INTRODUCTION
The genus Cymbastela Hooper & Bergquist, Reference Hooper and Bergquist1992 is represented by cup-shaped, thinly lamellate to flattened sponges, with a choanosomal skeleton comprising exclusively small to medium-sized oxeas, in a relatively distinct axial and extra-axial skeleton that lacks a tangential ectosomal skeleton (Alvarez & Hooper, Reference Alvarez, Hooper, Hooper and Van Soest2002). This definition was emended to include several species previously assigned to Cymbastela but allocated to a new genus Pipestela, erected to accommodate sponges bearing mainly oxeas as in Cymbastela, but lacking the characteristic cup-shape, and with distinctive loose skeletons of vaguely plumose and wavy spicule-fibres dissimilar to the highly spiculose and tightly meshed spicule tracts found in Cymbastela (Alvarez et al., Reference Alvarez, Hooper and van Soest2008). Cymbastela is a relatively small genus, with six species described from shallow tropical and subtropical waters of Australia and New Caledonia (Hooper & Bergquist, Reference Hooper and Bergquist1992; Van Soest et al., Reference van Soest, Desqueyroux-Faúndez, Wright and König1996; Alvarez & Hooper, Reference Alvarez, Hooper, Hooper and Van Soest2002) (Table 1). However, known and undescribed species have also been recorded by Michelle Kelly and others (Alvarez personal communication; Lori J. Bell, personal communication) from southern Papua New Guinea (Eastern Fields, Louisiade Archipelago), Vanuatu, Fiji, Palau, Chuuk Atoll, and Cebu, Philippines, extending the distribution to the west central Pacific and into South-East Asia.
Table 1. Comparative data on the spicular content and micrometries, and occurrence of all presently recognized species of Cymbastela Hooper & Bergquist, Reference Hooper and Bergquist1992, including the new species described here. Data compiled from the literature. Oxea sizes are expressed in micrometres. Alvarez et al. (Reference Alvarez, Hooper and van Soest2008) has transferred Cymbastella hooperi and C. terpenensis to the new genus Pipestela.

Axinella tricalciformis Bergquist, Reference Bergquist1961 was transferred to Cymbastela in Kelly et al. (Reference Kelly, Edwards, Wilkinson, Alvarez, Cook, de, Bergquist, Buckeridge, Campbell, Reiswig, Valentine and Gordon2008) as it has exclusively large oxeas in a spongin-rich plumoreticulate skeletal arrangement. This species was described by Bergquist (Reference Bergquist1961) as being relatively common throughout southern New Zealand (Dunedin, Stewart Island, Chatham Island and Auckland Island). Despite comprehensive collections around New Zealand since the early 1960s, it has not been re-collected (with the exception of a single beach washed specimen from the west coast of the North Island in 2005).
Another species provisionally assigned to Cymbastela by Kelly & Alvarez (unpublished data) is known to be very common on steep slopes between 5–25 m in Doubtful Sound, Fiordland. This species has a thin tough leathery lamellate–flattened–encrusting form (2–4 mm thick), and a regular dense plumoreticulate skeleton of robust oxeas (~200 µm long) embedded in the spongin. Although the morphological form resembles that of Pipestela, and P. hooperi (Van Soest et al., Reference van Soest, Desqueyroux-Faúndez, Wright and König1996) comb. nov Alvarez et al. (Reference Alvarez, Hooper and van Soest2008) in particular, this species does not have the loose reticulation of plumose and wavy paucispicular to multispicular tracts of relatively thin oxeas that define the genus (Alvarez et al., Reference Alvarez, Hooper and van Soest2008). A thorough comparison of this Fiordland species with P. hooperi and Cymbastela species is required before the true identity can be confirmed.
The discovery of a new species of Cymbastela in Sodwana Bay, north of Durban on the north-east coast of South Africa, extends the genus significantly further east, and raises questions about its current, apparently disjunct distribution. Species of Cymbastela have not been recorded in the relatively extensive older literature on collections from the western and eastern Indian Ocean and south-eastern South Africa (Ridley & Dendy, Reference Ridley and Dendy1887; Topsent, Reference Topsent1892; Lendenfeld, Reference von Lendenfeld1897; Burton, Reference Burton1931, Reference Burton1933, Reference Burton1936, Reference Burton1959; Lévi, Reference Lévi1963, Reference Lévi1967, Reference Lévi1964; Thomas Reference Thomas1973, Reference Thomas1979, Reference Thomas1981; Pulitzer-Finali, Reference Pulitzer-Finali1993; Van Soest, Reference van Soest and van der Land1994; Thomas et al., Reference Thomas, Gopinadha Pillai and Rajagopalan1997; Hooper et al., Reference Hooper, Kelly and Kennedy2000; Kelly, Reference Kelly and Richmond1997), or encountered during reasonably comprehensive collections off Madagascar, Mauritius, Seychelles, Tanzania (Zanzibar and Pemba Islands), and throughout South-East Asia by Michelle Kelly and others, and the Coral Reef Research Foundation, Palau.
