The Mediterranean species of Hornera Lamouroux, 1821 (Bryozoa, Cyclostomata): reassessment of H. frondiculata (Lamarck, 1816) and description of H. mediterranea n. sp.

ABSTRACT Hornera Lamouroux, 1821, a genus which includes large, rigidly erect, ramified and highly calcified cancellate cyclostome species, is represented in the Mediterranean Sea by two species, H. frondiculata (Lamarck, 1816), the type species of the genus, and a species previously left unnamed or wrongly attributed to the northern H. lichenoides (Linnæus, 1758), and classified under this name as threatened in the Mediterranean (Barcelona Convention, Annex II). On the basis of abundant material including large, well-preserved colonies collected by diving, the distinctive morphological and ecological features of these two species are detailed, leading to the description of a new species, H. mediterranea n. sp., and to a better characterization of H. frondiculata. The current state of knowledge of the range of the two Hornera species suggests that they are endemic to the Mediterranean. Their depth and habitat distributions span from 30 to 100 m on both dimly lit rocky walls and flat bottoms with coarse elements for H. frondiculata, and from 55 to 200 m only on flat sandy bottoms for H. mediterranea n. sp., but both species can coexist in the same microhabitat. The distribution of H. frondiculata in two separate habitats is reflected in a remarkable plasticity of colony shape and branching design which optimizes food particle capture according to local flow conditions.


INTRODUCTION
The genus Hornera Lamouroux 1821 is particularly speciose (93 species listed in http://bryozoa.net/index.html, accessed on 12/X/2020), with a large proportion (73 %) of fossil species spanning from the Eocene to the Pleistocene (Smith et al. 2008). Recent molecular phylogenetic data (Waeschenbach et al. 2009;Taylor & Waeschenbach 2015) have upset the traditional generic classification of Cyclostomata with, in particular, the grouping in the same clade of the Cancellata Hornera with some Tubuliporina genera, such as Entalophoroecia Harmelin, 1976, Diplosolen Canu, 1918, Cardioecia Canu & Bassler, 1922 and Frondipora Link, 1807. In the Mediterranean Sea and the North Atlantic, Hornera species are the largest vinculariiform (cf. Lagaaij & Gautier 1965), or 'multiserial rigidly erect' (basic growth form according to McKinney & Jackson 1989), cyclostomes. One may also assert that H. frondiculata (Lamarck, 1816) is the most beautiful bryozoan in the Mediterranean owing to its delicate branching pattern. Like some other large rigidly erect bryozoans occurring in the nearshore coastal zone [e.g. Reteporella septentrionalis (Jullien, 1903), Myriapora truncata (Pallas, 1766), Pentapora fascialis (Pallas, 1766)], it was soon noticed and depicted by early naturalists exploring the Mediterranean. One of them, Count L. F. Marsili, a military engineer and naturalist also known as Marsilli or Marsigli, who came to Marseille to resolve the old debate regarding the vegetal, animal or mineral nature of the precious red coral [Corallium rubrum (Linnaeus, 1758)], wrote the first oceanographic treatise (Marsilli 1725). In this book, several benthic organisms living with the red coral were depicted and, as already noted by Busk (1856), he gave the first representation of a large colony of H. frondiculata (pl. 33, fig. 163, reproduced here: Fig. 1A), placed in the category of 'madrépores rameux'. The finding of H. frondiculata by Marsili was not surprising considering that he sampled communities hosting the red coral with local red coral fishermen at sites off the Marseille-Cassis coast and Riou Island, where many specimens of the present collection were collected.
The genus Hornera Lamouroux 1821 is represented in the Mediterranean Sea by H. frondiculata, the type species of the genus, considered here to be probably endemic to this sea (see below), and a second species left unnamed or, in many cases, ascribed without significant justification to H. lichenoides (Linnaeus, 1758), a species mostly known from northern and Arctic seas (e.g. Busk 1875; Hayward & Ryland 1985;Kuklinski & Bader 2007).
As for many Hornera species, the morphological criteria characterizing the two Mediterranean species are not clearly defined. Consequently, the species names frondiculata and lichenoides have often been incorrectly used. For instance, the assertion that H. frondiculata is 'a common but highly variable species ' (Smith et al. 2008) is certainly based on erroneous identifications. The use of the taxon H. lichenoides, in particular, is subject to uncertainties, especially because the origin of the material on which Linnaeus (1758) erected the species Millepora lichenoides is unknown, as well as the real existence of a type. The purpose of the present paper is to give a precise description of the morphology and the ecology of the two Mediterranean Hornera species on the basis of abundant material. Furthermore, the taxonomic status which was originally proposed without a valid procedure by Waters (1904) for a second Mediterranean Hornera species, i.e., H. mediterranea n. sp., will be formally assigned to specimens so far ascribed to H. lichenoides or named Hornera sp.
Another problem related to the alleged presence of H. lichenoides in the Mediterranean involves the Mediterranean Action Plan and the Barcelona Convention for the protection of the marine environment and the coastal region of the Mediterranean (UNEP 2011). In Annex II of this convention are listed species endangered or threatened in the Mediterranean. A single bryozoan, H. lichenoides, is included in this list (see Rosso 2009;Rosso et al. 2010), and thus put at the same level as the iconic and ZOOSYSTEMA • 2020 • 42 (27) highly threatened monk seal Monachus monachus (Herman, 1779). Apart from the obvious problems of nomenclature already evoked, the choice of this species (by an unknown expert) without any indication of specific threats is quite surprising. In addition to the poor knowledge of the morphological features of the Mediterranean H. lichenoides, the ecological requirements, the frequency of occurrence and the vulnerability of this species have never been documented. In contrast, H. frondiculata is better known, at least its general features, is more frequent at shallower depth in the nearshore zone exposed to multiple threats, and may be considered as much more attractive to collectors. One of the aims of the present paper is to offer evidence to support an eventual revision of this list.

