et sp. nov. (A) SM X.41087,
holotype, overall view, ×4; (B), SM X.41087, detail showing
diagnostic spicule morphologies, ×30. (C) SM X.41088, paratype,
showing detail of disarticulating oscular margin (?), ×20. All
photographs taken under water.
Figure 12 Cut-away reconstruction of Granulispongia obscura gen.
et sp. nov.; scale bar =5 mm.

Caradoc sponge fauna from Central Wales 45
Figure 13 Indeterminate protospongioids, all from locality 2a. (A) SM X.41907, sp. A, ×7.5; (B) SM X.41099, sp. B, ×15; (C) SM X.41186, sp.
C, ×23; (D) SM X.41187, sp. D, ×5. Photographs (A), (B) and (D) taken under water.
OCCURRENCE. Locality 2a.
DESCRIPTION. Tall conical sponge preserved as a mass of
irregularly distributed spicules, partly hidden by iron oxides
that mark soft tissue. Height 18 mm, with maximum width
of 8 mm, at apex. Osculum occupies essentially entire upper
surface. Base 3 mm wide, rounded, with some scattering of
spicules; widening evenly thereafter. At least five orders of
spicules present, but no distinct zero-order. Spicules entirely
hexactines or simple derivatives. Hexactines have maximum
observed ray length of 0.48 mm, corresponding to basal ray
diameter of 0.038 mm. Smaller size orders reduce by a factor
of approximately one-third at each order. No visible orna-
ment or modification of rays, which are straight and smoothly
tapering.
Arrangement of spicules is unclear over most of sponge
body. Regular quadruling is absent and there is no consistent
orientation, but spicules tend to subdivide the irregular spaces
rather than clustering randomly. Spicule orientations are ex-
tremely variable, but near-45◦angles between spicules are
very common, suggesting that orientation is not completely
irregular. No parietal gaps.

46 J. P. Botting
Base preserves no anchorate structure, but may be
broken.
REMARKS. The lack of zero-order spicules and simplicity of
the hexactines, combined with the single-spicule thickness
of the skeletal wall, suggests a position within the Protospon-
giidae, but the irregular mesh supports a slightly more derived
position. Heminectere, for example, preserved a much more
regular grid, while those taxa described with a less regular
arrangement generally possess more derived features (e.g.
spicule morphology and parietal gaps in Hemidiagoniella
caseus gen. et sp. nov.; thicker wall in Dilatispongia tumidus
gen. et sp. nov.). The new sponge should be regarded as a
derived member of an early lineage of the Protospongiidae,
whose loss of a regular mesh is convergent with that in the
rossellimorphs, discussed below.
Indeterminate protospongioid sp. B (Fig. 13B)
MATERIAL. Two small mesh fragments, SM X.41098–9.
DESCRIPTION. Each fragment consists of parts of six grid
squares, with some scattered spicules nearby, but SM
X.41098 has grid parameters three times larger than than
of SM X.41099 and represents a much later growth stage.
The spicules are primarily hexactine, with possibly some
stauracts; extent of proximal and distal rays unknown. There
are at least four spicule orders, perhaps also with zero-order
spicules: large, diagonally-orientated and irregularly posi-
tioned spicules do not preserve ends of rays. The majority
of observed spicules are first- and second-order, with very
few of smaller sizes. Length of suspected zero-order spicule
rays at least 5.4 mm, with basal ray diameter of 0.12 mm.
Maximum observed first-order ray length is 4.3 mm, with
basal ray diameter of 0.09 mm. In specimen SM X.41099,
basal ray diameter is 0.04 mm for first-order spicules, with
ray length at least 1.4 mm. First-order spicule rays overlap
extensively in both specimens.
OCCURRENCE. Locality 2a only.
REMARKS. Little discussion can be made of this species,
beyond the fact that it differs from all others in the fauna. The
large, slender spicules and partial tract development suggest
a position on the Cyathophycus lineage, supported by the
presence of zero-order spicules.
Indeterminate protospongioid sp. C (Fig. 13C)
MATERIAL. SM X.41186; single specimen, very early growth
stage.
DESCRIPTION. Globular sponge body of diameter 1.9 mm,
showing faint impressions of hexactine-based spicules (prob-
ably stauracts) in orthogonal square grid. Grid is poorly pre-
served over most of the specimen, and each part visible only
under specific lighting, but appears at least moderately reg-
ular, with consistent orientation over most of the flattened
surface. In the centre, grid square size is 0.15 mm, but in
places there is sub-division to 0.06 mm. Individual spicules
are too small to allow diagnosis of size orders. Basal tuft
aligned at 90◦to the sponge body, assuming that the conver-
ging spicule rays at the top of Fig. 13C represent one pole
of the sponge. At least one basalia spicule is hexactine with
a hypertrophied ray, 1.4mm long. Basalia converge distally,
but perhaps resulting from transport. Only four basalia are
certainly present, but faint traces suggest the presence of at
least two more.
REMARKS. The sponge is presumably a juvenile and the grid
is insufficiently developed to allow comparisons with other
taxa. The regular grid and sub-spherical morphology recalls
Cyathophycus loydelli sp. nov., but the basalia are distinct
hexactine spicules, as seen in Hemidiagoniella gen. nov. and
the asthenospongiids. In all the latter species, however, the
grid is only weakly consistent and it seems unlikely that it
represents the juvenile of one of the other species described in
this paper. The adult would probably resemble Protospongia
tetranema Dawson & Hinde, 1888, although perhaps with a
greater number of basalia.
Indeterminate protospongioid sp. D (Fig. 13D)
MATERIAL. SM X.41187.
OCCURRENCE. Locality 2a.
DESCRIPTION. Single, partial specimen showing neither ter-
mination and disarticulating on one side. Length of fragment
23 mm and maximum width 15 mm. Irregular but coherent
mesh of fine hexactines, some with perpendicular rays re-
duced, and others with full basal diameter in the distal or
proximal ray. It is unclear which surface is visible, since
spicules are small and clinopentactines not evident. The mesh
consists of locally consistent orientations among some struc-
tural spicules, but these are overlain by many smaller spicules
in orthogonal, diagonal, or irregular orientations. Positions
of spicule centres are not regular, such that first-order grid is
difficult to specify over a wide area and, where present, is not
sub-divided by second-order spicules in the centres of inter-
vening spaces, but at random positions. First-order squares
appear expanded, containing more structural complexity and
more spicules than in quadruled patterns.
Largest (zero-order) spicules with ray length of 2.4 mm
and basal ray diameter of 0.07 mm; these are oriented diagon-
ally, and sporadically. First-order spicules with ray length of
1.6 mm and basal diameter 0.04 mm. At least six subsequent
size orders, the smallest with ray length 0.1 mm; reduction
of ray length by one-third at each order. Ray diameter is
constant over most of length. All spicules with straight rays,
but rare examples are non-orthogonal, with opposing rays
parallel, to angles of 80◦. Small granular bodies also present,
but not abundant; origin uncertain.
REMARKS. The limited data prevent a full description of this
species, which in many respects is very similar to Gran-
ulispongia obscura gen. et sp. nov. Differences are the relat-
ively small size and slenderness of spicules compared with
body size, the absence of the curved spicule morphology and
the relative rarity of granules. The grid structure is better
preserved in this specimen than in G. obscura, but represents
a similar grade of organisation. Few other described taxa
could be confused with this species, although it is possible
that similar fragmentary material is known, but remains un-
described or in open nomenclature. The difficulty with taxa
of this type is that they lack any clear distinguishing fea-
tures and statistical studies are often impossible due to small
sample sizes.

