Polyaxone monaxonids: revision of raspailiid sponges with polyactine megascleres ( Cyamon and Trikentrion)

Abstract Among the thousands of non-tetractinellid (monaxonid) Demospongiae species, less than twenty possess polyactine (usually three- or four-claded) megascleres. These are currently assigned to two closely related genera, viz. Cyamon Gray and Trikentrion Ehlers, both members of the raspailiid subfamily Cyamoninae. The two genera are considered valid on account of differences in the shape and the ornamentation of the polyaxone spicules. Cyamon predominantly has four-claded equiangular spicules with all cladi spined or rugose, whereas Trikentrion usually has a majority of three-claded spicules on which spines are found only on a single basal clade. Nevertheless, the differences between the two genera appear to overlap in several known and newly discovered species, necessitating a revision of the two groups. Two new species of Cyamon were found to occur on inshore sandstone platforms off the coast of Mauritania. One of the new species, Cyamon amphipolyactinum sp. n., possesses unique small ‘double’ polyactine spicules in addition to the usual calthrops-like polyactine megascleres characteristic for Cyamon. The second new species, Cyamon arguinense sp. n.,possesses polyactine megascleres of which only one of the cladi is spined the remaining three or more cladi being smooth, a feature that is considered characteristic of sponges of the genus Trikentrion. The type species of Cyamon, Cyamon vickersii (Bowerbank) appears to have been misinterpreted as a Caribbean species, because circumstantial evidence strongly indicates an Indian Ocean origin. This has the consequence that specimens recorded subsequently under the name Cyamon vickersii from various Western Atlantic localities are reassigned to Cyamon agnani (Boury-Esnault), a species originally described from Brazil. A new species, reported as Cyamon vickersii sensu Burton & Rao from the east coast of India, and available to us only as a single thick section mounted on a glass slide, is named Cyamon hamatum sp. n. The Cyamon membership of the only deep-sea species, Cyamon spinispinosum (Topsent) is drawn in doubt due to considerable morphological deviation from mainstream Cyamon. The type species of Trikentrion, Trikentrion muricatum (Pallas), is extensively described and discussed, and a neotype is assigned. West African Trikentrion laeve (Carter) is for the first time since its original description properly redescribed from the type material. The specimen recorded by Burton as Trikentrion laeve from Congo turned out to be different from the original material of Carter and is assigned to a new species, Trikentrion africanum sp. n. All species of both genera considered valid are reviewed, mostly based on the examination of type or other original specimens. Our revision shows the existence of twelve species of Cyamon and six species of Trikentrion. A key to the species is provided and remarks on the geographic distribution of both genera are made. Based on our study, the differences between Cyamon and Trikentrion are re-evaluated. Only one character absolutely distinguishes the two genera, the presence (Trikentrion) or absence (Cyamon) of trichodragmata. A further discriminating character is the possession of short thick styles (most Cyamon species) versus thick oxeas (many Trikentrion), but this is complicated by absence of the oxeas in three Trikentrion species. Although spination of the polyactine spicules in itself cannot serve to distinguish the two genera with certainty, those of Trikentrion are usually recognizable by excessive hook-like spines against a finer spination in Cyamon. Possibly, the polyactine spicules of both groups are non-homologous, with Cyamon polyactines derived from styles and Trikentrion polyactines from oxeas, but this remains to be further investigated.


Introduction
The revision presented below was inspired by the recent discovery of two new species, evidently belonging to the sponge genus Cyamon Gray, 1867 (Demospongiae, Poecilosclerida, Microcionina, Raspailiidae, Cyamoninae), growing on shallow-water sandstone ridges off the coast of Mauritania. Cyamon species are unusual among raspailiid sponges in possessing polyactine megascleres (mostly four-or three-claded) with all cladi spined. Most species of Cyamon are rare encrusting sponges recorded from seemingly random localities across the warmer waters of the globe (Hooper 2002). Next to typical raspailiid ectosomal skeletal features, they share a plumose arrangement of smooth choanosomal styles and a basal mass of polyactines. To date ten species have been described (see Van Soest et al. 2012), usually recorded only once, from shallow waters of the Western Atlantic, Eastern Pacific, Indian Ocean and Indonesia, with a single species from deep-sea North Atlantic localities. A presumed sister genus, Trikentrion Ehlers, 1870, with only four species (see Van Soest et al. 2012), together occurring likewise circumglobally, has broadly similar polyactine spicules, with only one of the cladi spined. The two Mauritanian species were assigned to Cyamon because of the encrusting habit and stylote condition of the choanosomal megascleres. While one of them possesses unique and unprecedented 'double' micro-polyactines, it is the second species that appeared to be the most intriguing as it was found to possess polyactine spicules with only a single cladus spined, thus overlapping with the alleged spination in the polyactines of the sister genus Trikentrion. This raised the question whether the two genera could be part of a single diverse genus, rather than being separate morphological groups. Cyamon shares with members of the genus Trikentrion the polyactines and the raspailiid ectosome, but the choanosomal spicules in the type species Trikentrion muricatum and other Trikentrion species are smooth oxeas, and the polyactines in Trikentrion are scattered throughout the choanosome and replace the oxeas entirely in some species. The polyactines of Cyamon and Trikentrion appear distinctly different at first glance, with mostly equiangular geometry in Cyamon and sagittal (Y-shape, Tshape) condition in Trikentrion. Authors with experience of these sponges favour the hypothesis that the polyactine spicules derive from echinating acanthostyles, but there is no firm evidence for this and the spicule types remain unique in the family Raspailiidae and among the non-tetractinellid demosponges. Preliminary DNA sequence information confirmed the raspailiid affinity of at least Trikentrion (Erpenbeck et al. 2007), so there is at present no reason to take a different view.
Below, we describe four new species and review previously described species of both genera, pointing out aspects that appear to have been overlooked. We propose the synonymy of several previously accepted species, indicate a serious misinterpretation of the origin of the type species of Cyamon and provide extensive data on the type species of Trikentrion, including designation of a neotype. We demonstrate that the distinguishing characters of the two genera are eroded by intermediate conditions in new, but also in already known taxa, and discuss the remaining characters available for unambiguous genus assignment. We provide keys to the species and review the geographic distribution. We will refrain from taking decisions affecting the genus-and subfamily classification until such time that sufficient independent molecular support may become available. Recently, molecular evidence was presented that Raspailiidae, currently assigned to the suborder Microciona of the order Poecilosclerida (Hooper, 2002) is probably not closely related to the chela-bearing Poecilosclerida (Erpenbeck et al. 2007a; Morrow et al. 2012). While we acknowledge that this evidence will likely lead to alteration in the near future of the classification of the raspailiid sponges, including Cyamon and Trikentrion, we think it is currently premature to adopt these changes. More confirmation from additional studies and additional taxa is necessary to reassign Raspailiidae.

Material and methods
Specimens of Cyamon and Trikentrion present in the collections of the Zoological Museum of Amsterdam and the Rijksmuseum van Natuurlijke Historie at Leiden (together now the Naturalis Biodiversity Center) were available from West Africa (two new species from a locality off Mauritania shown in Fig. 3, old collection specimens from Ghana), from the West Indian region (Curaçao and Colombia), the Seychelles, Indonesia and North Australia. We obtained loans of type material of most species from the collections of BMNH, USNM, MNHN, SMF, and LACM. One of us (JH) addition-ally examined fragments of Cyamon vickersii (Bowerbank, 1864) and Trikentrion flabelliforme Hentschel, 1912 obtained on loan from ZMB and SMF respectively. Non-type material of species of both genera was obtained on loan from BMNH and USNM (see below for abbreviations), and one of us (JLC) examined fresh material of Cyamon koltuni Sim & Bakus, 1986 and Cyamon (=Trikentrion) catalina Sim & Bakus, 1986. Details of collection numbers and localities are provided below with each species.
