A highly diverse siliceous sponge fauna (Porifera: Hexactinellida, Demospongiae) from the Eocene of north-eastern Italy: systematics and palaeoecology

A siliceous sponge fauna, consisting of more than 900 specimens, is described from an early Lutetian tuffite horizon in the Chiampo Valley, Lessini Mountains, north-eastern Italy. Thirty-two taxa (15 Hexactinellida, 17 Demospongiae) are determined and illustrated, belonging to 24 genera, two of which are new (Rigonia gen. nov. and Coronispongia gen. nov.). Among these, 10 new species are proposed: Stauractinella eocenica sp. nov., Rigonia plicata gen. et sp. nov., Hexactinella clampensis sp. nov., Camerospongia visentinae sp. nov., C. tuberculata sp. nov., Toulminia italica sp. nov., Coronispongia confossa gen. et sp. nov., Cavispongia scarpai sp. nov., Corallistes multiosculata sp. nov. and Bolidium bertii sp. nov. Of the genera identified at Chiampo, 14 range back to the Cretaceous, three to the Jurassic and one to the Triassic, while six are still extant. The studied fauna shows affinities with sponges from the Eocene of Spain and the Cretaceous of Germany. The sponge fossils are uncompressed and bodily preserved, but the original siliceous skeleton is dissolved and substituted by calcite. Delicate attachments can be nevertheless documented: some sponges attached to a hard substrate by encrustation, while others were anchored on soft sediments by root-like structures. The presence of different modes of attachment suggests heterogeneous substrate conditions. Small, possibly young, sponges are recorded too. The sponge fauna is essentially autochthonous and lived in the middle-outer part of a carbonate ramp, where it formed clusters. This study extends the geographical and stratigraphical range of many sponge taxa, including Camerospongia, Toulminia, Ozotrachelus and Bolidium, previously documented only from the Cretaceous. The Recent calcified demosponge genus Astrosclera is reported here in the Cenozoic for the first time, having been reported previously in the Triassic only. Additionally, this study documents the second worldwide occurrence of the Recent sphinctozoan genus Vaceletia in the Palaeogene, formerly recorded exclusively in Australia. http://zoobank.org/urn:lsid:zoobank.org:pub:B3466955-8E20-429A-89BE-42BAEB4002E8


Introduction
Sponges are among the oldest extant multicellular animals. They appeared in the Precambrian and were already well established by the Cambrian, becoming major reef builders during the Palaeozoic and Mesozoic (Brunton & Dixon 1994;Wood 1998). Siliceous sponges play a significant role in marine ecosystems (Van Soest et al. 2012); for example, they contribute to the silicon cycle (Maldonado et al. 2005(Maldonado et al. , 2010Chu et al. 2011;Tr eguer & De La Rocha 2013). However, little is known about the abundance and diversity of sponges in the geological past: due to the low preservation potential of many taxa, their fossil record is rather incomplete. Only sponges with a rigid skeleton, such as lithistids (demosponges with desmas), Hexactinosida and Lychniscosida (both representatives of Hexactinellida), and sponges with a massive calcareous skeleton (a polyphyletic assemblage within demosponges and Calcarea), have a more or less continuous fossil record that is nevertheless inadequately studied, especially for the Cenozoic (Pisera 2006). For siliceous sponges (lithistids and hexactinellids), the fossil record is very rich in the Jurassic (e.g. Quenstedt 1877À8; Schrammen 1936Schrammen , 1937Leinfelder et al. 1993;Pisera 1997) and the Cretaceous of Europe (e.g. Schrammen 1910Schrammen , 1912Moret 1926;Reid 1958Reid , 1959Reid , 1961Lagneau-H erenger 1962). Many diverse sponge faunas are known from the Miocene of the Mediterranean area (e.g. Pomel 1872; Moret 1924;Brimaud & Vachard 1986a, b;Matteucci & Russo 2011 and references cited therein). Palaeogene sponges in contrast have a rather scarce record (Pisera 1999). Eocene bodily preserved sponge faunas are known from five areas. In Europe, Bartonian sponge faunas are known from Spain (Ebro and Pamplona Basin À Pisera & Busquets 2002;Astibia et al. 2014 and references cited therein) and France (Aquitanian Basin, Basque Country À d'Archiac 1846, 1850. In both cases hexactinellids are the dominant group, but lithistids occur in Spain as well. In North America, Bartonian sponges (hexactinellids, 'soft' demosponges, lithistids and calcareous sponges) have been reported from the Castle Hayne Formation, North Carolina, by Rigby (1981) and Finks et al. (2011). Late PaleoceneÀearly Eocene sponge faunas were reported by Buckeridge et al. (2013), Kelly et al. (2003) and Kelly & Buckeridge (2005) from New Zealand, where hexactinellids are more abundant than lithistids. The Priabonian Pallinup Formation in south-west Australia also yields sponges, mostly lithistids (Pickett 1983;Gammon et al. 2000;Pisera & Bitner 2007 and references cited therein). In the late Eocene Oamaru diatomite in New Zealand there are isolated sponge spicules (Hinde & Holmes 1892;Edwards 1991).
Eocene siliceous sponges from the Chiampo Valley have been reported only by Menin (1972), Visentin (1994), Frisone, Fornasiero et al. (2014) and . Matteucci & Russo (2005) provided illustrations and preliminary determinations of 23 species. The poor knowledge of the Chiampo sponge fauna is in contrast with the richness of public collections in the museums of Vicenza and Venice and the University of Padua. These collections have never been studied or catalogued. The aim of the present study is to describe the taxonomy of the Chiampo sponges in order to shed a new light on the diversity of siliceous sponges of Eocene age. By assessing the sponge diversity of the Eocene of the Chiampo Valley, this study contributes to knowledge of sponge distribution through time, and provides new information on their role in Eocene marine ecosystems.

Geological setting
The study area is located in the eastern Lessini Mountains, a portion of the Prealps of north-eastern Italy, on the west side of Chiampo Valley (Fig. 1). The Lessini Mountains are a triangle-shaped tableland, and occupy some 800 km 2 in the western Venetian Region, at the transition between the Prealps and the Po Plain. They belonged to the Cenozoic Lessini Shelf, a resurrected carbonate platform with scattered reefs, lagoons, islands and volcanoes circumscribed northwards by lands and surrounded by deeper water marine settings (e.g. Bassi et al. 2008;A. Bosellini 1989;F.R. Bosellini & Papazzoni 2003). The studied sites were located within a NNW-trending extensional structure known as the Alpone-Agno or Alpone-Chiampo graben (e.g. Barbieri et al. 1982Barbieri et al. , 1991, bounded to the west by the Castelvero normal fault. The full extent of the graben is unknown because the outcrop area is truncated to the north and east. Zampieri (1995) proposed that the Alpone-Chiampo graben was 20 km wide and at least 35 km long. The area belongs to the Veneto Volcanic Province (VVP), identified by principally mafic and ultramafic rocks erupted during the PaleoceneÀOligocene, mainly in submarine environments. Large volumes of mainly subaqueous volcanics and their penecontemporaneous reworking products (hyaloclastites and tuffites) accumulated in the graben. The eruptive centres of the eastern Lessini Mountains were aligned with the Castelvero fault (e.g. Piccoli 1966). Several magmatic pulses occurred, separated by periods of magmatic inactivity during which marine sedimentation took place (De Vecchi & Sedea 1995 and literature therein). As a result of alternating volcanic activity and sedimentation, intercalated within volcanic rock successions as thick as 200 m are thick beds of limestone, locally called 'Chiampo limestone'. This unit was quarried until the 1990s as a building stone. It belongs to a lowerÀmiddle Eocene informal unit named 'Nummulitic limestone', widespread in the western part of Veneto (e.g. Fabiani 1915). This lithostratigraphical unit is not well constrained, and includes limestones with Nummulites of different ages and depositional settings (Bassi et al. 2013;Papazzoni et al. 2014). Beccaro et al. (2001) interpreted the 'Nummulitic limestone' of the studied sites as belonging to the outermost facies of a carbonate ramp and the volcaniclastic debris as transported by sediment gravity flows (debris flows and turbidites). In the study area, there is evidence of palaeocurrents: volcaniclastic sediments are often lens shaped with a channelized morphology (Mart on et al. 2011), and transport sedimentary structures are visible in the field. Pelagic fossils (e.g. pteropods) are commonly found, and locally (Lovara quarry) there are accumulations of plankton and nekton fossils (e.g. planktonic foraminifera, shark teeth) (Beccaro et al. 2001). Volcaniclastic debris, tuffites or reworked tuff beds exhibit a faunal association more complete than usual. Some volcaniclastic levels are extremely rich in very well-preserved fossils (e.g. threedimensional crustaceans complete with appendages and ventral parts), that belong to several endemic species (e.g. Beschin et al. 1991;De Angeli & Garassino 2006). In a few sections, sponges constitute the most common macrofaunal element, especially in a relatively thin horizon of volcaniclastics within the 'Nummulitic limestones' of a few quarries on the western flank of the Chiampo Valley. Sponges are absent or rare in all other Eocene localities of the Venetian Prealps.

