First records, rediscovery and compilation of deep-sea echinoderms in the middle and lower continental slope of the Mediterranean Sea

This study provides a compilation of all available information on deep-sea echinoderms from the middle and lower slopes of the Mediterranean Sea, with the aim of providing a unified source of information on the taxonomy of this group. Previous records of species are updated with new data obtained from 223 trawl hauls conducted in 11 cruises within the northwestern Mediterranean Sea between 800 m and 2845 m depth. Valid names, bathymetric ranges and geographic distributions are given for all species. The new data report, for the first time, the presence of the Atlantic echinoid Gracilechinus elegans (Duben and Koren, 1844) in the Mediterranean Sea. We also report the presence of the endemic holothurians Hedingia mediterranea (Bartolini Baldelli, 1914), dredged only once previously in 1914 in the Tyrrhenian Sea, and Penilpidia ludwigi (von Marenzeller, 1893), known previously only from three samples, two in the Aegean Sea and one in the Balearic Sea. Additionally, the deeper limits of the bathymetric distribution of four species have been expanded: the asteroid Ceramaster grenadensis (Perrier, 1881) to 2845 m; the echinoid Brissopsis lyrifera (Forbes, 1841) to 2250 m; and the holothurians Hedingia mediterranea and Holothuria (Panningothuria) forskali Delle Chiaje, 1823, to 1500 m and 850 m, respectively.


INTRODUCTION
The deep Mediterranean Sea has a wide variety of geological and ecological settings. Their faunal composition and local biodiversity are largely unknown (Danovaro et al. 2010). The western Mediterranean deep basin is no exception. It has a complex assemblage of markedly different habitats  including sedimentary slopes, submarine canyons and seamounts . The specific geomorphological characteristics of these habitats (e.g. the elevation of seamounts, the walls and axes of the submarine canyons, the inclination of the continental slopes, etc.) and associated abiotic processes (e.g. variation in oceanographic currents, hard vs. soft substratum, food availability) result in large-scale heterogeneity of the continental margin seafloor (Carpine 1970, Emig 1997, D'Onghia et al. 2003. This high habitat heterogeneity plays a major role in the establishment and maintenance of diverse faunal communities (Levin et al. 2010), which, to date, are still largely unexplored in the deep Mediterranean Sea (Bienhold et al. 2013, Mecho et al. 2014).
The shallow Mediterranean marine fauna, inhabiting the shelf and upper slope areas, have been studied from ancient times. Consequently, they are well known at many levels (taxonomic, ecological, and biological) (Riedl 1983, Bolam et al. 2002, Danovaro and Pusceddu 2007, Coll et al. 2010. Nevertheless, because of the difficulties in sampling the deep sea, the bathyal and abyssal fauna of the Mediterranean Sea remains poorly studied (Pérès and Picard 1956a, Fredj 1974, Galil and Goren 1995, Danovaro et al. 2010, Tecchio et al. 2011a. The description of the benthic fauna occurring deeper than 800 m in the Mediterranean started on the 19 th century. Cruises carried out by the R. N. Washington (1881-1882) and S.M.S. Pola (1890-1898 provided the first extensive descriptions of bathyal and abyssal Mediterranean fauna (Marenzeller 1893, Bartolini Baldelli 1914 including many new species of non-crustacean invertebrates. From the late 1920s to 1960s the number of deep-sea Mediterranean research cruises decreased, resulting in limited new information (Pérès andPicard 1956a,b, Pérès 1958).
Since the late 1970s, improvements in sampling methods and equipment have allowed a second period of deep-sea scientific exploration and investigation below 1000 m depth, conducted by ships such as the 'Bambu', 'Mango', and 'Ruth Ann' in Italian waters, as well as the R.V. 'Jean Charcot' in the Alboran Sea or the R.V. 'Garcia del Cid' in the Balearic Sea.