Origin Of material
Most examined specimens of the two Hornera species were collected by the author and collaborators by diving, dredging or trawling, and stored at the Station Marine d'Endoume, Marseille (SME). Diving enabled the collection of unbroken large specimens that revealed their genuine growth-forms, and supplied information on their microhabitat and local abundance. A collection of specimens of both species from Sicily was loaned by A. Rosso, University of Catania. Other examined material includes specimens stored at the Muséum national d'Histoire naturelle, Paris (MNHN), specimens collected in Tunisia by S. Sartoretto, Ifremer, Toulon, and underwater photos of H. frondiculata by two amateur naturalists, D. Ader and E. Driancourt.

Specimen repOSitOrieS
Most studied specimens are deposited at the MNHN, including the type material of H. mediterranea n. sp. Some fertile specimens of both species were sent to the Natural History Museum, London (NHMUK). Rosso's material is located at the Paleontological Museum of the University of Catania, under the codes 'PMC. Rosso Collection I.Ps-H. B-45' and 'PMC. Rosso Collection I.Ps-H. B-46', respectively for H. frondiculata and H. mediterranea n. sp.

methOdS Of Study
Morphological traits of specimens were observed with stereomicroscopes and scanning electron microscopes (SEM) after gold-palladium coating: Hitachi S-570 (SME) and Quanta 200, FEI (Plateforme Microscopie Electronique Timone, Aix-Marseille University). Measurements were carried out with an eyepiece micrometer and from scales of SEM photos. Macro and underwater photos were taken with an Olympus Pen EPL7 equipped with a 60 mm macro lens.