Abbreviations of institutions cited in the text: Terminology: We employ the collective word 'polyactine' for the spicules previously named acanthotriaenes by Hooper (2002) because the suffix -triaene suggests astrophorid affinities and also the triaene condition is only one of a range of cladi numbers in this spicule type (2-8). Other terms used in the literature (e.g. quadriradiates, cf. Carter 1879, acanthotetractine, cf. Hentschel 1912, pseudotetracts, cf. Dendy 1922pseudactines, cf. Burton and Rao 1931;tetraxons, cf. De Laubenfels 1936) are equally unsuitable to capture the nature and variation of this spicule type. The cladi are subdivided into basal and lateral (see below). Furthermore, the style categories are indicated with the adjective 'long thin' for the extra-axial or peripheral long styles protruding from the surface and causing the hispidation of many species, 'short thin' for styles that form a bouquet or sheath around the long thin or thick styles in many species. In Cyamon we apply the term 'short thick' styles for the often subtylote styles that singly or in bundles form the choanosomal skeleton supported by the polyactines in many species. Spicules of Trikentrion are called 'oxeas' only when they represent choanosomal megascleres; reduced diactinal conditions of the polyactine spicules (also occurring in certain Cyamon species), recognizable by being roughened at one of the apices and usually swollen or crooked in the middle, are termed diactines or two-claded polyactines, not oxeas. Not all Cyamon and Trikentrion species appear to possess the full spicule complement of long thin, short thin and short thick styles/oxeas, so in individual species additional terms may be employed, notably 'long subtylostyles', which characterize Cyamon quinqueradiatum (Carter, 1880) and one of the new species. Several Trikentrion species lack choanosomal oxeas at all.

AHF-NHMLA
Microscopic preparation: dissoluted spicule preparations for measurements and SEM observations were made by dissolving a small fragment of the sponge in concentrated HNO 3 or in undiluted household bleach, subsequent rinsing at least five times in distilled water, the last time in ethanol 96%, and finally pipetting a spicule suspension on stub or slide to be dried in a stove. Thick sections of the sponge made for the study of the skeletal structure were air-dried on a hotplate or in a stove and embedded in Canada balsam. Measurements of spicules (minimum-average-maximum) were made of 25 spicules of each category for each individual, unless otherwise stated (e.g. long thin spicules were often broken so the required number of spicules could not be measured).

results
We present the results in the following seven sections: a refined description and illustration of the type material of the type species of Cyamon, C. vickersii, in which we argue that its original locality has been misinterpreted, followed by a description of recent (1993) Seychelles material considered to belong to C. vickersii; description of two new Cyamon species from West Africa; descriptions and illustrations of all species assigned to Cyamon previously, including a new species based on misidentified material; a refined description of the specimens of the type species Trikentrion muricatum (Pallas, 1766) including assignment of a neotype; descriptions of the remaining species, including proposed synonymies and the description of a new species of Trikentrion based on misidentified material; we provide a key to the recognized species of Cyamon and Trikentrion; we make summary remarks on the geographic distribution of the two genera.
Definition (emended): Cyamoninae with skeleton consisting of a basal mass of polyactine spicules of which one or more cladi are spined or rugose in mature condition, supporting a plumose choanosomal skeletal arrangement of single or columnar groups of styles or subtylostyles with pointed ends outwards. Additional longer and shorter thin styles may be present in peripheral regions.
Remarks. The styles are usually smooth, but in Cyamon spinispinosum (Topsent, 1904) both shorter and longer styles are spined (see below). In the type species, and several other species, thin short styles take the form of angulated and/or centrotylote strongylostyles, some of which have one end faintly or more heavily spined (see below). Polyactine spicules are genuinely polyaxone, with axial canals visible in all cladi. They are predominantly calthrops-like and have four cladi, but this may vary between two and eight cladi in some species. Usually, one of the cladi differs from the others by having a pointed spined apex, whereas the other cladi frequently have rounded ends, with prominent spined bulbs in several species, or they are occasionally entirely smooth, differing frequently also in length (either longer or shorter) from the other cladi. The spined pointed cladus is termed 'basal', under the assumption that it is homologous to the shaft of an ancestral echinating acanthostyle. The remaining cladi are here termed 'lateral', based on the assumption they are lateral proliferations of the acanthostyle head. One of the new species described below, has the polyactine spicules in two distinct categories, the smaller one of which is 'amphipolyactine' (see below).  Sim & Bakus, 1986: transferred to Trikentrion Cyamon dendyi De Laubenfels, 1936 Cyamon hamatum sp. n.: new species based on misidentified material of Cyamon vickersii sensu Burton and Rao, 1931 Cyamon incipiens (Topsent, 1928 as Acantheurypon): j. syn. of C. spinispinosum Cyamon koltuni Sim & Bakus, 1986: valid species Cyamon neon De Laubenfels, 1930: valid species Cyamon quadriradiatum (Carter, 1880 as Microciona): species inquirenda Cyamon quinqueradiatum (Carter, 1880 as Microciona): species inquirenda Cyamon spinispinosum (Topsent, 1904 as Hymeraphia): valid species, atypical, possibly not a Cyamon Cyamon toxifera Arndt, 1927: mixture of C agnani and Clathria (Microciona) ferrea (De Laubenfels, 1936) Cyamon vickersii (Bowerbank, 1864 as Dictyocylindrus): valid species, type species, type locality proposed to be Indian Ocean, Central West Atlantic specimens transferred to C. agnani. Trikentrion africanum sp. n.: new species, formerly T. laeve sensu Burton, 1948 Trikentrion catalina (Sim & Bakus, 1986 as  Trikentrion Ehlers, 1870 shares the polyactines with Cyamon. According to the latest treatment of both genera (Hooper, 2002) the polyactines of Cyamon would have all the cladi spined, whereas those of Trikentrion would have only the basal cladus spined. If this distinction between Cyamon and its close relative Trikentrion in the cladus spination would be maintained, then four species originally described as members of Cyamon would need to be transferred to Trikentrion, C. quinqueradiatum, C. neon de Laubenfels, 1930, C. argon Dickinson, 1945 andC. catalina, as well as one of the new species described below. We will demonstrate below and in the Discussion that cladus spination does not coincide with other more compelling differences with Trikentrion and consequently we will not transfer (all) the mentioned taxa.
The species considered valid members of Cyamon are listed in Table 1 and their properties in Table 2. The holotype was extensively described by Carter (1879) (his illustrations are reproduced in Fig. 1B), and redescribed by Hooper (2002). The specimen is now (2012, see Fig. 1A) a dry, macerated, wedge-shaped sponge, glued to a label containing the text Bk. 1887, Dictyocylindrus vickersii, lodged in a round box. There are five microscopic slides: three thick sections (one is reproduced in Fig. 1C), and two spicule mounts. A photo was made (Fig. 1D) of the contents of one of the spicule slides  (Bowerbank, 1864) showing characteristic polyactines and one centrotylote strongylostyle. All microscopic slides are labeled with texts in Bowerbank's and Carter's handwritings.

Description of the type material of the type species of Cyamon
Description. The specimen consists of a barely coherent mass of columns, fragile, crumbly. Size approx. 3 × 2.5 × 0.6 cm. Colour now dark red-brown.
Skeleton: a branched columnar structure built by bundles of short thick styles supported at the base and along the column by masses of polyactines. The remaining spicules are not readily visible in the sections, so their positions are derived from Carter's drawings (Fig. 1B): the columns are echinated by long and short styles and wavy strongylostyles. Spicules (Fig. 2): long thin styles, short thin (strongylo-)styles, short thick styles, polyactines.
Remarks. Contrary to most other authors referring to Cyamon vickersii, we have become convinced that this species does not occur in the Western Atlantic. The evidence for this is two-fold.
(1) There is considerable uncertainty about the origin of the type specimen. Bowerbank (1862: 831), when he first drew attention to the polyactine spicule, described it as follows: Spiculated inequi-angulated triradiate, with cylindrical entirely spined radii (Plate XXXVI. fig. 15). -From a fragment of a sponge presented to me by Mr. Vickers of Dublin, who thinks it probably came from the West Indies. This spiculum is an external defensive one. The triradiate rays are imbedded immediately beneath the dermal membrane, and the spicular ray is projected through it at right angles to its plane; they are very numerous.