Caradoc sponge fauna from Central Wales 47
Order UNCERTAIN
REMARKS. Overlap between extant and palaeontological tax-
onomy is negligable among the ‘Rossellimorpha,’ and there
are major difficulties with paraphyly. The species below cer-
tainly fall within the order Reticulosa, as proposed by Reid
(1958) (but regarded by Krautter 2002 as polyphyletic), but
in modern terms should be classified among the order Lyssa-
cinosida, as discussed in detail in the section on phylogeny,
below. The term ‘Rosselimorpha’ is available for these taxa,
but is almost certainly polyphyletic, referring to a structural
grade and should not be encouraged. A full revision is re-
quired before the biological and palaeontological systems
can be reconciled.
Family UNCERTAIN
Genus HEMIDIAGONIELLA gen. nov.
TYPE SPECIES.Hemidiagoniella tenax gen. et sp. nov.
DIAGNOSIS. Thin-walled protospongiid (single spicule layer)
with dense mesh of acanthohexactinal spicules with reduced
distal ray, in locally regular sub-quadruled arrangement,
but minimal lateral consistency; orientation of highest-order
squares is essentially independent of those three squares
away. Several spicule orders present.
ETYMOLOGY. Latin, half-diagonal, referring to the variable
orientation of the reticular grid.
OCCURRENCE. Only certainly known from the type locality,
except for a single specimen, possibly of H. tenax, from loc-
ality 1. Isolated spicules closely resembling Hemidiagoniella
have been recorded in the uppermost murchisoni Biozone of
Bach-y-Graig stream section, Llandrindod (SO 073 611).
OTHER SPECIES.H. caseus gen. et sp. nov.
REMARKS. The skeletal structure of Hemidiagoniella rep-
resents a previously unrecorded variation on the standard
protospongioid arrangement. Most genera related to a Pro-
tospongia-like form either lose the regularity of the mesh
entirely, or develop overlapping spicule rays to form tracts,
as in the dictyosponges. In this case, the local regularity is
largely retained, but the arrangement of first-order spicules
has become irregular. A slight tractose development in H.
caseus is related to parietal gap development and is gener-
ally curved in irregular orientations.
Although isolated acanthohexactines are known from
older deposits, this is the earliest example where provenance
can be determined. The recognition of surface ornament on
in situ spicules is often impossible due to weathering of pyrit-
ised fossils and low resolution of surface features in external
moulds. The recognition of spicular ornament in hexactinel-
lids allows more useful comparisons with disarticulated ma-
terial and some similarity exists between Hemidiagoniella
and the indeterminate specimens of Watkins & Coorough
(1997) and others, as well as numerous modern lyssakids
(e.g. Tabachnik 2002a,b,c).
Hemidiagoniella tenax gen. et sp. nov. (Figs 14A–C,
15)
TYPES. Holotype: SM X.41100; large partial specimen
with excellent spicular mould preservation. Paratypes:
SM X.41101–21; partial specimens showing features of mor-
phology or structure.
OTHER MATERIAL. SM X.39426, SM X.41022–41139; num-
bers refer to slabs–sometimes multiple specimens.
DIAGNOSIS.LargeHemidiagoniella in which the spicular
mesh is homogeneous, with little or no tract development.
Spicule sizes decrease by approximately one-third at each
order. Spicules with greatly reduced proximal and distal rays,
distal rays apparently only present as small nodes; slight
ventral curvature of lateral rays. Overall form based on high
cone with flexure point at maximum width, at three-quarters
of sponge height, then contraction to around three-quarters
of maximum width at apex.
ETYMOLOGY. Latin tenax, tough, persistent, referring to the
remarkable degree of articulation of most specimens.
OCCURRENCE. Only certainly known from locality 2a, with
single possible specimen from locality 1.
DESCRIPTION. Specimens of this species are typically pre-
served intact and fully articulated, although they dominate
in a spiculitic bed (locality 2a) which may represent many
disarticulated specimens. Due to large size (greater than
10 cm high; maximum observed width 68 mm), no complete
specimens are known, but sufficient fragments to allow re-
construction of entire form. Sponge body is steep conical
with rounded base and flexure point at around three-quarters
of height, after which diameter contracts slightly. Expan-
sion approximately 20◦from axis up to the flexure, nar-
rowing at 10–20◦thereafter. Osculum comprises the entire
apex.
Skeleton is a dense single-layer mesh of acantho-
clinopentactines with a very small distal node representing
the sixth ray. Curvature of tangential rays is greatest at the
node, straightening and inflexing distally to nearly paral-
lel surface. Local reticulation is regular, with quadruling
including almost entirely diagonally-arrayed second-order
spicules; subsequent orders may be diagonal or orthogonal
relative to local grid. Reticulation irregular on scale greater
than two or three first-order squares, deviating to inclined
orientations. First-order grid sometimes confused by pres-
ence of first-order spicules in diagonal second-order posi-
tions, allowing curvature and realignment of primary grid.
Five, probably six, size orders present, spicules similar on all
scales. Zero-order spicules present in small numbers, typic-
ally arranged diagonally.
Maximum observed ray length of first-order spicules is
2.7 mm, but often difficult to assess due to inclination of rays
to surface. Corresponding basal ray diameter of 0.08 mm;
aspect ratio of spicule rays appears to be constant at 35–40:1
for all size orders.
Several discrete prostalia in the basal region formed
by elongating distal ray of hexactines; basalia project sub-
perpendicularly to sponge surface, but limited in length to a
few millimetres.
REMARKS. The broad spicule arrangement and morpho-
logy are sufficient to distinguish this species from all
other fossil genera. Comparison with H. caseus is given
below.

48 J. P. Botting
Figure 14 Hemidiagoniella gen. nov. (A–C)H.tenax gen. et sp. nov. (A) SM X.41101, paratype, showing basalia and proximal morphology, ×10;
(B) SM X.41102, paratype, ×2.8; (C) SM X.41100, holotype, showing detail of mesh, ×15. (D)–(G)H. caseus gen. et sp. nov. (D) SM X.41140,
holotype, detail of upper region, showing spicule arrangement, ×5; (E) SM X.41140, basal region, ×2.5; (F) SM X.41141, paratype, showing
parietal gaps; (G) SM X.41188, paratype, near-complete but disarticulating, ×5. All photographs taken under water.

Caradoc sponge fauna from Central Wales 49
Figure 15 Cut-away reconstruction of Hemidiagoniella tenax gen. et
sp. nov.; scale bar =10 mm in overall view and 1mm in detailed view.
Hemidiagoniella caseus gen. et sp. nov. (Figs 13D–G,
16)
TYPES. Holotype: SM X.41140; complete, but disarticulat-
ing at edges. Paratypes: SM X.41141–2, fragments; and SM
X.41188, complete but disarticulating.
OTHER MATERIAL. HCM.1998–21/254 (Hereford City Mu-
seum); small fragment, lacking diagnostic parietal gaps, but
tentatively referred to H. caseus.
ETYMOLOGY. Latin caseus, cheese, after the inconsistently-
sized holes irregularly arranged in the sponge wall. Noun in
apposition.
DIAGNOSIS. Broad conical Hemidiagoniella with partial me-
andering tractose development and areas of sparse spic-
ule cover representing parietal gaps. Spicules include large
zero-order, thin-rayed acanthohexactines. Hexactines with
reduced proximal and distal rays, distal rays reduced more
than proximal; lateral rays almost coplanar.
DESCRIPTION. Outline not preserved sharply in rare complete
specimens, but broad conical with rounded base. Osculum
entire width of sponge, but margin unclear. Width of holotype
45 mm, at height 50 mm.
Skeletal mesh of acanthose hexactines with reduced ray,
approximating clinopentactines; proximal ray also reduced,
but present. Largely irregular, although with very local or-
dering within part of one first-order square, often rotated
from orthogonal alignment. Structure is perforated by irreg-
ular parietal gaps up to 4 mm in width; largest spicule rays
cross the gaps and smaller size orders cluster at the margins,
further distorting any mesh regularity. Where arrangement
is clear, it is based on quadruling, often with diagonal low
orders. There is no coherent first-order grid.
Largest spicules hypertrophied, with maximum ob-
served ray length in the holotype being 10.0 mm and basal
diameter 0.25 mm, while the corresponding dimensions for
the next largest are 5.3 mm and 1.3 mm. This marked size
Figure 16 Cut-away reconstruction of Hemidiagoniella caseus gen.
et sp. nov.; scale bar =10 mm in overall view and 1mm in detailed
view.
difference and the sporadic placement suggests that these are
best interpreted as zero-order spicules. In addition to these,
there are at least six further orders in large specimens. Ray
length to basal diameter ratio appears to be around 40:1 in all
size orders, but is often impossible to measure due to inclined
rays.
No prostalia or basalia have been recorded, but this
may be due to slight marginal disarticulation in the known
specimens.
DISCUSSION. The broad spicule arrangement and morpho-
logy are diagnostic of the genus and would be difficult to
confuse with other described taxa. H. caseus is distinguished
from H. tenax by the presence of parietal gaps, reduced mesh
regularity, and large spicule size relative to body dimensions;
and in disarticulated remains by the hypertrophied zero-order
spicules.
Genus SOLUSRECTUS gen. nov.
TYPE SPECIES.Solusrectus rosetta gen. et sp. nov.
DIAGNOSIS. Thin-walled protospongioid with regular ex-
ternal orthogonal mesh comprising almost exclusively
clinopentactines, some with central ornament, in partly quad-
ruled array. Spicules with reduced proximal rays and distal
rays as slight nodes only; slight proximal recurvature of lat-
eral rays, with centre forming boss. Inner layer of sparse
oxeas, dominantly aligned parallel to the vertical component
of the dermal mesh and normally arranged in pairs.
ETYMOLOGY. Latin, ‘only vertical,’ after the occurrence of
monaxons in the subdermal layer.
OCCURRENCE. Known only from the type locality.
REMARKS. This is an extremely interesting genus, in that it
shows characteristics of both the protospongiids and early
dictyospongioids. It resembles Heminectere minima gen. et
sp. nov. in possessing a part-quadruled wall of dominantly
pentactines, with some ray overlap indicating the beginning
of tract development. The weakly-developed inner layer is,
however, typical of Cyathophycus,asisthesizeandar-
rangement of zero-order spicules. These features suggest a

50 J. P. Botting
Figure 17 Solusrectus rosetta gen.etsp.nov.(A)–(D) SM X.41143, holotype. (A) detail of mesh in apical region, inluding paired monaxons
(i) and apparent pentaradiate (ii), ×10; (B) detail of inner layer, showing several pairs of vertically-aligned monaxons, under water, ×20;
(C) part of proximal region of skeleton, showing pyritic film over soft tissue and strong replacement of spicules, under water, ×5; (D) detail of
individual rosette clinopentactine from distal region, ×40.
relationship to the early stages of Cyathophycus evolu-
tion, but the morphologically derived clinopentactines are
more complex than the simple hexactines of Cyathophy-
cus, suggesting convergent acquisition of either clinopentact-
ines or the inner monaxon layer between lineages (see sec-
tion on phylogeny, below). The detailed morphology of the
clinopentactines is also very close to that described for some
supposed brachiospongiids (Webby & Trotter 1993). Mor-
phologically, however, the inner layer is sufficient to classify
the genus with the Dictyospongioidea.
Solusrectus differs from Cyathophycus in the nature of
the monaxial layer and it differs similarly from Cyathophy-
cella, while also having a less banded distribution of pentact-
ines. No other bilaminar reticulosans are likely to be confused
with Solusrectus, if the inner layer is preserved.
Solusrectus rosetta gen. et sp. nov. (Figs 17, 18)
TYPES. Holotype: SM X.41143, near-complete sponge with
some disarticulation.
OTHER MATERIAL. SM X.41144–45; fragments probably at-
tributable to S. rosetta.
DIAGNOSIS. As for genus.
Figure 18 Interpretative cut-away reconstruction of Solusrectus
rosetta gen. et sp. nov.; scale bar =10 mm in overall view, 1mm in
detailed view. Regularity of the mesh is slightly enhanced, allowing
for the slight disarticulation of the holotype.
ETYMOLOGY. Latin for rosette, after the distinctive ornament
on clinopentactine bosses. Noun in apposition.
OCCURRENCE. Locality 2a only.