Specimens collected by these expeditions have stimulated a number of publications and new records of species (Carpine 1970, Parenzan 1970, Reyss 1971, Fredj 1974. Nevertheless, most of this deep sea literature focuses on the dominant groups such as fishes and crustaceans, the commercial use of Mediterranean marine resources and the management of these resources (Sardà et al. 1994, Moranta et al. 1998, Company et al. 2004). Thus, both fish and crustaceans are well known taxonomically in comparison to other megafaunal groups, such as ascidians, sponges, echinoderms, sipunculans and echiurans (Monniot and Monniot 1975, Alvà 1987a, Uriz and Rosell 1990, Villanueva 1992, Pancucci-Papadopoulou et al. 1999, Quetglas et al. 2000. In this context, Mediterranean Echinodermata from middle and lower slopes have been poorly studied, particularly in comparison with the Atlantic Ocean where echinoderms are important in terms of abundance, biomass and ecosystem function (Billett, 1991). The large number of investigations conducted in the Atlantic Ocean, have resulted in a good taxonomic knowledge of the echinoderms (Mortensen 1903, 1943, Koehler 1921, Hérouard 1923, Hyman 1955, Sibuet 1979, Borrero Perez et al. 2003. In contrast there have only been a few studies on the taxonomy of Mediterranean deep-sea echinoderms (Marenzeller 1893, Bartolini Baldelli 1914, Tortonese 1954, Sibuet 1974, Alvà 1987b. Most reports provide only species lists; morphological descriptions are of secondary importance (Cherbonnier and Guille 1967, Alvà A total of 223 deployments were completed (Table 1) resulting in a total swept area of 10.3 km².
Of these hauls, 119 samples were obtained by a single warp otter-trawl Maireta system (OTMS, Sardà et al. 1998) with a net length of 25 m and a cod-end mesh size of 40 mm. A SCANMAR system was used to estimate the width of the mouth of the net. An average horizontal opening of 12.7±1.4 m was calculated. As the SCANMAR system can only operate down to 1200 m depth, the same value for the mouth's width of the net was used also for deployments deeper than 1200 m. The height of the trawl mouth was estimated to be 1.4 m (Sardà et al. 1998). In addition, 49 hauls were conducted with an Agassiz dredge, made of a square steel frame with a mouth width of 2.5 m and a mouth height of 1.2 m, and fitted with a 12 mm mesh net. Further, 55 samples were obtained with an epibenthic sledge, which consisted of a rectangular steel frame with three nets attached at different heights (10-50 cm, 55-95 cm and 100-140 cm above the bottom) with a mesh size of 300 µm (only one epibenthic sledge sample contained echinoderms). A total of 1503 individuals belonging to 11 species were sampled (Table 2). Of these, 196 were asteroids, 494 echinoids and 813 holothurians. The classes Crinoidea and Ophiuroidea were absent from all samples.

Specimen identification
The echinoderms were sorted, weighed, counted and fixed with 4% formalin diluted with seawater and neutralized with borax on board ship. After 30 days, the samples were transferred to 70% alcohol in the laboratory for further examination. Some specimens were fixed in absolute ethanol on board to allow for molecular analyses (not included in this study). All specimens are stored in the Biological Reference Collection of the Institute of Marine Science, Barcelona (Spain).
In the laboratory, all specimens were classified to species level. For microscopic examination of holothurian spicules small pieces of soft tissue (i.e. skin, tentacles, and gonads) were dissolved in bleach solution and mounted on glass slides for identification. The taxonomic results were compared to previous taxonomic studies. The nomenclature was checked against the World Register of Marine Species (WoRMS). The identification of the echinoid Gracilechinus elegans (Düben and Koren, 1844) was based on taxonomic descriptions from the Atlantic Ocean (Mortensen 1903, 1943, Koehler 1927, Minin 2012. This species has not been cited previously in the Mediterranean Sea. Its geographic distribution was compared with data in the Atlantic Ocean and other echinoids records from the Mediterranean Sea.

Synthesis of taxonomic information on deep-sea Mediterranean echinoderms
A comprehensive table was created of all the echinoderms recorded previously in the Mediterranean Sea at depths greater than 800 m (Table 3). Table 3 was constructed based on Tortonese (1965) and Koukouras (2007). New data acquired during the PROMETEO, DOSMARES and PROMARES cruises was added (see above).

RESULTS
Description: Shape pentagonal to stellate, very variable (Fig. 3A, B). Body flattened dorsoventrally. Oral and aboral surface composed by more or less tabulate hexagonal plates covered by little granules. Marginal plates thick and massive, from 18 to 22; sampling methods could take them out. R =from 6 to 45 mm. r = from 3 to 25 mm. R/r = 1.54 to 2.53. Colour variable, from cream, pale-yellow to pale pink. Polygonal madreporite, well defined, larger than surrounding plates. Adambulacral plate with 4 to 6 furrow spines, outside these a series of usually four clubshaped spines and outer spines similar to internal ones. Pedicellariae valvate, scarce on aboral side, larger and more numerous on oral side near ambulacral furrow. One of the specimens collected in the present study had 6 arms ( Fig. 3C) Note: Similarities were observed between the genus Litonotaster described by Halpern (1969Halpern ( , 1970. Nevertheless, owing to 1) the absence of the characteristic flat and thin abactinial plates of the genus Litonotaster, and 2) the presence of tabulate abactinial plates covered by granules, the marginal plates disposition and in agreement with available literature, we consider our specimens to be Ceramaster grenadensis. Litonotaster has not been reported in the Mediterranean Sea. Great intraspecific morphological variations have been signalled for Ceramaster grenadensis in the Mediterranean (Halpern 1970, Tortonese 1972, Sibuet 1974, Alvà 1987. It is likely that a revision of the genus Ceramaster is needed. Order BRISINGIDA Fisher, 1928 Family BRISINGIDAE G.O. Sars, 1875 Genus Hymenodiscus Perrier, 1884 Hymenodiscus coronata (G.O. Sars, 1872)  Brissingella coronata Tortonese, 1965: 194-196, fig Description: Diameter of disc 11 mm. 9 to 13 long and slender arms. Colour orange to reddish.