A lack of records from major South-East Asian and Indian Ocean regions, despite relatively comprehensive collections undertaken, may also be due in part to the difficulty of separating species of Cymbastela from fan-shaped Niphates and Amphimedon (order Haplosclerida: family Niphatidae). Species in these genera also contain oxeas in a reticulate arrangement that, at times, can appear almost plumoreticulate. A recent assignment of species previously recognized as Aphimedon and Siphonochalina, in the haplosclerid families Niphatidae and Callyspongiidae, to the new axinellid genus Pipestela, exemplifies this exact problem (Alvarez et al., Reference Alvarez, Hooper and van Soest2008). Colour would appear to be the primary distinguisher between these difficult groups; Niphatidae and Callyspongiidae are dominated by colours in the blue–lilac–pink spectrum, whilst axinellid sponges are dominated by colours in the yellow–orange–brown spectrum. Many species of Cymbastela also have green and maroon tinges conferred by symbiotic algal and cyanobacterial symbionts (Hooper & Bergquist, Reference Hooper and Bergquist1992). Cymbastela sodwaniensis sp. nov., described here, extends the known distribution of the genus to the subtropical south-east coast of South Africa, and broadens our current understanding of the skeletal arrangement of the genus.
MATERIALS AND METHODS
Sample collection
Material described here was collected during a survey that documents the sponge fauna of the east coast of South Africa. Upon collection, specimens were preserved in 70% ethanol, and prepared for histological examinations as outlined in Samaai & Gibbons (Reference Samaai and Gibbons2005). Spicule dimensions are given as the mean length (range of length measurements) × mean width (range of width measurements) of 20 spicule measurements, unless otherwise stated. Primary type material has been registered and deposited in the South Africa Museum and fragments of both the holotypes and paratypes are in the collections of T.S. The prefix Ts indicates specimen sample numbers for the collection of Toufiek Samaai housed at Marine and Coastal Management (MCM).
Sampling locations
Sodwana Bay and Aliwal Shoal are situated on the east coast of South Africa in the KwaZulu-Natal province, north and south of Durban, respectively. Sodwana Bay (Figure 1A) is in the marine reserve that now forms part of the World Heritage site formerly named The Greater St Lucia Wetlands Park, and known since May 2007 as iSimangaliso Wetland Park. Aliwal Shoal is situated about 5 km off the town of Umkomaas on the KwaZulu-Natal south coast; about 45 minutes drive from Durban (Figure 1B).

Fig. 1. Map of collection localities on the east coast of South Africa. (A) Sodwana Bay is situated in the iSimamgaliso Wetland Park and forms part of the greater Maputuland reef complex. Location of Sodwana bay reefs on the coastline of northern KwaZulu-Natal, South Africa (adapted from Celliers & Schleyer, 2002); (B) Aliwal shoal is situated about 5 km off the town of Umkomaas, south of Durban on the KwaZulu-Natal coastline.