The part of the sentence we placed in roman lettering contains the only factual information on the origin of the specimen, which was subsequently named Dictyocylindrus vickersii by Bowerbank (1864: 267) with the same sentence and figure repeated. Bowerbank's slides of the type material in BMNH marked as Bk 1887 were labeled prudently "West Indies ?" (see Fig. 1C), but first Gray (1867: 546) and later Carter (1879: 292) omitted the question mark. Carter did an extensive redescription of the Bowerbank material (see Fig. 1B), which properly established the characters of the species. Shortly before that (Carter, 1876: 391) he alluded to a  (Bowerbank, 1864), SEM images of spicules of the holotype BMNH 1877.5.21.188, A long thin style A1 details of apices of long thin style B short thin (strongylo-)style B1details of apices of short thin (strongylo-)style C short thick style C1 details of apices of short thick style D four-claded (left) and three-claded (right) polyactines. specimen with quadriradiate spicules obtained from Thomas Higgin from Grenada (Caribbean Sea), which he thought to belong to the same species. Higgin (1877: Pl. 14 Fig. 9) figured the spicule. However, both authors mentioned only long styles in addition to the polyactines, which is, as we know now, insufficient to characterize Cyamon species. As we described above, and was also clearly pictured by Carter himself (1879: Pl. 27 Fig. 6c, see also our Fig. 1B), C. vickersii should possess undulated or crooked centrotylote thin styles or strongylostyles. We will demonstrate below that none of the Western Atlantic specimens of Cyamon we examined possess such spicules, in stead of which they have straight thin styles without centrotylote swelling or undulations. Nevertheless, from the time of Carter onwards it was assumed, that Bowerbank's type came from the West Indies. Subsequent reports of Cyamon from Western Atlantic localities all employed the name C. vickersii, and ignored the peculiar shape of the short thin styles.
(2) Dendy (1922) and Thomas (1973) reported Cyamon vickersii from the Seychelles. Their descriptions exactly match the properties of Bowerbank's type specimen, including the undulating short thin centrotylote styles. They especially mention the spination on the pointed ends of many of the undulating styles, precisely as we found in the type (see Fig. 2B, B1). De Laubenfels (1936: 80) also was of the opinion that the Seychelles material differed specifically from the Western Atlantic material. Because he believed that C. vickersii was West Indian, he proposed the name Cyamon dendyi for the Seychelles material. Below, we describe and illustrate ( Fig. 3) material obtained from the Seychelles, in which we demonstrate beyond doubt that it belongs to Cyamon vickersii.
To conclude: specimens identical or similar to the type of C. vickersii are reported from the Seychelles. Specimens recorded from the Western Atlantic are dissimilar to the type of C. vickersii, a.o. by lacking the characteristic undulating spicules. For the Atlantic representatives, the name Cyamon agnani (Boury-Esnault, 1973) is available (see below).  (Hooper, 2002: Fig. 17) but could not be found in the collection of the Natural History Museum in 2011 (Ms Emma Sherlock, in litteris).

Description of ZMA material of
Description. Strawberry-shaped sponge (Fig. 3A), forming a single semiglobular mass with microlobate surface. Color red or orange-red (alive), dark brown-red in alcohol. Consistency firm, barely compressible. Specimens now looking clathrate due to loss of thin surface membrane, still present in places. Size of largest specimen 3 × 2 × 2 cm.
Skeleton: condition described as columnar, consisting of hillock-like masses of polyactines, variable in thickness up to 2 mm, supporting thick plumose bundles of thick styles, which in turn are peripherally surrounded by short thin strongylostyles. Rare long thin styles are not present in all slides.
Long thin styles (Fig. 3B), very rare, invariably broken in small pieces, largest piece found in our slides 300 × 12 µm; according to Dendy they can reach 1700 × 14 µm. We reconstructed a long style from several pieces found on the SEM stub (Fig. 3B).
Distribution. So far known with certainty from several localities throughout the Seychelles (Mahé and the Amirante Islands).
Ecology. Sandy bottoms at 30-50 m surrounding reefs and atolls. Discussion. The ectosomal strongylostyles in Cyamon vickersii are reminiscent of those found in the type species of the Axinellidae genus Reniochalina (Reniochalina stalagmitis Lendenfeld, 1888), which Alvarez and Hooper (2009) suggested were indicative of a possible close relationship between Reniochalina and the Raspailiidae. This close relationship was further confirmed from molecular evidence (Erpenbeck et al. 2007b) showing affinities of R. stalagmitis with the raspaillid species Axechina raspailioides Hentschel (1912), indicating the strong morphological apomorphy of these ectosomal spicules for the Raspailiidae. Burton and Rao (1932) reported C. vickersii from South India (21 miles WSW from Mangalore), stating their specimen answered to Dendy's (1922) material. We were able to examine a slide made by Burton (BMNH 1931.1.1.19a, the specimen is presumably in the collections of the Indian Museum), and found it to be close but nevertheless distinct from C. vickersii proper. See below for a description and illustration, as Cyamon hamatum sp. n. Gray's (1867: 546) suggestion that the unnamed spicule without locality pictured in Bowerbank, 1864: figure 88 also belongs to C. vickersii is debatable as the spicule with its single cladus spined conforms more likely to Trikentrion. Description. Encrusting a sandstone flake accompanied by several other encrustations (position of holotype indicated by arrow in Fig. 4A). Lateral size of holotype approximately 4x3 cm, thickness up to 3 mm. Color red in life, light orange brown in alcohol. Surface irregularly grooved and venous. Consistency soft, easily damaged.

Description of new species from Mauritania
Skeletal structure: A basal mass of polyactine spicules pierced by erect single or bundled thick styles, alternated by long thin styles protruding beyond the surface. At the periphery, the long styles are surrounded by bouquets of thin (tylo-)styles. . Cyamon amphipolyactinum sp. n., holotype ZMA Por. 22412, A shape (arrow) encrusting a fragment of sandstone (scale 1 cm) B long thin style B1 details of apices of long thin style C short thin style C1 details of apices of short thin style D short thick styles showing size variation D1 detail of head of short thick style E polyactines (three-, four-, five-, and seven-claded) and one amphipolyactine showing size differences E1 detail of bulbous end of lateral cladus F amphipolyactines full-grown and spined (left) next to incipient smooth spicule (right).
Spicules: of five types, long thin styles, short thin styles, short thick styles, large polyactines and small double polyactines.
Large polyactines (Figs 4E, E1), in full-grown condition with all cladi ending in prominent, heavily spined knobs (Fig. 4E1) except one, the basal cladus, which is bluntly pointed. Cladi are less heavily spined towards the centre and at low magnification appear smooth. Growth stages may be partly or entirely without spines, but they are recognizable as unfinished by their irregularly undulating surface. The number of cladi varies between three and seven. In the largest spicules the cladi may be occasionally bifid. Basal cladi usually slightly shorter than the remaining cladi. Overall length of cladi regardless of condition is 18-51 × 3-10 µm.
Etymology. The name is an adjective that reflects the possession of unique small double polyactines, unprecedented in Cyamon and sponges in general. Distribution (Fig. 5). So far known only from the sandstone ridges of coastal flats of the Banc d'Arguin, Mauritania, West Africa.
Discussion. The new species stands out among all described Cyamon and Trikentrion species by having unique double micro-polyactines. Further striking characters of the new species are the prominent heavily spined bulbous knobs of the large po-lyactines, which are only similarly developed in Californian Cyamon koltuni Sim & Bakus, 1986, and the high frequency of five-claded polyactines, which has been to that extent reported only for C. quinqueradiatum (Carter, 1880) and C. koltuni. The structure of the skeleton and the overall spiculation is shared with the type species of the genus, Cyamon vickersii and its close relative Cyamon agnani. Differences are the sizes of the spicules and the less prominent bulbous knobs on the cladi of the polyactines in the latter two species.
Skeleton: columnar bundles of megascleres issuing from a basal layer of polyactines. Columns consist of a single long subtylostyle sheathed in a tight bundle of fusiform centrotylote styles; bundles separate, interconnected only near the substratum.
Spicules of three types: subtylostyles (assumed to be homologues of the long thin styles), centrotylote styles (assumed homologues of the short thin styles), polyactines (short thick styles apparently lacking).
Etymology. The name is an adjective referring to the type locality: the Mauritanian nature reserve Banc d'Arguin, one of the richest faunal areas of the west coasts of Africa (cf. Wolff et al. 1993). Distribution (Fig. 5). So far known only from the sandstone ridges of coastal flats of the Banc d' Arguin, Mauritania, West Africa.
Ecology. In shallow-water (12-18 m), highly sedimented environments, in the company of many other sand dwelling sponges such as Ciocalypta and Polymastia (cf. Van Soest, 1993: Pl. I fig. a).