Very difficult to collect intact, usually the disc and the arms are broken and separate (Fig. 4).
Note: Description taken from (Clark and Downey 1992).
Gracilechinus elegans, known in the Atlantic, has been reported for the first time in the Mediterranean Sea in the present study. It was sampled in the Blanes Canyon at 1500 m depth ( Fig. 2). Other specimens were observed and collected with the ROV during the PROMARES cruise (Mecho, pers. obs.), in the lower Palamós Canyon and Blanes Canyon areas (1200 and 1500 m). Brissopsis lyrifera was found over a wide bathymetric range in the present study (from 900 to 2250 m; Fig. 2). It was abundant in some canyons between 900 and 1500 m, (Table 2). In contrast, only 5 small specimens of B. lyrifera were collected on the open slope at depths between 1750 and 2250 m depth (Table 2).
Description: Diameter test = from 38.5 mm to 48.3 mm; h= 25.6 mm to 34.7mm.
Test low, from conical and flattened above to slightly flatten on both sides, usually the height of the test is more than a half of the diameter (Fig. 5A). Colour whitish pink -pink, sometimes a few green (Fig. 5B, C). Long primary spines usually flat at the end. One primary tubercles present on each plate, forming a very regular series from oral to aboral side, usually secondary ones form a short longitudinal series from the middle to the oral side. A small tubercle was present between the pores and the primary tubercle, but not between the pores and the end of the plate. Some miliary tubercles are present giving a rough appearance to the test. Three pairs of pores very clearly and disposed in a strong angle. The boundary between the areas was more straight than sinuous. Periproct (Fig. 5D) covered by large irregular plates, one of them with a spine. The plates surrounding the anus are irregularly club-shaped and smaller than the other plates. Ocular plates not in contact with the periproct. No spines on the buccal plates where pedicellariae were present and abundant. Tridentate pedicellariae have the valves flat, narrow and mesh-worked, with the edge sinuate (from 500 to 650 µm long). In some cases small individuals had flatter valves than larger individuals (Fig. 5E). These valves have a narrow area near the base (Fig. 5F). Globiferous pedicellariae (500 to 550 µm) have usually 1 or 2 lateral teeth on either side of the blade and a more or less round to rectangular shape ( Fig. 5G -I). Ophicephalus pedicellariae, broad, sinuate and with small teeth in the edge, and an intricate mesh-work.
Note: Mortensen (1903) reported this species from the Mediterranean, but he later discarded this identification (Mortensen 1943). Alvà (1987b) described another species, Gracilechinus alexandri, in the Mediterranean Sea. Both G. elegans and G. alexandri have many similar characteristics, making their true identification difficult (Mortensen 1903, Ramírez-Llodra and Tyler 2006, Minin 2012. Furthermore, juvenile G. alexandri, have characteristics that might be confused with G. elegans. It is possible that the specimen of G. alexandri reported by Alvà (1987b) was a juvenile and was a misidentification of G. elegans. The specimen is no longer available for a comparison. In our specimens, the presence of one or two teeth on the globiferous pedicellariae, their narrow base and their mesh-work are similar to that described in the literature (Mortensen 1903, Minin 2012. The tubercular pattern, the periproct, the shape of the ocular and genital plates and their disposition, allowed us to classify these specimens as G. elegans. Mortensen (in 1903, pag.144, pl. XX, fig.9) found a small form for G. elegans with tridentate pedicellariae that had more flattened and truncate blades without mesh-work. This characteristic and the overlapping range in the number of teeth in the globiferous pedicellariae (from 1 to 4 in G. elegans and from 2 to 5 in G. alexandri) could lead to a misidentification if only one individual was available, as appears to be the case in Alvà (1987b).