Caradoc sponge fauna from Central Wales 51
DESCRIPTION. Overall form of sponge uncertain; the holotype
is a partial specimen showing broadly cylindrical form, dis-
articulating on one side. Width 36 mm and preserved length
65 mm, but probably originally much longer. Skeleton bil-
aminar, with outer layer of moderately regular, quadruled
clinopentactines. Inner layer of sparse monaxons arranged
dominantly vertically, parallel to the outer layer.
Clinopentactines mostly sub-orthogonal, but with fre-
quent spicules at 45◦to first-order squares, which are defined
by partial ray overlap in the Cyathophycus pattern. Sub-
division of first-order squares by up to six orders of smaller
clinopentactines, in locally regular quadrule arrangement.
Clinopentactines with reduced proximal ray and distal ray
very reduced or absent. No clear relationship between spicule
size and the development of distal ray. Lateral rays initially
convex, forming a prominent boss, then inflexed to become
parallel to the body wall. In many spicules, the boss is orna-
mented by a ring of 8–10 small tubercles, arranged in a circle
around the distal ray. In some spicules, there appear to be
further tubercles over the proximal parts of the lateral rays,
but preservation is poor. Rays are generally straight, with
occasional spicules showing significant curvature of one or
more rays and tapering gradually over their length.
Zero-order spicules are distributed sporadically over the
sponge and are invariably oblique (usually close to 45◦)to
the primary mesh. They appear to be morphologically similar
to first-order spicules, but in some cases lack any distal ray
or ornament. Due to their size, the boss is raised slightly
above the sponge surface, but the rays appear to lie below the
lower-order spicules that cross them.
Inner layer poorly developed, comprising fine oxeas in
dominantly vertical orientation. Although not clearly seen,
the oxeas regularly occur in parallel pairs, even in disarticu-
lated areas; in one region, three sub-vertical pairs and three
single oxeas lie sub-vertically and transversely aligned. A
few sub-horizontal oxeas are also seen and the layer is pre-
sumed to be a partial orthogonal array.
A single, second-order ?pentaradiate is seen (Fig. 17A).
This is assumed to be teratomorphic, in the absence of further
material.
REMARKS. The precise arrangement of the partially-
preserved inner layer must be regarded as tentative, although
it is clear that such a layer exists and that it was composed
primarily of orthogonal oxeas parallel to the primary mesh.
This feature is of great interest in taxonomic and phylogen-
etic terms, as it seems to represent a simple (and perhaps
primitive) variation on the inner spicule layer of Cyathophy-
cus and the dictyosponges. This is discussed more fully in
the section on phylogeny, below.
The similarity to other sponges is discussed under the
genus and there seems to be little chance of confusion in well-
preserved material. If the inner layer is not visible, confusion
is possible with the other clinopentactine-bearing sponge
Hemidiagoniella tenax gen. et sp. nov., but the regularity
of the dense dermal mesh and extensive spicule ornament of
H. tenax distinguishes them.
Family ? DICTYOSPONGIIDAE Hall, 1884
REMARKS.Cyathophycus was placed in the Hintzespongiidae
by Finks (1983) and followed by Rigby (1995), Rigby &
Mehl (1994) and many subsequent workers (e.g. Krautter
2002), based on the presence of two spicule layers in the
body wall. Previous interpretations placed them either in the
Protospongiidae (de Laubenfels 1955), or the Dictyospon-
giidae (Hall 1884; Rigby 1986, by implication). The sim-
ilarity to the Hintzespongiidae relates to the inner layer of
irregularly-oriented hexactines, which is compared with the
monaxon layer of Cyathophycus.However,inCyathophy-
cus, the layer is only irregular if parietal gaps are present and
it is frequently composed of monaxons, despite the emen-
ded diagnosis of Rigby & Mehl (1994), which states that
the inner layer consists of irregularly-arranged hexactines or
their derivatives, surrounding parietal gaps. However, Rigby
(1995) described specimens of C. reticulatus Walcott, 1879
(the type species) and C. pseudoreticulatus Rigby, 1995 and
found no trace of the inner spicule layer in either species.
Walcott (1879, 1881) also makes no mention of a second-
ary layer in C. reticulatus and the spicules are referred to
as ‘thread-like’ (Walcott 1881: 395), presumably implying
monaxons. Dawson & Hinde (1888, 1889) also found no in-
ner layer and the present species possesses an inner layer
with no parietal gaps, but a regular cross-hatch of grid-
parallel monaxons. In contrast, Rigby & Chatterton (1994)
found a possible dermal layer with parietal gaps in their new
species, C. mackenziensis, although spicules could not be
discerned.
The species assigned to Cyathophycus so far include
a wide diversity of morphologies and structures, the only
fully consistent features being the partial bundling of spicule
rays in the regular orthogonal grid of the primary layer. The
secondary layer is either present or absent and, if present,
is dermal or gastral, consisting of regular or irregular mon-
axons (or hexactines?), or perhaps no megascleres at all.
None of these species are closely similar to Hintzespongia,
in that the nature of neither spicule layer has been shown
to correspond closely to that of the latter. The inner layer of
Cyathophycus is considered here to be non-homologous with
that of Hintzespongia, because of the absence of zero-order
spicules in the latter and the lack of either dermal quadrul-
ing, or a clear description of hexactinal spicules in the inner
layer of Cyathophycus. The earlier view, that Cyathophycus
represents a primitive member of the Dictyospongioidea, is
preferred. This is also discussed in detail in the section on
phylogeny, below. The genus Cyathophycus requires revision
and clarification.
Genus CYATHOPHYCUS Walcott, 1879
REMARKS. The morphology of Cyathophycus loydelli is
closer to that of dictyosponges than is that of most other
Cyathophycus species, in that circular parietal gaps are not
developed. Parietal gaps existed in some dictyosponges, but
they were formed by vertical and horizontal separation of
the tracts, rather than irregular, discrete regions. The mor-
phology of the first true dictyosponge is unknown, but both
open conical and sub-globose forms have been described,
the conical being much more diverse. Cyathophycus was
probably close to the ancestral dictyosponge, but the pre-
cise significance of the genus is not clear. The earliest dic-
tyospongioids, such as Quadrolaminiella and perhaps Pa-
laeophragmodictya, differed in some structural character-
istics and Mehl (1996), Mehl-Janussen (1999) has argued
convincingly for polyphyly. In this case, Cyathophycus

52 J. P. Botting
would be a natural structural and stratigraphic precursor
of the Dictyospongiidae (sensu stricto). The presence of
diagonally-oriented zero-order spicules with reduced distal
rays in both C. loydelli and Reticulicymbalum tres (Botting,
in press a)supports this interpretation. Pending a revision cla-
rifying the diagnosis of Cyathophycus,C. loydelli is included
within it.
Cyathophycus loydelli sp. nov. (Figs 19, 20)
2003 Cyathophycus sp. nov. Botting, p. 42.
In press b Cyathophycus sp. nov. Botting.
TYPES. Holotype: SM X.39398; showing both skeletal layers,
but lacking basalia; locality 2b. Paratypes: SM X.39394–95,
SM X.39399–406, SM X.39427; varied specimens, mostly
complete, from localities 1, 2a–c.
OTHER MATERIAL. SM X.39407–39425, SM X.39428–51.
Specimen numbers refer to slabs, containing in total approx-
imately 200 specimens, often crowded on bedding planes
and in all growth stages. Primarily from locality 1, but with
additional specimens from all others.
DIAGNOSIS. Sub-spherical Cyathophycus lacking parietal
gaps, with maximum spicule ray length around 1.2 mm and
at least five spicule orders, including zero-order. Root tuft
small rope of fine monaxons.
ETYMOLOGY. After D. K. Loydell, who donated the first
specimens from the type locality and thus initiated the in-
vestigation.
OCCURRENCE. All sponge localities at the Llanfawr Quarries,
but not known elsewhere.
DESCRIPTION. Body morphology truncated spheroid with
broad osculum occupying two-thirds of total width. Max-
imum observed diameter 60 mm and minimum recognised
1.7 mm, although smaller specimens could have been over-
looked. Detailed description of skeletal dimensions and
growth was provided by Botting (2003) and only a summary
is required here.
Skeletal mesh comprises eight to 12 first-order squares
in compressed equatorial half-circumference at earliest
growth stages, with strongly overlapping rays. Second-order
spicules also present at early growth stages, sub-dividing the
primary grid but relatively inconspicuous. Homogeneous ex-
pansion of spicules and grid with size, with appearance of
third and fourth-order spicules by a diameter of 4 mm. At
around 6 mm, the second-order spicules have enlarged to ap-
pear part of the first-order grid, resulting in doubling of the
number of squares/circumference. Further expansion leads
to an additional increase of around 50%, and a final value
of around 30–35 squares/compressed half-circumference.
Spicules expand continuously to a maximum ray length of
1.2 mm, which only first-order spicules have been observed
to have reached, at a body diameter of around 40 mm. In the
described population, very few individuals reached this size,
with most being up to 15 mm in diameter.
All dermal spicules are hexactines, smooth with evenly
tapering rays and no obvious modification of the distal ray,
except for some reduction; it is difficult to establish the length
of prostalia due to their orientation in the rock, but basal
distal ray diameter was significantly less than that of lateral
rays.
The inner layer is intermittently preserved, since almost
all specimens are external moulds with some replacement of
the dermal layer, but is visible between the dermal spicules in
some specimens. Individual spicules cannot be distinguished
due to dense packing, except in slightly disarticulating spe-
cimens, where they are oxeate monaxons. Arrangement is in
dense orthogonal bands parallel with the dermal grid. The
spatial relationship between vertical and horizontal monax-
ons could not be established. No evidence of parietal gaps
has been observed in either layer, although it is possible that
they exist in the monaxon layer.
The root tuft is not preserved articulated with any spe-
cimen and appears to have been more fragile than the body,
although several isolated examples have been recovered from
the near-monospecific beds at locality 1, some in close asso-
ciation with body fossils and these are referred to C. loydelli.
The tuft consists of perhaps 70–100 fine monaxons, each
0.02 mm wide and several millimetres long. Individual spic-
ules do not appear to run the entire tuft length and termina-
tions are oxeate where visible. The tuft width is constant at
approximately 1 mm and length at least 10 mm, up to 34 mm
observed maximum.
DISCUSSION. No other species of Cyathophycus has a
sub-spherical form. Several taxa with similar spicule arrange-
ments could potentially be confused with C. loydelli in poorly
preserved material, such as Heminectere conica (Rigby &
Harris 1979) and H. minima gen. et sp. nov., but these lack
the inner monaxon layer and differ markedly in that they
possessed only a rudimentary attachment, if any, based on a
few pronounced basalia (Rigby & Mehl 1994).
Order LYSSAKIDA Zittel, 1877
Genus DILATISPONGIA gen. nov.
TYPE SPECIES.Dilatispongia tumidus get. et. sp. nov.
DIAGNOSIS. Moderately thin-walled, but with multiple spic-
ules present within wall thickness; arrangement based on or-
thogonal quadruling, but poorly preserved beyond very local
scale. Dermalia clinopentactines, with short distal boss and
further clinopentactines and probably hexactines within wall.
Secondary wall component of meandering tracts of similar
spicules, producing irregular array of rounded regions with
lower spicule density. Tractose bands appear to permeate
entire wall thickness. Spicules of several orders, including
zero.
ETYMOLOGY. Latin dilatare, broad, expanding; referring to
the increased wall thickness without the development of sep-
arate layers.
OCCURRENCE. Known only from the type locality.
REMARKS. This new genus is potentially significant in un-
derstanding the development of thick-walled hexactinellids.
Current ideas essentially follow those of Rigby (1986), who
advocated an origin for lyssakid amphidiscosan hexactinel-
lids through the bilaminar Hintzespongioidea. Those without
a distinct dermal layer, the Malumispongiida (Rigby 1967),
were presumed to have arisen by loss of the outer quadruled
layer, from where the later variations of dermal and gastral