Material and methods
The studied sponge material comes from two adjacent quarries both located in the municipality of Chiampo (Vicenza): Cengio dell' Orbo: 45 32' 25.56 "N, 11 15' 44.47" E (called 'Boschetto di Chiampo' in Beschin et al. 1991;Beccaro et al. 2001 and other references);and Lovara: 45 32' 11.87 "N, 11 15' 58.92" E (part of which is named 'Zanconato' in, for example, Ancona 1966;Visentin 1994). The quarries have been closed since the 1990s and the sponge-bearing level is now inaccessible. A single small section (less than 3 m thick) that yielded only a few fossil sponges was found near Cengio dell'Orbo quarry.
The study material consists of more than 900 specimens, housed in six Italian public museums: Museo di Storia Naturale di Venezia (MSNV); Museo di Archeologia e Scienze Naturali 'G. Zannato', Montecchio Maggiore (MCZÀPAL); Museo Civico 'D. Dal Lago', Valdagno (MCV); Museo 'Padre Aurelio Menin', Chiampo (MMC); Museo di Geologia e Paleontologia dell'Universit a degli Studi di Padova (MGP-PD and IG-PD); and Museo Naturalistico Archeologico, Vicenza (IGÀVI). Of the material studied, only 261 specimens were taxonomically useful. The others were either too poorly preserved or could not be adequately prepared. The study material was mainly collected by amateur palaeontologists between the 1960s and 1990s. The amateur palaeontologists frequently used mechanical preparation techniques to remove attached volcaniclastics, and therefore the sponge surface was often smoothened, destroying important characters of the outer surface (e.g. Figure 2. Simplified stratigraphical sections of Lovara and Cengio dell'Orbo quarries modified from Beccaro et al. (2001), with sponge-bearing horizons (asterisks). Facies codes are provided on the left of the stratigraphical column. The sponge-bearing horizon at Lovara quarry has a question mark as sponges were not found during fieldwork in this study. The data thus rely on labels of museum specimens and personal communications (Antonio De Angeli pers. comm.). rim around canal openings, small outgrowths, papillae). As methods of study depend on sponge preservation (Finks 2003b), the petrology of the specimens had to be investigated. Therefore, some specimens were etched in dilute acetic acid to reveal strongly calcified spicules. As the cement around the spicules is a carbonate too, the preparations chosen for the present study were polished hand sections and thin sections. In order to dissociate spicules from the entire specimen, various etching procedures were also tested on some sponge fragments, but none of them provided isolated spicular skeletons. The treatment of the sediment (method explained in ) also did not yield any spicules. Initially, spicules were searched for in each specimen under a binocular microscope. In some specimens with no evident megasclere on the surface, a polished section was prepared, and in many cases, a fused or articulated skeleton was recognized. Selected specimens were chosen for preparation of 52 thin sections, which confirmed that both the spicules and cement are made of calcite sparite. Veils of micrite and peloidal micrite always form coatings around spicules.
Reflected light observations on entire specimens were performed with a Leica MZ 125 optical binocular microscope. Thin sections were studied with petrographic microscopes, under transmitted optical (Leica DM EP T and Zeiss Axiophot) and fluorescent (Leica 5000B) light, all at Padova University. Classification and terminology for sponges generally follows Kaesler (2003Kaesler ( , 2004, Hooper & Van Soest (2002) and Boury-Esnault & Rutzel (1997).
Diagnosis. Globular to compressed subglobular sponges with rounded terminal osculum. Spongocoel is divided by radial folds of the wall. Dermalia, atralia, gastralia are mostly pentactines while choanosomal megascleres are mostly diactines and hexactines.
Derivation of name. For the Eocene age of the type locality.
Description. The holotype (MCZ-PAL 3795) is 9.1 cm high, 9.9 cm wide and 12.5 cm long. The smallest specimen (MSNVEÀ22855) is 2.1 cm high and 2.8 cm in diameter while the largest (MMC 35) measures 18 and 20 cm, respectively. Wall thickness varies from 1.35 to 4 cm. The osculum is generally rounded (Fig. 3A), ranging from 0.75 to 12 cm in diameter. Some specimens show on the external surface rounded canal openings of 4À6 mm in diameter (Fig. 3C). The external surface of some specimens is eroded so that the interior of the spongocoel is visible (Fig. 3B, C). The spongocoel is divided by radial folds of the wall forming large radial to irregular chambers that can be up to 9 cm long and 4 wide, visible on eroded specimens. Dermalia (Fig. 3D) are mostly pentactines. Their tangential rays are usually parallel to the sponge surface and are 2À8 mm long, while proximal rays are 1.6À4.0 mm long (Fig. 3DÀF). Choanosomal megascleres are mostly diactines and hexactines. Their length is difficult to measure precisely as they are seen only in sections. Nevertheless, their dimensions are extremely variable, ranging from 2.4À4.8 mm for diactines, to 0.4À4 mm for hexactines and their other derivates ( Fig. 3G, H). Gastralia are pentactines, with four rays tangential to the spongocoel wall and the proximal ray towards the choanosome (hypogastralia). Proximal rays are 0.8À1.2 mm long (Fig. 3I).
Remarks. As reported by Pisera & Busquets (2002), Pomel (1872) erected numerous species of his genus Laocoetis ( D Craticularia Zittel, 1877) from the Miocene of Algeria based mostly on slight differences in general growth form. Moret (1924), however, did not take into account skeletal differences, and synonymized all Pomel's species into Laocoetis crassipes Pomel. Aside from shape differences, there are differences in basic skeletal structure among Pomel's species and differences in the shapes of canal openings on both surfaces. At least three different Laocoetis species were distinguished in Pomel's material by Pisera & Busquets (2002). One has round to elongated canal openings on both surfaces, another has canal openings on the outer surface with rectangular shapes and rounded canal openings on the inner surface, and the third has rounded canal openings on the outer surface and rectangular ones on the inner surface. We assigned our material to Laocoetis patula Pomel, 1872 emended Pisera & Busquets (2002) because of the rectangular canal opening on the outer surface and the rounded to oval openings on the inner surface. Unfortunately, in many specimens internal canals opening are not observable, as the spongocoel is encrusted by sediment or poorly preserved. Nevertheless, the clear presence of rounded canal openings on the inner surface of some specimens supported the assignment to L. patula with reasonable confidence. Craticularia stellata Lagneau-H erenger, 1962, from the Lower Cretaceous of south-east France and north-east Spain, shows the same canal opening pattern: rectangular in the outer surface, rounded in the inner surface. Nevertheless, our material differs in general shape and spiculation from C. stellata, which has a cup-like or narrow cylindrical shape while our material is either cone-like or platy. Craticularia stellata shows spiny hexactines on canal openings of the outer surface. This feature was not observed in our material but, as spicules are strongly calcified, we cannot exclude that this is due to poor preservation. Additionally, some Chiampo specimens show rounded external pores, but we suspect that this could be due to abrasion caused by preparation. On the reverse of a few specimens, due to poor preservation, only the internal surface with rounded openings is visible, making the assignment uncertain. In fact, Laocoetis crassipes Pomel, 1872 is characterized by rounded canal opening on both surfaces. Moreover, the tuberose base, present in some specimens, is nearly identical in shape and in the longitudinal canals to L. crassipes (Pomel 1872;Moret 1924; Miocene of southern Spain: Ott d'Estevou & T ermier 1978; Pliocene of Sardinia, Italy: Matteucci 1989). Bulbose attachments are an adaptation to the substrate and do not have, in our opinion, a taxonomic value. The tuberose base observed in our material is also similar to those present in the species described by Malfatti (1901) from the middle Miocene of Emilia Romagna (northern Italy): C. manzonii, C. globularis and C. ranzorei. These species are no longer considered valid since Moret (1924) synonymized them with 'Craticularia crassipes'. Regarding the Eocene, d'Archiac (1850) reported Laocoetis samueli from the Biarritz area (southwestern Aquitaine Basin, Basque Country). The same species (that has oval canal openings both on the outer and inner surfaces) occurs in the Pamplona Basin (western Pyrenees) (Astibia et al. 2014). Finks et al. (2011) reported poorly preserved Laocoetis sp. cf. L. crassipes from Caste Hayne (North Carolina, USA). Pisera & Busquets (2002) reported L. patula from the Ebro Basin (Spain). In our material the 'hemispherical enlarged nodes' on the surfaces of the dictyonal skeleton reported by Pisera & Busquets (2002) were not observed, but this could be due to poor skeletal preservation.