Posterior petals shorter than the anterior ones, diverging and well separated. Globiferous short, ending in two long teeth. Tridentate pedicellariae of various forms, with 3 more or less leafshaped blades. Rostrate pedicellariae blade slender. Note: Differences from Brissopsis atlantica mediterranea (Mortensen 1913) are evident in the posterior petals: diverging and well separated in B. lyrifera and confluent on the base, as opposed to nearly parallel in B. atlantica mediterranea (Lacour and Néraudeau 2000).
CLASS HOLOTHUROIDEA de Blainville, 1834 The Holothuroidea was the most abundant echinoderm class sampled in this study, with a total of 813 specimens and 7 species (Table 2). Three species belonging to the Order Aspidochirotida    (Table 2).
Description: Large species, up to 30 cm long (Koehler 1927). Body nearly cylindrical with both ends flattened (Fig. 7A). Mouth subventral surrounded by 20 peltate tentacles. Scattered small tube feet all over the body, more abundant near the anterior and posterior ends. Dermis usually covered by shells, skin very fragile and thin in fresh specimens. On preservation, the dermis becomes thicker and more wrinkled. Characteristic ossicles are round tables (± 100 µm) that more or less regular with small peripheral holes around a central hole, and with central spire built by four rods, ending in a crown of several thorns (Perrier 1898) (Fig. 7B, C). Hermaphroditic species (Hyman 1955), gonads constituted by one branched tuft attached to left side of the dorsal mesentery, with some tubules male and some female, not found ripe at the same time (Mortensen 1927). Two respiratory trees, gelatinous and transparent. The species produces a substance which gels in formaldehyde and alcohol when preserved. Specimens usually eviscerate during capture.
Note: The presence of a second Mesothuria species of the genus in the Mediterranean Sea, Mesothuria verrilli (Théel, 1886) was discarded by Gebruk et al. (2012).
Distribution: Mediterranean Sea and North East Atlantic Ocean (Pérez Ruzafa et al. 1987). New depth range: 20 -850 m depth (present study). The previous maximum depth reported for this species in the Atlantic Ocean was 348 m (Pérez Ruzafa et al. 1987). The previous Mediterranean Sea maximum depth was 193 m (Pérez Ruzafa et al. 1987).
Description: 60 mm long. Cylindrical body flattened ventrally (Fig. 9A). Numerous tube feet in three or four rows. Conical papillae on its dorsal surface. Subventral mouth with about 20 stumpy, branched tentacles. Calcareous deposits scarce, as small discs in skin (Fig. 9B) and branched and curved rods in tube feet and tentacles. Colour, usually black with white spots, sometimes brown with a yellow ventral side. Cuverian tubules are present.
Description: Up to 200 mm long. Mediterranean Sea forms smaller. "Sausage" shaped, with a small tail (Fig. 10A). Terminal, mouth surrounded by 15 pink digitate tentacles with two small prolongations (Fig. 10B). Skin rough and thick, coloured from grey to dark purple, due to phosphatic deposits (Fig. 10A, B). Ossicles tables have few perforations and a small solid spine (500 to 700 µm). Rosettes and racquet-shape plates and anchors present (Fig. 10C). Fusiform rods (±1000 µm) always present in tail, usually also on body wall (Fig. 10D). Calcareous ring with posterior bifurcate projections on radial plates. Two long and slender respiratory trees. Ossicles and body shape could vary, but fusiform rods of the tail are diagnostic. Colour varies with the age and growth of the animal. In the early stages they are grey-white and, when grown to the adult size, the colour turns darker from the accumulation of phosphatic deposits.
Note: In the Mediterranean Sea, the maximum depth of distribution for this species was 1050 m depth (Tortonese 1965, Sibuet 1974, Cartes et al. 2009, Ramírez-Llodra et al. 2010 Family Caudinidae Heding, 1931 Genus Hedingia Deichmann, 1938 Hedingia mediterranea (Bartolini Baldelli, 1914) Tortonese, 1965  Description: Fresh specimens pale violet or white, acquiring a yellowish white colouring when conserved (Fig. 11A, B). Body divided on two regions, an elongated body and a long caudal appendage (more than a half the length of the body). Body oval, without podia. Rough skin due to calcareous plates. Anterior region wrinkled and cylindrical, with a terminal mouth. Skin without phosphatic deposits. Fifteen tentacles without digitations. Anus situated at the end of the caudal appendage. Five subdivided muscular bands visible by transparency. Ossicles very similar to H. albicans; tables (from 150 µm to near 250 µm) present all over the skin with very irregular holes and a central spine with three spiny columns (Fig. 11C -E). Smooth plates on anal papillae (Fig.   11F, G). Two respiratory trees (right and left), low ramified and attached along the mesentery.
