Taxonomic notes on Poecillastra sponges (Astrophorida: Pachastrellidae), with the description of three new bathyal southeastern Pacific species

Three new species of Poecillastra are described here from bathyal waters off central Chile. P. antonbruunae n. sp. is diagnosed by its two categories of oxeas, the smaller one only up to 621 μm in length, and calthrops as the sole triaene category; P. sinetridens n. sp. by the possession of oxeas larger than 3000 μm long, microxeas which can be over 300 μm long, and absence of triaenes; and P. maremontana n. sp. by its three categories of oxeas, three categories of streptasters (one of spirasters and two of plesiasters), one category of calthrops and centrotylote microxeas. Comments upon other Poecillastra and an identification key for Pacific species of the genus are given. Evolutionary hypotheses are discussed for the genus given the distribution of spicule morphotypes, as well as the recurrent phenomenon of spicule loss in Astrophorida. Five species of Pachastrella are transferred here to Poecillastra on the basis of their possession of microxeas instead of microrhabds, viz. P. cribrum Lebwohl, 1914; P. dilifera de Laubenfels, 1934 [sensu Dickinson (1945)]; P. fusca Lebwohl, 1914; P. incrustata Bergquist, 1968; and P. scrobiculosa Lebwohl, 1914. We recognize 28 species of Poecillastra, 19 of them occurring in the Pacific Ocean.

las, 1888; Uriz, 2002).In Pachastrellidae Carter, 1875, tetraxons are a widespread character, and their absence is described here for the first time.
Pachastrellidae is an important source of new bioactive molecules of potential commercial value, such as Dercitine, from Dercitus sp., with a marked antitumoral activity (Burres et al., 1989); Psammapline A, from Poecillastra sp., with antibacterial activity (Kim et al., 1999); and a citotoxin from Pachastrella sp.(Gunawardana et al., 1989).Nevertheless, the relatively small numbers of species known from several large basins and the deep-sea dwelling habit of many species suggest that there is a need for improved taxonomic effort.
Six species of Astrophorida were previously known from Chile, viz.Asteropus ketostea (de Laubenfels, 1950), Asteropus simplex (Carter, 1879), Geodia amphistrongyla Lendenfeld, 1910, Geodia magellani (Sollas, 1886), Stelletta phrissens Sollas, 1886, and Stelletta vosmaeri (Sollas, 1886).In the present study we describe three new species of Poecillastra from central Chile (32ºS-35ºS).One of these species is atypical in lacking triaenes.This study further determines the status of the Pacific Ocean as the richest ocean basin for this genus.A brief review of Poecillastra species worldwide is undertaken, comparing them to the new species described here, and possible biogeographic scenarios are outlined taking into account the distribution of spicule morphotypes within the genus and their primary suggestion of monophyly of the species group sharing these characters.

MATERIALS AND METHODS
Specimens were collected off the central Chilean coast from 290 to 450 m depth.Figure 1 illustrates the collecting localities for the three specimens studied.These were collected by distinct oceanographic/ fisheries surveys, and were deposited in the following institutions: United States National Museum (USNM, Washington, D.C., USA), Museum d'Histoire Naturelle (MHNG, Geneva, Switzerland), and the Huinay Scientific Field Station (HSFS, Huinay, Chile).
Procedures for the study of spicules and skeletal architecture under light microscopy and scanning electron microscopy (SEM) followed Hajdu (1994).SEM analysis was carried out at the Instituto de Biofísica Carlos Chagas Filho and the Museu Nacional, both at the Universidade Federal do Rio de Janeiro, Brazil.
Twenty spicules of each category for each specimen were randomly selected and measured, except for oxeas (50 measurements) or in cases in which more measurements were necessary to precisely establish size ranges for the other spicule types.An exhaustive search for maximum and minimum ranges was done.Spicule dimensions are presented as minimum-mean-maximum length and width, in micrometres.