Discussion. The single spined cladus of the polyactine spicules is an alleged feature of the genus Trikentrion, but in all other characters (growth form, monaxone spicules and skeletal arrangement) this is a typical Cyamon. It reminds strongly of Indian Ocean Cyamon quinqueradiatum, with which it shares the shape and up-per length of the subtylostyles, the lack of differentiated long and short thick styles, and the size and single cladus spination of the polyactines. Differences are the predominantly five-claded polyactines and the shape and size of the stylote spicules in C. quinqueradiatum. Long subtylostyles with prominent heads are shared with Indian Ocean C. quadriradiatum but that species has all the cladi of the polyactines densely spined.
The new species was collected in the same dredge sample as C. amphipolyactinum sp. n. (see above), but on a different sandstone flake (these provide hard substratum for sponges that would otherwise be buried in the sand). The two species differ sharply in the shape, size and ornamentation of the polyactines as well as in the shape and size of the styles. Remark. In view of the proposed major change in the status of Cyamon specimens reported from the Western Atlantic, description of the available material is presented in two sections, first the holotype of Cyamon agnani, subsequently other specimens known from the area and proposed to be assigned to C agnani. Description. Small hispid crust, color ochre. Detachable skin. The material borrowed from MNHN measured a few mm 2 encrusting a small piece of coral.
Skeleton: basal layer of polyactines, upon which megascleres are erected individually. Spicules: long thin styles, short thick styles, polyactines. Long thin styles, curved, variable in length, possibly in two size categories, but difficult to establish due to broken condition of most spicules, longest complete spicule 960 × 7 µm ( Fig 7A).
Short thin styles were not mentioned in Boury-Esnault (1973), but there were a few small broken styles and one complete spicule measuring 210 × 4 µm (Fig. 7C).
Discussion. The Cyamon nature of this material was previously detected by Mothes et al. (2004), who examined the present type material. Their conclusion was corroborated by Van Soest (2009) in his discussion of Timea species of the West Atlantic region. Mothes et al. (2004) proposed to assign T. agnani to the synonymy of Cyamon vickersii, but as explained above, that species differs in spiculation and geographic distribution. Despite the scanty available type material and the poor representation of short thin styles, it looks as if the categories, sizes and shapes of the spicules are broadly similar between the type of C. agnani and Caribbean and Carolinian specimens recorded as C. vickersii (see for details below). It is proposed here to consider all these Western Atlantic specimens as members of a widespread Cyamon agnani.   8A1), with hispid, bumpy surface (preserved condition). Size 3 × 2.5 cm in lateral expansion, 3-5 mm in thickness. Colour (alive) red, (alcohol) red-brown. Consistency soft. Skeleton (Fig. 8A2): basal mass of polyactine spicules penetrated by single short thick styles erect with heads embedded in the substrate. Long thin styles also erect on the substrate with rare short thin styles arranged around the peripheral protruding apices. This 'raspailid' feature was only observed in a few places.
Distribution. Greater Caribbean, Gulf of Mexico, South Carolina, N.E. Brazil. Ecology. Encrusting dead corals and other limestone substrates, 0-70 m. Discussion. Topsent (1889) records thinly encrusting specimens of the species under the name Trikentrion wickersi. This was apparently a common species on the Campeche Bank in the Mexican part of the Gulf of Mexico. His specimens were violet or blackish brown in color (preserved) and he observed that next to four-claded spicules also five-claded and three-claded occurred, though rarely. His drawings of the polyactines conform closely to those of our material, but no spicule sizes were given. Topsent (l.c.) believed that the similarities between Cyamon and Trikentrion were too great to keep them as separate genera, but his choice of Trikentrion as the valid name for the group is incorrect as Cyamon is the older name.
De Laubenfels (1950: 68, fig. 30) also reported the species from Bermuda (as Cyamon vickersi), depth not given. The specimen was probably not a Cyamon, because the drawings of the polyactine spicules appear to be rather those of a Timea aster with proliferated rays. The Bermuda occurrence must thus be considered suspect. Little (1963) recorded C. vickersii as an orange encrustation from the Gulf of Mexico, depth 11 m. His description is obviously copied from De Laubenfels (1936), as he gives exactly the same measurements of the spicules and also omitted to mention the short thick styles. We were able to examine this specimen, USNM 221078, thanks to a loan from the Smithsonian Institution. It has long thin styles 1050-1563 × 9 µm, short thin styles 330-345 × 2-3 µm, short thick styles 270-332 × 13-20 µm, polyactines (three-, four-, and five-claded) with basal cladi 36-60 × 7-12 µm and lateral cladi 33-61 × 7-10 µm.
We investigated the type material of Cyamon vickersii var. toxifera Arndt, 1927 (the name should be corrected to toxiferum to match the gender of the genus), ZMA Por. 00828, from Spaanse Water, Curaçao, and discovered that the toxas forming the basis of Arndt's variety are clearly foreign. They form part of the spiculation of a microcionid sponge, readily identified as Clathria (Microciona) ferrea (De Laubenfels, 1936 as Fisherispongia) by its characteristic polytylote subtylostyles (see also description of Curaçao material of that species in Van Soest, 1984). This discovery means that the name C. (M.) ferrea is threatened by Arndt's variety. The material is so scanty, that any trace of Cyamon polyactines has now (2012) disappeared from the sample. De Laubenfels (1936: 80) elevated Arndt's variety to specific rank; needless to say that this is unwarranted.
The spicule complement and the shape of the polyactines is broadly similar in the Brazilian type of C. agnani and specimens recorded from Caribbean and Carolinean waters as C. vickersii, but the latter may have long thin styles up to twice as long. The short thick styles and the polyactines also are on average clearly longer and more robust in Caribbean specimens. The geographic separation caused by the Amazonian outflow could be a barrier to gene flow between these shallow-water sponges, and the differences may thus have a genetic background. On the other hand, the Brazilian type material is only a single small specimen and variation in Brazilian waters may turn out to be as large as that in the Caribbean. Thus distribution and ecology for this species may be summarized as: tropical waters of Brazil, the Greater Caribbean and Gulf of Mexico, South Carolina, known from 0-70 m depth, usually encrusting dead corals and other limestone substrates. Description. The holotype is an encrusting sponge of 6 cm long and 3 cm wide growing over a haplosclerid sponge (Hentschel, 1912). The fragment of less than 0.5 × 0.5 cm and 1 mm in thickness (see Fig. 9A) examined by us was mixed with the haplosclerid in such a way that the microscopic slides were thoroughly contaminated with it. We have to rely on Hentschel's remarks about shape and surface characters. The surface is hispid due to the long styles protruding from the sponge, which was grey coloured in alcohol, but shows a pale brownish colour in our fragment. Consistency not mentioned by Hentschel, but crumbly describes it best.
Skeleton: the usual basal mass of polyactinal spicules upon which relatively long styles are erected surrounded in the periphery by bundles of thin centrotylote styles. Thick short styles are singly erect on the substrate, buried in the basal mass of polyactines.
Spicules: long thin styles, centrotylote thin styles, short thick styles, polyactines. Long thin styles (Figs 9B, B1), relatively rare, smooth, almost always broken in the slides so we cannot show a complete SEM image of them, heads smooth and not distinguished in width from the shaft, the other end gradually pointed. Longest style approximately 1620 × 16 µm, whereas Hentschel mentioned 1760 × 9-12 µm. Hentschel suggested a faint tyle, but we did not observe this.
Distribution. Only known from the Arafura Sea. Ecology. Deeper water on hard substrate. Discussion. The heavy spination of the polyactines appears to be a distinct feature of this species. Hooper's (1991) redescription denies the occurrence in this species of centrotylote ectosomal thin styles, wheras these spicules appeared common in the fragment of the holotype examined by us. These spicules are comparable to those of Cyamon arguinense sp. n., rather than to those of Cyamon vickersii because they do not have the characteristic crooked shape and also are not rugose at the pointed end. The polyactines of this species appear somewhat similar to those of Cyamon quadriradiatum as described and drawn by Carter (1880). However, details and sizes of the other spicules differ between the two: long styles are much longer and thinner in C. aruense and there is apparently no further category of short thick styles in C. quadriradiatum. Since both are ill known, we must have more data and further specimens to establish these species as distinct. The holotype (Fig. 10A1) was received on loan from the Smithsonian Institution, but in view of the small crust and previous studies of it, including SEM examination (Sim & Bakus, 1986: Fig. 3; Lee et al. 2007), and the presence of additional material, it was decided that no further sampling of it was necessary. We report the occurrence of a second specimen of this species from Mexican Pacific waters, from which we obtained our data for the description below.