The coral reefs within Sodwana Bay form the central complex of the Maputaland reefs with a northern complex adjoining Kosi Bay and a southern complex comprising Red Sands Reef and Leadsman Shoal (Schleyer & Celliers, Reference Schleyer and Celliers2003a, Reference Schleyer and Celliersb). Substantial water movement occurs on the reefs with the coast occasionally swept by an inshore meander of the southward flowing Agulhas Current (Schumann, Reference Schumann and Schumann1988). The structure and bathymetry of the reefs are variable (Schleyer & Celliers, Reference Schleyer and Celliers2003a, Reference Schleyer and Celliersb). Two-mile Reef ranges in depth from 6 m to 10 m on its shallowest pinnacles, 14–19 m on the reef flats and 24–27 m at the edge of the fore reef. Spur and groove outcrops occur at 18–24 m at Four-mile Reef. Five-mile Reef is a large flat and delicate reef with a maximum depth of 21 m. Seven-mile Reef is a small table-like feature with a prominent drop off from 17–22 m. Nine-mile Reef consists of shallow platforms, ranging in depth from 6–18 m with steep drop-offs from 12–22 m (Ramsay, Reference Ramsay1996; Schleyer, Reference Schleyer1999). With the exception of a few isolated pinnacles at Nine-mile Reef none of the reefs break the surface (Schleyer & Celliers, Reference Schleyer and Celliers2003a, Reference Schleyer and Celliersb). The main body of the Aliwal Shoal reef is 1.5 km and 0.5 km wide midway along its length. The depth of the reef varies between 4 m on the pinnacle to 28 m on its outer edges. The shoal is part of an offshore reef system running along much of the KwaZulu-Natal coast, which had its origin in a line of coastal dunes formed during a period when the sea level was lower than current levels. Aliwal Shoal lies inshore of the Agulhas Current, a warm southerly current, and the reefs are shallow enough for the penetration of natural light (Schleyer, M.H. unpublished report, Reference Schleyer2001; M & CM and EKZN, unpublished report, 2002).
RESULTS
TYPE SPECIES
Pseudaxinyssa stipitata Berquist & Tizard, Reference Bergquist and Tizard1967: 189.
DIAGNOSIS
Cup-shaped growth form. Surface smooth with choanosomal or ectosomal spicules projecting slightly; with or without specialized skeleton of smaller oxeas. Choanosomal skeleton generally differentiated in axial and extra-axial regions; axial skeleton condensed or vaguely reticulated; extra-axial skeleton plumoreticulate with well-defined multispicular plumose tracts enveloped with well to poorly developed spongin fibres, compared with laminar, branching or tubular growth forms with a loose reticulation of mostly paucispicular tracts encased sometimes in poorly developed spongin fibres, in Pipestela. Megascleres are oxeas often with telescoped points (modified from Alvarez & Hooper, Reference Alvarez, Hooper, Hooper and Van Soest2002 and Alvarez et al., Reference Alvarez, Hooper and van Soest2008).

Fig. 2. Cymbastela sodwaniensis sp. nov. (A) Holotype specimen SAM-H 5111 in situ; (B) paratype specimen SAM-H 5112 in situ; (C) paratype specimen SAM-H 5113 in situ; (D) specimen Ts 884 in situ.

Fig. 3. Cymbastela sodwaniensis sp. nov. Holotype SAM-H 5111; (A&B) details of skeletal architecture at 50 × and 100 × magnification; (C) oxeas.
MATERIAL
Holotype: SAM-H 5111 (Ts 851): Five-mile Reef (potholes), Sodwana Bay, KwaZulu- Natal, South Africa, 27º29′42″S 32º41′25″E, 20 m depth, 3 November 2003, collected by T. Samaai.
Paratypes: SAM-H 5112 (Ts 847): Ribbon Reef—5 mile, Sodwana Bay, KwaZulu-Natal, South Africa, 27º29′17″S 32º41′52″E, 20 m depth, 7 November 2003, collected by T. Samaai; SAM-H 5113 (Ts 853): Deep Sponge Reef, Sodwana Bay, KwaZulu-Natal, South Africa, 27º31′0.18″S 32º41′0.65″E, 30 m depth, 6 November 2003, collected by T. Samaai; SAM-H 5114 (Ts 629): Aliwal Shoal, Chunnel, KwaZulu-Natal, South Africa, 30º15′41″S 30º49′37″E, 14 m depth, 13 April 2003, collected by T. Samaai.
Other material: Ts 884: Ramsay Reef—7 mile, Sodwana Bay, KwaZulu-Natal, South Africa, 27º26′48″S 32º42′55″E, 18 m depth, 3 November 2003, collected by T. Samaai; Ts 937: Seven-mile Reef, Sodwana Bay, KwaZulu-Natal, South Africa, 27º27′29″ S 32º42′51″E, 20 m depth, 7 November 2003, collected by T. Samaai; Ts 630: Aliwal Shoal, Chunnel, KwaZulu-Natal, South Africa, 30º15′41″S 30º49′37″E, 18 m depth, 13 April 2003, collected by T. Samaai.
EXTERNAL MORPHOLOGY
Sponge typically vase-shaped (Figure 2A, B) with undulating rounded margins, up to 5.5 cm high, 7 cm maximum width, elevated on a short stalk, 0.5 cm long, 2 cm wide. Some sponges have multiple undulating lamellae with incised margins (Figure 2C, D). Cup walls are thin and somewhat fleshy, 0.1–0.8 cm diameter. Surface of cup predominantly smooth with nodules and occasionally additional lamellae, exterior of cup microscopically hispid and smooth, interior of cup with numerous small flush oscules, 0.5–1 mm in diameter. Texture is firm, flexible and slightly compressible. Colour in life orange-red; in preservative beige.
