A new deep-water Tethya (Porifera, Tethyida, Tethyidae) from the Great Australian Bight and an updated Tethyida phylogeny

1 South Australian Museum, North Tce, Adelaide, South Australia; SARDI Aquatic Sciences, 2 Hamra Ave, West Beach, South Australia. 2 Queensland Museum, Southbank, Queensland, Australia; School of Biological Sciences, University of Queensland, St Lucia, Queensland, Australia. 3 Center for Marine Drugs, State Key Laboratory of Oncogene and Related Genes, Department of Pharmacy, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China; Centre for Marine Bioproducts Development, College of Medicine and Public Health, Flinders University, South Australia. 4 Pharmacognosy, Department of Medicinal Chemistry, BMC, Uppsala University, Uppsala, Sweden; Institut Méditerranéen de Biodiversité et d’Ecologie marine et continentale, CNRS, Aix Marseille Université, IRD, Avignon Univ., Station Marine d’Endoume, chemin de la Batterie des Lions, 13007 Marseille, France.


Introduction
The family Tethyidae Gray, 1848, reclassified from the order Hadromerida Topsent, 1894 (order abandoned) to order Tethyida Morrow & Cárdenas, 2015, contains 14 genera of which the genus Tethya Lamarck, 1815 is by far the most speciose with 93 accepted species . Tethyidae are defined by 'stylote megascleres mainly stronglyoxeas, generally in radiate tracts and two categories of euasterose microcleres, micrasters and megasters, sometimes rhadbs' (Sarà 2002). Occurrence and length of a stalk are included as key morphological characteristics in distinguishing some genera of Tethyidae (Sarà 1994(Sarà , 2002Sarà & Burlando 1994); although the development of the stalk in the family has been noted as being an adaptation to deep-water, soft sediment habitat (Sarà & Burlando 1994). The taxonomy of the Tethyidae was rearranged by Sarà (1994) into eight genera, with the new genera Burtonitethya Sarà, 1994 andTethycometes Sarà, 1994 joining Halicometes Topsent, 1898, as stalked genera in the family. Cladistic analysis by Sarà & Burlando (1994) placed Tethyidae into three clades (a) the genus Tethya, (b) genera with stalks and (c) massive and encrusting genera.
The slope of the Great Australian Bight (GAB) was explored in 2010 as part of a preliminary exploration of deep-water communities (Currie & Sorokin 2011) and again in 2015 as part of systematic benthic surveys to understand unexplored communities in the light of current oil and gas exploration activity in the area (MNF 2015;Williams et al. 2018). The surveys resulted in the discovery of several new benthic species including sponges. Multiple specimens of a small stalked tethyid were found at 1000 m; here, we describe this new species, using morphological characters and molecular markers (COI, 28S (D3-D5) and ITS1-5.8S-ITS2) to consider how it fits into the family.
The description of this new species is an opportunity to revisit the molecular phylogeny of this group. Heim et al. (2007) and Heim & Nickel (2010) produced the first phylogenetic analyses of Tethya, combining COI and morphology. Since then, although large Demospongiae trees were produced that led to the creation of the order Tethyida by Morrow & Cárdenas (2015), no phylogenetic studies truly focused on the Tethya, Tethyidae or Tethyida. We therefore felt it was time to present an updated molecular phylogeny for COI and ran the first comprehensive 28S phylogenetic analyses focusing on the Tethyida.