Abbreviations  Definition.Pachastrellidae, whose microscleres always include streptasters in the form of spirasters and plesiasters, as well as microxeas in a single category (Maldonado, 2002).
Diagnosis.Poecillastra with two categories of oxeas, the smaller one only up to 621 µm in length.
Description.The single known specimen is lamellate, 0.4 cm high, and 6×5.2 cm in diameter (Fig. 2A).Both surfaces are rough to the touch, hispid, with projecting oxeas (ca.1000 µm beyond the surface), and with a great number of pores that seem to be oscules (0.5-1 mm in diameter) on one side, while the other side has fewer pores.The sponge is firm and incompressible.Colour is dark beige in ethanol.
Skeleton: Ectosomal skeleton is composed of abundant spirasters and microxeas, but also contains oxeas and calthrops.The smaller oxeas are principally present in the ectosome, although not forming a crust.Choanosomal skeleton confused and crossed by aquiferous channels, with oxeas and calthrops more abundant than in the ectosome.
Etymology.The new species name "antonbruunae" is in honour of the ship "Anton Bruun", whose crew collected this species.We opted for female gender termination as ships are usually referred to as "she".
Distribution.Known only from its type locality, off Constitución, central Chile.
Remarks.The new species is closely related to Poecillastra schulzei (Sollas, 1886, sensu Sollas, 1888) in its spicule composition (see Table 1).Nevertheless, the status of P. schulzei and its alleged synonyms has been the subject of a long debate, still in need of some clarification.Poecillastra schulzei was considered to be conspecific with P. compressa (Bowerbank, 1866) by some authors (Topsent, 1894;Burton, 1930).Recently, however, Maldonado (2002) re-examined the type of the former and determined its validity on the basis of the possession of two categories of oxeas, as opposed to only a single category in P. compressa.Poecillastra tenuilaminaris, suggested by Burton (1959) to be a likely synonym of P. schulzei, has been considered a valid species by Maldonado (2002) on the basis of its lack of slender ectosomal oxeas.According to Maldonado, Poecillastra schulzei is an Indo-Pacific/Southern Ocean species (but see below), and P. tenuilaminaris is a North Pacific one.
Poecillastra laminaris (Sollas, 1886) and P. eccentrica Dendy and Burton, 1926, from the Indo-Pacific region, were also regarded as possible junior synonyms of P. schulzei by Maldonado (2002).The proposed likely synonymy of tropical species of Poecillastra, P. eccentrica and P. laminaris with the originally subantarctic P. schulzei, though supported by re-examination of the relevant type material (Maldonado, 2002), implies accepting a considerable morphologic variability and a marked eurybathyal distribution in the concept of "P.schulzei".Poecillastra eccentrica was originally reported with much larger calthrops, a varied set of triaenes, smaller asters, and completely lacking rough microxeas (Dendy and Burton, 1926).Maldonado (2002) had the opportunity to examine microscopic preparations of this species, but unfortunately provided very few arguments to support this proposed synonymy.Re-examination of the holotype of P. eccentrica (BMNH 1926.10.1.95;two spicule slides and two thick section slides) revealed the presence of microxeas as well as larger oxeas, both overlooked by Dendy and Burton (1926).We concluded from the confirmed presence of relatively larger calthrops that P. eccentrica is a valid species.Conversely, P. laminaris was originally described with trichoxeas representing a second category of much larger and slender oxeas, a trait not present in P. schulzei.Additionally, P. laminaris possesses relatively smaller asters, and occurs in a much deeper and more distant habitat.Consequently, this proposed synonymy should also be verified through re-examination of type material, which has not been possible here.We consider P. schulzei to be a subantarctic species, characterized by possession of two categories of oxeas rather well differentiated in terms of length and thickness (Table 1), triaenes in two forms (larger orthotriaenes and smaller calthrops), microxea, metasters and spirasters.It ap-proaches the new species as both possess two categories of oxeas and calthrops as megascleres, as well as spirasters as microscleres.However, they differ in the presence of orthotriaenes and metasters in P. schulzei.Additionally, oxeas I in the new species are slightly smaller and thinner.