Caradoc sponge fauna from Central Wales 53
Figure 19 Cyathophycus loydelli sp. nov. (A) SM X.39399, paratype, detail showing spicule morphology, ×25; (B) SM X. 39398, holotype, ×9;
(C) SM X.39405, paratype in ventral view, under water, ×6; (D)and(F) SM X.39404, paratype. (D) detail showing mesh arrangement, ×12;
(F)×6; (E) SM X. 39402, paratype, ×8.
layers could have been derived. Dilatispongia shows an ad-
ditional or alternative route to the lyssakid condition, which
is clearly derived directly from a reticulosan precursor, as
shown by the presence of partial quadrules. The presence in
the fauna of three other protospongioid species dominated
by clinopentactine dermalia suggests a significant diversific-
ation of this lineage, or convergence. Given that there is no in-
dication of more than one hexactine-based spicule layer in the
other taxa, it is assumed that the development of specialised
spiculation and reduction of quadruling preceded the increase

54 J. P. Botting
Figure 20 Cut-away reconstruction of Cyathophycus loydelli sp.
nov.; scale bar =5 mm. Only the upper right region shows all spicule
components; the lower part shows just the primary mesh, for clarity.
in wall thickness. Continuation of the loss of regularity, com-
bined with increasing wall thickness, could have led imme-
diately to such lyssakid groups as the Brachiospongiidae and
Malumispongiidae.
Dilatispongia tumidus gen. et sp. nov. (Figs 21, 22)
TYPES. Holotype: SM X.41146. Paratype: SM X.41147;
small, poorly preserved fragment, but showing diagnostic
features.
OTHER MATERIAL. SM X.41148–9; assignment not abso-
lutely certain, but SM X.41149 has associated fragmentary
root tuft.
DIAGNOSIS. As for genus.
ETYMOLOGY. Latin tumidus, swollen, after the spicule mor-
phology.
OCCURRENCE. Locality 2a only.
DESCRIPTION. The holotype shows the approximate form to
be high conical, expanding upwards at a constant angle of 20◦
from the axis, to a maximum observed width of 43 mm and
height of greater than 68 mm (extrapolated to 80 mm). The
base and details of the oscular margin are unknown, although
a complex of disarticulating monaxon ropes from the type
locality (SM X.41149) shows the base of what is probably
this species. An isolated root tuft complex (SM X.41158,
below) may represent the structure under ideal preservation.
Wall thickness is difficult to judge due to compression, but
is greater than a single layer; the compressed thickness is
approximately 0.5 mm, having fractured on several levels
and it probably represents a live wall thickness of at least
2 mm.
Internal and external fracture surfaces are distinguished
by spicule morphology, with boss and reduced distal ray
present externally and lateral rays slightly recurved to-
wards proximal ray, internally. Spicules are dominantly
clinopentactines; no other morphologies definitely present.
Lateral rays typically slightly inflated, with maximum dia-
meter around one-tenth of ray length from base. Rays occur at
irregular angles and opposite rays are frequently non-parallel;
inter-ray angles vary from 30–150◦, although they are usu-
ally close to orthogonal. Rays are smooth and usually almost
straight except for proximal curvature. Spicules are relatively
blunt, with ray aspect ratio of 12–18:1, larger spicules typic-
ally having smaller ratios. The largest spicules have a basal
ray diameter of up to 0.33 mm and ray length of at least 4 mm.
At least four orders of spicules are present, excluding zero-
order, which are occasional and distributed largely diagonal
to the primary grid. There is no pattern to spicule position
within the wall thickness, although the bosses of zero- and
first-order spicules are often dermal, with the recurved rays
projecting into the wall.
There are two components to spicule arrangement. The
entire surface comprises a weak quadrule-based array, of-
ten with roughly alternating orthogonal–diagonal orienta-
tions at each size reduction. This pattern is only clear for
parts of one first-order square, while adjacent regular regions
are typically misaligned. Spicules smaller than third-order
are rarely present in geometric arrangement. Superimposed
on the mesh is a series of meandering tracts, composed of
densely-packed spicules of primarily the same type; some
regions may also contain monaxons, but none have been
confirmed and many are clearly hexactine-based. The tracts
contain the majority of small spicules, as well as many lar-
ger orders, particularly including those with strongly non-
orthogonal rays. Tracts vary between 0.4–1.3mm wide, ex-
cept at junctions, where they can be wider. The tracts curve
and branch repeatedly, often with small side branches that in-
tersect other tracts. The resulting open spaces are 1–3 mm in
diameter, often elongated or otherwise irregular. The tracts
appear to occur on the same levels as the mesh spicules,
throughout the wall thickness; there is no division into two
layers.
REMARKS.D. tumidus is among the most distinctive of Pa-
laeozoic hexactinellids described from Britain; no others
possess the combination of weak grid and tract develop-
ment that mimics parietal gaps. The obvious comparison
is with Hintzespongia bilamina Rigby & Gutschick, 1976,
which also possessed a quadruled grid and tracts/parietal
gaps. However, H. bilamina consisted of two distinct layers,
the spicules were stauracts and perhaps hexacts, rather than
clinopentactines, the quadruled layer was much more regular
and the tracts much less dense. In poorly preserved mater-
ial, the tract development may not be obvious and confusion
with taxa possessing large parietal gaps (e.g. Ratcliffespon-
gia Rigby, 1969 or Stephenospongia Rigby, 1986) is unlikely.
Superficially, Rufuspongia Rigby & Mehl, 1994, may appear
similar, although in this genus there are three distinct layers
producing complex apertures, rather than tracts intermingled
with a grid and the spicules are primarily diactinal. No other
described Palaeozoic sponge has the same structure as D. tu-
midus, although the development of an irregular spicule tract
network is common to many extant lyssacinosidans, albeit
usually without the pseudo-quadruled mesh.

Caradoc sponge fauna from Central Wales 55
Figure 21 Dilatispongia tumidus gen.etsp.nov.(A) SM X.41146, holotype, under water, ×4; (B) SM X.41149, probable D. tumidus,showing
scattered basal tufts, under water, ×6.4.

56 J. P. Botting
Figure 22 Cut-away reconstruction of Dilatispongia tumidus gen. et sp. nov.; scale bar =10 mm in overall view and 1mm in detailed view.
Root tufts
Hexactinellid root tufts are generally treated cursorily in
taxonomic studies, since they lack distinctive features and
are difficult to assign to known taxa. This is probably true for
Recent taxa, which show wide variability and many complex
morphologies, often convergent. However, the early Palaeo-
zoic development of root tufts may be of interest in broad-
scale phylogenetic and ecological studies and they should
be treated in more detail. Of particular importance is their
general conservatism, suggesting that differences may be im-
portant at a high taxonomic level. This does not exclude the
possibility of convergence.
A variety of anchoring structures have been recovered
from Llanfawr (Fig. 23), including some preserved in
association with the body fossils. The asthenospongiids
and Hemidiagoniella show a concentration of pronounced
prostalia in the basal region. None of the 20 specimens of
Heminectere minuta show any sign of basalia, although the
sponge base is sharply pointed and they were probably dir-
ectly embedded in the sediment.
Xylochos palindromica possessed a dense cluster of en-
larged monaxons, while Pirania llanfawrensis shows a radial
array of fine projecting spicules in the basal region. Cyatho-
phycus loydelli, unusually, preserves the body in generally
better condition than the root, but scatterings of monax-
ons are common around sponge margins and the root was
shown, above, to comprise a moderately dense tuft of fine
oxeas. The root tufts of Solusrectus rosetta, Granulispon-
gia obscura and the indeterminate protospongioids are
unknown.
Isolated root tufts fall into three groups, with some vari-
ation within them: 1, (SM X.41151–41155) fairly dense tufts,
as seen in C. loydelli; 2, (SM X.41156–41157; U.C.W. (Uni-
versity College of Wales) 19486) divergent splays of mon-
axons with a common base; 3, (SM X.41158–41160) com-
plex clusters of ropes or splays, which, assuming they are
not clusters of individual root tufts, suggests a broad base.
Several examples of what appear to be individual anchor-
ing spicules have also been observed, although these may be
isolated prostalia of Acutipuerilis spinosus.
The hypothesis that complexity of body morphology
broadly parallels complexity of the root tuft in early hex-
actinellids is consistent with the phylogenetic hypotheses
presented below and some trends relevant to phylogenetic
reconstructions can be established (Fig. 24). The simplest
hexactinellid tufts appear to have been simple modified hex-
actines as seen in various small protospongiid species, with
one ray greatly elongated (e.g. Protospongia tetranema and
P. mononema Dawson & Hinde, 1889). This is clearly seen in
the asthenospongiids, which developed a number of prostalia
in the basal region, generally not in contact. It is possible
that this represents the ancestral state of root tufts, although
other early hexactinellids exist in which no root structures are
known (e.g. H. minuta); in these cases, it may be secondarily
lost, or represent a lineage prior to the evolution of a dis-
tinct root structure. Hemidiagoniella also possessed spicules
of this type. Clusters of basal prostalia were developed in
primitive species of Cyathophycus, these clusters becoming
denser in later species, with the individual spicules becoming
diactinal. The type 2, above, may represent an intermediate
stage in which spicules are still divergent, but with closely
clustered bases. The root tuft of forms allied with Solusrectus
are anticipated to have been of this type.
Superficially, there is a contradiction in Dilatispongia,
which shows a cluster of fairly dense tufts, but lacks the
derived structural features of bilaminar taxa. This is discussed
in the section on phylogeny, below.
Demosponges often lack root structures, but the mon-
axon cluster of the simple hazeliid Xylochos palindromica is
very similar to type 2 above, suggested to be intermediate
stages of the Cyathophycus lineage. The dense tuft of very
fine monaxons in Pirania llanfawrensis may ultimately be
derived from this.