Field collection
Tethya irisae sp. nov. was first collected, as only one specimen, in the GAB in 2010 (Currie & Sorokin 2011). More specimens were collected in November and December 2015 as part of systematic epibenthic SOROKIN S.J. et al., A new deep-water Tethya (Porifera: Demospongiae) 3 surveys of the central GAB slope. Stations were sampled by beam trawl along five longitudinal transects and over depths ranges from 200 m to 5000 m (Fig. 1). Benthic specimens were collected under Australian Commonwealth Area Permit No. AUCOM2015-284 and Commonwealth Marine Reserve Permit No. CMR-15-000344. Tethya irisae sp. nov. was found at six out of eight of the 1000 m depth stations. Specimens were photographed on board and fixed in 70% ethanol. Specimens for molecular analysis were fixed in ethanol (> 95%). Most specimens (including the holotype) are lodged at the South Australian Museum (accession prefix SAMA), Adelaide, South Australia; two specimens were deposited at the Queensland Museum (accession prefix QM), Brisbane, Australia; four specimens from lot SAMA S2039 and a thick section are deposited at the Museum of Evolution, Uppsala, Sweden (UPSZTY 178608). Collection information of specimens examined in this study is archived in the open access PANGAEA data repository (https://doi.pangaea.de/10.1594/PANGAEA.894720).

Light microscope preparation
To examine the skeleton, thick sections in resin were prepared from a specimen from collection lot SAMA S2039, following the method described by Boury-Esnault et al. (2002). Thin sections were also  (Williams et al. 2018). Cruise SS2010_T02 (Currie & Sorokin 2011). Taxonomy 529: 1-26 (2019)   4 made, with the following protocol: sections were cut perpendicular to the surface with a sharp blade and laid onto a slide, covered with a weighted coverslip and dried on a hot plate. Sections were covered with mounting media (Durcupan™) and dried overnight in an oven at 50 o C. Spicule slides were prepared by dissolving a small amount (~ 2 mm 2 ) of sponge tissue in 3.9% sodium hypochlorite. The resultant spicules were rinsed with distilled water three times and with 95% ethanol twice before mounting on a microscope slide with DPX™ mountant.

SEM preparation
Scanning Electron Microscope (SEM) tissue preparations were made by dissolving the tissue in 12.5% sodium hypochlorite to remove the soft tissue. They were then rinsed twice in distilled water, rinsed twice in 70% ethanol and then finally twice in 98% ethanol and then air dried. SEM preparations were sputter coated in gold to improve resolution. The scanning electron micrograph photos were taken using a Hitachi TM-1000 SEM and plates assembled in Adobe Photoshop. Morphometric measurements of the spicules were done using the same Hitachi TM-1000 SEM.

Spicule terminology
Spicule terminology follows that suggested by Bergquist & Kelly-Borges (1991) for the genus Tethya.

Molecular studies
Whole genomic DNA was extracted from sponge tissue frozen at -80°C. A conventional hexadecyltrimethylammonium bromide (CTAB)-based protocol (Taylor et al. 2004) was used for isolating DNA. Briefly, the sponge tissues were ground under liquid nitrogen. The CTAB extraction buffer was applied to lyse tissues and then combined with polyvinylpyrrolidone (PVP) and β-mercaptoethanol to help remove phenolic compounds and tannins in the extract. To separate the proteins and polysaccharides from nucleic acids, phenol: chloroform: isoamyl alcohol (25:24:1) was utilised before DNA was precipitated with chilled isopropanol. The mitochondrial cytochrome c oxidase subunit 1 (COI) Folmer fragment was amplified using the universal primers LCO1490 and HCO2198 (Folmer et al. 1994). The D3-D5 region of 28S rRNA gene was amplified by primers NL4F and NL4R (Nichols 2005). To amplify ITS, we used primers originally designed for a unicellular eukaryote, ITSRA2 (5′-GTC CCT GCC CTT TGT ACA CA-3′) and ITS2.2 (5′-CCT GGT TAG TTT CTT TTC CTC CGC-3′) (Adlard & Lester 1995), to amplify a 753 bp long sequence including ITS1-5.8S-ITS2 and small fragments of the surrounding 28S and 18S. The thermocycler was programmed following Yang et al. (2017). Sequences were assembled and blasted using Geneious® ver. 8.1 (created by Biomatters, http://www.geneious.com).