Regarding the shape, there are fossil (e.g. Reid 1964;Pisera 1997) and Recent (e.g. L evi 1986) Laocoetis species that are cone-like when small and then become more plate-like. The variety of shapes in the studied material can thus be interpreted to represent different ontogenetic stages. Laocoetis represents the longest living genus of Hexactinellida, ranging from the Late Jurassic to the Recent (Mehl 1992;Reiswig 2002a;.
Description. Tubular-branched craticulariid up to 16.1 cm high and 12 cm in diameter ( Fig. 5A) that is composed of three to nine cylindrical to subconical tubes ( Fig. 5B) 1.3À3.5 cm in diameter. Rounded to subrounded terminal oscula are 0.7À1.8 cm in diameter. Wall thickness is 4À6 mm. Rectangular canal openings on outer surface are in quadrate arrangement and measure 0.7À0.8 £ 1.5À2 mm. They are separated by skeletal bridges 1À1.5 mm wide (horizontally) and 1 mm high (vertically). Canal openings of the inner surface are rectangular as well, measuring 0.5 £ 1.0À1.5 mm, and are separated by skeletal bands 0.5À1 mm wide horizontally, 1 mm wide vertically. Radial canals are 0.7À0.9 mm wide and cross almost entirely the sponge wall (Fig. 5C). The skeleton is euretoid with mainly quadrate dictyonal meshes 200À399 mm in size, dictyonal strands diverging toward both surfaces.
Remarks. In describing Craticularia emiliana from the Miocene of Italy, Malfatti (1901) considered fragments of sponges that could correspond to singular tubes of our material. Nevertheless the external canal openings are mainly rectangular, while in the Miocene material they are subrounded. In our material, radial canals nearly completely cross the sponge wall, while in Malfatti's material they are shorter. The species was illustrated and preliminarily described by Matteucci & Russo (2005) as Paracraticularia sp. This genus, originally described by Schrammen (1937), was synonymized by Reid (2004b) with Craticularia Zittel, 1877. We agree with his opinion that the branching habitus has no taxonomic value at the genus level. Our material is very similar, in general morphology, to Laocoetis fittoni (Mantell, 1822) (Malfatti, 1901). A, lateral view of MSNVEÀ23020 with branched habitus and canal openings in quadrate arrangement (craticulariid pattern); B, top view of MSNVEÀ23020 with rounded tubes; C, detail of MSNVEÀ22903 showing quadrate dictyonal meshes and radial canals almost entirely crossing the sponge wall. Dictyonal strands are diverging toward both surfaces. Description. Sponges with star cross section that are 4.6À8.8 cm high, 4.2À15.3 cm wide, 2.8À10.2 cm long, and usually composed of 5À6 wings. These wings are united in the lower part, but may be separate, forming elongated branches directed upward and outward, in the upper part (Fig. 6A, B). The largest specimen (MCZ-PAL 3784b) is 5.5 cm high, 15.3 cm wide and 8.5 cm long, and shows irregular fins instead of wings (Fig. 6C). Radial fins may be considered an adaptation to improve stability, due to large dimensions (Finks 2003a) (see 'Sponge autecology' below). Along the edges of the wings, there are papilliform outgrowths (4 mm in height) with rounded oscula 0.6À2 mm in diameter (Fig. 6A, D). The wings are 9À17 mm thick and have walls 2À7 mm thick. Canal openings on the outer surface are rounded to subrounded, 0.2À0.4 mm in diameter, either irregularly distributed ( Fig. 6E) or in quincuncal arrangement. The skeleton is euretoid with node-to-node beam connection and regular meshes, from square (140À200 mm) to triangular (120À200 mm) (Fig. 6F). In transverse section, both canal systems (which can be interpreted as epirhyses and aporhyses sensu Reid 1963) perforate the wall in opposite directions and generally terminate immediately below the opposite surface (Fig. 6G).
Remarks. The presence of irregularly distributed canal openings on the external surface together with the pattern of canal systems supports the assignment to Guettardiscyphia thiolati (d'Archiac, 1846). The papilliform outgrowths observed in some specimens also complies with Reid's redefinition (1961, p. 743) of the genus Guettardiscyphia de Fromentel, 1860 that includes occurrence of accessory parietal oscula along margins of flanges or branches.
The genus is widespread in the Cretaceous of Europe: United Kingdom (Reid 1961(Reid , 1968, Germany (e.g. Schrammen 1910 as Guettardia), France (Lagneau-H erenger 1962) and Czech Republic ( Zitt et al. 2006). Within the material studied, some specimens may belong to either Guettardiscyphia thiolati (d'Archiac, 1846) or Pleuroguettardia iberica Pisera & Busquets, 2002, which are homeomorphic species/genera very similar in general morphology but belonging to different families, having different canalization patterns, i.e. Cribrospongiidae (cribrospongiid or irregular pattern) and Craticulariidae (quadrangular, craticulariid pattern). This pattern is often disrupted in both genera, especially near parietal gaps, so assignment of poorly preserved/fragmentary material to either of the genera may be difficult (Pisera & Busquets 2002), and assignment is rather difficult without detailed study (Astibia et al. 2014). Matteucci & Russo (2005) assigned some specimens from Chiampo Valley to Pleuroguettardia sp. and Pleuroguettardia aff. iberica. The last species is also known from the Eocene of the Vic Marls Formation (Pisera & Busquets 2002).
Diagnosis. Fan-shaped sponges, consisting of a thin, strongly plicated wall forming deep, narrow furrows and rounded ridges on one side. On the other side, a labyrinthine pattern of folds has large openings leading to the cavaedial spaces. The walls are perforated by deep, straight and closely spaced canals. The choanosomal skeleton has triangular meshes.
Derivation of name. In honour of Francesco Giuseppe Rigoni (Vicenza, Italy) who inspired this work.
Remarks. Our material is similar in its fan-like shape to the tretodyctid Ramalmerina Brimaud & Vachard, 1986b from the Miocene of Spain, except for the lack of schizorhyses and the absence of oscula at branch bifurcations. In R. fisheri Brimaud & Vachard, 1986b (p. 426), dictyonalia form mainly rectangular and large meshes (700 £ 200 mm), while in our material they are triangular and small (110À150 mm).
The genus Auloplax Schultze, 1904 was traditionally considered to belong to the family Tretodictydae, but Reiswig (2002b) transferred it to Dactylocalycidae because it completely lacks schizorhyses. Reiswig & Kelly (2011, p. 136), in a study on Recent hexactinellids from New Zealand, proposed moving Auloplax Schulze, 1904 from Dactylocalycidae to the resurrected family Auloplacidae, presenting a new diagnosis for this monogeneric family. This latter interpretation fits our material: body consisting of several vertical plates or fans composed of conjoined thin-walled tubes, which split with an acute angle and remain tightly connected side-by-side. Specifically, Auloplax breviscopulata Reiswig & Kelly, 2011 has a choanosomal skeleton with oval apertures similar to our material. Choanosomal skeleton beam length is comparable: approximately 110 mm for both genera. Nevertheless, there are fundamental differences in the skeleton framework between Auloplax and Rigonia: firstly, the wall of Auloplax has very shallow canalization, while in Rigonia it is deeply canalized; secondly, the choanosomal skeleton of Rigonia has mainly triangular meshes, while in Auloplax rectangular and polygonal forms are also common.
Derivation of name. From the Latin 'plicatus' ( D folded).
Description. Fragment of a group of two branching tubular sponges 10.6 cm high, 4À4.1 cm in maximum diameter, with rounded (1.6 cm in diameter) to elliptical (2.5 £ 0.9 cm in size) terminal oscula (Fig. 8A). On the external surface, irregularly distributed rounded to subrounded canal openings 1À2 mm in diameter are separated by skeletal bridges 0.5À2 mm wide (Fig. 8B). The sponge wall (9À14 mm in thickness) is deeply canalized, with cleft-like to labyrinthine cavities (schizorhyses) (Fig. 8C).
The dictyonal skeleton has mainly triangular meshes 120À180 mm in size. The nodes are swollen (Fig. 8D).