Caradoc sponge fauna from Central Wales 57
Figure 23 Root tufts. (A) U.C.W. 19486, type 2, ×3; (B) SM X.41154, type 1, ×10; (C) SM X.41155, type 1, ×7; (D) SM X.41160, type 3 with fine
tufts, ×5; (E) SM X. 41158, type 3, ×4.5. Photographs D and E were taken under water.
Phylogenetic considerations
Our understanding of early sponge phylogeny is complicated
by their relatively poor record. Although there are now a
large number of described species from the late Precambrian
to the Silurian, most are recorded from very few localities,
implying that the record is still extremely incomplete. This is
unlikely to be due to ubiquitously strict endemism, based on
the moderately wide dispersal of modern taxa. As a result,
there are large gaps in the records of many groups, particu-
larly those from high-energy environments (but see Botting
in press a). Many genera are long-ranging, obscuring strati-
graphic indications of relationship. There is also believed
to be a high level of convergence at all taxonomic levels,
as seen in modern taxa and, in most cases, few characters.
Microscleres, essential in modern taxonomy, are rare in the
Palaeozoic record, almost exclusively isolated and, thus, of
little phylogenetic use despite some dedicated attempts (Reid
1968). However, the record is gradually improving and the
current fauna adds considerably to our knowledge of early
hexactinellid and demosponge evolution.
Particular problems include the relationship of reticu-
losan hexactinellids to early demosponges (Botting in press
b) and the origin of extant hexactinellid groups, such as the
hexactinosans. Some primitive living forms, such as the Ros-
sellidae and Euplectellidae, potentially allow some connec-
tions with extinct groups, but their histories are very uncer-
tain. Even the origins of extinct Palaeozoic clades such as
the Brachiospongiidae are still unclear. The fauna described
here appears to fill some gaps relating to the origins of the
rossellimorphs, brachiospongioids and related groups.
Several structural grades of reticulosan are represented
in the fauna. The simplest is that of Heminectere minima,
involving a single layer of simple stauracts or clinopentact-
ines in regular orthogonal array, with moderately ordered
quadruling and partial first-order ray overlap. This type of
morphology may be regarded as similar to the most primit-
ive hexactinellids, since Mehl (1991, 1996) has argued that

58 J. P. Botting
Figure 24 Proposed schematic evolution of hexactinellid root tufts in early Palaeozoic; generic names are used as examples and do not
indicate direct relationships. This scheme is an outline and many exceptions are expected, particularly in extant taxa. s.s., sensu stricto.
Protospongia itself is derived, while Botting (2003) has
shown that the Protospongia spicule arrangement could have
been derived from a more tractose arrangement by pera-
morphic growth to large size. Although less regular mor-
phologies are present in the Lower Cambrian (e.g. Steiner
et al. 1993), later hexactinellids all appear to derive from
reticulosans and these other morphologies must be regarded
as either early representatives of derived clades, or relatives
of extinct stem group taxa from which the protospongioids
may have evolved. This view may require modification, but
it will be employed as a null hypothesis for the purposes of
this paper. Hemidiagoniella tenax and H. caseus each pos-
sessed a single layer of acanthohexactines, with very short
distal rays that therefore approximated to clinopentactines
and each retains a locally ordered, dominantly orthogonal ar-
ray. Although it is impossible to state with certainty that they
shared a near ancestor with Heminectere, the few characters
available support a fairly close relationship.
The development of a second spicule layer is critical
in early hexactinellid phylogeny and several routes appear
to have led to a bilaminar structure. Mattaspongia apaches
(Rigby 1970) developed a dermal monaxon layer, while
the otherwise similar Cyathophycus–dictyosponge lineage
developed a parallel gastral layer and Hintzespongia,Di-
erespongia (Rigby & Gutschick 1976) and allies evolved a
second hexactine-based layer. None of the Llanfawr species
developed two distinct, hexactine-based layers, but Solus-
rectus rosetta and Cyathophycus loydelli possessed a gastral
monaxon layer parallel to the external mesh. Of these, Solus-
rectus shows a simpler gastral layer, with sparse distribu-
tion of monaxons, typically in vertical pairs, while Cyatho-
phycus possessed a dense cross-hatched layer. The dermal
layer of Cyathophycus was also further from the loose, part-
quadruled array of Heminectere, instead showing strong tract
development at the first order.
This suggests that Cyathophycus represents a more de-
rived part of the reticulosan lineage. However, the dermal
spicules of Cyathophycus are simple hexactines, in contrast
to the derived morphologies of Heminectere,Hemidiagoni-
ella and Solusrectus. This requires either convergent acquisi-
tion of clinopentactines, or redevelopment of the distal ray in
Cyathophycus. However, the acanthose ornament of Hemidi-
agoniella and Solusrectus must also have been independently
derived, unless the monaxial layer was convergent between
Solusrectus and Cyathophycus, or secondarily lost in Hemidi-
agoniella. Both the latter hypotheses appear unlikely. A pre-
liminary hypothesis is that spicule ornamentation and ray
suppression is genetically simpler than the production of a
new spicule layer and more likely to be convergently derived.
This requires that Solusrectus is a sister group of Cyathophy-
cus and later forms, with Hemidiagoniella being more dis-
tant. In this scenario, it seems likely that the clinopentactinal
spicules of Heminectere represent the ancestral morphology
in this lineage. The subsequent modification to ornamented
clinopentactines and hexactines appears to have been a re-
current phenomenon and may have been a genetic tendency
within the lineage.
The rosette ornament of the clinopentactines in Solus-
rectus is similar to some specimens described by Webby
& Trotter (1993) and Acanthocoryna Finks (1960), who re-
ferred them to the brachiospongiids. The Brachiospongiidae
are characterised by a loose, irregular skeleton of hexactine-
based spicules, comprising a thick wall, with a morpholo-
gically distinct spiculation in the dermal layer. This con-
trasts with the Malumispongiidae (Rigby 1967), in which
the dermal layer is not distinct. The Brachiospongiidae are
known primarily from North America, particularly the Or-
dovician and Silurian. Precursors are uncertain; Rigby (1986)
suggested that they derived from the Hintzespongiidae by
loss of the quadruled layer, but with no direct evidence. An