terminOlOgy
The surface of colonies of Hornera presents a great diversity of aspects due to layers of secondary calcification that conceals to  (1920) and Mongereau (1972), and their use by the latter in the diagnosis of the genus Hornera, these terms introduce more complexity than clarity in the descriptions. Therefore, following Borg (1926Borg ( , 1944, they are not used here, except in some cases for indicating equivalence.  ; 1976:  223, table I, 229, table III -Mongereau 1972: 329 (part), pl. 5, figs 1-3 -Zabala 1986: 686, fig. 213 -Zabala 1993: 571, fig. 3 -Rosso 19871996: 209, table 5, pl. 1c, g;2005: 263,    deScriptiOn Colony erect, strongly calcified, firmly attached to a substrate by a broad expansion of secondary calcification, branching dichotomously many times with short, variously directed ramifications without anastomoses, and with the further addition of small, secondary lateral branches growing at right angle (Fig. 3F). Resulting growth-form varying from sub-planar to convoluted rosette shape, both reaching large size, up to ca. 15 cm in width and height, with narrow spacing between secondary branches, pale salmon pink in colour when alive ( Fig. 2A, G). Autozooid apertures distributed on frontal side in 5-8 (6.7 in average) alternating longitudinal (linear) rows ( Fig. 3A, D, E). Peristomes short, longer on lateral sides of branches, with distal edge typically lacking, leaving a U-or V-shaped notch, while lateral edges may be prominent and distinctly tapered, particularly in lateral rows ( Fig. 3C, D). Horizontal part of autozooid tubes pierced with large, round pores (12.5-15 µm) down to the base of raised peristome, clearly visible in the apical zone of branches, where autozooid walls remain apparent in frontal view (Fig. 3C). Secondary calcification increasing greatly from the branch tips to the basal parts of the colony, rapidly masking the external features of autozooids. In an initial stage, secondary calcification forming thick longitudinal ridges surrounding the peristomes (Fig. 3C). These ridges ('nervi') increasing in thickness and soon joining with flat or convex transversal bridges of calcified layers which partially cover the autozooids, but leaving oblong or rounded windows ('sulci') within which some mural pores remain visible (Fig. 3D). In a further stage, frontal side, except for raised peristomes, entirely covered with a thick layer of secondary calcification densely punctuated with small, round pustules often aligned transversally and, proximally to each peristome, interrupted by 3-5 large, irregularly shaped holes ('vacuoles') ( Fig. 3E). Dorsal side of branches convex, with surface structured by a network of longitudinal ridges branching and anastomosing to produce long, concave, spindle-shaped areas pierced with 2-6 large pores, covered with small pustules (Fig. 4E). Fertile colonies frequent, the large ones bearing a great number of gonozooids of the same colour as branches, but clearly denser (Fig. 4A). Gonozooid chamber large, developed on the dorsal side from an enlarged tube migrated from the frontal side ( Fig. 4B), clearly longer than wide when placed between two bifurcations, or roughly triangular or heart-shaped when adjacent to a branch fork; a prominent crest along the upper midline of the chamber, extending on both sides of the ooeciostome (Fig. 4A). Brood chamber wall made of foliated crystallites overlapping according to the direction of wall growth, pierced with mural pores (10.6-14.4 µm), which are rapidly closed by pointed radial spines during the development of the gonozooid (Fig. 4D). External relief of the gonozooid formed by a dense network of small, reticulated ridges, spreading perpendicularly towards the upper crest, bearing a line of small, round pustules, and delimiting spaces ('cancelli-like cavities', Taylor & Jones 1993); ooeciostome large, much broader than the peristomes of autozooids (x 3.8 in average), placed at the middle of the upper crest, curved laterally, with a wide elliptical aperture opening towards the space delimited by the closest lateral branch (Fig. 4A, B). Ancestrula and early astogenetic stages not observed. remarkS

Taxonomic issues
The authorship of H. frondiculata has long been attributed to Lamouroux (1821) and this designation was maintained by d'Hondt (1994: 302), though he noted that two specimens of this species kept at the MNHN had a handwritten label signed by Lamarck naming them 'Retepora frondiculata, Méditerranée'. There is no indication that Lamouroux had the opportunity to examine these specimens in Lamarck's collection. However, his knowledge of the species Retepora frondiculata created by Lamarck (1816)