Gonads long and unbranched tubules extending to the posterior end of the body , disposed in two tuffs attached to the mesentery on the upper part and free for the rest of their length in the coelom (Fig. 11H). Calcareous ring with 5 radial pieces, each with two posterior bifurcated projections and five interradial pieces (Fig. 11I, J). Tentacular ampullae long and digitate.
Note: Only one specimen has been reported previously in the Mediterranean Sea, dredged by R.N.
Description: Small species 5 -20 mm in length. Fragile animals with skin usually broke.
Digestive tract visible by transparency (Fig. 12A). Body elongated ovoid, with ventral side flattened. Six pairs of tube feet on the posterior half of the flattened ventral sole. Three pairs of papillae are present on the dorsal side, two pairs on the anterior part of the body and one pair on the posterior part. Ten tentacles surrounding the mouth (Fig. 12B), each divided into six to eight marginal lobes. Tentacles spicules curved rods with spines (130 -300 µm) at their ends and in the middle on the external side of the curve (Fig. 12C). Calcareous ring with five interlinked pieces, usually visible by transparency. Each piece has four pair of arms radiating from the centre (Fig.   12D). Arched rods with one or two spines and four spiny legs ossicles present (Fig. 12E, F).
Papillae spicules smooth rods (Fig. 12G). Marenzeller (1893) reports males and females, describing gonads as one tuft slender and ramified for males and short and less ramified for females.
Note: Penilpidia ludwigi has been reported twice in the eastern Mediterranean Sea basin (Marenzeller 1893, Fiege andLiao 1996) at depths from 755 to 4766 m. Its presence was reported in the north-western Mediterranean Sea from sediment traps at 22 m above the bottom at depths between 1200 and 1700 m in the Palamós Canyon (Pagés et al. 2007). Although, a specimen has been reported from a depth of only 48 m on the south-western coast of Portugal (Cunha de Jesus and Cancela da Fonseca 1999), there is some doubt about this identification owing to depth (very shallow) and substrate (i.e. rocky area), as well as the poor condition of the specimen. Gebruk et al. (2008Gebruk et al. ( , 2013) described a related species in the North Atlantic and included a re-description of the genus and its species.
Description: Typically U-shaped (Fig. 13A). Two opposite siphons, oral and anal. Body wall thorny, due to the presence of intricate scales, also visible with naked eye. Eight digitiform tentacles, of very unequal size, one on each side, being larger than the others. Calcareous plates visible with naked eye (Fig. 13B). Plates subcircular. Strong short spire placed near the edge of the plate (Fig. 13C). The plates are perforated by many small holes giving an irregular shape.
Calcareous deposits in tentacles. Calcareous ring with eight plates. Lateral interradial plates with anterior bifurcated projections (Fig. 13D, E). The projections are often asymmetric.
Note: Differs from Y. talismani by the bifurcated projections of the calcareous ring and the size of the plates (Gage et al. 1985, Alvà 1991.

General remarks
This study provides a thorough review of all citations and distribution information of deep-sea echinoderms in the Mediterranean Sea. The literature review showed that for some species only very limited biological/ecological data were available, and in many cases only species lists were provided (Tortonese 1979;Pérez-Ruzafa and López-Ibor, 1988). This paper provides new information of specimens collected in the last years, including new records, extensions of geographic and bathymetric distributions. Our new data include information from areas with complex topography such as canyons, which previously have been sampled inadequately. We have collected together information of echinoderms living deeper than 800 m.
Our results report, for the first time, the presence of the echinoid Gracilechinus elegans (Düben and Koren, 1844) in the Mediterranean Sea. In addition, there are new records of two species considered previously as "rare" in the Mediterranean Sea. At present, there is no consensus regarding what determines a 'rare species' (Cunningham and Lindenmayer 2005). In our study, taking into account all published information, we considered "rare" those species that have been reported less than five times in the whole basin. Based on this, two 'rare' holothurians, Hedingia mediterranea (Bartolini Baldelli, 1914) Tortonese, 1965and Penilpidia ludwigi (von Marenzeller, 1893 endemic to the Mediterranean Sea were identified. Additionally, we note greater bathymetric ranges for four species. The depth range of the asteroid Ceramaster grenadensis (Perrier, 1881), previously dredged in the Mediterranean Sea down to 2400 m (Carpine 1970, Tortonese 1979, Alvà 1987a, was extended to 2845 m. The echinoid Brissopsis lyrifera (Forbes, 1841), previously dredged around 1500 m (Sibuet 1974, Tortonese 1979, Cartes et al. 2009 Below, we discuss the results by Class. At the beginning of each section, if appropriate, we discuss first any new records and those of rare species. We then compare our results with the published literature, as detailed in Table 3.