Caradoc sponge fauna from Central Wales 59
alternative scenario is that they evolved via aDilatispongia-
type form, from the main protospongioid–dictyospongioid
lineage; in this case there is already some thickening of a
homogeneous spicule layer. The production of a thick wall
represents a novel development in growth pattern, which
potentially allowed fused skeletons and radial addition to
be utilised (Botting 2003), and is not a trivial transition.
Therefore, in the absence of evidence for thickened walls in
the Hintzespongioidea, derivation from a Dilatispongia-type
form is preferred.
The presence of a cluster of root tufts in Dilatispon-
gia tumidus is, at first sight, inconsistent with the scheme,
because a single dense tuft of this type appears only with
Cyathophycus. This must be convergent and represents one
of two developmental paths from the asthenospongiid con-
dition: (i) concentration of basalia into a single tuft and
(ii) development of several tufts corresponding to the po-
sitions of individual basalia. Later hexactinellids appear to
have largely retained this distinction, with the amphiscophor-
idan family Hyalonematidae possessing a single, sometimes
hypertrophied tuft (Tabachnik & Menshenina 2002), and the
Monoraphidae possessing only a single, hypertrophied basal
spicule (Tabachnik 2002d); the Pheronematidae generally
have a single, but more diffuse tuft. Among the hexastero-
phoridan Hexactinosa there is more variation, including a
variety of cementing attachments to hard substrates, or ri-
gid stalks. The Lyssacinosida (Euplectellidae, Leucopsacidae
and Rossellidae) generally possess either isolated basalia or
several tufts. The Euplectellidae, in particular, possess basalia
that are modified hexactines or their derivatives (Tabachnik
2002a), as seen in Hemidiagoniella and the asthenospon-
giids. The Leucopsacidae possess a fused basal disc or stalk
(Tabachnik 2002b), while the Rossellidae include a variety of
forms, including modified hexactine basalia (e.g. Anoxycalyx
(A.) ijimai; Tabachnik 2002c: 1455), and clusters of tufts (e.g.
Anoxycalyx (Scolymastra)joubini; Tabachnik 2002c: 1457).
The non-lyssacinosidan Hexasterophora also tends towards
more plastic morphology than either of the other groups.
Taken straightforwardly, this discussion suggests that
the Amphidiscophora were derived from the later stages of
the Cyathophycus–dictyosponge lineage, while the Lyssa-
cinosida evolved from a more primitive stage, perhaps close
to Hemidiagoniella and Solusrectus. Assuming monophyly
of the hexasterophoridans, this implies that the Hexactin-
osa are a later development from this lineage. However, the
development of coloniality in those with single tufts would
be expected to result in multiple tufts and concentration of
several tufts into one is also near-inevitable in taxa with nar-
row morphologies. Therefore, the number and nature of tufts
cannot be taken as an infallible indicator of relationship, par-
ticularly in later forms, but is nonetheless an interesting and
potentially useful character in primitive hexactinellids.
This scenario is consistent with structural considera-
tions regarding rossellimorph ancestry. The term refers to
Recent hexactinellids, particularly lyssacinosidans, that are
thin-walled, typically with pronounced dermal clinopentact-
ines or stauracts. The earliest examples ascribed to the
Rossellimorpha are the Devonian Rufuspongiidae (Rigby
& Mehl 1994), but here there are three distinct layers of
hexactine-derived spicules. Although they lack the quadrul-
ing in one layer of the Hintzespongiidae (Rigby & Gutschick
1976), it is possible that they were derived from this lineage,
or represent a convergent development.
The Rossellidae sensu stricto are defined on hypodermal
pentactines that include the development of prostalia and
basalia, small dermalia and a chonaosomal skeleton chiefly
of monaxons (Tabachnik 2002c). They are similar in many
ways to the Euplectellidae, except for the rarity of hexact-
ines or pentactines in the latter (Tabachnik 2002a), and to
the Leucopsacidae (Tabachnik 2002b). These comprise the
modern understanding of the extant Lyssacinosida and all
are united by the loose arrangement of spicules, which are
typically acanthose or otherwise modified. These features
are also found in Solusrectus rosetta and Hemidiagoniella
spp., which possessed a thin wall of ornamented (acanthose)
pentactines, some of which were extended into basalia, as dis-
cussed above, smaller size orders of pentactines and hexact-
ines which could be homologous with the rosselid dermalia
and, in the case of S. rosetta, an inner layer of relatively
loosely arranged diactines, possibly representing the ances-
tral choanosomal skeleton. Although the geometric arrange-
ment of the spicules is more striking in these Ordovician
forms, the critical features are all present and these taxa are
suggested as representing stem group Rossellidae. The other
extant Lyssakida would probably have been derived from this
group, since they do not share the dominance of clinopentact-
ines.
The presence or absence of zero-order spicules (sensu
Botting 2003) also appears to be a useful, although previously
overlooked, feature. The poor distinctions between size or-
ders prevent this being useful in modern hexactinellids, but
they are normally clearly discernable, if present, in Lower
Palaeozoic reticulosans. Zero-order spicules are most prom-
inent in the Asthenospongiidae, but also occur in all other
members of the Solusrectus–Cyathophycus lineage, as well
as Dilatispongia. However, they are certainly not present
in the simplest protospongioids, such as Heminectere and
perhaps not in Granulispongia, although this is uncertain –
first-order spicules appear slightly hypertrophied, but the reg-
ularity of their arrangement is unknown.
Among other described taxa, zero-order spicules are ab-
sent from Protospongia hicksi (e.g. Rigby 1966), Diagoniella
spp. and Hintzespongia bilamina Rigby & Gutschick, 1976,
but present in Protospongia spina Mehl et al., 1993 and in the
structurally complex, bilaminar Dierespongia palla Rigby &
Gutschick, 1976. They also appear to be absent from the
Lower Cambrian Sansha sponges (Steiner et al. 1993), al-
though there is too little regularity in the skeletal mesh to
allow certain recognition, and from later dictyosponge-like
forms such as Ammonella (Keupp & Mehl 1995). The tri-
laminar Rufuspongia Rigby & Mehl, 1994, which has been
described as the earliest known rossellid (Mehl-Janussen
1999), is composed dominantly of diactines and the attach-
ment structure is unknown. There do not appear to be zero-
order spicules, but these could be disguised by the complex,
irregular arrangement and reduced spicule rays. I am uncer-
tain whether this is a genuine rossellid, or a member of an
independent lineage, perhaps related to Hintzespongia.
Although the position of many groups is still largely
obscure, comments can be made on the status of some. The
coarse network of dermal and gastral spicules in the pel-
icaspongiid Larispongia magdalenae Carrera,1998, is struc-
turally very similar to the dense regions of the mesh of D. tu-
midus and the group is probably a closely related lineage. The
presence of probable zero-order spicules, especially in the
basal region, is consistent with derivation from a group with

60 J. P. Botting
Rufuspongia
Granulispongia
Cyathophycus
Dilatispongia
Hintzespongia
Asthenospongia
Xylochos
Hemidiagoniella
Protospongia
Acutipuerilis
Demospongea
Rossellidae
Heminectere
Brachiospongiidae
Dictyospongia
Solusrectus
1
2
3
4
5
6
7
??
?
?
?
Figure 25 Proposed phylogeny of early hexactinellids, based on the
present fauna. Listed characters represent critical structural
innovations; 1, second hexactine-based layer; 2, zero-order spicules;
3, hypertrophication of zero-order spicules as prostalia; 4, thickening
of single-layer wall; 5, inner monaxon layer; 6, discrete root tuft
developed; 7, loss of outer hexactine-based layer. For full discussion
of the occurrence of these features, see the text.
enlarged, separate basalia. Wongaspongia Rigby & Webby,
1988, also shows evidence of enlarged and irregularly ori-
entated spicules within the endosomal skeleton, but has a
bilaminar wall, both layers hexactinal. The structure of the
inner wall is again similar to D. tumidus, with a combina-
tion of sinuous tracts and very weak reticulation, suggesting
that this lineage diversified significantly during the Ordovi-
cian, including the development of additional spicule layers.
There is little similarity of these layers to the bilaminar wall
of Hintzespongia Rigby & Gutschick, 1976, which lacks
zero-order spicules, implying that multiple hexactinal layers
have been convergently acquired.
Although the relationships suggested here (Fig. 25, dis-
cussed below) are only an outline hypothesis and will doubt-
less be subject to major revision, they are supported by appar-
ently consistent structural features and are based on import-
ant new material. They differ from previous interpretations
in attributing significance to attachment structure and pres-
ence of zero-order spicules as much as to traditional char-
acters, such as spicule ray overlap, which may be unreliable
(Botting 2003). A bilaminar wall appears to have evolved
several times in different forms, and the evolutionary em-
phasis has been shifted away from the Hintzespongiidae to
the Cyathophycus lineage, which is shown to be much more
diverse than previously recognised.
According to this scheme, the ancestral hexactinellid
would have possessed simple hexactine-based spicules, pos-
sibly pentactines, in multiple but similar size orders, and
would have lacked a discrete attachment structure. The
skeletal arrangement would have been regular and unilam-
inar and either tractose or quadruled, or some combination
of the two. Several of the small protospongiid species, such
as H. conica, approach this condition, although none fulfil
it entirely. Major emergent clades from this group are the
Diagoniella lineage (possibly a sister group to other hexact-
inellids), the Hintzespongiidae and the Dictyospongioidea.
The latter was derived through intermediary stages that ap-
parently gave rise to the Rossellidae and, perhaps, the Bra-
chiospongioidea and thus includes the common ancestor of
extant hexactinellids.
There are significant differences between this phylo-
genetic interpretation and those published previously. Reid
(1958) argued for a fundamental separation of Amphidisco-
phora and Hexasterophora, but this requires extensive ghost
ranges and has little direct evidence beyond the modern mu-
tual exclusion of microsclere types. He erected the Reticulosa
as an independent lineage perhaps derived from the Hexas-
terophora during the Precambrian and including the majority
of Palaeozoic hexactinellids.
Rigby (1986) proposed the Hintzespongiidae as be-
ing close to the ancestral state for thick-walled taxa such
as the Brachiospongiidae and regarded the Cyathophycus–
dictyosponge lineage as an evolutionary dead end. This has
been reversed in the present scheme, on the basis that a bil-
aminar wall of two thin layers is a separate development to a
single thick wall. This view is augmented by the similarity of
Dilatispongia gen. nov. to the thinner-walled brachiospon-
gioids and the absence of zero-order spicules in Hintzespon-
gia.
The more detailed hypothesis of Mehl (1996), modi-
fied by Mehl-Janussen (1999), is also contradicted in sev-
eral points. Most critically, the brachiospongiids are there
regarded as stem group amphidiscophorans and the dic-
tyospongioids as hexasterophorans, but the reverse is pro-
posed herein. Mehl-Janussen derived the ‘Rossellimorpha’
from the Brachiospongiidae, via the Malumispongiidae, but
distinguished this group from the extant Rossellidae and
allies, which were regarded as Mesozoic derivatives of
the Hexactinosans. The split between the amphidiscophor-
ans and hexasterophorans was interpreted as a Precambrian
event, with the Sansha fauna (Steiner et al. 1993) yielding our
closest approximations to the ancestral state of hexactinel-
lids. Krautter (2003) essentially followed Mehl’s scenario,
advocating a Precambrian separation of the extant subclasses
and regarding the taxon Reticulosa as polyphyletic, due to
the inclusion of hexaster- and amphidisc-bearing forms.
Mehl’s (1996) scheme emphasises stratigraphic occur-
rence and the occurrence of oxyhexasters in Griphodictya
epiphanies Hall & Clark, 1898. In contrast, stratigraphic oc-
currence is not regarded as critical herein, due to the incom-
pleteness of the record and relatively conservative morpholo-
gies through long intervals. The presence of oxyhexasters is
regarded as of ambiguous significance. Kling & Reif (1969)
described the co-occurrence of paraclavules (closely related
to amphidiscs) and microhexactines with terminal branches
from their Itararella gracilis, with a hexaster that they re-
garded as foreign. It is not, however, clear that these mi-
crohexactines differ significantly from their simple hexaster
and co-occurrence of the two microsclere morphotypes is
a less contrived interpretation. Since true amphidiscs are
also known from the Carboniferous and microhexactines
almost identical to the hexaster of Kling & Reif (1969),
occur in some modern amphidiscophorans (e.g. Hyalonema