Skeleton: a basal mass of polyactine spicules (Fig. 10B), upon which with styles are erected (Fig. 10C), no discernible skeletal organization due to thinness.
Distribution. Southern California, Pacific coast of Mexico.
Ecology. Under rocks and in caves in shallow water. Discussion. The enhanced bulbous endings of the polyactines is distinctive and is only matched by those of C. amphipolyactinum sp. n., but that species differs clearly by possessing a smaller category of amphipolyactines. It is generally similar to C. agnani, differing from that species in the sizes of the styles and the very prominent bulbous endings of the cladi of the polyactines.   11A) with irregular conulose-villose surface. Size of specimen 4 × 3 cm in lateral expansion, 2 cm in thickness. Color (alcohol) red brown.
Skeleton: columnar, with thick short styles at the center of a mass of polyactines, with long thin styles protruding from this skeleton surrounded by shorter centrotylote styles.
Distribution. Southern Californian Bight (San Pedro, Santa Catalina island, La Jolla). Ecology. On hard substrate, at depths 0-36 m. Discussion. Cyamon neon is unusual among Cyamon species by it possession of polyactines with smooth or barely spined cladi, the shape of many of the polyactines mimicking those of Trikentrion, and the occurrence of diactinal polyactines. The latter spicules are shared with Cyamon argon, which in most respects is similar to C. neon. For a comparison between the two species see below in the remarks to C. argon. The only other Cyamon species in the area is C. koltuni, which differs substantially in the bulbous endings of the cladi of the polyactines and absence of the short thin styles. Description. Shape upright, bilobed thick branches (Fig. 12A), spreading out upwards, with longitudinal grooves and covered in rounded spiny projections and conules. Height and diameter 3.5 cm, stalk approximately 1.5 cm. Colour (preserved) red-brown. Consistency tough, barely incompressible.
Skeleton: axial-columnar, with surface projections formed by the outwardly directed columns (Fig. 12B) branching off from the axial region. Columns have a core of short thick styles and polyactines crowned at the surface by long thin styles accompanied by (rare) short thin centrotylote styles.
Distribution. Pacific coast of North Mexico. Ecology. In kelp forest, 18-27 m. Discussion. As pointed out above, this species is close to Cyamon neon, and if more data on variation would become available, it is possible, in view of the nearness of both type localities that the two might be part of a single variable species. The following characteristics are similar between the two: long thin styles of 1000+ µm in length, the possession of short thin centrotylote styles with spined pointed apex (shared with C. vickersii), smooth evenly curved short thick styles of 400-500 µm in length, polyactines consisting predominantly of three-claded polyactines with all cladi smooth except for the apices, short basal cladus compared to long lateral cladi, and the frequent occurrence of diactinal polyactines. However, there are also clear differences, which presently preclude synonymization of the two: shape bush-like in C. argon, massively encrusting in C. neon, thickness of short thick styles in C. argon twice that of C. neon, basal cladi of the polyactines distinctly spined in C. argon whereas these are only rugose or even smooth in C. neon, and finally the size (length but also thickness) of the lateral cladi in three-claded polyactines which are usually well over 200 µm long and 20 µm thick in C. argon, whereas those of C. neon are on average around 150 × 10 µm.
Skeleton (Figs 14A-C): bundles of subtylostyles and styles standing erect on the substratum, in the basal layer supported by polyactine spicules.
Spicules: predominant spicules are longer and shorter subtylostyles with a minority of thin styles and polyactines. Subtylostyles, presumably a mixture of undifferentiated long thin styles and short thick styles, with prominent heads, usually lightly and gradually curved, in a large size range, which makes determining an average size meaningless: 129-1989 × 3-33 µm.
Thin styles, tapering gradually to thinly pointed curved ends, size range limited, 492-698 × 3-5 µm. Dendy believed these spicules to be growth stages of the subtylostyles, but we regard them, like Carter, as a separate spicule category.
Distribution. Only known from the Gulf of Manaar. Ecology. Deep water (not specified). Discussion. As pointed out above, Mauritanian Cyamon arguinense sp. n. shares many features with Indian Ocean C. quinqueradiatum, including the smooth lateral cladi and the lack of differentiation of the long thin and short thick styles. Although the Cyamon nature of this species has never been challenged, it is nevertheless obvious from the original description and drawing by Carter (1880) and the subsequent record of Dendy (1905) that the polyactines of this species have only their basal cladi spined, an alleged prominent and discriminating feature of the genus Trikentrion. We have confirmed single cladus spination by examining a series of slides of Dendy's material. The structure of the skeleton with longer and shorter styles originating from a basal mass of polyactines is characteristic for Cyamon. This indicates that emphasis on a single spined cladus versus all cladi spined as a difference between Cyamon and Trikentrion is wrong. See further discussion below. Among the species of Cyamon the present species also stands out by the extreme length variation of the structural subtylostyles, assuming these are homologous with the 'short thick styles' of many other Cyamon species, and perhaps related to it, the absence of a category of long thin styles. The thin styles observed above are assumed by their size to be homologous to the peripheral short thin styles surrounding the long thin styles in other species. Description. (From Carter, 1880). Thinly encrusting, hispid, color when dry dark brown. Spicules (Fig. 14E, left) of three kinds, long thick styles with a globular tyle, size given as 1042 × 41 µm, short thin 'crooked' styles, length 347 µm, and robust four-claded polyactines with all cladi entirely spined, length of cladus given as 76 µm.
Distribution. Gulf of Manaar, Southeastern India. Ecology. No data.
Discussion. This species needs redescription, but the long thick styles in combination with the densely spinous polyactines appear sufficiently distinct. Nevertheless there is a resemblance to the polyactines of Cyamon aruense, see above.

Description. partly from Burton and Rao 1931:
The single representative is a portion of a dull brown spherical mass. It agrees with the specimen described by Dendy (l.c.) except that the longest ray of the pseudactines bears a few recurved rays on the shaft and a crown of spines at the apex; the basal rays of these spicules have spines at the apex only; and the styli are very scarce. Locality. -21 miles S.W.W. of Mangalore, S India (4 th May 1888).
The slide (Fig. 15A) contains thick sections of the skeleton, showing the usual columnar structure of thick styles and polyactines (Fig. 15B). The slide allows the recognition and measurement of the spicule complement.
Short thin styles, wavy outline, faintly centrotylote, under light microscopy mostly looking smooth but occasionally some spines are visible on the pointed end and also in at least one spicule two spines on the rounded end, 272-313.2-355 × 2.5-3.4-5 µm Short thick styles, smooth, curved rather strongly near the rounded end: 421-495.6-604 × 16-19.9-31 µm.
Etymology. The adjective hamatus (L.), means provided with hooks. Distribution. South India. Ecology. No data. Discussion. It is with some hesitation that we decided to name this scanty material as a valid new species. Although measurements of the megascleres conform to or are close to those of C. vickersii, the shape and spination of the polyactines is distinctly different, as Burton & Rao already observed. With their strong hooks on the basal cladi and the peculiar short crowned lateral cladi the polyactines are different from any other known Cyamon.   (Topsent, 1904), ZMA Por. 19422, from SE Rockall Bank, North Atlantic A shape encrusting deep sea coral (scale = 1 cm) B short style, lightly spined at the head B1 details of apices of short style C long style, coarsely spined C1 detail of head of long style D various shapes of polyactines E detail of the cladome of a seven-claded polyactine.
Discussion. This is a deviating Cyamon with several unique features not shared by the majority of the species. Both monaxone megascleres are partially heavily spined, and the raspailiid feature of a long thin style surrounded by short thin styles is absent. The polyactines resemble echinating acanthostyles by their long basal cladus and crown of short irregular lateral cladi. These spicules may be assumed to bridge the gap between the polyactines with more or less equal length cladi and acanthostyles with heavily knobbed and spined heads such as found in some myxilline genera (Hymedesmia Bowerbank, 1864, Discorhabdella Dendy, 1924 and in the raspailiid genus Eurypon Gray, 1867. Additionally, it occurs in cold deep-sea habitats unlike all other Cyamon species. It is likely that this species does not belong in Cyamon, but we will await additional (molecular) evidence before removing it from the genus.
Definition (emended). Cyamoninae with reticulate skeleton containing polyactine spicules of which the basal cladi are provided with hook-like spines in mature condition, and if present choanosomal oxeas. Microscleres trichodragmas. Additional longer and shorter thin styles are usually present in peripheral regions.