Class Asteroidea
Our results for the Class Asteroidea were based on two typical bathyal species, Hymenodiscus coronata (G.O. Sars, 1872) and Ceramaster grenadensis (Perrier, 1881). The depth range of C. grenadensis has been expanded to 2845 m. Where their depth ranges overlapped (1500 m -2250 m depth) the two species co-occurred perhaps facilitated by their contrasting diets; H. coronata is a suspension feeder and C. grenadensis a secondary consumer (Carlier et al. 2009).

Class Echinoidea
This study reports for the first time the presence of Gracilechinus elegans (Düben and Koren, 1844) in the Mediterranean Sea. While Mortensen (1903) reported this species from the Mediterranean, he discarded the record in a later publication (Mortensen 1943). The lack of observations of G. elegans in the Mediterranean Sea could be caused by misidentification of congeneric species. For instance, adults of G. elegans are similar to juveniles of G. alexandri (see G. elegans description above). The only specimen of G. alexandri reported from the Mediterranean Sea (Alva, 1987b) was not available for comparison. Another species that could lead to misidentification in the Mediterranean Sea is Gracilechinus acutus var. norvegicus (Düben and Koren, 1844). The possibility of hybridization between species should be taken into account.
Hybridization has been described for other species of the same genus in Atlantic (Shearer et al. 1911). Hybrids themselves may be responsible for some failures in identification. Molecular studies of Mediterranean Sea and Atlantic Ocean specimens may be able to determine the species more clearly in the future, including hybridization and phylogenetic differences.
Brissopsis lyrifera was present in canyon muddy sediments below 900 m, as suggested originally by Carpine (1970). Large and dense aggregations of dead and live Brissopsis were observed by ROV in canyons. The gregarious behaviour of this species has been reported in previous studies (Laubier andEmig 1993, Ramírez-Llodra et al. 2008). Many echinoid tracks were visible on the sediment, suggesting a 'herd' in movement, similar to what has been observed for other bathyal echinoids (Salazar 1970, Gage andTyler 1991). Although the number of collected specimens was too low to conduct population structure analyses, we observed that smaller specimens appeared to occur at greater depths. This contrasts with the results of Ferrand et al. (1988) who proposed the recruitment of smaller individuals at shallower depths. Our results are in agreement with Harvey et al. (1988), who suggested a possible 'dwarfism' for this species at greater depths. Brissopsis lyrifera is usually reported from the upper slope (250 -400 m depth) on the Mediterranean continental margin (Tortonese 1965, Carpine 1970, Ferrand et al. 1988, Koukouras et al. 2007, Ramírez-Llodra et al. 2008, Cartes et al. 2009). The abundance of this species has decreased greatly in recent years on the upper and middle continental slopes at depths down to 1000 m (Mecho, pers. obs.) which may be related to intensive commercial trawling activity down to depths of 900 m (Ramírez-Llodra et al. 2010. Local fishermen have noted a large decrease of B. lyrifera in their by-catch in the last decade.
No specimens of the closely related species Brissopsis atlantica var. mediterranea Mortensen, 1913 were found.
Eight other species of echinoids have been reported from the Mediterranean Sea at depths below 800 m ( Table 3). Two of these species, Stylocidaris affinis (Philippi, 1845) and Cidaris cidaris (Linnaeus, 1758) are common in the deep sea and have been sampled frequently below 800 m in the Mediterranean Sea (Alvà 1987a, Cartes et al. 2009). However, these two species were absent from our samples. Other species that occur mainly at shallower depths, such as Spatangus purpureus O.F. Müller, 1776, and Gracilechinus acutus Lamarck, 1816 were also not sampled in the recent cruises, even though they have been reported previously at depths greater than 800 m.