SKELETON
The choanosomal architecture is plumoreticulate, with poorly differentiated axial and extra-axial skeletal regions (Figure 3A, B). The central region of choanosomal skeleton only slightly axially condensed; predominately reticulate, with plumose paucispicular ascending tracts curving gradually towards the surface (Figure 3B). Peripheral skeleton clearly diverges into plumose ascending multi- or paucispicular tracts interconnected by uni- or paucispicular transverse tracts (Figure 3B). These ascending multispicular tracts are encased in a sheath of transparent spongin that is clearly visible under high magnification. The ectosome is not specialized, with plumose brushes of choanosomal oxeas protruding through ectosome for short distances, producing microscopic surface conules. Tracts may or may not be interconnected by paucispicular transverse fibres, but usually hidden by diverging spicules.
SPICULES
Megascleres: Single category of smooth oxeas present; oxeas relatively small, with uniform thickness, usually curved midway, sharply pointed or fusiform, 185 (176–200 µm) × 8.5 (8–9 µm) (Figure 3C).
SUBSTRATUM, DEPTH RANGE AND ECOLOGY
Restricted to shallow coastal coral reef waters, depth 15–30 m. Substrate is predominately rock or coral, and the sponge is found in areas sheltered from strong currents.
GEOGRAPHICAL DISTRIBUTION
Sodwana Bay and Aliwal Shoal, KwaZulu-Natal, South Africa.
ETYMOLOGY
Named for the type locality.
REMARKS
Cymbastela sodwaniensis sp. nov. is most similar to C. cantharella (Lévi, Reference Lévi1983) from New Caledonia, which typically forms a golden brownish orange vase or cup. In both species the oxeas are non-telescoped and fusiform, and both species have poorly differentiated axial and extra-axial skeletons. The oxeas are larger in C. cantharella (143–245 µm), and C. sodwanienses sp. nov. is brighter orange with thicker smoother walls that are less fibrous in texture. Specimens of C. cantharella do not usually have external buttresses or concentric lamellae, and lack prominent oscules on the concave face.
The spicule dimensions, external morphology and colorations are compared for all nine species of Cymbastela, including the new species in Table 1. The majority of other Cymbastela species form large flattened cups with concentric striations (e.g. C. stipitata), or infilling buttresses and with quite variable spicule lengths (Table 1). Many species are olive to dark brownish green (e.g. C. stipitata) with a pronounced stalk, and C. marshae and C. notiaina have much smaller oxeas than C. sodwaniensis sp. nov. The former species has a cavernous subdermal region. Cymbastela coralliophila differs from other members and the new species, in having a flattened growth form, with flabellate lamellae, and C. concentrica has a plumoreticulate extra-axial choanosomal skeleton. The new species is thus considered clearly distinct from all possibly more closely related species of Cymbastela.
DISCUSSION
A comparison of C. sodwaniensis sp. nov. to described species of Cymbastela (Table 1) indicates that it is more closely related to species of Cymbastela than to any other Axinellidae (see Hooper & Bergquist, Reference Hooper and Bergquist1992; Van Soest et al., Reference van Soest, Desqueyroux-Faúndez, Wright and König1996; Alvarez & Hooper, Reference Alvarez, Hooper, Hooper and Van Soest2002; Alvarez et al., Reference Alvarez, Hooper and van Soest2008). The possession of smallish oxeas, albeit non-telescoped, in a plumoreticulate arrangement with some axial compression in the stalk, indicates strong affinity with Cymbastela as it is known from the Indo-Pacific. Coloration and morphology strongly link it to Axinellidae in general.
Cymbastela has a relatively low diversity of morphological characters but differences can be recognized in the variation in size and shape of oxeas, their arrangement in the skeleton, and the amount of spongin that supports the tracts. External coloration differs across the genus although it may be obscured in some species by the presence of cyanobacteria that confer a maroon or dark brown coloration.