Remarks. The presence of a euretoid skeleton with schizorhyses supports assignment to the family Tretodictyidae. The occurrence of swollen dictyonal nodes strongly suggests affinity to some living genera: Anomochone Ijima, 1927, Psilocalyx Ijima, 1927and Cyrtaulon Schulze, 1886(Reiswig 2002b. The Psilocalyx body, however, is composed of small tubes (e.g. P. nitidus Schrammen, 1936), which are not present in our material. On the other hand Cyrtaulon lacks the central spongocoel. The skeleton of our specimen is very similar to that of Anomochone but with less plicate morphology. However, in the branching habit the studied specimen is similar to another tretodictyid, genus Sclerothamnopsis Wilson, 1904, but this genus lacks swollen nodes. Sclerothamnopsis collina Brimaud & Vachard, 1986b from the Miocene of southern Spain has a branching habit too, but the diameter of the branches is significantly smaller (10 mm) and the spiculation is unknown. Type locality and stratum. Cengio dell'Orbo quarry, Chiampo, Italy, Eocene, Lutetian volcaniclastics.
The body is composed of anastomosing tubes and lamellae. Tube openings range from subcircular (5À8 mm in diameter), to elongate (10À18 mm on the longer axis). The tube walls on the outer surface are 2À4 mm thick without an intradictyonal canal system (Fig. 10D). In cross section, the tube walls are either meandriform or anastomosing and form lamellae 4À6 mm thick, or bowlshaped structures (Fig. 10E). Meshes of the dictyonal skeleton are rectangular (300À375 £ 360À425 mm node to node) to quadrate (250À425 mm node to node), and have lychniscs (Fig. 10F).
Remarks. The studied specimens show two morphological varieties: one with more meandriform openings (Fig. 10A) and thinner walls, the other with mainly subcircular openings and generally thicker tube walls (Fig. 10B). These two morphotypes may belong to two different species. However, considering that one large specimen (MCZ-PAL 1380) shows both morphological features and wall thickness variability could also depend on fossil preparation, we regard them as conspecific. Traditionally, meandriform sponges have been classified as Plocoscyphia Reuss, 1846, but Reid (1962 clearly demonstrated that different sponges occur under this generic name (Pisera & Busquets 2002, p. 341). The studied specimens are homeomorphs of Exanthesis Regnard in Moret, 1926 and Robinia Finks, Hollocher & Thies, 2011, but they lack the labyrinthine canal system characteristic of these genera. The studied specimens are also morphologically nearly identical to Callicylix farreides Schrammen, 1912, but differ in the absence of a spongocoel. They show also affinities with genera Brachiolites Smith, 1848 and Centrosia Schrammen, 1912. In determining these specimens, we encountered problems of material preservation and difficult classification. Generally, callodictyonid sponges are poorly described in the literature and a revision of the family is needed. In fact, there are numerous fossil genera that are similar, and there are no clear-cut differences between them. Specifically, in our material the poor preservation and the lack of diagnostic characters (e.g. cortex, peripherical capsule, lychniscs sculptures) make the assignment difficult, if possible at all, in some cases. Material. One specimen: MCZ-PAL 1379. Occurrence. Cengio dell'Orbo and Lovara quarries, Chiampo, Italy, Eocene, Lutetian (Matteucci & Russo 2005; this study).
Remarks. The pattern of canal openings was difficult to observe due to problems in removing the embedding sediment. A thin section was prepared from the base of the specimen (to protect morphology), in which some lychniscs were observed but the internal canalization was not seen. The genus Ventriculites is characterized by straight, unbranched canals that are perpendicular to the wall (e.g. Swierczewska-G»adysz 2012). These characters could not be observed in the studied material. As a consequence, the assignment is uncertain. Due to the conical shape and large canal openings on the external surface, our material is similar to Cretaceous representatives of the genus. Many species, though, show more everted conical shapes and external furrows which are not observed in the studied specimen. The genus Ventriculites has a rich fossil record from the Cretaceous of Europe (e.g. Hinde 1884Hinde [1883; Lagneau-H erenger 1962; Reid 1962Reid , 1968 Swierczewska-G»adysz 2006, 2012; Olszewska-Nejbert & Swierczewska-G»adysz 2013). Pomel (1872) erected but did not illustrate three Ventriculites-like genera from the Miocene of Algeria: Stelgis, Cladostelgis and Pleurostelgis. Reid (2004c) synonymized these with Ventriculites. Another record (without illustration) of the genus is Ventriculites poculum Zittel from the Paleocene of Austria (K€ uhn 1930).
Camerospongia visentinae sp. nov. (Fig. 12) Diagnosis. Low-conical Camerospongia with a rounded to elliptical terminal osculum, around which a small elevated rim is present. The top surface is either flat or slightly inclined, and covered by a siliceous membrane. On the lateral surface, there are large rhomboidal to irregular canal openings, in quincuncal or irregular arrangement. The base may be supplied with a ledge. Description. Low-conical sponges (Fig. 12A, D) that are 3.3À7.6 cm high, 4.7À12.9 cm wide and 3.6À9.8 cm long. The holotype (MCZ-PAL 3784c) is 3.9 cm high, 5.3 cm wide and 3.8 cm long. The conical spongocoel extends to the base of the sponge. The terminal osculum is elliptical, ranging from 1.0À1.3 £ 2.3À6.2 cm in dimension ( Fig. 12B, C, E) to rounded 1.8À1.9 cm in diameter (Fig. 12F), and supplied with a narrow elevated rim 2À4 mm in height. The top surface of the sponge is covered with a smooth siliceous membrane. This membrane may also appear at the base of the sponge. The uppermost part of the sponge has a larger diameter than the rest, and forms an overhanging ledge over the lateral surface (Fig. 12A). The lateral surface of the sponge displays densely distributed, large, longitudinally elongated, rhomboidal to irregular canal openings that are 1.5À2.5 mm in diameter (Fig. 12G) and quincuncally or irregularly arranged. The base, when present, is encrusting, sometimes supplied with a ledge 3 mm wide (Fig. 12D). The sponge wall reaches up to 18À22 mm in thickness. The wall is pierced by large radial canals, 0.6À1.2 mm in diameter (Fig. 12H). The choanosomal skeleton has mainly quadrate meshes, with beams 400 mm (node to node) in length. On the external side of the sponge wall, beams are 300À500 mm long and meshes are more irregular in shape.

Remarks. The investigated specimens differ from
Camerospongia fungiformis (Goldfuss, 1831) in having a low-conical shape, a flat or slightly inclined À rather than convex À top surface, a ledge at the base, and large rhomboidal to irregular canal openings, in quincuncial or irregular arrangement. Camerospongia tuberculata sp. nov. has a cylindrical shape, low conical outgrowths and no ledge at the base. Although the genus Camerospongia has its roots in the Jurassic (Gaillard 1983;Pisera 1997), its fossil record is mainly from the Cretaceous of Europe: Germany (Schrammen 1912 Type locality and stratum. Cengio dell'Orbo quarry, Chiampo, Italy, Eocene, Lutetian volcaniclastics. Description. Cylindrical sponge slightly narrowing towards the base, 2.4À2.8 cm in diameter and 1.7À4.1 cm high. The holotype (MSNVEÀ22973) is 2.5 cm in diameter and 3.5 cm high. The uppermost part of the sponge has a larger diameter than the rest and forms an overhanging ledge over the lateral surface (Fig. 13A, C). The terminal osculum is large and rounded (0.7À1.1 cm). The surface  around the terminal osculum is flat and smooth, suggesting the original presence of a siliceous membrane (Fig. 13B, D). On the lateral surface there are outgrowths 4À5 mm in diameter and 3À4 mm tall. The spongocoel is narrow and cylindrical, running through the entire sponge. The canals are 0.8À1.6 mm in diameter (Fig. 13E). The skeleton has lychniscs with node-to-node connections and quadrate (200À300 mm) to rectangular (150 £ 300 mm) meshes (Fig. 13F).
Remarks. The specimen resembles the Jurassic Multiloqua fungiformis (Goldfuss, 1833)  Description. Narrow to irregularly conical sponge (Fig. 14A, F) that is 3.1À7 cm high, 1.7À3.1 cm wide and 2À4 cm long. The holotype (MSNVEÀ22972) is 4.6 cm high, 3.1 cm wide and 2.5 cm long. The terminal osculum is elliptical, 0.2À0.5 £ 0.7À1.5 cm wide (Fig. 14C, E, H). Some specimens have a second osculum on one side (Fig. 14B) that is elliptical and up to 1 £ 0.3 cm wide. The spongocoel is narrow and elliptical in cross section. The wall is up to 15 mm thick and composed of tubes that are 3À5 mm in diameter. Low-conical lateral outgrowths 5 mm in diameter and 3À5 mm in height are present, and have rounded ends (Fig. 14A, F). An external siliceous membrane on the upper part of the sponge is developed in the holotype (Fig. 14C). The external surface of all specimens displays, at the sides, a characteristic originally siliceous membrane showing horizontal concentric lines (Fig. 14A, D, G). The choanosomal skeleton is composed of lychniscs. In transversal section, meshes of the dictyonal skeleton are quadrate (400À500 mm, node to node) to rectangular (400 £ 500 mm).