Remarks. Polyactine spicules are genuinely polyaxone, with axial canals visible in all of the predominantly three, occasionally four-or two cladi. As will be demonstrated below, none of the specimens of the type species we were able to examine, including the neotype, possess the raspailiid synapomorphy of peripheral long styles surrounded by short styles, despite Hooper's (2002) description of the type species where such spicules were mentioned. Possibly, but unlikely, these spicules are present in living condition, because we only had dry old specimens available and the peripheral skeleton may have become abraded. It seems likely that Hooper's (2002) description was based on a contaminated spicule slide. All other Trikentrion species do have the long and short styles as a peripheral skeletal feature, and in that sense the type species appears a deviating representative of the genus.
Trikentrion differs from Cyamon in its possession of choanosomal oxeas (whereas Cyamon has styles), but several species, T. catalina, T. helium Dickinson, 1945 and T. africanum sp. n., are lacking these spicules. The polyactines of Trikentrion differ from those of Cyamon in having only the basal clade provided with strong hook-like spines, with the lateral cladi smooth; also the shape is often Y-or T-shaped. As demonstrated above, these differences are not entirely exclusive, because Cyamon arguinense sp. n. and C. quinqueradiatum also have only the basal cladus spined, whereas Y-and T-shaped polyactines occur in C. neon and C. argon. Finally, all species of Trikentrion described below have abundant trichodragmas, which are entirely lacking in Cyamon species.

Remarks.
The identity of the sponge named Spongia muricata by Pallas, which is assumed to be the type species of Trikentrion, is not straightforward. The first use of the name combination stems from Linnaeus (1759: 1348), who described it as: S. ramosissima, poris cylindricis subulatis prominentibus aequalibus multifidis hispidis, without further indication of where it had been collected or by whom. The Latin name muricata is generally considered to mean spined (after the name of a mollusk (Murex) yielding a purple dye, cf. Brown 1985), for sponges a hardly distinguishing feature. The description speaks of cylindrical pores, which is quite vague, and this character does not occur in any specimen discussed in this paragraph and below. Pallas (1766: 389), employed the name combination also, but indicated and described the sponge figured in Seba's (1734Seba's ( -1765 volume 3 pl. 99 fig. 7 as representing his Spongia muricata. Pallas did not refer to Linnaeus' name, nor did his description remind in any aspect of Linnaeus' description. Seba's figure is here reproduced in Fig  17A, and Pallas's description in Fig. 17C. Pallas also quoted Elmina on the coast of Guinea (now Ghana) as the locality of the specimen based on Seba's information.
In the same year (or perhaps one year before), Ellis (1765Ellis ( -1766, pictured a similar sponge (a branched tuberculated sponge here reproduced in Fig. 17B), stating that it originated from the Cape Coast Castle in Africa (which could very well be the same locality Elmina), but not naming it. In his 1767 edition, Linnaeus again described Spongia muricata, replacing the first word of the 1759 edition, S. ramosissima by the text S. foraminulata ramosissima angulata tenax, followed by the same words as previously (reaffirming the unrecognizable shape of the sponge). He also added that it originated from O. Aethiopico (Indian Ocean). He now gave three sources for his record of this species, viz. Mus. Tessin 118, plate II figure 1, Seba's volume 3 plate 99 fig. 7, and Pallas' record. Finally, Gmelin in Linnaeus, 1788: 3821, admits that the species is from Guineae littorea, quoting a.o. Pallas (1766) and Linnaeus (1767), but remarkably omitted any reference to Linnaeus (1759). Linnaeus' and Pallas' (Seba's) specimens of Spongia muricata have never been identified in later collections (but see below), and their identity remains a matter of speculation. In 1794, Esper extensively described Spongia muricata and his figure is here reproduced in Fig. 17D. This time, the specimen, stated to be from Guinea, from cliffs near Elmina (= Ghana), was still extant in the collections of the University of Erlangen (Germany) in 1870, when Ehlers revised some of Esper's specimens (Ehlers, 1870). He detected the triactine spicules and erected the genus Trikentrion for it. His description included some measurements of the spicules: oxeas 354-414 × 16 µm, and polyactines, with basal cladi 95 µm and lateral cladi 72 µm, which data conform closely with those presented below for the species. However, since Ehlers' redescription, the whereabouts of the Esper material is unknown and it must be assumed lost. Lamarck (1814: 448) misinterpreted Spongia muricata and his material was assigned to Raspailia hispida (Montagu, 1818) by Topsent (1932: 107). Possibly, Spongia echidnea Lamarck, 1814 is a junior synonym of Spongia muricata Pallas, because the redescription and figured specimen of Topsent (1932: 108, as Ptilocaulis echidnaeus) reminds rather strongly of it. However, Topsent fails to mention the presence of polyactine spicules.
To conclude: the identity of Spongia muricata is not unequivocal, primarily due to the unrecognizable description of Linnaeus (1759) and the likelihood that he used the name for an unknown species from the Indian Ocean. Pallas' description in combination with Seba's figure make it likely that his Spongia muricata indeed is what we now know as Trikentrion muricatum, but uncertainty reigns due to the fact that only Esper's, not Pallas', material was shown to possess the synapomorphy of the polyactine spicules. It appears highly necessary to fix Spongia muricata as a Trikentrion, by assigning a neotype. In the absence of any topotypical fresh material of the species we are forced to choose dry old collection material.
A likely candidate is the assumed type of Trikentrion muricata housed in the Natural History Museum, London, BMNH 1872.10.19.1 (see Fig. 18A), with schizotype ZMB 7160, on the basis of which Carter (1879) redescribed and illustrated the species Trikentrion muricatum, and which subsequently formed the basis of the Systema Porifera entry of the genus and its type species. This is not likely to be Seba's specimen, nor Esper's because the locality data (though from Ghana as well) do not indicate Elmina. In addition to this specimen, the Natural History Museum collections incorporate a schizotype of Spongia muricata, Coast of Guinea, BMNH 1954.2.20.93, which appears unimportant for the present choice of neotype because it is not a Trikentrion, but an unidentified species of Axinella Schmidt, 1862.  (Pallas, 1766) The choice of a neotype again is complicated due to a recent discovery in the collections of the Naturalis Biodiversity Center at Leiden (NBC) of four old collection specimens, RMNH Por. 306 and 309, and ZMA Por. 02545 and 02546, which are sufficiently similar to Seba's and Esper's plates to raise the suspicion that they could belong to one of the original specimens of Spongia muricata.
RMNH Por. 309 (see Fig. 18B) is labeled Raspailia xerampelina (Lmk) ? type (Spongia ---Lmk) without further information, and this specimen bears an overall strong likeness to Seba's plate. RMNH Por. 306 (see Fig. 18C) is labeled Raspailia hispida (Mont.) type van Spongia muricata Lmk, Mus. Parijs, Kust van Guinée (translation: type of Spongia muricata Lamarck, from the Paris Museum, Coast of Guinea). If the specimen is compared to the plate of Spongia muricata of Esper one is compelled by the overall likeness of the two (though it is not an exact likeness). ZMA Por. 02545 (see Fig. 18D The skeleton and spicules of all five specimens conform with the descriptions of Ehlers (1870) and Carter (1879).
The reason for the names on the labels of the specimens of the NBC and the referral to the Paris Museum is explained in Holthuis (1995): during the French occupation in 1795 of the Republic of Holland in the Napoleontic period, Dutch collections were confiscated and relocated to the Paris Museum. Some time after the end of the emperorship of Napoleon in 1815, negotiations between The Netherlands and France resulted in a donation of specimens, notably duplicates from Lamarck's collection, to the then founded Rijksmuseum of Natural History at Leiden. Dozens of sponge specimens labeled with Lamarck's names are incorporated in the RMNH collections, but because the redescription of Lamarck's sponges by Topsent (1931Topsent ( , 1932Topsent ( , 1933 was initiated after the transfer of specimens to Leiden, there is often little correspondence between the identities of the MNHN and RMNH specimens bearing labels with the same original Lamarck names. Topsent (1932) identified Lamarck's Spongia muricata as Raspailia hispida, and this was duly taken over by past curators of the Leiden specimens, who apparently were unaware of the discrepancies between the Paris and Leiden specimens. It is possible, that the Lamarck specimen redescribed in Topsent (1932 as Raspailia hispida), is not Lamarck's original specimen, because this may have ended up in the Leiden or Amsterdam collections.