Etymology
Named after the golden-winged Greek goddess Iris, grandchild of Tethys, who could reach all parts of the cosmos including the deep sea; and in memory of marine naturalist Iris Sorokin.

DNA barcoding
COI (MH518072), 28S (D3-D5) (MH511148), ITS1-5.8S-ITS2 (MH511149). All sequences came from the same individual from lot SAMA S2913, although this is a different individual than the type specimens. A tissue sample from this voucher is deposited at the Australian Biological Tissue Collection at the South Australian Museum, Adelaide (ABTC145318).

Description
A small, spherical to oval, stalked, sponge (Fig. 2). The sponge body is 11-14 mm diam., with the stalk approximately the same length as the diameter of the sponge. The surface is covered in polygonal platelike tubercules (2-3 mm diam.) separated by grooves (0.5 mm wide, 0.25-0.5 mm deep). The sponge is firm to hard and spiculose. Grey/white in life and in ethanol. There is a single raised apical osculum. No sign of any budding.

Skeleton.
A stalk of dense megascleres supports the sponge. There is a 'nucleus' where the stalk meets the centre of the sponge body, and although the stalk may divide and/or flatten and thicken externally it emanates from the same point at the base of the sponge. From the nucleus dense bundles (0.3-0.7 mm in diameter) of megascleres radiate through the choanosome to the surface tubercles; the bundles slightly fan out in the tubercles. The cortex is a dense layer of micrasters and oxyspherasters interspersed with megascleres emerging through the tubercules, making the surface microscopically hispid ( Fig. 2D-E). The cortex is well developed and follows the contours of the tubercules, 1-1.7 mm thick. Large cortical canals are visible between tubercles (Fig. 2E). A thin fibrous layer is below the cortex, it has micrasters in a much lower density. Large oxyspherasters are especially found at the base of the cortex. The megascleres of the stalk are covered in a layer of micrasters and regularly interspersed with shortrayed oxyspherasters. The choanosome is rich in sediment-like particles; there are some micrasters and rare oxyspherasters. Foraminifera (Globigerina d'Orbigny, 1826) and Radiolaria are common in the cortical canals and the choanosome.

Ecology and distribution
Found on the continental slope in the Great Australian Bight at a depth of 1000 m, in soft sediment (clay/ silt).