Remarks. The smooth surface around the terminal osculum supports the assignment to the family Camerospongiidae Schrammen, 1912 Schrammen, 1912 from the Cretaceous of northern Germany, but the latter are more elongated and never exceed 2À3 mm in diameter. This is the first record of Toulminia outside the Cretaceous. As a consequence the last appearance of the genus is now moved forward to the Lutetian (Eocene).
Derivation of name. From the Latin corona ( D crown) for the general shape.
Diagnosis. Conical to cup-shaped sponge composed of branching and radiating tubes, with a deep spongocoel. The dictyonal skeleton is lychniscosid, canalized and generally irregular. The canal openings on the outer surface are rounded to elongated.
Remarks. The assignment of this new genus to a family is a critical issue, as there are different points of view in the literature. In the Treatise (Reid 2004c), the only truly canalized lychniscosid family is Dactylocalycidae Gray, 1867. This family has a complex history and was traditionally considered to belong to the Hexactinosida. According to Reiswig (1991), Reid's (1958Reid's ( , 1962 transfer to the Lychniscosida was surprising. Reiswig (1991Reiswig ( , 2002c, who studied the problem in detail, returned the family to Hexactinosida due to the lack of lychniscs in the dictyonal skeleton. Recent Lychniscosida include only two families: Aulocystidae and Diapleuridae (Reiswig 2002c). Taking into account the lack of loose spiculation and of many diagnostic characters À due to the poor preservation of our material À we tentatively locate the studied specimens in the Recent family Diapleuridae Ijima, 1927(Reiswig 2002d). This decision is made on the basis of the similarity in skeletal canalization and the irregular framework of dictyonalia.
Wall is up to 30 mm thick. The wall is composed of a branching and anastomosing network of tubes 5À7 mm thick (Fig. 15A, C, F). The tubes circumscribe external cavedial spaces of approximately the same thickness. Tubes can be rounded at their ends (Fig. 15B, D). In the larger specimens, the wall is folded and fused to form a series of radial parietal tubes around the spongocoel (Fig. 15G). Radiating tubes are up to 15 mm wide and 40 mm long and have, distally, a circular opening (5À10 mm in diameter). The spongocoel is conical to cup shaped, and 3À6 cm deep. The dictyonal framework is irregular or, locally, with triangular to quadrate meshes (beams are 200 mm long, node to node). The intradictyonal canals are rounded (1À3 mm) to elongate (up to 5 mm in length) (Fig. 15E, H). The dictyonal skeleton is lychniscosid (Fig. 15I).  Reiswig & Wheeler (2002) in the case of the tortuous taxonomic history of the genus Myliusia. Our specimens are also very similar in shape to Myliusia cancellata Brimaud & Vachard, 1986b from the Miocene of Spain, but differ in spiculation (in Myliusia there are no lychniscs). Brimaud & Vachard (1986b) doubtfully synonymized the species with Tretostamnia favosa Pomel, 1872 from the Miocene of Algeria, suggesting also a revision of the taxon that was briefly described by Pomel (1872, pp. 70À71, pl. 2 bis, fig. 1). Matteucci & Russo (2005) assigned some of their specimens to the lychniscosid genus Brachiolites sp. This assignment should be treated with caution for the reason that an intradictyonal canal system was observed, while Brachiolites generally lacks channels (Reid 1962, p. 34;Reid 2004c  Description. Conical sponge (Fig. 16A, C, D) 6.1À11.8 cm high, 4.1À14.5 wide and 2.4À9.5 cm long. The holotype (MSNVEÀ22003) is 11.8 cm high, 14.5 cm wide and 9.5 cm long. The body is usually composed of two to four (up to 12) cylindrical branches with generally circular terminal oscula 6À14 mm in diameter (Fig. 16B, E), and wall 4À6 mm thick. The outer surface has rounded to roughly elongated knobs (Fig. 16A, D) 3À5 mm high, 6À8 mm wide and 9À21 mm long. The knobs are sometimes fused and separated by meandering furrows 3À6 mm wide. Some specimens show a small tubular stalk (Fig. 16C) 17 mm in diameter. The dictyonal skeleton is preserved only in small sponge fragments. As a consequence only a few meshes could be observed. Meshes are mainly quadrate, beams are 300 mm long node to node. (2005)  Remarks. Demosponges with desmas have been traditionally described as lithistid demosponges (from the order Lithistida Schmidt, 1870). Due to the fact that they have been found to be a polyphyletic group sharing just one common character that developed independently several times, i.e. the articulated choanosmal spicules called desmas, this taxon has been abandoned, and the families have been distributed mostly among Astrophorida and Spirophorida (Schuster et al. 2015). The term 'lithistid sponges' thus has no taxonomic significance, but exclusively a morpho-functional meaning. Occurrence. Cengio dell'Orbo quarry, Chiampo, Italy, Eocene, Lutetian.
Remarks. The typical stalked and subglobular habitus, the presence of a shallow central depression and the rounded canal openings on the outer surface support assignment to the genus Siphonia Goldfuss, 1826. The piriform shape appears in different 'lithistid sponges', e.g. Scytalia curta Moret, 1926 (with rhizoclones desmas) and Melonella radiata (Quenstedt) (didymoclones). Our specimens resemble also in shape another tetracladine, Phyllodermia Schrammen, 1924. Siphonia and Phyllodermia differ in the ectosomal spicules, which are not preserved in our material. As a consequence, the assignment of our specimens to Siphonia is tentative. Menin (1972) and Visentin (1994) already reported the presence of Siphonia sp. in the Lovara and Cengio dell'Orbo quarries. Catullo (1856) reported Siphonia from the Cretaceous and Eocene of north-eastern Italy, but recently Matteucci & Russo (2011) demonstrated that none of the specimens described by Catullo belong to siliceous sponges. Manzoni (1882) reported a doubtful Siphonia from the Miocene of northern Italy.
Description. Subcylindrical to cylindrical sponge fragments (Fig. 18A, B) that are 4.2À4.5 cm in diameter, with wall 13À20 mm thick and a central spongocoel 1.5À2.5 cm in diameter. Rounded canal openings 0.7À2 mm in diameter are irregularly distributed on the external surface. The choanosomal skeleton shows radial and longitudinal canals 0.5À1 mm wide (Fig. 18C), and is composed of tetraclone desmas (Fig. 18D), probably smooth, that are up to 450À569 mm in size.
Remarks. This sponge strongly resembles in shape the Cretaceous tetracladine genus Rhoptrum. It also shows affinities to another Cretaceous tetracladine, Phymatella Zittel, 1878. These two genera are very similar in general shape, canalization and canal opening pattern. The main difference is in ectosomal spicules, which are not preserved in our material. As a consequence the assignment is uncertain. On the other hand, Phymatella is characterized by a tubular spongocoel with lateral chambers (e.g. Corallistes multiosculata sp. nov. (Fig. 19) Diagnosis. Club-or double club-shaped sponge. Numerous small circular oscula are found on the flat or slightly convex top; the lateral surface is finely porous. The base is an encrusting disc. The desmas are strongly arched and tuberculated dicranoclones. Description. Small club- (Fig. 19A) or double clubshaped (Fig. 19D) sponge 2.3À5 cm high, 1.1À4.3 cm wide and 1.2À1.8 cm long. The holotype (MSNVEÀ22912) is 3.9 cm high, 1.8 cm wide and 1.5 cm long. On the top, numerous circular oscula are present (Fig. 19B, E) which are 1À1.5 mm in diameter and 1À2 mm apart from one another. The lateral surface is finely porous with rounded canal openings measuring 0.3 mm in diameter (Fig. 19D). A basal encrusting disc up to 2 cm in diameter is sometimes present (Fig. 19A). In cross section, vertical canals 0.8À1 mm in diameter are seen that lead to the openings visible on the top surface (Fig. 19C). Desmas are dicranoclones 120À200 mm in size (Fig. 19F). Some fragments of monaxial spicules (Fig. 19G) 200À400 mm in length, 20 mm wide were also observed. Because these spicules were observed in thin section only, they could be genuine monaxial À oxeas, strongyles or styles (tips are not clearly visible) À or fragments of triaene rhabdomes.  Remarks. This sponge is very close in shape and oscula organization to Meta sp. (Pomel, 1872), synonymized without figures by Moret (1924)  ?Corallistes sp. (Fig. 20) Material. One specimen: MCZ-PAL 3707.