None of the alleged synonyms of P. schulzei cited above approaches our new species as closely as Sollas's original material.The other new species described from this region are discussed further below.
(Fig. Diagnosis.Poecillastra with three categories of oxeas, three categories of streptasters (one of spirasters and two of plesiasters), one category of calthrops and centrotylote microxeas.
Description.The single known specimen is lamellate and consists of fragments (Fig. 3A), the largest one 7×4.7×0.5 cm (height×length×thickness). The surface is hispid with several scattered oscules (up to 1.5 mm in diameter).These oscules are more visible on one side of the sponge than the other.The colour is white in ethanol.
Skeleton: Ectosomal skeleton has no specialization, apart from a dense concentration of all microscleres around the pores.Choanosomal skeleton confused with scattered oxeas I, II and III.Calthrops, plesiasters, and spirasters are scattered through the ectosome and choanosome.
Distribution.Currently known only from its type locality -Juan Fernandez seamounts, Chile.
Remarks.Aside from this new species, only two other species of Poecillastra possess three categories of oxeas: P. japonica (Thiele, 1898) and P. tenuilaminaris sensu Green and Bakus (1994) (see Table 1).
Poecillastra japonica has larger oxeas, approximately the same length as those in this new species but 5-6 times stouter.The intermediary sized oxeas are also much stouter in P. japonica than in the new species, and the smaller oxeas are larger than 2500 µm, whereas in the new species the smallest oxeas are only 315 µm.The new species has two plesiasters, with P. japonica reportedly having only a single category.
Green and Bakus (1994) described a specimen of P. tenuilaminaris from California with three categories of oxeas, similar to our new species, whereas previous records noted only one category (Sollas, 1888;de Laubenfels, 1932).Nevertheless, oxeas in the new species are much longer, the largest category reaching over 4600 µm in length, while the oxeas in the Californian sponge are smaller than 2750 µm in length.
Three categories of oxeas, one of them larger than 4600 µm, three categories of streptasters (one of spirasters and two of plesiasters), one category of calthrops and centrotylote microxeas distinguish Poecillastra maremontana n. sp.from all other species of the genus.
Diagnosis.Poecillastra without calthrops or any other kind of triaenes.Oxeas always larger than 3000 µm long, and microxeas can be over 300 µm long.
Description.The single specimen consists of two fragments, the largest one 2.5×1.5×1cm in size (Fig. 4A).The irregular surface is rough, sometimes hispid with large oxeas projecting through it (3000 µm be-yond the surface).The sponge is soft and compressible.The colour is beige in ethanol, with some darker spots.
Skeleton: Ectosomal skeleton has no specialization.It is a tangential layer of microxeas with a marked concentration of spirasters and plesiasters, the former being more abundant.Choanosomal skeleton confused.Spirasters and plesiasters are also present, with plesiasters being more abundant in this region.Microxeas are found in a smaller quantity.
Etymology.The new species is named "sinetridens" due to its unique character of lacking calthrops or any other type of triaenes, which occur in all other species of the genus (sine, Latin for without; tridens, trident).
Distribution.So far known only from its type locality, off Zapallar, central Chile.
Remarks.The possession of asters is a widespread feature within the order Astrophorida.Sponges belonging to the Pachastrellidae typically have streptasters, microxeas (in most cases) and microrhabds as microscleres (see Table 1).Megascleres include a varied set of tetraxons (Maldonado, 2002).Most genera of Pachastrellidae have calthrops and/or triaenes.The occurrence of microxeas in a single category and streptasters including plesiasters indicates that this new species belongs to Pachastrellidae.The apparent absence is assumed to be a secondary loss of triaenes in this new species.This is unique and as such its allocation to Poecillastra requires further discussion.