Remarks
The morphology as well as molecular markers confirm that our new sponge is a Tethya. Table 2 shows morphological comparisons between other species of Tethya from Australia and New Zealand. The external appearance of Tethya irisae sp. nov. is similar to T. fissurata from Port Jackson, New South Wales, Australia, which is spherical with polygonal tubercules and has a stalk. However, T. fissurata differs from T. irisae sp. nov. in body size (~ 4 cm diam.), tubercule shape, and number of oscula (2-4). Tethya fissurata has megascleres with stepped ends unlike T. irisae sp. nov., which are smooth and T. fissurata lacks the short-rayed oxyspherasters seen in T. irisae sp. nov. Although we do not know the exact depth at which T. fissurata was collected, Port Jackson (viz. Sydney Harbour) is not deeper than 45 m (Johnston et al. 2015), so this is presumably a shallow species. Tethya bullae is a deep-water (100 m) sponge that is of comparable size to T. irisae sp. nov., although it has prominent raised tubercules rather than the flat plate-like tubercules of T. irisae sp. nov. (Fig. 4). The holotype from the Australian Museum does not include a stalk but the description and photograph in Bergquist & Kelly-Borges (1991) shows "basal flattened branched rooting processes". The long-rayed oxyspherasters of T. irisae sp. nov. are similar to those of T. bullae. The short-rayed oxyspherasters in T. irisae sp. nov. do not fork as those of T. bullae. Tethya irisae sp. nov. has lightly spined acanthooxyspherasters compared to the completely spined acanthooxyspherasters of T. bullae. In addition to T. fissurata and T. bullae, other Tethya with rooting processes/stolons are shown in Table 2 (descriptions in bold text). It is difficult to tell how similar the rooting processes are to each other but these species differ in spicule forms and dimensions from T. irisae sp. nov. For example: species with megascleres < 2000 µm (T. acuta Sarà & Sarà, 2004, T. bergquistae Hooper in Hooper & Wiedenmayer, 1994, T. burtoni Sarà & Sarà, 2004, T. dendyi Sarà & Sarà, 2004, T. robusta (Bowerbank, 1873, T. seychellensis (Wright, 1881), T. stolonifera Bergquist & Kelly-Borges, 1991); species with megasters not of a 'spheraster' form (T. amplexa Bergquist &Kelly-Borges, 1991, andT. fastigata Bergquist &Kelly-Borges, 1991); species with very different micrasters (T. ingalli Bowerbank, 1858, T. flexuosa Sarà & Sarà, 2004 and T. monstrosa (Burton, 1924)).
In addition, T. irisae sp. nov. is collected at the start of the bathyal zone (~1000 m). The deepest of the Tethya is T. compactus Bergquist, 1961 (402 m), which has very different external morphology.
It occurred to us that when using the key to genera of Tethyidae (Sarà 2002), Tethya irisae sp. nov. appears closest to the monospecific genus Burtonitethya, a tethyid with a stalk of equal length to the diameter of the sponge. The type of Burtonitethya (B. gemmiformis), was collected from the Andaman Sea at an unknown depth (Sarà 1994). Burtonitethya gemmiformis was originally assigned to Tethya (labelled as Tethya gemmiformis Burton & Rao, 1957 on the NHM microscope slide) but was re-assigned to a new genus Burtonitethya by Sarà (1994) on account of the stalk, the conspicuous nucleus with strongyles, the reduced lacunar cortex, the specialised surface tubercules and the giant oxyaster megasters. Our new species clearly differs from this species in having different microscleres and does not have the giant megasters present in B. gemmiformis. As there is no specimen of the type species of Burtonitethya and thus no potential to sequence the sponge, we cannot test if Burtonitethya is a junior synonym of Tethya.
As seen above, the genus Tethya shows many different modes of attachment including basal stolons, basal roots, curved peduncles, flattened rooting processes as well as attachment discs and narrow skirts of tissue. Our results suggest that the stalk may not be a good genus-defining character within the family. Heim et al. (2007) in their analysis of Tethya species, for which they used morphological characters and molecular markers, suggest that characters pertaining to ecological influences may have developed several times. Similarly, we suggest that some of the external morphological characters used to separate genera of Tethyidae are homoplasious, probably appearing several times in different clades of Tethya and we question whether they should be grouped as definitive characters in morphological identifications. In the same way the genus Amphitethya Lendenfeld, 1907(Family Tetillidae Sollas, 1886 was created based on its stalk, but phylogenies show it is a Cinachyrella Wilson, 1925 (Szitenberg et al. 2013;Schuster et al. 2017).    Sarà (1990); 9 Burton (1924).
Notes. * unknown, or not noted in sources. a Maximum depth of Pt Phillip Heads is 50 m (Barton et al. 2012); b Maximum depth of Pt Jackson is 45 m (Johnston et al. 2015). Tethya diploderma Schmidt, 1870 is listed as occurring in New Zealand by Kelly et al. (2009), but is not included as this is currently shown as inaccurate by van Soest (2019). Tethya compacta Bergquist, 1961 is not included in the list of Australia and New Zealand species by Sarà & Sarà (2004) due to its synonymisation by Bergquist with T. aurantium (Pallas, 1766), however they note that the taxonomic status of T. compacta "remains uncertain"; it is included here as it is shown as accepted by van Soest (2019). The occurance of T. aurantium in New Zealand is shown as inaccurate (van Soest 2019) and is not included here.