Caradoc sponge fauna from Central Wales 61
(Cycliconema)apertum apertum; Tabachnik & Menshenina,
2003), the mutual exclusion of microscleres taxa appears not
to be absolute. Given the rarity of early fossil microscleres,
it is likely that some reticulosan groups possessed both types
and the subclass division of modern hexactinellids originated
during the Palaeozoic; this was also the conclusion of Finks
(1970) and Bergquist (1978). The phylogeny proposed here
emphasises instead the wall structure, details of spicule or-
dering, spicule specialisation and root tufts, based largely on
information not previously available.
Palaeoecology
The Llanfawr fauna is unusual in preserving several associ-
ations within a short sequence, showing variations in com-
munity structure. By far the most abundant and ubiquitous
species is Cyathophycus loydelli, with the associated Xy-
lochos palindromica also widespread. Most other taxa are
rare and usually occur at only one locality.
Three associations can be recognised:
1. Near-monospecific assemblage of Cyathophycus loydelli,
with rare specimens of other taxa, including Xylochos
palindromica,Pirania llanfawrensis and Granulispongia
obscura. Locality 1; associated fauna is diverse, trilobite-
diplograptid dominated. Perhaps locality 3, but fauna is
too limited for discussion.
2. Near-bispecific assemblage of Cyathophycus loydelli and
Xylochos palindromica, with rare additional hexactinel-
lids. Locality 2b; associated fauna of molluscs, dendroid
and uniserial graptolites and chitinozoans, plus rare soft
tissue preservation of possible arthropod and problemat-
ica. Locality 2c; associated fauna of trilobites, brachio-
pods and diplograptids, with rare other taxa.
3. Diverse assemblage of almost all taxa, with Hemidiag-
oniella tenax and Heminectere minima most abundant;
includes a sub-assemblage dominated by H. tenax and
Xylochos palindromica. Locality 2a; diverse associated
fauna of trilobites, graptolites, brachiopods, conulariids
and others.
It is difficult to recognise the environmental differences
that may have led to these ecological variations. The only
strikingly different associated fauna is that of locality 2b,
where a dark mudstone yields abundant chitinozoan masses
and frequent dendroids and molluscs, but in which trilobites
are very rare. This is also the most unusual sponge preser-
vation in the quarries, with limited pyritisation and showing
a dark stain over the region of soft tissue. This suggests a
less reducing (sulphidic) environment, but also exclusion of
bioturbators and aerobic bacteria. The unusual, diverse fauna
is more characteristic of quiet shallow marine settings (e.g.
Loduca & Brett 1997) and may represent fluctuating phys-
ical or chemical conditions on tidal or seasonal timescales. It
is, however, unclear which parameters could have been fluc-
tuating in a relatively deep-water setting, unless it was the
influence of seasonal current variations within the restric-
ted Welsh Basin. The entire 10 cm bed is sedimentologically
homogeneous, but interpretation as a single event bed is ex-
cluded by the internal planes showing chitinozoan masses.
The interpretation of this horizon remains enigmatic.
It is clear, however, that sponges were not strongly
affected by these changes, since the same assemblage oc-
curs here as in locality 2c, which represents a normal shelf
fauna. The main structural difference between Cyathophy-
cus loydelli and Xylochos palindromica and the remainder
of the fauna, is the integrity of their skeletons and resistance
to disarticulation. Locality 1, in particular, records cross-
bedding in spongiferous horizons, indicating rapid sediment-
ation above normal levels. There is no obvious obrution de-
posit at locality 2a, except perhaps for the Hemidiagoniella
tenax–dominated bed, a species which also retained unusual
skeletal cohesion. Throughout the rest of locality 2a, most
sponges are at least partly disarticulated and spiculites occur
at some points. However, the taphonomy and non-poriferan
fauna is almost identical to that of locality 1, indicating that
chemical conditions were similar. The only apparent dif-
ference between the sites is turbulence and this may there-
fore have been the main factor dictating sponge distribution
within the sequence. It is possible that the apparent rarity
of other taxa in locality 1 is due to their complete disar-
ticulation, but this seems unlikely because of the obrution
horizons. The high abundance of C. loydelli, with crowded
bedding planes, is also consistent with high ecological
dominance.
Bodzioch (1994) discussed attachment structures in re-
lation to environmental turbulence in Cretaceous hexact-
inellids, concluding that fused basophytous structures were
found in sponges occupying the most turbulent conditions,
followed by basal tufts and stalks. A similar dependence of
distribution on attachment structure was reported by Mehl
et al. (1993), with discussion of modern hexactinellid distri-
butions in relation to topography-enhanced currents. Fused
basal structures had not developed in this group by the Middle
Ordovician and in the Llanfawr faunas, strength of the body
wall appears to have been a more critical parameter. The
development of meandering tracts in the large Dilatispon-
gia tumidus also suggests a strength benefit, since parietal
gaps were not fully developed. Occupation of more turbu-
lent environmental niches could therefore represent the main
impetus for the convergent development of thick, rigid or
otherwise strengthened walls in essentially every group of
sponge during the Lower Palaeozoic. This implies a broad
onshore migration during sponge evolution, in contrast to
the classical onshore–offshore trend characteristic of most
metazoan groups during the Phanerozoic.
Acknowledgements
This work was funded by a Junior Research Fellowship at
Christ’s College, Cambridge and was initiated under a PhD
studentship at the University of Birmingham. Thanks to Lucy
Muir for helpful discussions, proof-reading and assistance
with fieldwork, and to Reg Knill and John Davies for access
to the site. Reviews by Marcelo Carrera and Manfred Krautter
are appreciated.
References
Bergquist, P. 1978. Sponges. Hutchinson, London, 268 pp, 12 pl.
Bett, B. J. & Rice, A. L. 1992. The influence of hexactinellid sponge
(Pheronema carpenteri) spicules on the patchy distribution of mac-
robenthos in the Porcupine Seabight (bathyal NE Atlantic). Ophelia
36: 217–226.