In view of the uncertain history of the NCB specimens and the more precise data available for the Natural History Museum, London specimen, we here designate BMNH 1872.10.19.1 as the neotype of Spongia muricata, the type species of the genus Trikentrion.
It is a pleasure to be able to announce that material of Plectronella papillosa Sollas, 1879, since long known to be a junior synonym of Trikentrion muricatum through its excellent description by Sollas, but otherwise never redescribed, has been discovered in the collection of the Bristol Museum and Art Gallery, in the form of 2 slides labeled No. 30 Ah.200.1,200.3 (see Fig. 19B), containing cross sections of the skeleton and dissoluted spicules. We can confirm that P. papillosa is a junior synonym and that details in the slides conform closely to those of T. muricatum (see Fig. 19C-D).   Description. Wide basal holdfast upon which are erected groups of cylindrical branches, more or less in one plane, each branch usually with one or two dichotomous secondary branches, often also with anastomosing branches. Size of neotype (Fig. 18A) 13.5 × 12 × 5 cm of the whole group of branches, diameter of individual branches 1-1.5 cm. Sollas' specimens ( Fig. 19A as Plectronella papillosa) were described as being 20 × 20 cm, with branch diameter 2-3 cm. The other specimens are similar in size, but slightly smaller. Surface densely covered with broad, laterally flattened papillae, 1-4 mm in size (reminding of the surface projections of Ptilocaulis Carter, 1883). In some specimens the papillae are partially abraded (e.g. RMNH Por. 306, see Fig. 18C) giving the sponge a less striking aspect. Consistency (dry) hard, incompressible, crumbly. No live color has been reported in the literature, but color plates of Seba (Fig. 17A) and Esper (Fig. 17D) show a light orange brown color. Skeleton (Fig. 20A): predominantly a wide-meshed reticulation of tracts of robust smooth oxeas, with little axial and extra-axial specialization. The polyactines are common in peripheral regions. No longer or smaller peripheral styles have been found in any of the examined specimens.
Ecology. Depth range: no definite data, but probably shallow water, growing on rocks. Discussion. The species must have been of common occurrence off the coast of Ghana in 18 th century as there are a fair lot of specimens available from that age and region in several natural history museums. Curiously, no fresh material is known to exist, so the species remains ill-known. T. muricatum differs substantially from all other Trikentrion species described below in the lack of peripheral styles. Further differences are robust oxeas, up to twice as long and thick as those of the two other oxea-bearing species (Trikentrion laeve Carter, 1879 and T. flabelliforme), while the three remaining species (Trikentrion helium, T. catalina and T. africanum sp. n.) lack the oxeas entirely. Description. Multi-branched bush (Fig. 21A), with single stalk of 1.5 cm high, 0.8 cm diameter, from which cylindrical branches spread out dichotomously, ending in approximately 26 smaller terminally rounded branches. Size of entire specimen, which is broken in two unequal parts, 4.5 × 5.5 × 3 cm. Surface optically smooth, but microhispid, with punctate appearance. Consistency (dry) crumbly compressible, colour beige-purplish.
Discussion. This is the first redescription after Carter's report, which is accurate but deficient in omitting the trichodragmas and short thin styles. This is also the first depiction of habit of the specimen and with the details provided here the species is now at least properly described, but it remains ill known. Carter (1879) differentiated this species from Trikentrion muricatum by emphasizing the presence of ectosomal long styles, which appear lacking in T. muricatum. The shape and surface characteristics of the two are also quite different, and sizes of the oxeas and polyactines are considerably smaller than in T. muricatum. Carter (1882) reported this species from Australia, but from his description it is clear that it concerns the species later described as Trikentrion flabelliforme Hentschel, 1912. The two differ significantly in shape (T. flabelliforme being thinner or thicker bladed, lacking rounded branches forming a three-dimensional bush). The three other species of Trikentrion differ by lacking oxeas. Burton (1948) reported this species from the République du Congo, more to the south, but this specimen lacks oxeas and has a different shape. It is assigned to a new species (Trikentrion africanum sp. n.) below.  , 5.4134°S, 134.6677°E, depth 13 m, 23-12-1899. RMNH Por. 978, preserved in alcohol, Siboga Exped. Stat. 273, same data;ZMA Por. 14022 and 14023, preserved in alcohol, East Point, Darwin, Northern Territories, Australia, 10 m, 29-11-1987, coll. J.N.A. Hooper nrs 8 and 9; ZMA Por. 16049, dry old collection material without data. Description. Two distinct shapes, flabelliform (Fig. 22A), 6-26 × 4-19 cm high and wide, 0.2-1.4 cm thick) and digitate (Fig. 22B), up to 15 cm high, with flattened branches of up to 1.5 cm thickness (summary of many specimens described in Hooper, 1991). Flabelliform specimens may have blades at right angles (see Fig 22A). Frequently, the digitate specimens are infested with zoanthids (Fig.  22B). Surface optically smooth, microhispid, with characteristic pattern of fine meandering grooves. Texture firm. Colour orange-red, blood-red (shallow water) to beige (deep water).  (Fig. 22D), with loose extra-axial and spongin-rich axial spicule tracts cored by oxeas, echinated by triactine polyactines; at the surface protruding long thin styles are surrounded by bouquets of short thin styles (Fig. 22C).
Distribution. Arafura Sea, N and W Australia. Ecology. Shallow subtidal to offshore deeper water. Discussion. The species was erroneously attributed to Carter 1882: 294, allegedly as Trikentrion laeve var. flabelliforme, by Hooper (1991). This is a manuscript name because Carter did not name his Australian flabellate specimen, merely referring it to T. laeve. That species is West African and described above as a distinct new species.
The two 'growth forms' are rather distinct, but distribution, skeleton, and spicules are similar and overlapping enitirely, making it impossible to separate the forms further. The digitate form is often overgrown with a zoanthid species, both in Australian (Hooper, 1991) and Indonesian (RMNH Por. 978) specimens. The shape of T. flabelliforme reminds of Californian Trikentrion catalina and T. helium, but spiculation in these species differs substantially by their lack of proper choanosomal oxeas. Comparative variation in shape is also recorded for T. helium (see below).
The apices of the oxeas and the polyactines show minute spines, which is here interpreted as a unique feature. It violates the rule that in Trikentrion only the basal, not the lateral cladi of the polyactines have spines, but there is little correspondence with the lateral cladus spination in Cyamon.
This is the only Trikentrion species that appears to be widespread and common. Chemistry of T. flabelliforme includes unique indoles (Capon et al. 1986).
Distribution. The holotype was collected in the Southern Californian Bight (Mexican Pacific). Luke (1998) records several specimens from La Jolla, California (USA). If specimens of Gómez et al. (2002) belong to this species, it occurs in the Sea of Cortez and further south along the Mexican Pacific coast.
Ecology. Rocks and reefs at depths of 15-28 m. Discussion. The trichodragmas were not cited in the original description. T. helium shares the dominance of three-claded polyactines with relatively long lateral cladi with Trikentrion catalina (see below), to which it seems closely related. This species differs quite strongly from the other Trikentrion species by its possession of numerous diactinal or style-like reduced polyactines, which resemble, but clearly are not proper, oxeas like those of Trikentrion muricatum and T. flabelliforme. The spicules are recognizable as polyactines by the substantial difference between the smoothly rounded end, resembling the ends of the lateral cladi of the three-claded polyactines, and the dissimilar pointed end which shows an irregular surface and is connected to the other end by a swollen, often irregular middle part. Their lengths coincide with the added lengths of a lateral and a basal clade of the three-claded forms. Such reduced diactinal polyactines are also common in Cyamon neon.
The specimens described by Gómez et al. (2002) under the name Trikentrion catalina were branching erect rather than bladed, but branches were typically flattened, 2-4 mm in thickness. We reassign these specimens to T. helium, because they apparently possess oxea-like polyactines [described as oxeas but confirmed as reduced polyactines by one of us (JLC)], whereas in T. catalina there are neither oxeas forming the main skeleton as in T. flabelliforme nor diactinal polyactines as in T. helium. The difference in shape between the type of T. helium and Gómez et al.'s specimens is here considered to be mere variation (comparable to variation in T. flabelliforme, see above) but further studies might reveal there is more specific diversity along the Pacific coast of Mexico. Description. Flabelliform sponge (Fig. 25A), measuring 15 × 8 by 0.4 cm, attached to rocks by a 3 × 0.6 cm stalk. Surface hispid. No oscules apparent. Consistency firm and leathery. Color reddish orange (alive), pale beige (alcohol). Skeleton (Fig. 25B): choanosome densely packed with three-claded polyactines; ectosome with long, relatively thick styles surrounded by dense bouquets of short thin styles; trichodragmata commonly observed especially in the peripheral parts.