Morphological features
The general shape and branching type of colonies of H. frondiculata are typical, and cursory examination, even underwater, allows a correct identification of the species. Old drawings of large colonies, such as that represented by Marsilli (1725 pl. 33, fig. 163; here Fig. 1A), can therefore be assigned to this species with confidence. The shape of colonies shows a marked habitat-related plasticity, from nearly planar on rocky walls to strongly contorted on coarse detrital sandy bottoms ( Fig. 2 A-D, see below in Discussion). However, the detailed structure of the branches is similar in both growth-forms. Lateral branches growing at right angles are frequent in colonies of H. frondiculata (Figs 2E, F, 3F) regardless of their shape; they apparently appear to develop subsequently to the distal growth and bifurcation of the branches from which they are budded. Delayed budding of lateral branches is assumed to be an adaptive strategy to increase the fragmentation of empty spaces between laterally adjacent branches and improve the filtering activity of a colony according to its microenvironment (see below, Discussion). Lateral branches and typical notched peristomes are present on a colony of H. frondiculata collected by Abdelsalam (2014, fig. 2) on the Mediterranean coast of Egypt, but the stem and main branches are exceptionally thick and irregularly ramified. This unusual growth-form and remarkable calcification may be induced by peculiar features of the habitat, e.g. shallow depth (20-25 m) and proximity of the Nile delta and the mouth of the Suez Canal. The particular shape of peristomes characterized by a deep distal U-shaped notch (Fig. 3C, D), is a constant and highly discriminating trait of H. frondiculata. The main variability in the peristome shape concerns the  (27) lateral edges, more or less projecting and sometimes clearly triangular (Fig. 3D).  fig. 7), the carinated shape of the gonozooids, with a ridge along the upper side of the chamber ('carina', Borg 1926), is very typical. This prominent longitudinal ridge, which starts at the opposite sides of the long axis of the chamber and ends on both sides of the ooeciostome may result from the suture of two lateral valves, as suggested by stages in the development of the gonozooid (Fig. 4C). The tubular origin of the gonozooid from a zooid of the frontal side, well described by Borg (1926), is evident from SEM examination (Fig. 4B).  fig. 5).     deScriptiOn Zoarium erect, firmly fixed on small, discrete substrata by layers of secondary calcification expanding widely (Fig. 5D), white in colour, bushy, reaching large size (> 10 cm), but often smaller, ramified dichotomously many times without anastomoses, with slender, nearly cylindrical, isodiametric branches, bent in several directions, often with long segments between two bifurcations (up to 1-1.5 cm) (Fig. 5A-C); lateral branches growing at right angle present, but not frequent (Fig. 5E). Frontal side occupied by autozooids opening alternatively along 4-7 longitudinal rows on the frontal side (Fig. 5A, D), with wall pierced by round pores (about 8-11 µm), scattered all around the tube except above the base of the peristomes (Fig. 6B); peristomes short, longer on branch sides (180-210 µm), with aperture entire, ellipsoidal, slightly broader distally, with long axis oriented longitudinally in medial rows, more obliquely  (27) on lateral rows (Fig. 5B). External aspect of autozooids varying markedly from the branch tips to the base of the colony according to the increase with age in the amount of secondary calcification; four schematic stages perceivable in this progression ( Fig. 5C-F): (stage 1) apical area of branches ( Fig. 6A-C), autozooids with raised peristome and primary frontal wall fully exposed, 8-10 mural pores and both sides slightly thickened by a smooth longitudinal ridge, a large empty space (40-60 µm) bordered with the base of the lateral ridges at the proximal end of frontal wall, (stage 2) (Fig. 6D) thickening and broadening of the lateral ridges that tend to cover the whole frontal wall, leaving only few mural pores visible, which can be included within narrow, longitudinally oblong windows, (stage 3) in older parts (Fig. 6E), peristomes emerge from a thick cover of secondary calcification formed by thick, convex, longitudinal 'mouldings' covered with transverse lines of pustules, which border 1-2 large, oblong windows, (stage 4) in more basal parts (Fig. 6F) peristomes hardly emerging from a uniform mass of secondary calcification, which is densely punctuated by small pustules distributed transversally, and interrupted by small, irregularly shaped windows, 1 to 3 per zooid. Dorsal side markedly convex, entirely covered by layers of secondary calcification deposited straight from the branch tip, deeply striated with narrow, longitudinal, anastomosed ridges with rounded surface covered with tiny pustules aligned transversally, leaving long, narrow spindle-shaped empty spaces between them, open or closed by a wall pierced by 2-5 small pores (10-13 µm wide) ( Fig. 7A-C). Gonozooid chamber on the dorsal side, globular with ovoidal or roundish outline, broader than the branch on which it is placed (Fig. 7E); basal part made of a tube migrated from the frontal side and markedly widened (W = 220-260 µm) before building the floor of the gonozooid across a large part of the branch width (Fig. 7F); cover of fully grown gonozooid densely reticulated by a complex network made of stratified layers of anastomosed strings of secondary calcification converging towards the top of the gonozooid and forming a low crest towards the ooeciostome (Fig. 7F, G); areas between the strings irregularly shaped and sized, some closed by calcified, porous wall; primary gonozooid wall pierced with rounded pores closed by a diaphragm made of converging pointed processes (Fig. 7H). Ooeciostome a short tube opening laterally, slightly curved downwards, placed just above the basal tube of the gonozooid and seemingly prolonging it, frequently with a low crest on the upper midline, ooeciopore oval, a little broader than the autozooid peristomes (x 1.5 in average) (Fig. 7F). Frequency of fertile colonies and number of gonozooids on them relatively low ( Table 2); floor of gonozooids sometimes remaining on branches after loss of the upper parts (Fig. 8). Ancestrula and early astogenetic stages not observed.