Five out of the 12 currently valid genera of Pachastrellidae (Maldonado, 2002) are ruled out on the basis of their microscleres: absence of microxeas in Pachastrella Schmidt, 1868 and Thenea Gray, 1867; and possession of different kinds of microscleres in Acanthotriaena Vacelet, Vasseur and Lévi, 1976 (raphides), Dercitus Gray, 1867 (toxas), and Stoeba Sollas, 1888 (microrhabd-like sanidasters).Four genera are excluded on the basis of their megasclere morphology: Ancorella Lendenfeld, 1906 has strongyloxeas transitional to strongyles; Brachiaster Wilson, 1925 has short-shafted triaenes and/or mesotriaenes that become mesotrider desmas; Cladothenea Koltun, 1964 has cladotyles;and Triptolemma de Laubenfels, 1955 has exclusively short-shafted mesotriaenes with diversely branched cladi.Vulcanella Sollas, 1886 possesses specialized cribriporal oscula and one or more microsclere categories bearing an annulate decoration (Maldonado, 2002).Characella Sollas, 1886 also contains species with streptasters and microxeas as microscleres.However, the fact that the streptasters in Characella are never spirasters, and that its microxeas are always in at least two size categories, also excludes the assignment of the new species to this genus Lerner et al. (2004) and Esteves and Muricy (2005) described new species recognizable, among other traits, by the absence of microscleres (Guitarra sepia Lerner, Hajdu, Custódio and van Soest 2004 and Stelletta anasteria Esteves and Muricy, 2005).While this assumed secondary loss of triaenes is a negative attribute, and unwisely applied at the genus level, we contend it is a valid species trait.The decision to allocate the new species in Poecillastra stems from its possession of a set of microscleres which comprises spirasters, plesiasters and acanthose microxeas in a single category, precisely matching the characteristic of other species in this genus.
At the species level there are five species that have the same spicule set as Poecillastra sinetridens n. sp., aside from the presence of calthrops absent in the new species, viz.one category each of oxeas and microxeas, and two categories of streptasters.These are P. antarctica Koltun, 1964 (as P. compressa a., see below); P. compressa; P. eccentrica; P. laminaris (sensu Lévi and Lévi, 1989 [not sensu Sollas (1888), with two categories of oxeas]); and P. tenuilaminaris (not sensu Green and Bakus, 1994 [with three categories of oxeas]).
Poecillastra compressa sensu Boury-Esnault and van Beveren (1982) has oxeas which can be considerably shorter (ca.1000 µm) than those of the new species.It also has a third category of asters reach-   Sollas's (1888) sponge differs in having two categories of oxeas, the smaller one up to 3500 µm long only, while the new species has oxeas always longer than 3000 µm, and sometimes longer than 5000 µm.Additionally, Sollas's material was reported to possess trichoxeas, clearly differentiating it from the new species.
Finally, P. tenuilaminaris (sensu Sollas, 1888;Lebwohl, 1914) has smaller oxeas and microxeas and was also reported from greater depths off Japan, so it differs both morphologically and biogeographically from the new species.Additional records of P. tenuilaminaris need further comments.Dendy's (1916) description appears to concern a distinct taxon.He outlined seven points of divergence between the Indian and Sollas's (1888) original description (from Japan).Significant among these are the presence of long hair-like oxeas, the smaller size of the metasters which show no tendency to pass into plesiasters, and the rarer tetraxons and microxeas in Dendy's specimen.Further distinction stems from the possible shallow water habitat of his specimen, collected in the tropical islands of the Amirantes Archipelago, where the fauna has very little in common with that of Japan.Dendy's sponge appears to more closely resemble P. laminaris from Indonesia, which also possesses a second category of large hair-like oxeas.No striking morphological distinction is apparent when the two descriptions are compared, and the main trait setting them apart is the relatively deep water habitat of P. laminaris.Both specimens need to be compared in greater detail to correctly allocate Dendy's specimen.Poecillastra tenuilaminaris has also been recorded from California in a series of publications (de Laubenfels, 1932;Dickinson, 1945;Green and Bakus, 1994).Judging from each of the descriptions, these appear to conform to the Japanese population, making the species highly disjunct.A thorough revision of morphological characters, especially a detailed SEM study of asterose microscleres is required to verify the status of these populations.