Results of the phylogenetic analyses
28S and COI trees had similar topologies (Fig. 5). The monophyly of the Tethyida was not supported (28S, bootstrap of 12; COI, bootstrap of 33) with Timeidae sister to a moderately (COI) to poorly supported (28S) Tethyidae + Hemiasterellidae clade. Timea sp. from the '3PP cave', La Ciotat, France, (Chombard 1998) did not group with the rest of the Timeidae, but its paraphyletic position was poorly supported. The Hemiasterellidae (Adreus, Axos, Liosina) seemed to group in a moderately to wellsupported clade (28S, 69; COI, 98). The 28S tree suggested that the Australian Laxotethya dampieriensis (Tethyidae) and a Tethyida sp. from Ireland with no obvious genus assignment (C. Morrow, pers. comm.) had more ambiguous positions: in RaxML analyses they grouped together (poorly supported), while in MrBayes analyses Tethyida sp. branched between the Timeidae and the rest of the Tethyida.  Laubenfels, 1936, Stellitethya, Xenospongia andTectitethya Sarà, 1994) were mixed with Tethya species, especially in clade 4. T. irisae sp. nov. grouped with Tectitethya in clade 4 (in RaxML and MrBayes analyses) but this is poorly supported. T. wilhelma was originally described from a tropical aquarium in Germany but its original geographical location is unknown. In Tethya clade 2, we noticed several species were very close genetically to T. wilhelma: with COI, Tethya sp. from Israel (Mediterranean Sea) had only 1 bp difference with T. wilhelma; with 28S, T. taboga from Panama had 0-2 bp difference with T. wilhelma (uncertainty due to two ambiguous base pairs) while Tethya sp. 2 from Saudi Arabia had 1 bp difference with T. wilhelma.

Discussion
Tethya irisae sp. nov. is a new and distinctive sponge from the slope of the Great Australian Bight (GAB). This is the 28 th Tethya species reported from Australia, and the deepest, being found in the bathyal zone at around 1000 m deep. With uncertainties in the validity of genera divisions based on morphological characters e.g., possession of a stalk, a reappraisal of the genera within Tethyidae based on molecular sequencing is needed.