Mediterranean ecoregions References
remarkS

Taxonomic issues
The species name mediterranea was introduced by Waters (1904: 94, 1905: 15) for a specimen from Naples, first assigned by him to H. lichenoides (Linnaeus) because of similarities in the gonozooid, but differing from the latter by colony and autozooid features. However, this new species name fails to comply with article 12 of the ICZN as Waters did not give a real description of this taxon, nor a figure, and has not deposited type material, nor specimen bearing this name in the museums known to house his material (NHMUK, Museum of Manchester). Therefore, although there is a strong presumption that Waters designated under the name H. mediterranea a specimen belonging to the species described here, this specific name is considered to be a nomen nudum, and thus available. In tribute to A. W. Waters, the species name mediterranea is given here for the second Hornera species present in the Mediterranean. Records in the literature of Mediterranean bryozoans that can be referred with some confidence to H. mediterranea n. sp. are the following which were originally cited as (i) H. lichenoides (Linnaeus) (Calvet 1931;Zabala 1986;Zabala & Maluquer 1988;Harmelin & d'Hondt 1992;Di Geronimo et al. 1993;Rosso & Di Geronimo 1998) (Harmelin 1976(Harmelin , 1978Rosso 2009;Abdelsalam 2014). These records are recent specimens, but there are also fossils from the Plio-Pleistocene (Zabala 1986;Saguar & Boronat 1987;Zabala & Maluquer 1988;Rosso 1989;Rosso & Di Geronimo 1998;Abdelsalam 2014).

Morphological features
Colonies of H. mediterranea n. sp. are readily distinguishable from those of H. frondiculata. They are typically formed of    (27) narrow, often curved, subcylindrical branches, irregularly bifurcating in three dimensions. These colonies are fragile and easily fragmented, but sampling by diving has shown that they can reach a large size (> 10 cm, Fig. 5C) in favourable sites, such as that of the holotype. Lateral branching is present, but less common than in H. frondiculata. The autozooids differ from those of H. frondiculata in their larger diameter (Table 2), smaller mural pores, and shorter peristomes with an ellipsoidal aperture, slightly broader distally. The development of secondary calcification on the frontal side follows the same succession of stages as in H. frondiculata, with a similar pattern of thickening, that can be divided into 4 stages (Fig. 6). The main difference concerns the windows ('lacunes'), which are less numerous and smaller in the last stages of calcification in H. mediterranea n. sp. (Fig. 6F). The thickening of the convex dorsal side is typically achieved by distinct ribs with rounded outline, covered by transverse lines of small pustules, a structure resembling that of H. brancoensis Calvet, 1906 from Cape Verde Islands. Within depressions between these ribs, pores are smaller than in H. frondiculata, in which they are included in spindle-shaped depressions. As in H. frondiculata, the gonozooid is broader than the branch on which it is developed, but its relative size is smaller, its shape is rounded or oval, and it is not carinated. Unlike H. frondiculata, H. mediterranea n. sp. produces few gonozooids ( Table 2). The occurrence on branches of H. mediterranea n. sp. of vestiges of brood chambers consisting of floors more or less covered by secondary calcification (Fig. 8A), or limited to the basal tube of the chamber (Fig. 8B), can be diversely interpreted. These remains might be signs of aborted growth due as well to strong limitations in time of suitable conditions for the development of gonozooids as to vulnerability to particular adverse conditions. However, according to Batson et al. (2020), similar vestiges of gonozooid floors observed in Hornera colonies from New-Zealand would result from the resorption of brood chamber walls.   Di Geronimo et al. 1997;Rosso & Di Geronimo 1998;Di Geronimo et al. 2003;Rosso 2005) Jackson 1989). They differ in their general shape, plasticity potential, branching pattern and width of branches. The absence of anastomoses between adjacent branches, in contrast to some other Hornera species, such as H. antarctica Waters, 1904(Borg 1944, may increase the fragility of the colonial skeleton, particularly when it is formed by narrow branches with widely spaced successive bifurcations. However, thick layers of secondary calcification on the frontal and dorsal sides, a generic trait, improve the strength of branches and their resistance to breakage of branches. Hence, colonies of H. mediterranea n. sp., which have narrower branches with bifurcations often widely spaced (Fig. 5A, B1), are clearly more fragile than those of H. frondiculata. These differences in the architectural design of colonies are species-specific and/ or linked to local environmental conditions, and may offer good models for testing the influence of biotic and physical drivers on the habitat distribution and the capacity for food acquisition (e.g. Abelson et al. 1993;Helmuth & Sebens 1993;Eckman & Okamura 1998;Okamura et al. 2001).