In spite of the greater or lesser diagnostic trends noted above when P. sinetridens n. sp. was compared with other species, the main immediate diagnostic feature is the absence of triaenes.

Spicule loss
Tetractinellid sponges (Astrophorida and Spirophorida, sensu Borchiellini et al., 2004) are mainly characterized by the possession of tetractinal megascleres and microscleres (asterose forms and derivatives) (Hooper and van Soest, 2002), though examples of whole evolutionary lineages which have lost one or more of these diagnostic features are known (e.g.Asteropus spp., Jaspis spp.).According to Hooper and van Soest (2002), microscleres are the major taxonomic characters in tetractinellids, but even these may be totally absent (e.g.Stelletta anasteria).
Hooper and van Soest (2002) have recognized eight genera belonging to Astrophorida with a reduction or total loss of triaenes.Six of them belong to Ancorinidae (Jaspis Gray, 1867; Asteropus Sollas, 1888; Dorypleres Sollas, 1888;Holoxea Topsent, 1892;Melophlus Thiele, 1899;and Rhabdastrella Thiele, 1903), one to Pachastrellidae (Characella Sollas, 1886), and Lamellomorpha Bergquist, 1968, which is incertae sedis as to family allocation.Most of these genera have a tangential layer of megascleres in the ectosomal skeleton (e.g.Asteropus, Holoxea, Jaspis, Lamellomorpha and Melophlus; cf.Hajdu and van Soest, 1992), which may substitute for any possible structural integrity shortcoming arising from the loss of triaenes.Currently, this recurrent pattern is not judged synapomorphic, so the development of a variously neat tangential layer of oxeote megascleres in these genera might be understood as a possible convergent response to solve the likely parallel problem of triaene loss.Koltun (1964) did not mention the presence of microxeas in the original description of P. compressa antarctica, which would prevent the classification of this subspecies in Poecillastra.However, the reexamination of its holotype (ZIL 10862) revealed the presence of these spicules (Table 1), straight to very curved.Remarkably, no single morphological character observed by us in the fragment of the holotype re-examined matches the description of Koltun, which makes us wonder whether we have observed the same sponge.It is noteworthy that the micrometric values obtained here for the oxeas and plesiasters (metasters sensu Koltun, 1964) were much larger than values originally reported by Koltun.In spite of this confusion, these larger spicules serve the purpose of establishing the status of P. compressa antarctica as distinct from typical P. compressa.We prefer to raise Koltun's subspecies to full species status, P. antarctica, because the meaning of subspecies ranking in sponge classification is absolutely not agreed upon (Lopes et al., in press), and emphasizing the markedly allopatric distribution of both.P. compressa sensu Boury-Esnault and van Beveren (1982) fits much better in P. antarctica, due to its possession of oxeas up to 5075 µm in length, calthrops of up to 1247 µm and one category of plesiasters of up to 83 µm.Other records of Poecillastra compressa also need revision.Pulitzer-Finali (1983) reported an alleged P. compressa from the Mediterranean possessing oxeas as small as 300 µm in length; while Pulitzer-Finali (1993), working on materials from the northern Kenyan coast, reported oxeas up to 3500 µm in length and calthrops of up to 650 µm.These two specimens may belong to different species and should be re-analyzed in detail, which has not been possible here.

Comments upon other Poecillastra
Poecillastra crassiuscula (Sollas, 1886) and P. incrustans Sollas, 1888 were suggested as junior synonyms of P. compressa by Maldonado (2002).We disagree with this proposal as oxeas in P. incrustans are thinner and have a smaller size range than in P. compressa (Table 1).P. crassiuscula has larger oxeas and tetraxons.In addition, tetraxons (calthrops/orthotriaenes) are also larger in P. crassiuscula.It would be interesting to revise the microscleres of the three species in SEM to figure how micromorphologic features vary among them, but we did not have access to these materials in this study.Boury-Esnault et al.'s (1994) redescription of P. compressa, is the only one in which SEM was used for the species above, and only a single spicule of each microsclere category was illustrated.