Description. Subglobular sponge 2.8 cm high, 4.9 cm wide and 3.1 cm long. The body is composed of two subglobular parts (Fig. 20A), with respectively three and five oscula on the top. The oscula are rounded (2 mm in diameter) or elliptical (2À3 £ 4À5 mm), and separated by 2.5À3 mm (Fig. 20B). In cross section, vertical canals of 2.2À2.4 mm in diameter open on the top surface. Desmas are poorly preserved, tuberculate, and 40À60 mm thick, and resemble dicranoclones (Fig. 20C).
Remarks. Desmas are too poorly preserved for a more precise assignment, and their size could not be measured due to poor preservation. The specimen is similar to Corallistes multiosculata sp. nov., from which it differs in general shape and larger, less numerous oscula. The specimen is also similar in general shape and oscula organization to Meta gregaria Pomel, 1872. Nevertheless, Pomel's species is elongate, while our material has a subglobular shape.
Description. Conical to cylindrical (Fig. 21A, F) sponges with rounded or, rarely, flat top, 1À7.4 cm high, 1.5À3.5 cm in diameter. An encrusting base may be present (Fig. 21F). The terminal osculum is rounded, 0.2À0.5 cm in diameter, and the wall is 6À19 mm thick. Radial canals are present on the top surface and run from the osculum margin to the top edge (Fig. 21B, D). The external surface has small (0.2À0.3 mm in diameter) rounded canal openings. The spongocoel is narrow, deep and cylindrical. Canals run horizontally or bend downward (0.3À0.7 mm wide) toward the sponge wall (Fig. 21C). Desmas are sphaeroclones (Fig. 21E), with 3À5 clones, approximately 200 mm in length (arm to arm).
Remarks. Our specimens are identical to Pachytrachelus conicus illustrated in Schrammen (1910). Some larger specimens that may belong to the same species show a flat top and cylindrical shape, but the type of desma is unrecognizable. The larger specimens seem more similar to Phyllodermia houzeti Ott d'Estevou, Termier & Termier, 1981 from the Miocene of southern Spain that have tetraclones. Another morphologically similar species is Cucumaltina placocephalus Brimaud & Vachard, 1986a from the Miocene of southern Spain; however, this has rhizoclone desmas. All the studied specimens are conical/cylindrical in shape, display radiating furrows on the top, and have a central rounded osculum and a deep, narrow spongocoel. Nevertheless, without well-preserved desmas, we cannot completely reject the idea that our material belongs to a different taxon. This is particularly true for the larger, cylindrical specimens. The species was reported previously only from the Upper Cretaceous of Germany. As a consequence, the range of the species is extended to the Lutetian (Eocene).

Order uncertain
Remarks. The systematic position of both Recent and fossil demosponges with rhizoclone desmas is a matter of debate (Schuster et al. 2015): Recent sponges with rhizoclones show affinities to Spirophorida (due to presence of sigmaspire microscleres in many taxa), but their relation to Astrophorida remains unclear.
Diagnosis. Massive compound sponge, with lobated branches arising from a common, encrusting base. Various small, rounded oscula at the top of each branch. Two systems of canals are present, one vertical and larger, the other subhorizontal, finer and more meandriform. Desmas are rhizoclones with rounded lumps.  (Fig. 22A, B) 2À3.4 cm thick and 2.5À6 cm long. The holotype (MSNVEÀ22815) is 8.3 cm high, 16.2 cm wide and 15 cm long. On top, various small, rounded oscula are present (Fig. 22C), 2À3 mm in diameter. The distance between oscula is 2À5 mm. Two systems of canals are seen in cross section, one vertical with canals 1.4À1.6 mm in diameter and the other subhorizontal and more meandriform, with canals 0.24À0.80 mm wide (Fig. 22D). Desmas are rhizoclones 200À250 mm in length, covered with rounded lumps (Fig. 22E, F).
Remarks. Although the general morphology, shape of desmas and small oscula at the top of each branch fit with Bolidium descriptions, we were not able to observe in our material small pores on the external surface, probably due to inadequate preparation.
The compound shape, with Jereica-like branches arising from a common base and numerous small oscula on the top, is similar to Polyierea dichotoma Roemer, 1864 (p. 36, tab. 14, fig. 1) from the Late Cretaceous of northwest Germany, but this species has tetraclone desmas. The exhalant system is very similar to that of Jereica Zittel, 1878, but our material is different because oscula are widely spaced, while in Jereica sp. they are closer and more numerous.
Rhizoclones of the Cretaceous species B. arbustum Hurcewicz, 1968 are similar to those of Jereica polystoma (Roemer, 1864). Also similar are their arrangement in strands and the exhalant part. Because of these common features, Hurcewicz (1968) suggested a close relationship between Bolidium and Jereica. Schrammen (1910) reported B. palmatum (Roemer, 1864) from the Cretaceous of Germany but without illustration.
This genus is known only from the Cretaceous of Europe: Germany (Roemer 1864;Zittel 1878;Schrammen 1910), Poland (Hurcewicz 1968) and Greece (Mermighis & Marcopoulou-Diacantoni 2004). This is the first record of Bolidium from the Cenozoic and thus the range of the genus is extended to the Lutetian (Eocene).
Description. Conical to sub-cylindrical sponge (Fig. 23A, C), 6À15 cm in height and 6À8.4 cm in diameter. The spongocoel is deep, tubular or slightly conical, running throughout the sponge. A circular osculum is present (Fig. 23B), 3À4 cm in diameter. The wall thickness is 17À20 mm. The majority of the specimens bear, on the outer surface, irregular to cylindrical outgrowths (Fig. 23B, C) 6À12 mm high and 9À14 mm wide. Irregularly distributed canal openings, 0.4À0.5 mm in diameter, are situated on the outer surface. The skeletal framework is compact with radial canals (0.3À0.4 mm in diameter), and is composed of rhizoclone desmas (Fig. 23D).
Description. Short, cylindrical sponge (Fig. 24A), 4.5 cm in height and 4 cm in diameter. It has a shallow, bowl-like terminal depression (Fig. 24C), 1 cm in diameter. No central spongocoel is present but a bunch of vertical canals occur which open in the terminal depression with small rounded openings (diameter 0.5 mm). There is another system of irregularly radial canals that open on the outer surface with rounded, irregularly distributed openings (Fig. 24B). Desmas are rhizoclones (Fig. 24D).
Remarks. Our specimen is identical in shape and terminal depression to Jerea acerra Pomel, 1872. The specimen illustrated by Matteucci & Russo (2005) shows on the external surface elongated outgrowths not observed in our material. The side canal openings have an irregular shape in their material while they are rounded in the material studied here. The general shape of our specimen is also similar to that of Moretispongia micropora Lagneau-H erenger, 1962 from the Aptian (Cretaceous) of Spain, but our specimen lacks the typical rimmed canal openings on the sides. The exhalant system is very similar to a Recent rhizomorine from New Caledonia, Jereicopsis graphicophora L evi & L evi, 1983. Jereopsis clavaeformis (Pomel, 1872) was previously recorded from the Miocene only. The first occurrence of genus Jereopsis in the Eocene, suggested by Matteucci & Russo (2005), is thus here confirmed. Description. Discoidal to cup-like (Fig. 25A) sponge, 3.2À7 cm in diameter. The wall thickness is 5À7 mm. The upper surface has numerous raised, rounded canal openings (Fig. 25B), irregularly distributed and 0.5À0.7 mm in diameter. The distance between canal openings is 2À3 mm. In thin section, a dense skeleton of heavily calcified desmas (Fig. 25C) was observed, but their shape was unrecognizable.
Remarks. The surface of all studied specimens appears to have been smoothed by mechanical preparation. Although typical rhizoclones were not observed, and characteristic raised canal openings were visible in one specimen only (MCZ-PAL 3802), we attribute all specimens to Verruculina ambigua (Pomel, 1872) for the general shape, the dense skeleton with desmas, and the pattern of canal openings on the upper surface. The smaller specimens are discoidal, while the larger ones are vase-shaped. This feature was observed by Ott d' Estevou et al. (1981) as well. The species has been found in the Miocene of Algeria (Pomel 1872;Moret 1924) and southern Spain (Ott d'Estevou et al. 1981;Brimaud & Vachard 1986a). This is the first record of the species for the Eocene. Verruculina albanyensis Chapman & Crespin, 1934, from the Eocene of Western Australia (see also Pickett 1983), differs from the studied material in having a thicker wall (8À9 mm), and larger (1.75 mm in diameter) and less numerous canal openings on the upper surface.