Fossil records
Plasticity of colony shape is an advantageous feature allowing species to occupy different environments and habitats (Jackson 1979). Striking examples of intraspecific high plasticity of growth forms according to habitat features are found among cyclostomes, such as Platonea stoechas Harmelin, 1976or 'Cardioecia' watersi (O'Donoghue & de Watteville, 1939) (Harmelin 1975. Hornera frondiculata presents an obvious plasticity with colony shapes recalling those of retiform Reteporella species living in the same habitats. Their colony shapes exhibit a similar range with two opposite branching designs, i.e., planar vs complexly folded. Widely flared cupshaped colonies resulting from planar branching are typical of R. mediterranea (Hass, 1948) and H. frondiculata (Fig. 2C, D) when they co-occur on deep, dimly-lit rocky walls. On the other hand, the complex growth-shape with branches or colony parts bent in all directions is shared by H. frondiculata ( Fig. 2A, B) and R. grimaldii (Jullien, 1903) when both live at the surface of flat bottoms, free or attached to a tiny substrate. The unifacial location of lophophores in deep water colonies with planar shape growing parallel and close to the substratum can be inferred as an adaptation for exploiting the food resource from a boundary layer in which the flow is steady and unidirectional. In contrast, the multifacial location of lophophores in colonies with complex three-dimensional designs is probably an adaptation to life in a turbulent boundary layer generated by an unsteady flow, partly due to the topographical roughness of the bottom covered with coarse elements. Colonies of H. mediterranea n. sp., which thrive on the same flat bottoms, are also bushy. The relative fragility of their narrow branches is not a limitation considering the deeper distribution of this species and may be, on the contrary, an advantage in facilitating the local multiplication of clones (see below). The production of lateral branches (Figs. 2E-F, 3F, 5E), more frequent in H. frondiculata than in H. mediterranea n. sp., is another expression of the plasticity of the colony shape. This type of ramification occurs preferentially in proximal parts of large colonies and seemingly after the bifurcation of branches distal to them. In most cases, they grow perpendicular to the mother branch and split empty spaces between adjacent branches with, sometimes, drastic changes in the direction of growth (Fig. 2E-F). This pattern allows expanded lophophores to be adequately spaced and can be considered as a functional alternative to retiform branching and a remarkable example of colony integration to increase the efficiency of filtration of food particles by lophophores. Budding of lateral branches in Hornera species recalls the similar branching process observed in some Tubuliporina (Harmelin 1976;Jablonski et al. 1997), particularly among species of Annectocyma (Hayward & Ryland, 1985). In the latter, lateral budding involves skeletal resorption from pseudopores (Harmelin 1976, fig. 7; Batson et al. 2020, table 1: 'window resorption'). In Hornera, the process allowing the proximal budding of lateral branches is expected to be different considering their skeleton free-walled structure, and has to be investigated more precisely. Whatever its ontogeny, lateral branching contributes to the high potential of Hornera species of remodelling their colony shape by resorption and ZOOSYSTEMA • 2020 • 42 (27) re-budding stressed by Batson et al. (2020) in their synthesis of resorption of mineralized parts in bryozoans.
reprOductive pOtential The two Hornera species apparently differ in their reproductive output, as suggested by the rarity of gonozooids in H. mediterranea n. sp., even if floors of erased brood chambers are taken into account, and their much greater abundance in H. frondiculata (Table 2). The question of why these two co-occurrent Hornera species differ so much in the energy allocated to reproduction remains open. Does this result from a difference in filtering efficiency, i.e., in available energy, or from more indirect causes? The presumed low reproductive output of H. mediterranea n. sp. may be compensated by asexual reproduction through the fragmentation of its fragile, narrow branches. Strong turbulent bottom jets and shocks by mobile benthic animals are expected to be natural sources of fragmentation, but human actions, e.g. through trawling, may be now predominant. It is clear that the presumed high fertility of H. frondiculata is not reflected by features of its population. This species has none of the characteristics of an r-selected species (e.g. Pianka 1970), i.e., numerous small individuals, short-lived, subject to a high mortality rate in an unpredictable environment, nor those of an invasive species. The relative rarity of H. frondiculata colonies within upper circalittoral communities (i.e., above 40 m depth) despite their high potentiality in sexual reproduction may imply strong limitations either in survival or dispersal of larvae, recruitment success or ability of post-recruitment stages to cope with the interspecific competition, possibly coupled with a low tolerance for temperature variations (see below). Unfortunately, there was no ancestrula nor juvenile colonies in the studied material of both Mediterranean Hornera species. A different sampling methodology would be required to get information on early stages of these species, which could allow comparisons with the early astogeny in other Hornera species (Batson et al. 2019).
depth diStributiOn and thermal regime The two Hornera species live in the Circalittoral biozone (sensu Pérès & Picard 1964) and can co-exist in the same type of habitat within a relatively broad depth range (Table 4). However, they differ in the average depth and upper limit of their depth distribution (Table 2), which are clearly deeper for H. mediterranea n. sp. This disparity could be driven by the vertical distribution of habitats suitable to this species. However, detrital sandy bottoms with coarse elements, on which H. mediterranea n. sp. thrives in Provence with large specimens, also occur at depths shallower than the upper limit of the distribution of this species in this region (55 m). This upper depth limitation may be linked to a low tolerance to seasonally variable temperatures. The deep layers (below about 55-60 m) are characterized by a constant temperature of around 13°C while the upper layers are exposed to strong seasonal variations (Bensoussan et al. 2010) with, in some regions (e.g. Marseille area), rapid wind-induced temperature drops during summer (Millot 1979). For H. frondiculata, a relationship between depth range in different regions and this type of thermal stress has previously been hypothesized (Harmelin 1988).
vulnerability Of the Hornera SpecieS in the mediterranean The epifaunal 'facies' marked by an abundance of large erect bryozoans at the surface of coarse detrital sandy bottoms, which may include both Hornera species, once covered widespread patches in the 40-80 m depth zone (Marion 1883). This type of habitat has regressed during the last century because of pollution, silting and trawling, particularly off urbanized coasts, such as the gulf of Marseille or the bay of La Ciotat. In the latter, it has been shown that the restoration of this assemblage of large erect bryozoans after the diminution of stressors has been very slow (Picard & Bourcier 1976). The vulnerability of deep coralligenous outcrops, where H. frondiculata can live, is mainly due to silting and pollution (Weinberg 1978;Hong 1983;Ballesteros 2009). Collection or breakage of colonies by divers is deemed to be much more limited considering the depth of the sites where Hornera species live and a change in the behaviour of divers, who are now more aware of the need to respect the marine biota.