Poecillastra rickettsi was suggested as a likely synonym of P. tenuilaminaris by Maldonado (2002).Above we have already used the occurrence of hairlike oxea to argue for the unlikelihood of Dendy's (1916) specimen being properly identified in P. tenuilaminaris.P. rickettsi has the largest hair-like oxeas described from any species of Poecillastra (Table 1), being over 4 times larger than the largest oxeas reported from P. tenuilaminaris (3870 µm; Dickinson, 1945), and as such, deserving status as a valid species.
Cross-checking assemblages for shared species allows one to recognize a triad of species corroborated by two congruent traits.Poecillastra ciliata, P. tenuirhabda and P. tuberosa share the possession of dichotriaenes as well as two categories of microxeas, which is suggestive of likely phylogenetic affinity.No additional congruent traits were observed.
Figure 5 shows the approximate distribution of each of these recognized morphological assemblages (A-E).Group (A) reflects an apparent relictual Tethyan distribution, and group (B) is restricted to the Mediterranean.The other groups are notorious for their largely disjunct distributions.This is the case too for the species triad highlighted above, occurring in the Philippines, Saint Paul Isl. and South Africa, and spreading across the southern Indian Ocean and western Pacific.Widespread Indo-West Pacific species are known from a variety of sponge genera (Hooper and Lévi, 1994;van Soest, 1994;van Soest and Hajdu, 1997;Fromont et al., 2010).In many of these, several records punctuate the known distribution range of species [e.g.Acarnus bergquistae van Soest, Hooper and Hiemstra., 1991;A. bicladotylus Hoshino, 1981;A. wolffgangi Keller, 1889;Clathria procera (Ridley, 1884); cf.van Soest et al., 1991;Hooper and Lévi, 1994], so that discontinuity is not so striking.
Group (D) considerably matches the distribution observed in group (A), but records for the Mediterranean (rudiastra, symbiotica) and Central Indian ocean (nana) in (A) are replaced by a record for Korea (wondoensis) in (D), thus rendering group (D) more widespread, and less likely explained as a relictual Tethyan distribution.Group (C) has a nearly amphi-Pacific distribution, with an off-shoot in the southern Indian ocean.The complete absence of any Atlantic records here suggests that any phylogenetic signal present in this assemblage may be a consequence of trans-Pacific evolutionary tracks (e.g.Sluys, 1994;Hajdu and Desqueyroux-Faúndez, 2008).

Poecillastra vs. Characella
The number of categories of microxeas along with the form of streptasters seem to be the most obvious characters to distinguish Characella and Poecillastra.The latter always includes streptasters with twisted shafts (amphiasters, metasters, plesiasters, spirasters and/or transitional forms), while in Characella streptasters have straight shafts (amphiasters, sanidasters and/or transitional forms) (Maldonado, 2002).Maldonado (1996Maldonado ( , 2002) ) still emphasized the location of triaenes as diagnostic.These supposedly occur in the ectosome and choanosome in Poecillastra, but are restricted to subectosomal locations in Characella.
An identification key for the 20 species of Poecillastra occurring in the Pacific Ocean is given below.Superior), CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico), FAPERJ (Fundação do Amparo à Pesquisa do Estado do Rio de Janeiro), and the Fondation Claraz, Geneva, for providing grants and/or fellowships.

Table 1 .
-Comparative table of micrometric and distributional data for the species of Poecillastra of the world, including the three newly described eastern Pacific species.Measurements are expressed in µm, as minimum-mean-maximum length/width value whenever available.

Table 1 (
cont.).-Comparative table of micrometric and distributional data for the species of Poecillastra of the world, including the three newly described eastern Pacific species.Measurements are expressed in µm, as minimum-mean-maximum length/width value whenever available.