Remarks. The presence of rhizoclones, small canal openings on both surfaces, small radial canals and a leaflike habitus support the attribution to the genus Platychonia Zittel, 1878. This is a typical Jurassic genus but it was reported also from the Eocene of Australia, with the species Platychonia tertiaria Chapman & Crespin, 1934 (p. 117, pl. 11, fig. 22). Unfortunately, the Australian species is poorly illustrated and preserved. The attribution to a rhizomorine lithistid is dubious as the desmas are described as globular and having "4 to 7 or more radiating arms" (Chapman & Crespin 1934, p. 117); thus, they could possibly be sphaeroclones. The leaf-like habitus is similar to that of other rhizomorine sponges: Phlyctia expansa and Histiodia undulata from the Miocene of Algeria, described by Pomel (1872) and revised by Moret (1924) and Pisera & Busquets (2002). Our material is different from Phlyctia in lacking the fibrous divergent skeletal structure, and from Histiodia in lacking external longitudinal furrows. In both Miocene genera, radial canals are missing. Another rhizomorine genus, Chonellopsis, has a very similar morphology but its canal openings are on the upper surface only (Schrammen 1937, p. 96), while the studied specimens show openings on both sides. Our material is comparable to the Cretaceous Chonella tenuis Roemer, 1864(see Schrammen 1910, 1912, but it lacks the concentric growth lines, and the wall is thicker (12 mm in our material, 4À6 mm in C. tenuis).
Description. Vase-shaped and thick-walled (8À14 mm) sponge (Fig. 27A), 3À9 cm in height and 3.7À7.7 cm in diameter. Deep spongocoel. The surface is smooth, without canal openings. There are two systems of canals, one radial and larger, the other finer and meandriform and descending from the top (Fig. 27B). Desmas are possibly rhizoclones (Fig. 27C).
Remarks. This specimen is left undetermined because it is poorly preserved. The habitus resembles that of the Jurassic rhizomorine genus Hyalotragos but the studied specimens lack the diagnostic vertical canals.
Description. Cylindrical fragment (Fig. 27D) 7 cm high and 4.4 cm in diameter, having a thick wall. One large, circular osculum is present on the top (Fig. 27E). Rounded canal openings are visible in some parts of the surface. The skeleton framework is fibrous and composed of desmas that are possibly rhizoclones (Fig. 27F).
Remarks. Desmas are faintly visible in thin section. The skeleton framework is similar to that of Phlyctia expansa from Spain (Pisera & Busquets 2002), but this latter species is flat.
Remarks. Due to the poor preservation of the material, the taxonomy of this sponge remains undetermined.
Remarks. Due to the poor preservation of the spicules, this taxon cannot be determined.
Remarks. Our specimen is morphologically identical to the Upper Cretaceous Discodermia gleba Schrammen, 1910 (p. 98, tab. 15, fig. 2). Schrammen indicated a 'chestnut' size while our specimen is wider. Due to poor preservation, the type of desmas is not recognizable; thus, a taxonomic assignment is not possible.
Remarks. The upper surface of the studied specimens is very similar to that of Pliobolia vermiculata Pomel, 1872 described from the Miocene of Algeria, but the main canal openings and the radial furrows are less pronounced in our material. Nevertheless, the presence of a spherulitic microstructure together with astrorhizae-like canals on the external surface supports the assignment to genus Astrosclera (Vacelet 2002a and literature cited therein). The stratified internal structure observed in thin section can be interpreted as growth rings similar to those observed by W€ orheide (1998). The lack of megasclere (e.g. diagnostic verticillated acanthostyles) precludes determination at the species level. Astrosclera is regarded as a living fossil, with a first record in the Upper Triassic of Turkey (A. cuifi W€ orheide, 1998) (W€ orheide et al. 2002). The only Recent species, A. willeyana, is restricted to cryptic and lightreduced environments of the Indo-Pacific, with a depth range of 1À185 m (Hartman 1980b). Astroscleridae are the main representatives of the calcified demosponges (also known as 'coralline sponges') that were classified within class Sclerospongiae. This class was subsequently abandoned because it was shown to be polyphyletic (Van Soest & Hooper 2002 and literature cited therein). This is the first record of Astrosclera in the Cenozoic, which bridges the gap between the Triassic (hitherto the youngest known fossil) and the extant forms.
Description. Cylindrical, externally and internally segmented (Fig. 30A, D) fragments 1.5 to 1.7 cm high and 0.7 to 1.3 cm in diameter. The domal upper surface bears a circular terminal osculum 1.8 mm in diameter (Fig. 30B). The spongocoel is cylindrical, 1.8 mm in diameter. The sides of the spongocoel have a continuous wall (endowall) 0.15À0.2 mm thick, running along the whole specimen. Sponge walls are 2À3 mm in thickness. Internally, the sponge consists of a series of annular, crescent-shaped chambers which extend throughout the full width of the sponge wall (Fig. 30D). They are 0.4À0.6 mm high and traversed usually by pillars 0.05À0.1 mm thick. Subpolygonal canal openings, 0.1 mm in diameter, occur on the external surface (Fig. 30C).
Remarks. Our material is morphologically almost identical to the specimens from Pallinup Siltstone of south-west Australia (Eocene, Priabonian) described by Pickett (1982). Chambers are smaller in our specimens, and, curiously, more similar to the living species Vaceletia crypta (Vacelet, 1977)   encrustation (e.g. basiphytous hexactinellid Aphrocallistes); others anchor on soft sediments by a basal spicule tuft (e.g. hexactinellid Hyalonema). In our material, some specimens of Ozotrachelus conicus and Corallistes multiosculata (lithistids) show a basal subcircular disc, 2À3 cm in diameter (Fig. 31A, B). In Recent sponges, basal discs are associated with the presence of a hard substrate. In some specimens, such as the lithistid Bolidium bertii and some indeterminate rhizomorines, an encrusting base has been observed (Figs 22B, 27J). There is more direct evidence that the sponges studied were attached to a hard substrate at least in their initial life stages. Some specimens of hexactinellids and lithistids are still attached to larger foraminifera (Fig. 31F, G). Red algae and larger foraminifera (Nummulites sp.) were found incorporated in the skeletons of the sponges (Fig. 31H, I). In one case, a Hexactinella sp. was found encrusting red algal nodules (Fig. 31J). In other cases, small cavities (0.8À1.6 cm high, 0.7À1.6 in diameter; Fig. 31K) are found at the base of complete sponges that are interpreted as originally hard objects that dissolved during diagenesis, or were organic in nature and decayed leaving an empty space. Tuber-like or root-like basal structures were observed instead in some hexactinellids. Three specimens of Laocoetis patula, for example, show a tuber-like basal part that is 2.5À4.6 cm in diameter and 3.6À5.8 long (Fig. 31C, D). Delicate, root-like structures of 4À4.5 mm in diameter were observed in a specimen of Guettardiscyphia/Pleurogettardia, although their delicate tips are always broken (Fig. 31E). These basal structures differ from those of Laocoetis, because they are more delicate and smaller in proportion to the complete sponge body. As for the basal disc, these structures are also often lost after breaking off.

Sponge clusters
In spite of the fact that the material available for this study was collected decades ago, and thus it is now impossible to establish whether some of the sponges were found in life position, six slabs bearing numerous sponge specimens were found in museum collections (Fig. 32A, B). These sponge aggregates are mainly composed of hexactinellids (hexactinosan and lychniscosan), with the most abundant species being the lychniscosan Callicylix eocenicus. Other hexactinellids, such as Camerospongia visentinae, Guettardiscyphia/Pleurogettardia and Laocoetis patula, were also identified. The individuals in these specimens apparently grew one on top of the other (Fig. 32E). We interpret these structures as natural sponge clusters that could not be transported en masse, and therefore infer that at least some of the sponges of Chiampo Valley were collected in life position.

Small specimens
More than 150 specimens among the over 900 of this study are less than 3 cm high, some 1.5 cm or smaller. For the majority of these small specimens, taxonomic attribution was impossible due to the paucity of diagnostic features. Nevertheless, a few can be assigned to described taxa with reasonable confidence, including Laocoetis patula (Fig. 32C, D), Stauractinella eocenica, Callicylix eocenicus and Toulminia italica. These taxa show a wide range in dimensions. Considering entire specimens only, the smallest specimen of Toulminia is half the size of the largest (height range 3.1À7 cm), the smallest specimen of Callicylix is one-third the size of the largest (3.6À12.7 cm), the smallest specimen of Stauractinella is one-eighth the size of the largest (2.1À18 cm) and the smallest specimen of Laocoetis is one-fourteenth the size of the largest (1.1À16 cm). Although we cannot exclude the possibility that many of the small specimens were adults, those for which a taxonomic assignment was possible and which belong to taxa represented by large-sized specimens should be considered young forms. We suspect that smaller specimens would have been recovered if the outcrops were accessible for unbiased sampling.