Two deep 'rare echinoid species' are reported in the literature from the Mediterranean Sea: Hemiaster expergitus Lovén, 1874, sampled only three times (Cherbonnier 1958, Tortonese 1972, Koukouras et al. 2007 and Asterechinus elegans Mortensen, 1942, an Indo-Pacific species recently found in the eastern Mediterranean in association with sunken wood (Bienhold et al. 2013). These two species were not sampled in the present study. Three other species,

Class Holothuroidea
The holothurian Hedingia mediterranea was first described by Bartolini Baldelli (1914) in the Tyrrhenian Sea. Its presence has not been reported since in the Mediterranean. It is possible that specimens reported as H. mediterranea have been misclassified as sipunculids because of the similar body shape between the two groups. Some studies have cited H. mediterranea as endemic to the Mediterranean Sea (Koehler 1921, Tortonese 1963, Parenzan 1970, Fredj 1974, Koukouras et al. 2007, Matarrese 2010, but only by referring to the original record of the type specimen. Accordingly, we consider the individuals sampled in this study as a truly 'rediscovered' species and extending both its geographic range to the north-western Mediterranean Sea and its bathymetrical distribution. One sample collected in the Blanes Canyon at 1200 m included 4 individuals and another at 1500 m in the same area had 5 specimens, suggesting a greater presence of this species in canyons. Pawson (2001) considered the Bartolini-Baldelli specimen as Hedingia albicans (Théel, 1886) Deichmann, 1938. This species is known from several locations in the North Atlantic. However, no explanation was provided for the synonymy of H. albicans and H. mediterranea. The information available does not allow us to classify if the Mediterranean specimens (classified as Hedingia mediterranea) are the same or distinct species than the Atlantic species (classified as Hedingia albicans). In the present study we continue to classify the species as H. mediterranea following Tortonese (1963). A molecular comparison between species of Hedingia would help to resolve the taxonomic discrepancies.
The only species of Elpidiidae present in the Mediterranean Sea is Penilpidia ludwigi. This is also considered to be 'rare' species, because it has been reported only three times previously, twice from the eastern Mediterranean Sea (Marenzeller 1893, Fiege andLiao 1996) and once from the deep western Mediterranean Sea (Pagés et al. 2007). However, when it does occur it may be found in abundance. Pagés et al. (2007) collected 150 individuals. More than 200 individuals were collected in one epibenthic sledge sample suggesting that the species may occur in dense aggregations (Fiege andLiao 1996, Pagés et al. 2007), similar to those reported for other Elpidiidae in the Atlantic Ocean (Billett and Hansen 1982, Billett et al. 2001, 2010, Gebruk et al. 2003, Ruhl and Smith 2004. The presence of P. ludwigi in the Blanes Canyon sediment traps adds new faunistic records for this area. Pagés et al. (2007)

collected P. ludwigi in the Palamós
Canyon also with sediment traps moored at 22 m above the bottom. Our sediment traps sampled greater numbers in autumn and winter, coinciding with a stormy period in the north-western Mediterranean (Sanchez-Vidal et al. 2012). This may have resulted in greater re-suspension of bottom sediments and associated small fauna, such as P. ludwigi. Another factor that can cause resuspension of sediments, and thus the collection of small holothurians in sediment traps, are deep currents (Gebruk et al. 2013). In addition, swimming behaviour has been described in other Elpidiidae (Ohta 1985, Pawson and Foell 1986, Miller and Pawson 1990 and has been proposed also for P. ludwigi (Pagés et al. 2007). Swimming cannot be discarded as an explanation of the presence of this species in sediment traps. Pagés (2007) suggested that aggregations of P. ludwigi might occur during periods coincident with phytoplankton spring blooms and the flux of new organic matter to the seafloor. Although our sediment traps sampled greater numbers of specimens in autumn (similarly to the epibenthic sledge sample) and winter, these seasonal peaks of abundance may also indicate periodic recruitment of opportunistic species, as reported for other small species of Elpidiidae (Billett and Hansen 1982, Ohta 1985, Billett 1991, Billett et al. 2001, 2010. The Class Holothuroidea was the most speciose and most abundant of all the groups collected in our samples, similar to the north Atlantic deep sea (Billett, 1991;Gage and Tyler, 1991 Cartes et al. (2009) from 1600 m in the same region. Another species of the same genus, Mesothuria verrilli (Théel, 1886), has been reported from the Mediterranean Sea (Koukouras et al. 2007), but the presence of this species in the Mediterranean Sea was reviewed and discarded by Gebruk et al. (2012). Pseudostichopus occultatus Marenzeller 1893, a cosmopolitan aspidochirotid species, showed a restricted geographic and bathymetric distribution in our samples, occurring only between 2000 and 2200 m on the open slope, but in very high abundances.