From a biogeographical perspective, Cymbastela is most speciose around north-eastern Australia with eight species (Hooper & Bergquist, Reference Hooper and Bergquist1992; Alvarez & Hooper, Reference Alvarez, Hooper, Hooper and Van Soest2002). The genus extends east to New Caledonia with a single species C. cantharella (Lévi, Reference Lévi1983) also reported further east in Vanuatu and Fiji (unpublished data, Michelle Kelly and Lori J. Bell) and north in Palau and Chuuk Atoll. Cymbastela concentrica (Lendenfeld, Reference von Lendenfeld1887) has been recorded from southern Papua New Guinea and Fiji. The distribution of Cymbastela appears to be markedly disjunct; no species have been recorded west of the Philippines to date and there are no records for any South-East Asian locality, the Indian subcontinent or any of the areas within the western Indian Ocean despite contemporary collections from these regions. Other relatively well-known areas such as the Caribbean region, Mediterranean, and Brazil, also have no records of Cymbastela.
The closest species to sodwaniensis sp. nov. in terms of morphology and spiculation is C. cantharella from the Indo-Pacific, which supports a Tethyan origin for colonization within the western Indian Ocean, rather than a colonization via the longer trans-Pacific or Indo-West Pacific tracts (Hooper et al., Reference Hooper, Kelly and Kennedy2000). Evidence for this assumption stems from the fact that the fauna of the western Indian Ocean is thought to be similar to the Indo-Malayan regions (Hajdu, Reference Hajdu1995; Hooper et al., Reference Hooper, Kelly and Kennedy2000; Peixinho et al., Reference Peixinho, Fernandez, Oliveira, Caíres, Hajdu, Custódio, Lôbo-Hajdu, Hajdu and Muricy2007; Hajdu & Desqueyroux-Faúndez, Reference Hajdu and Desqueyroux-Faúndez2008).
Alternatively, as already pointed out by Alvarez et al. (Reference Alvarez, Hooper and van Soest2008) the accurate and clear recognition of species of Cymbastela depends upon a unique combination of plesiomorphic characters and it is quite conceivable that the genus (like Axinella) may be polyphyletic. Some Axinella (defined as usually possessing styles and oxeas in equal proportions) are known to have megasclere components dominated by styles with only very rare oxeas in some specimens (A. waltonsmithi Alvarez et al., Reference Alvarez, van Soest and Rützler1998 from North America), while we know of an undescribed South African species of Axinella, with a similar morphology to C. sodwaniensis sp. nov., that has only occasional styles amongst the dominating oxeas (M. Kelly and Lori J. Bell, unpublished data). Under this scenario, the species C. sodwaniensis sp. nov. may be an Axinella that completely lacks styles amongst the oxeas, and thus represent one extreme boundary in the definition of the genus. However, the exclusive presence of small to medium-sized oxeas in species of Cymbastela in general, and C. sowaniensis sp. nov in particular, the skeletal arrangement of differentiated axial and extra-axial regions, the latter with plumose spicule tracts in thin spongin sheaths or thick fibres, and the temperate to (sub)-tropical biogeographical distribution (see Table 1), confirm the integrity of the genus Cymbastela as presently recognized, and the recognition of C. sodwaniensis sp. nov. as a distinct species. It may be possible in the future to provide confirmation for this hypothesis through genetic analysis to determine whether these morphological variations between congeners represent edge effects in populations of Axinella or indicate the existence of a series of possible allopatric sibling species within the genus Cymbastela.
ACKNOWLEDGEMENTS
The authors are grateful to Vic Peddemors (Macquarie University, Australia), Cloverley Lawrence (Ezemvelo KwaZulu-Natal Wildlife, Durban), Kerry Sink (South African Biodiversity Institute, Cape Town) and Sean O'Donohue (University of KwaZulu-Natal, Durban), for help with the collection of this and many other specimens in our collections. We thank Peter Timm for provision of facilities during our stay in Sodwana Bay, as well as Blue Wilderness Dive Expeditions (Mark Addison) and The Whaler Dive Center, both situated in Umkomaas. We would like to thank W. Florence for comments on the manuscript. We are also very grateful to Belinda de Glasby and John Hooper for their insightful suggestions and discussions. The authors thank Marine & Coastal Management, Department of Environmental Affairs & Tourism, South Africa, for permits to undertake these collections. This work was supported by the National Research Foundation, South Africa, as a research grant to T.S. (Grant number FA 2004043000058), and in New Zealand by the Foundation for Research Science & Technology (Contract C01X0219) to the National Institute of Water & Atmospheric Research (NIWA). Ongoing research support to M.K. through Capability Funding to the Biodiversity Programme, from NIWA, is gratefully acknowledged. We also thank the three anonymous referees for their constructive comments.