62 J. P. Botting
Bodzioch, A. 1994. Palaeoecology of hexactinellid sponges from the epi-
continental Triassic of Poland. Pp. 35–44 in vanSoest,R.M.W.,
van Kempen, T. M. G. & Braekman, J. (eds) Sponges in Time and
Space. Balkema, Rotterdam.
Botting, J. P. 2000. Palaeoecology and systematics of Ordovician biotas
from Welsh volcaniclastic deposits. Unpublished Ph.D. Thesis: Uni-
versity of Birmingham, 369 pp.
— 2003. Growth patterns of Lower Palaeozoicsponges. Lethaia 43: 41–52.
—inpressa. Exceptionally-preserved Middle Ordovician sponges from
the Llandegley Rocks l¨
agerstatte, central Wales. Palaeontology.
—inpressb. Cyathophycus and the origin of demosponges. Lethaia.
Carrera, M. G. 1998. First Ordovician sponge from the Puna region,
northwestern Argentina. Ameghiniana 35: 205–210.
Chen, J.-y., Hou, X.-g. & Lu, H.-z. 1989. Lower Cambrian leptomitids
(Demospongea), Chengjiang, Yunnan. Acta Palaeontologica Sinica
28: 17–31.
—, — & Li, G.-x. 1990. New Lower Cambrian demosponges –
Quadrolaminiella gen. nov. from Chengjiang, Yunnan. Acta Palae-
onologica Sinica 29: 402–413.
Davies, J. R., Fletcher, C. J. N., Waters, R. N., Wilson,D., Woodhall,
D. G. & Zalasiewicz, J. A. 1997. Geology of the Country around
Llanilar and Rhayader. British Geological Survey, The Stationery
Office, London, 267 pp.
Dawson, J. W. & Hinde, G. J. 1888. New species of fossil sponges from
Little Metis, Province of Quebec, Canada. Peter Redpath Museum,Mc-
Gill University,Montreal. Notes on specimens – April, 1888. Dawson
Brothers, Montreal.
— & — 1889. On new species of fossil sponges from the Siluro-Cambrian
at Little M´
etis on the lower St. Lawrence. Transactions of the Royal
Society of Canada 7: 31–55.
De Laubenfels, M. W. 1955. Porifera. Pp. E21–212 in Moore, R. C.
(ed.) Treatise on Invertebrate Palaeontology. Geological Society of
America and University of Kansas Press, Lawrence, Kansas.
Evans, D. H. 1994. Siphonal and cameral deposits in Allumetoceras and
the position of the Tripteroceratidae (Nautiloidea). Palaeontologische
Zeitschrift 68: 321–337.
Finks, R. M. 1960. Late Paleozoic sponge faunas of the Texas region; the
siliceous sponges. American Museum of Natural History Bulletin 120:
1–160, 50 pl.
— 1970. The evolution and ecologic history of the sponges during Pa-
laeozoic times. Symposium of the Zoological Society of London 25:
3–22.
— 1983. Fossil Hexactinellida. Pp. 101–115 in Broadhead, T. W. (ed.)
Sponges and Spongiomorphs; Notes for a Short Course. Studies in
Geology, vol. 7. Palaeontological Society, University of Tennessee
Department of Geological Sciences, Knoxville, Tennessee.
Fortey, R. A. & Owens, R. M. 1978. Early Ordovician (Arenig) strati-
graphy and faunas of the Carmarthen district, south-west Wales.
Bulletin of the British Museum Natural History (Geology) 30:
225–294.
Furnes, H. 1978. A comparative study of Caledonian volcanics in Wales
and West Norway. Unpublished PhD thesis: University of Oxford,
368 pp.
Hall, J. 1884. Descriptions of the species of fossil reticulate sponges,
constituting the family Dictyospongiidae. New York State Museum
Reports 35: 59–481.
—& Clarke, J. M. 1898. A memoir on the Palaeozoic reticulate sponges
constituting the family Dictyospongidae. New York State Museum
Memoir 2: 1–197.
Hinde, G. J. 1883. Catalogue of the Fossil Sponges in the Geological
Department of the British Museum (Natural History). Taylor& Francis,
London, viii + 248 pp, 38 pl.
— 1887. Monograph of British Fossil Sponges. Monograph of the Palae-
ontographical Society, London 1: 1–92, pls 1–8.
— 1888. Monograph of British Fossil Sponges. Monograph of the Palae-
ontographical Society,London 2: 93–188, pl. 9.
Huff, W. D.,Merriman, R. J.,Morgan,D.J.&Roberts,B.1993.
Distribution and tectonic setting of Ordovician K-bentonites in the
United Kingdom. Geological Magazine 130: 93–100.
Hughes,C.P.1969. Ordovician trilobite faunas from central Wales. Bul-
letin of the British Museum of Natural History (Geology) vol. 1 18:
39–103.
Jones, O. T. & Pugh, W. J. 1948. The form and distribution of doler-
ite masses in the Builth–Llandrindod Inlier, Radnorshire. Quarterly
Journal of the Geological Society 104: 71–98.
Keupp, H. & Mehl, D. 1995. Ammonella quadrata Walther 1904 (Por-
ifera, Hexactinellida) of the Solnhofen plattenkalks: systematics and
environmental deductions. Extended Abstracts of the Second Interna-
tional Symposium on Lithographic Limestones,Ediciones de la Uni-
versidad Aut´
onoma de Madrid 93–94. (Abstract)
Kling, S. A. & Reif, W.-E. 1969. The Palaeozoic history of amphidisc and
hemidisc sponges: new evidence from the Carboniferous of Uruguay.
Journal of Paleontology 43: 1429–1434, 1 pl.
Krautter, M. 2002. Fossil Hexactinellida: an overview. Pp. 1211–1223
in Hooper, J. N. A. & van Soest, R. M. W. (eds) Systema Porifera.
Kluwer Academic/Plenum, New York.
Land, L. S. 1976. Early dissolution of sponge spicules from reef sed-
iments, north Jamaica. Journal of Sedimentary Petrology 46: 967–
969.
Lewis, H. P. 1940. The microfossils of the Upper Caradocian phosphate
deposits of Montgomeryshire, North Wales. Annals and Magazine of
Natural History series 11 5: 1–39.
Loduca, S. T. & Brett, C. E. 1997. The Medusaegraptus epibole and
lower Ludlovian Konservat-Lagerst¨
atten of eastern North America.
Pp 369–406 in Brett, C. E. & Baird, G. C. (eds) Paleontological Events;
Stratigraphic, Ecological and Evolutionary Implications. Columbia
University Press, New York.
M’Coy, F. 1852.A Synopsis of the Classification of British Palaeozoic
Rocks, by Adam Sedgwick, part 2. London, pp. i–viii + 185–406.
Mehl, D. 1991. Are Protospongiidae the stem group of modern Hexact-
inellida? Pp. 43–53 in Reitner, J. & Keupp, H. (eds) Fossil and Recent
Sponges. Springer-Verlag, Berlin.
Mehl, D. 1996. Phylogenie und Evolutions¨
okologie der Hexactinell-
ida (Porifera) im Palaeozoikum. Geologische Palaeontologische Mit-
teilungen Innsbruck 4: 1–55.
—, Rigby, J. K. & Holmes, S. B. 1993. Hexactinellid sponges from
the Silurian-Devonian Roberts Mountains Formation in Nevada and
hypotheses of hexactine-stauractine origin. Brigham Young University
Geology Studies 39: 101–124.
Mehl-Janussen, D. 1999. Die fr¨
uhe Evolution der Porifera. M¨
unchner
Geowissenschaftliche Abhandlungen, Reihe A 37: 1–72.
Owens, R. M.,Fortey, R. A.,Cope, J. C. W.,Rushton, A. W. A. &
Bassett, M. G. 1982. Tremadoc faunas from the Carmarthen district,
South Wales. Geological Magazine 119: 1–112.
Pulfrey, W. 1933. Note on the occurrence of sponge spicules associated
with the iron-ores of North Wales. Annals and Magazine of Natural
History series 10 11: 67–76, 1 pl.
Raiswell, C., Bartels, C. & Briggs, D. E. G. 2001. 3.4.6 Hunsr¨
uck Slate.
Pp. 346–348 in Briggs,D.E.G.&Crowther,P.R.(eds)Palaeobiology
II. Blackwell, Oxford.
Reid,R.E.H.1958. The Upper Cretaceous Hexactinellida. Mono-
graph of the Palaeontographical Society,London part 1, (1958.) 111,
i–xlvi.
— 1967. Hyalostelia Zittel and the Heteractinellida. Geological Magazine
104: 389–399.
— 1968. Microscleres in demosponge classification. University of Kansas
Paleontological Contributions 35: 1–37.
Rigby, J. K. 1966. Protospongia hicksi Hinde from the Middle Cambrian
of western Utah. Journal of Paleontology 40: 549–554.
— 1967. Two new early Paleozoic sponges and the sponge-like organ-
ism, Gaspespongia basalis Parks, from the Gasp´
e Peninsula, Quebec.
Journal of Paleontology 41: 766–775.
— 1969. A new Middle Cambrian hexactinellid sponge of western Utah.
Journal of Paleontology 43: 125–128.
— 1970. Two new Upper Devonian hexactinellid sponges from Alberta.
Journal of Paleontology 44: 7–16.
— 1971. Sponges of the Ordovician Cat Head Member, Lake Winnipeg,
Manitoba. Geological Society of Canada Bulletin 202: 35–79.

Caradoc sponge fauna from Central Wales 63
— 1978. Porifera of the Middle Cambrian Wheeler Shale, from the
Wheeler Amphitheater, House Range, in western Utah. Journal of
Paleontology 52: 1325–1345.
— 1983. Sponges of the Middle Cambrian Marjum Limestone from the
House Range and Drum Mountains of western Millard County, Utah.
Journal of Paleontology 57: 240–270.
— 1986. Sponges of the Burgess Shale (Middle Cambrian) British
Columbia. Palaeontographica Canadiana Monograph 2: 1–105.
— 1995. The hexactinellid sponge Cyathophycus from the Lower-Middle
Ordovician Vinini Formation of central Nevada. Journal of Paleonto-
logy 69: 409–416.
—&Chatterton,B.D.E.1994. New Middle Silurian hexactinel-
lid sponges from the McKenzie Mountains, Northwest Territories,
Canada. Journal of Paleontology 68: 218–223.
—& Gutschick, R. C. 1976. Two new Palaeozoic hexactinellid sponges
from Utah and Oklahoma. Journal of Paleontology 50: 79–85.
—& Harris, D. R. 1979. A new Silurian sponge fauna from Northern
British Columbia, Canada. Journal of Paleontology 53: 968–980.
—& Mehl, D. 1994. Middle Devoniansponges from the northern Simpson
Park Range, Nevada. Brigham Young University Geology Studies 40:
111–153.
—&Webby,B.D.1988. Late Ordovician sponges from the Malongulli
Formation of central New South Wales. Palaeontographica Americana
56: 1–147.
—, Hannum, C. & Frest, T. J. 1979. Hexactinellid sponges from the
Silurian Waldron Shale in southeastern Indiana. Journal of Paleonto-
logy 53: 542–549.
—, King, J. E. & Gunther, L. F. 1981. The new Lower Ordovician
protosponge, Asthenospongia, from the Phi Kappa Formation in central
Idaho. Journal of Paleontology 55: 842–847.
Salter, J. W. 1861. Sheet 32; Scotland.Memoirs of the Geological Survey
of Great Britain. Her Majesty’s Stationery Office, London.
Schmidt, O. 1870. Grundz¨
uge einer Spongien-Fauna des Atlantischen
Gebietes. Wilheim Engelmann, Leipzig, 88 pp.
Sheldon, P. R. 1987. Trilobite evolution and faunal distribution in some
Ordovician rocks of the Builth Inlier, Central Wales. Unpublished
Ph.D. Thesis: University of Cambridge, 256 pp.
Sollas, W. J. 1885. A classification of the sponges. Annals and Magazine
of Natural History series 5, 16: 395.
Steiner, M.,Mehl, D.,Reitner, J. & Erdtmann, B.-D. 1993. Oldest en-
tirely preserved sponges and other fossils from the Lowermost Cam-
brian and a new facies reconstruction of the Yangtze platform (China).
Berliner geowissenschaften Abhandlungen 9: 293–329.
Stridsberg, S. 1990. Internal shell destruction in some Silurian nautiloids.
Pal ¨
aontologische Zeitschrift 64: 213–220.
Tabachnik, K. R. 2002a. Family Euplectellidae Gray, 1867. Pp. 1388–
1434 in Hooper, N. A. & van Soest, R. M. W. (eds) Systema Porifera.
Kluwer Academic/Plenum, New York.
— 2002b. Family Leucopsacidae Ijima, 1903. Pp. 1435–1440 in Hooper,
N. A. & van Soest, R. M. W. (eds) Systema Porifera.Kluwer
Academic/Plenum, New York.
— 2002c. Family Rossellidae Schulze, 1885. Pp. 1441–1505 in Hooper,
N. A. & van Soest, R. M. W. (eds) Systema Porifera.Kluwer
Academic/Plenum, New York.
— 2002d. Family Monorhaphididae Ijima, 1927. Pp. 1264–1266 in
Hooper, N. A. & van Soest, R. M. W. (eds) Systema Porifera.Kluwer
Academic/Plenum, New York.
—& Menshenina L. L. 2002. Family Hyalonematidae. Pp. 1232–1263 in
Hooper, N. A. & van Soest, R. M. W. (eds) Systema Porifera.Kluwer
Academic/Plenum, New York.
Walcott, C. D. 1879. Fossils of the Utica Slate. Transactions of the Albany
Institute 10: 18–35.
— 1881. On the nature of Cyathophycus.American Journal of Science 22:
394–395.
— 1920. Cambrian geology and paleontology. IV. Middle Cam-
brian Spongiae. Smithsonian Miscellaneous Collections 67: 261–
364.
Watkins, R. & Coorough, P. J. 1997. Silurian sponge spicules from
the Racine Formation, Wisconsin. Journal of Paleontology 71:
208–214.
Webby, B. D. & Trotter, J. 1993. Ordovician sponge spicules from New
South Wales, Australia. Journal of Paleontology 61: 28–41.
Woodland, A. W. 1939. On an occurrence of sponge spicules in a deposit
of Arenig age in the Rhiw District (Caernarvonshire). Proceedings of
the Geologists’ Association 7: 19–21.
Zittel, Karl, A. 1877. Beitr¨
age zur systematik der fossilen Spongien.
Repertorium zum Neuen Jahrbuch und Zentralblatt f¨
ur Mineralogie,
Geologie und Palaeontologie 1–40, 4 pls.