Discussion. This species is assigned to Trikentrion, because of the flabellate shape resembling T. flabelliforme Hentschel (1912), the sagittal polyactines, three-claded with spines only on the basal cladus, and the possession of trichodragmas. It is similar to T. helium in the lack of genuine choanosomal oxeas, and the short basal cladi of the polyactines. Remarkably, when describing T. catalina, Sim & Bakus (1986) did not notice -they did not discuss T. helium -the similarities with their species. Specimens assigned to T. catalina by Gómez et al. (2002) are considered to be long to T. helium (see above). Description. Upright flattened branch with two or three short side projections (Fig. 26A), with wider base and a cut-off upper ending, possibly the specimen is only a fragment as base and apex look damaged. Length of holotype 6.5 cm, diameter 1.5 cm at the base, 1 cm higher up. Side projections only on one side of the branch, less than 1 cm long and 0.5 cm thick, with rounded apex. Surface uneven, somewhat hispid. No apparent oscules. Consistency firm. Colour (alcohol) red-brown.
Skeleton: a dense mass of polyactines, towards the periphery surrounding long thin styles and short thin styles, which are embedded in the skeleton more so than in other Trikentrion species. No oxeas present, but T-shaped polyactines with very short basal clade appear to have taken the position of oxeas.
Etymology. The name is anadjective referring to the type locality. Distribution. République du Congo. Ecology. Shallow water Figure 26. Trikentrion africanum sp. n., holotype BMNH 1939.2.20.9, A shape of holotype (scale bar = 1 cm) B details of long thin style C short thin style C1 details of short thin style D various shapes of polyactines E trichodragma E1 detail of trichodragma.
Discussion. Burton (1948Burton ( , 1956 assigned this material to the relatively unknown species Trikentrion laeve Carter without any morphological information. This is obviously wrong, a.o. because that species has abundant oxea megascleres, lacking in the present material. Carter's T. laeve was expressly differentiated from Trikentrion muricatum in its possession of long thin styles, which are indeed absent in T. muricatum. Both T. muricatum and T. laeve were described and illustrated by Carter to have a strong complement of oxeas (see also above). Their function appears to have been entirely taken over by the polyactine spicules in the present material.
The lack of choanosomal genuine oxeas is shared with Californian Trikentrion catalina and T. helium, but these species have flabelliform or bladed shape and much larger polyactine spicules.

Key to the species of Cyamon and Trikentrion
Below the species of Cyamon and Trikentrion considered valid are keyed out. See Table  1 for a summary of recognized species and Table 2 for a summary of their characters.

Geographic distribution of species of Cyamon and Trikentrion
With the new records from Mauritania, South Carolina and the reassigned Brazil record, the genus Cyamon appears to have a circumglobal warmer water distribution (Fig. 27), commonly observed in many shallow-water sponges (Van Soest, 1994b;Van Soest et al. 2012). Gaps in this distribution appear to be the NW Pacific (Japanese and Chinese waters) and the SW Pacific (Australian and New Zealand waters), and the absence in the Mediterranean is also noteworthy. An odd outlier occurrence is that of Cyamon spinispinosum (bathyal North Atlantic), while the concentrated occurrence in the tropical East Atlantic and the warm temperate North East Pacific (three species each) is striking. Species assignable to the genus Trikentrion are also found in the warmer waters of all three oceans (Fig. 28), but so far the genus is not recorded from the Central West  Atlantic. In contrast, West African waters appear to have a concentrated occurrence of Trikentrion species.
It is likely that more species of both genera will be discovered in the near future.

Discussion
The two genera were independently erected contemporarily (1867 vs 1870), but at first Cyamon was ignored (Higgin, 1877;Carter, 1879). Topsent (1889) attempted to synonymize the two on account of the polyactine spicules characterizing both, but he picked the junior name Trikentrion. Most subsequent authors kept the two genera as separate and in the latest comprehensive classification of the Porifera (Hooper & Van Soest, 2002) this was maintained. Different authors were not consistent in outlining the differences between the two. Mostly, the emphasis was laid on the single vs the overall spination of the cladi of the polyactines. Other features variously indicated as differences, such as growth form (encrusting vs erect), skeleton (plumose vs. reticulate) and choanosomal megascleres (styles vs oxeas) need critical reexamination in recorded specimens. Based on specimens described here as belonging to Cyamon species and Trikentrion species, we constructed a matrix of characters found in both putative genera (Table 3). From this table it is apparent that four features appear to be more or less consistently different between the two groups. (a) Shape, with the majority of Cyamon thickly or thinly encrusting, whereas only in two species the shape is rather more massive or lobate (the type, C. vickersii and C. argon), with Trikentrion never thinly encrusting, always erect and usually elaborate, thick branches or flabellate. In conclusion: this appears to be a fairly consistent difference, although shape is variable and probably not operational for a clear distinction. We do not consider the skeletal structure, plumose in Cyamon and reticulate in Trikentrion, as an independent character, but assume here that elaborate shape can only be achieved by reticulate organization of the skeleton. (b) The choanosomal megascleres, with Cyamon having thick, mostly short, terminally curved styles, and Trikentrion thick short oxeas if present (not present in T. helium, T. catalina and T. africanum sp. n.). The thick short oxeas and styles in the two genera could be expressions of the same spicule type, as both are similar in dimensions and tend to be entirely smooth (except Cyamon spinispinosum). Nevertheless, the conclusion is that possession of choanosomal 'true' oxeas distinguishes Trikentrion from Cyamon. (c) The presence (Trikentrion) or absence (Cyamon) of trichodragmas. This appears so far a clear and absolute difference between the two. (d) The spined cladi of the polyactines differ mostly also in spination: coarse recurved spines in Trikentrion, finer spined in Cyamon (excepting C. arguinense sp. n. and C. hamatum sp. n.). Most Cyamon species have spines on all cladi, whereas Trikentrion polyactines have spines only on the basal clade (or are entirely smooth). However, there is no absolute distinction because C. arguinense sp. n. and to a lesser extent C. quinqueradiatum, both with only spines on the basal cladi, bridge the gap between the polyactines of Cyamon and Trikentrion. Also, the diactinal polyactines of Cyamon neon and Trikentrion helium appear quite similar. Possibly, the position of the polyactines in the skeleton is different in the two genera: usually a basal or central concentration of these spicules in Cyamon and more peripheral or scattered throughout in Trikentrion, but more observations are necessary to confirm this feature.
The isolated occurrence of such unusual polyactine spicules in two genera that are otherwise likely to belong to monactine raspailiids could be interpreted as support for Uriz & Maldonado's (1995) hypothesis -based on research of crambeid sponges -that monaxone spicules have evolved from ancestral polyaxones by reduction. Circumstantial evidence (different shape and spination and different position in the skeleton) points towards the possibility that the polyactines of the two genera have a different evolutionary origin: Cyamon species could have derived their polyactines from styles, or perhaps acanthostyles (as is suggested by the polyactines of C. spinispinosum), whereas Trikentrion polyactines could have been derived from choanosomal oxeas. This would mean that the two genera do not share a common ancestor not shared also by other raspailiid groups and the subfamily Cyamoninae would then be artificial.
On the basis of the current state of our knowledge, with, for example, a compelling similarity of polyactines of Cyamon arguinense sp. n. and Trikentrion catalina (compare Figs 6D and 25D, lower right), such a hypothesis lacks sufficient support, and likewise cannot yet be interpreted as support for Uriz & Maldonado's (1995) theory. Phylogenetic relationships based on DNA sequence information for the present genera are still tentative. Erpenbeck et al. (2007b) confirmed that Trikentrion flabelliforme is a member of a restricted Raspailiidae clade, but is not clearly differentiated from other genera. We will have to await further sequence analysis, which could help to answer the questions whether Cyamon and Trikentrion are non-monophyletic and whether Cyamon spinispinosum is really a Cyamon.