Phylogeny of Tethyidae
The first phylogenetic analyses of Tethya, using COI and morphology (Heim et al. 2007;Heim & Nickel 2010) revealed four main clades: 1) the seychellensis-wilhelma complex, 2 + 3) the citrinia-actinia complex divided in two subclades (European species and western Atlantic species + eastern Pacific) and 4) the aurantium clade. Our COI and 28S analyses with extended datasets retrieve these four clades, but with a higher biogeographical diversity. The seychellensis-wilhelma complex now includes specimens from Israel, Vietnam, Panama, China and Queensland; the aurantium clade now includes species from the Mediterranean Sea, the Red Sea and Panama. All clades are well-supported in the COI tree except for clade 3, the western Atlantic/Pacific clade. This is precisely the group joined by the COI sequence of T. irisae sp. nov.; its position within this group, however, remains unclear. These same four clades are not as clear in our 28S tree, their inter-relationships are also different, and not supported at all. This may be due to the fact that our 28S alignment is a mix of different 28S domains and different sampling than COI, both of which may influence some of the groupings. The seychellensis-wilhelma and aurantium clades are well supported with 28S as well. On the other hand, the citrinia and actinia subclades are unclear, and this is probably due to the addition in this dataset of many different genera of Tethyidae (Tethytimea, Tectitethya, Stellitethya, Xenospongia Gray, 1858, Laxotethya). As suggested by Sarà et al. (2001) and Heim et al. (2007), Tethya wilhelma and T. gracilis Sarà, Sarà, Nickel & Brümmer, 2001, both described from aquaria in Germany belong to the seychellensis-wilhelma complex. There is only 1 bp difference between the COI of T. wilhelma and Tethya sp. (Mediterranean Sea, Israel) so this specimen should be re-examined to see if it could be conspecific with T. wilhelma. Heim et al. (2007) showed that the most reliable characters for Tethya taxonomy were morphometric spicule data, but none could actually make good morphological synapomorphies for the two Tethya clades supported with COI and 28S. New characters (e.g., chemical compounds, specialized cells, associated microbes) must be explored in order to find independent support for these groups.
External colour may be a reliable character to discriminate those clades, as shown previously in some calcareous sponges (Rossi et al. 2011). Indeed, most shallow water Tethya species have a yellow, orange to red surface colour, probably due to different carotenoids (Tanaka et al. 1982)  , or pink (e.g., Tethya bergquistae) but none of these species have been sequenced yet. We can probably dismiss the green colour. It is found in species that can also be orange; Laubenfels (1950) suggested the green colour of T. actinia in Bermuda was due to symbiotic algae (a specimen may "turn orange" when fixed in alcohol, as the chlorophyll is extracted). More problematic are species with varying colours, from yellow to orange and red (e.g., Tethya fastigata).
As for the few deep-sea species of Tethya, some have lost their colours (e.g., Tethya irisae sp. nov.) while others have retained them: e.g., Tethya levii Sarà, 1988 from New Caledonia is light orange, and groups in the actinia clade, in accordance with our hypothesis (P. Cárdenas, unpublished data). This groupingby-colour hypothesis should be further tested with the sequencing of new species of Tethya. Other genera of Tethyidae included in our dataset have usually irregular massive forms or are disc-shaped (instead of subspherical forms), and all have dark colours: black-brownish for Tectitethya spp., beige-gray for Xenospongia, and whitish-brown in ethanol for Laxotethya and Stellitethya (the live colour is unknown). Since all except Laxotethya are sister group to a bright orange Tethya sp. from South Australia (possibly in the actinia clade) (Fig. 5, 28S tree), we suppose the common ancestor of these other genera lost its yellow-orange colours, and so its capacity to produce carotenoids.
Our COI and 28S dataset include type species of four Tethyidae genera (of the 14 valid genera): Tethya (T. aurantium, COI), Tectitethya (T. crypta, 28S), Xenospongia (X. patelliformis, 28S) and Laxotethya (L. dampierensis, 28S). In addition to that, two other Tethyidae genera are represented in our 28S tree: Stellitethya and Tethytimea. All these genera are essentially defined by different skeletal structures and therefore body shape; all these genera have an indistinct or ill-defined cortex (vs a distinct thick cortex for Tethya) and an irregular massive or encrusting shape (vs (sub)spherical shape in Tethya). Our After the species name, locality of the specimen is given (when known), followed by the GenBank accession number(s). For 28S, we also indicated the 28S region that was sequenced as well as the first author + date of the publication where the sequence first appeared. Type species of genera are in red boxes while Tethya irisae sp. nov. appears in red.
28S tree suggests that Xenospongia, Stellitethya, Tectitethya and Tethytimea are grouping with Tethya (Fig. 5), while Laxotethya groups with Hemiasterellidae, albeit with no support. Tethytimea carmelita, Tectitethya and Stellitethya/Xenospongia evolved independently within Tethya thus suggesting that the loss of a distinct cortex and of the subspherical shape happened several times. More sequences from Australian Tethya are needed to understand the origin and relationships of these other Tethyidae genera. One clade that is moderately supported (bootstrap of 69) is the sister-group relationship of Xenospongia and Stellitethya, with a 5-6 bp difference in 28S (D3-D5). Both genera have a poorly defined cortex but different shapes: discoid for Xenospongia, massive irregular for Stellitethya. These two genera also share megasters reaching large sizes (> 150 µm), as in T. irisae sp. nov., grouping nearby (bootstrap of 62) with Tectitethya (which does not have very large megasters).