The presence of large aggregations of individuals near canyon axis could be related to food inputs (Morgan and Neal 2012). Submarine canyons act as conduits of organic matter from the shelf to bathyal/abyssal depths . The aggregations of P. occultatus may respond the periodic changes in food availability originating from canyon refluxes, as proposed for Mesothuria. To our best knowledge, the presence of Holothuria (Panningothuria) forskali Delle Chiaje 1823, at mid-bathyal depths has not been reported previously. The deepest records were at 345m off the Canary Islands (Pérez Ruzafa et al. 1987, Hernández et al. 2013. The specimen sampled in the present study came from the Blanes canyon at 850 m depth. Two species of the Order Molpadiida were collected. Molpadia musculus Risso, 1826, was present only in open slope areas. Hedingia mediterranea occurred mainly in canyon areas. Both species are deposit feeders and live infaunally. Molpadia musculus was reported as a typical canyon species in the Atlantic Ocean (Amaro et al. 2009) and in other Mediterranean Sea areas (Ramírez-Llodra et al. 2008, Cartes et al. 2009). Nevertheless, no specimens of M. musculus were found in our canyon samples. The high presence of H. mediterranea inside canyons suggests habitat specialization, but further sampling inside canyons is necessary to confirm this hypothesis.
The Order Dactylochirotida was represented by a single species, Ypsilothuria bitentaculata (Ludwig, 1893). The presence of this species only at middle slope depths is commonly reported (Pawson 1965, Gage et al. 1985. This species reported from the Mediterranean Sea only in the early 1990s (Alvà, 1991). Subsequently Ypsilothuria bitentaculata has been cited by other authors (Massin 1996, Cartes et al. 2009) and also as Y. talismani by Ramírez-Llodra et al. (2008). Little information is available for Ypsilothuria in the Mediterranean Sea. A detailed discussion on its taxonomy must await further sampling.
Of the holothurians species reported previously from the deep (>800m) Mediterranean Sea, only two species did not occur in our study (Table 3) profundicola (Kemp, 1905), has been reported at 900 m (Marenzeller 1893, Tortonese 1958. One species typical of shallower Mediterranean waters, Parastichopus regalis (Cuvier, 1817) has been cited at 834 m depth by Marenzeller (1893), but no other reports are known for these depths.
Finally, there are three other species, Panningia hyndmanni (W. Thompson, 1840), Pseudothyone raphanus (Düben and Koren, 1846) and Thyone gadeana Perrier R., 1898, which while they have maximum depth ranges extend to around 1000 m depth in Atlantic Ocean, occur no deeper than 300 m in the Mediterranean Sea.

Class Crinoidea
Crinoids were totally absent from our samples. Three species of crinoids have been cited from the bathyal Mediterranean seafloor (Table 3). Only one of them, the endemic crinoid Leptometra phalangium (J. Müller, 1841), has a maximum depth of distribution greater than 800 m. Stalked crinoids were not reported in the Mediterranean Sea (David et al., 2006).
There are some records of high abundances of Leptometra phalangium in upper slope areas (100 m to 400 m depth) (Pérès andPicard 1956a, Mifsud et al. 2009) as observed for the same genus in other areas (Fonseca et al., 2013). The deepest record for this species is 1292 m (Marenzeller 1893). However, despite these deeper records, not a single crinoid was collected in any of our hauls, or was observed during the ROV dives. Their occurrence at predominantly shallower depths (Hellal 2012) may explain the absence of these crinoids in our samples.

Class Ophiuroidea
Ophiuroids were also totally absent from our samples. Nine species of ophiuroids have been cited previously from the Mediterranean Sea at depths between 300 and 1219 m (Table 3), with only two species, Ophiura (Dictenophiura) carnea Lütken, 1858 ex M. Sars, and Ophiotreta valenciennesi (Lyman, 1879) cited below 800 m (Tortonese 1979, Mifsud et al. 2009). All nine species have been reported from depths greater 800 m in the Atlantic Ocean, but their maximum depth of distribution in the Mediterranean Sea are shallower. This may explain the lack of ophiuroids in our study.

Endemicity in echinoderms from the Mediterranean
There has been considerable debate as to whether the deep-sea fauna of the Mediterranean is truly endemic or is a sub-population of Atlantic species (Bouchet andTaviani 1992, Tyler 2003). The shallow Gibraltar Sill may be a significant barrier for the influx of larvae of echinoderms from the Atlantic and may act as an isolating mechanism once populations are established in the Mediterranean. The higher temperatures of deep water in the Mediterranean may mitigate against the immigration of species from the deep Atlantic. Our observations lean towards the concept that Mediterranean deep-water echinoderms are sub-populations derived from deep water in the Atlantic as exemplified by Gracilechinus elegans. However, increased sampling effort and molecular analyses are required before this aspect is fully resolved.