needS fOr the reviSiOn Of H. licHenoides
The status of H. lichenoides Linnaeus is not clear while this species name is commonly used. The origin of the material on which Linnaeus (1758) based his very short description of Millepora lichenoides is unknown, and there is no indication of the current existence of type material. Moreover, it is not certain whether the original material considered by Linnaeus was actually part of his own collection or belonged to another naturalist. However, H. lichenoides Auctt. was soon, and is still, considered as a northern species, present in the Arctic and northern seas (e.g. Smitt 1867;Busk 1875;Hincks 1880;Borg 1926;Hansen 1962;Kluge 1962;Hayward & Ryland 1985;Ryland & Hayward 1991;Bader & Schäfer 2005;Kuklinski & Bader 2007;Noël 2010;Rouse et al. 2018). The most obvious differences between H. lichenoides (Linnaeus) Auctt. and H. mediterranea n. sp. are the degree of secondary calcification, which can be huge in the former, forming a uniform cover hiding the boundaries between the autozooids, and the size of branches, clearly thicker and broader, with a greater number of autozooid rows (e.g.  fig. 24) have the same globular shape. Calvet (1931) considered that the erection of a new species, H. mediterranea, by Waters (1904Waters ( , 1905 for Mediterranean specimens of H. lichenoides was unjustified considering that there was no difference in the features of the gonozooid. The apparent resemblance in the shape of the gonozooids in the two species Harmelin J.-G. should be more clearly described in northern specimens of various origins, together with details of the zooidal features at different stages of secondary calcification. Currently, the question of the existence of one or several species under the name H. lichenoides Auctt. in the Atlantic and the northern seas remains open and should be thoroughly reviewed with modern tools. It is clear from the present material that the particular features of the colonies and the zooids of H. mediterranea n. sp. are constant, suggesting a genetic individualisation in a well-defined habitat.