Phylogeny of Tethyida
The Tethyida also include Hemiasterellidae and Timeidae. Since the ex-hemiasterellid genera Stelligera Gray, 1867 and Paratimea have been reallocated to the Stelligeridae Lendenfeld, 1898, order Axinellida (Morrow et al. 2012), the Hemiasterellidae include four genera: Adreus, Axos, Hemiasterella and the monospecific Leptosastra Topsent, 1904. The sister position of Hemiasterellidae Adreus and Axos with the Tethyidae has been repeatedly shown by previous COI (Morrow et al. 2012, 28S Cruz-Barraza et al. 2017) and 18S analyses . Liosina, a genus with only four species and a loose arrangement of oxeas/styles, had been tentatively assigned to the family Dictyonellidae, order Bubarida (van Soest et al. 2002). The grouping of Liosina paradoxa, type species, with the Hemiasterellidae was revealed for the first time by Morrow et al. (2012) with 28S. Our study confirms this grouping for the first time with COI sequences (L. paradoxa and Liosina blastifera), as well as a 28S sequence of L. blastifera. Its unambiguous grouping with the Hemiasterellidae suggests that species of Liosina are in fact Hemiasterellidae that have secondarily lost their euasters. Furthermore, Liosina often have a polygonal surface pattern, and/or pores in shallow grooves, a character present in most Tethyidae and Timeidae, which often gives rise to the characteristic surface tubercles. Since we have sequenced the type species of Liosina, we formally propose the reallocation of Liosina from Dictyonellidae to Hemiasterellidae. However, it is unclear in our trees whether Liosina is polyphyletic. The small polygons from Liosina, also called tubercles or microconules, are also present in the three Adreus species from Australia (A. australiensis (Ridley, 1884), A. axiferum (Hentschel, 1912) and Adreus sp.) thereby confirming their reallocation. These two branching species also share with the Tethyida the typical radial bundles of megascleres fanning out closer to the surface. Adreus australiensis and A. axiferum were previously grouped in Raspailidae (Hooper 1991) although they lacked echinating megascleres, before Erpenbeck et al. (2007) showed with 28S that they clustered instead with Hemiasterellidae and Tethyidae. These two species also secondarily lost their asters, while the vase-shaped Adreus sp. from Queensland (QM G315767) still has tylasters. This second case of aster loss in the Tethyida suggests that, similarly as in the Astrophorina (Cárdenas et al. 2011), more genera or species without asters, can be expected to be reallocated to the Tethyida once they are sequenced. Since this Adreus clade does not cluster with the clade of Adreus fascicularis (type species of the genus), they potentially represent a new genus in the Hemiasterellidae. To conclude, the Hemiasterellidae now include Adreus, Axos, Hemiasterella, Leptosastra, Liosina, and a potential new genus. So far, all GenBank Hemiasterella sp. sequences are doubtful and failed to cluster with the Hemiasterellidae (cf. Material and methods). The type specimen of Hemiasterella typus Carter, 1879, has not been revised and sequenced so that the phylogenetic position of Hemiasterella remains to be tested. We note, however, that H. typus does not share with most of the Tethyida 1) a surface with pores in grooves around tubercles/plates or 2) bundles of megascleres fanning out at the surface.
To sum up the main findings of the phylogenetic analysis.
1. Four Tethya clades were retrieved (as in previous analyses) for which no synapomorphies are currently known; we, however, discuss the possibility of using external colour to support some of these clades. 2. Despite unclear phylogenetic relationships amongst Tethyidae from Australia, our results suggest that Tethyidae genera Tethytimea, Tectitethya, Laxotethya, Stellitethya, and Xenospongia derive from species of Tethya, which may challenge their validity in the future. 3. Our results suggest that Hemiasterellidae is the sister-group of Tethyidae while the position of Timeidae is still ambiguous (not supported). 4. We show that asters have been secondarily lost at least twice in the Hemiasterellidae: in Liosina and a potential new genus from northern Australia. We formally propose the reallocation of Liosina from Dictyonellidae to Hemiasterellidae.