Among 261 identifiable specimens, the most abundant species is Stauractinella eocenica, followed by Callicylix eocenicus, Laocoetis patula, Camerospongia visentinae and Ozotrachelus conicus. Seventy-four percent of the specimens belong to Hexactinellida, and 24% to Demospongiae. In terms of species diversity, the Shannon index (Hammer et al. 2001) is 2.966, which should be considered high in comparison with other marine invertebrate Eocene communities (cf. Veto et al. 2007;Pearson et al. 2008;Yamaguchi et al. 2014). Due to the fact that this study was based only on museum collections (collected mostly by amateur palaeontologists) a strong sampling bias is expected, so further statistical analysis would be unreliable. Nevertheless, judging from the presence of clusters, which are pristine sponge associations, the dominance of hexactinellids over lithistids appears to be a genuine feature of this fauna.
A critical comparison of Chiampo fossil genera with other fossil sponge faunas was performed based on selected papers for which the systematic palaeontology is well documented. Data quality in palaeospongiology is generally a serious problem: difficulties in taxonomic assignments, discontinuous fossil records, and preservational and collection biases (Hartman et al. 1980;Pisera 2004) make sponges less than ideal organisms for palaeobiological analyses. Despite this, but keeping in mind all the limitations mentioned above, we attempted to make a comparison of faunal composition at the generic level using a binary dissimilarity analysis (R Core Team 2014; software package 'vegan', Oksanen et al. 2014). We recorded the presence/absence of hexactinellid and demosponge genera in nine bodily preserved sponge faunas (see Online Supplemental Material). Only valid genera were considered (Kaesler 2004). Records of either indeterminate species or species identified only above the genus level were ignored. The two new genera described in this paper were ignored. As non-lithistid (soft) demosponges have a low fossilization potential (Finks & Rigby 2003;Reid 2003;Pisera 2006), we interpret the absence of 'soft' demosponge in the Chiampo fauna as a taphonomical effect rather than a real feature of the faunal composition. As a consequence, we removed the 'soft' demosponge genera found in other faunas from the analysis. The cluster dendrogram resulting from this analysis (Fig. 33) clearly shows the affinity of the studied fauna with sponges from the Eocene of Spain and the Cretaceous of Germany. The three main clusters in the dendrogram can be interpreted as three major chronobiogeographical groups: Eocene south-west Pacific faunas, Miocene Tethyan faunas and CretaceousÀEocene Tethyan faunas.

Ancestors and extant descendants of the Chiampo fauna
Among the taxa identified in the Chiampo Valley fauna, some have ancestors in the Mesozoic. Three genera have their roots in the Jurassic: Stauractinella, Laocoetis (see Kaesler 2004) and Ventriculites (see H erenger 1942). The genus Astrosclera was documented in the Triassic (W€ orheide et al. 2002).
Six genera are still extant. Of these, four (Laocoetis, Anomochone, Astrosclera, Vaceletia) are recorded only in the Indo-Pacific Ocean, a fifth (Hexactinella) is also present in the Atlantic Ocean, and a sixth (Corallistes) is cosmopolitan and occurs also in the Mediterranean Sea (Van Soest et al. 2014).
Most extant descendants of Chiampo sponges live in rather deep waters. The only living Laocoetis species, L. perion L evi, 1986 from the Southern Indian Ocean, is known from 250À750 m water depth at temperatures of 9 to 15 C (L evi 1986; Tabachnick (Vacelet, 1977) which lives in semi-closed cavities of coral reefs and bathyal environments of the Indo-Pacific, at 10À530 m water depth (Vacelet 2002b;W€ orheide & Reitner 1996 and references cited therein). The only living species of Astrosclera, A. willeyana Lister, 1900, is distributed in the tropical Indo-Pacific and, as for Vaceletia, thrives in semi-closed microenvironments such as coral rubble, reef cavities, caves and deep cliffs, with a depth range of 1À185 m (Vacelet 2002a and references within).
In summary, the majority of living representatives of the studied Eocene sponges live in deeper shelf environments, which could also have been the setting for the Chiampo fauna. The presence of the lyssacinosid Stauractinella further strengthens this hypothesis. Recent lyssacinosids, with few exceptions, live in the bathyal or even the abyssal zone (e.g. Janussen et al. 2004;Van Soest et al. 2007;Janussen & Reiswig 2009), and this life environment is generally confirmed in the fossil record (e.g. Pisera & Busquets 2002; Swierczewska-G»adysz & Jurkowska 2013). Beccaro et al. (2001) found that the sponge-bearing sediment of the Cengio dell'Orbo quarry exhibits sedimentary structures indicative of mass transport, in the context of an outer ramp sedimentary environment. Therefore, the sponge fauna, together with sedimentological evidence and the composition of the associated fauna (including planktonic foraminifera, pteropods, shark teeth: Beccaro et al. 2001), points to a middle-outer carbonate ramp depositional environment.

Sponge autecology
Despite the fact that our study material comes only from museum collections, the large number of examined specimens (more than 900) allows for some autecological considerations based on functional morphology. Important insights can be obtained into the types of substrate on which the sponges lived. It is clear that some demosponges with desmas colonized a hard-bottom substrate. This is indicated by the presence of a basal disc, which today characterizes sponges from hard bottoms or attached to hard objects on a muddy substrate (Pisera 1997;Pomponi et al. 2001). The basal disc is lost in many specimens because the stalk is the most fragile part of the sponge. Tuber-like basal structures of some hexactinellids indicate anchorage in soft sediment, similar to the tuberlike basal part of the Jurassic hexactinellid Cribrospongia radicata (Quenstedt) that was interpreted as a structure preventing the sponge from sinking into mud by creating buoyancy. The mass of the basal root structure keeps the sponge in an upright position (Krautter 1998). Root-like structures in hexactinellids are seldom reported in the literature (e.g. Reid 1958). Krautter et al. (2006) described similar basal structures in Aphrocallistes vastus and Heterochone calyx, in the only known Recent hexactinellid reef, off British Columbia. When these two species increase in size, they produce root-like outgrowths that can attach to hard objects such as rocks and sponge skeletons. Krautter et al. (2006) interpreted the production of outgrowths as an intrinsic mechanism to optimize stability by helping the growing sponge to keep an optimal life position. Accordingly, the delicate root-like basal structure of Guettardiscyphia/Pleurogettardia from Chiampo could have had a stabilizing function as well.
In summary, there is compelling evidence that most of the Eocene sponges of the Chiampo Valley colonized a hard substrate, at least in their early growth stages, while sponges with root-like tubers are rare in this fauna. Generally, sponge larvae need to settle on hard objects (e.g. rocks, mollusc shells, rhodoliths) to begin metamorphosis (Bergquist 1978). In this process, they can form an envelope around hard objects , and in later stages they can incorporate sediment and living organisms like foraminifera (Cerrano et al. 2007;Guilbault et al. 2006). This explains the occurrence of small bioclasts completely enclosed in some of the Chiampo sponges. However, large/old specimens also retain structures indicative of a hard substrate. The presence of different modes of attachment in the Chiampo sponge fauna suggests heterogeneous bottom surface conditions. This feature was also well documented in the Miocene sponge fauna from southern Spain (Brimaud & Vachard 1985) in resedimented deposits.
A second group of observations testifies to the autochthonous nature of the Chiampo fauna, in which various growth stages are present with small and fully grown specimens being found together. The absence of size selection implies minimal influence of transport processes, and the abundance of small specimens indicates that reproduction must have been a frequent event (Klitgaard & Tendal 2001). Preservation of delicate encrusting bases, sponge clusters and the fact that sponges were reported in one site only strengthen the hypothesis that the studied sponge fauna is essentially autochthonous or parautochthonous, and most probably was rapidly buried.

Conclusions
The bodily preserved fauna from the Eocene of Chiampo Valley (northern Italy) is highly diverse, with 32 taxa that belong to 24 genera, including 10 new species and two new genera. It is dominated by siliceous sponges, most having solid skeletons (Hexactinosida, Lychniscosida and lithistid demosponges). The original siliceous skeleton of the sponges was dissolved and replaced by calcite. However, a veil of peloidal micrite is observed around skeletal elements which often permits the identification of spicules. A comparison of faunal composition at the generic level using a binary dissimilarity analysis shows the affinity of the Chiampo fauna with sponge faunas from the Eocene of Spain and the Cretaceous of Germany. The presence of the genera Camerospongia, Toulminia and Bolidium extends their last occurrences from the Cretaceous to the Eocene. Six extant genera live today in rather deep-water environments. The Recent calcified demosponge genus Astrosclera is reported here for the first time in the Cenozoic, and the second worldwide occurrence of the Recent sphinctozoan genus Vaceletia in the Palaeogene is also reported. The Chiampo Valley sponges colonized a mixed substrate, sometimes forming clusters. The sponge fauna is essentially autochthonous and inhabited the middle-distal portion of a carbonate ramp.