Revision of the Genus Schizoretepora (Bryozoa, Cheilostomatida) from the Atlantic-Mediterranean region

We examined the type specimens and historical collections holding puzzling Atlantic and Mediterranean material belonging to the genus Schizoretepora Gregory, 1893. We performed a detailed study of the colonial characters and re-describe the resulting species and those that have rarely been found or have poor original descriptions. As a result of this revision, nine species are found in the northeast Atlantic and Mediterranean. Six of them are re-described and illustrated: S. aviculifera (Canu & Bassler, 1930), S. calveti d’Hondt, 1975, S. imperati (Busk, 1884), S. sp. nov.? (= S. imperati sensu O’Donoghue & de Watteville 1939) (in open nomenclature, specimen lacks ovicells), S. pungens (Canu & Bassler, 1928) and S. solanderia (Risso, 1826). For S. dentata (Calvet, 1931), no material remains; furthermore, S. hassi Harmelin, Bitar & Zibrowius, 2007 and S. serratimargo (Hincks, 1886) have recently been described and redescribed, respectively. This new arrangement attains a coherent geographical distribution: S. imperati seems restricted to the eastern Atlantic, S. dentata and S. calveti are deepwater species from Atlantic islands, S. pungens and S. aviculifera dwell on the African coasts of the Western Mediterranean, S. hassi and S. sp. nov.? (=S. imperati sensu O’Donogue & de Wateville 1939) are confined to the Eastern Mediterranean, and S. solanderia and S. serratimargo live on the European coasts of the Mediterranean.

Schizoretepora imperati was first described by Busk from Challenger material collected along the Cape Verde coast (E Atlantic) (Busk 1884). Afterwards, it was also cited from Oran (Mediterranean) by Waters (1925), and from Tunisia (Mediterranean) and Eastern Atlantic coasts by Canu & Bassler (1928, 1930. In fact, it was Canu & Bassler (1925, 1928, 1930 who introduced a great deal of uncertainty when they described four new species after the surveys performed along the Atlantic coast of Morocco (Dragages Vanneau) and the Mediterranean Tunisian coast (material collected by P. Pallary in 1904). All the descriptions were supported by pictures taken with an optical microscope, a novelty at that time, thought to improve the accuracy of the descriptions. These photographs, however, lack the details illustrated in drawings. In any case, the descriptions of the four new species (S. aviculifera, S. granulosa, S. longisetae and S. pungens) were rather overlapping and Canu & Bassler (1925, 1928and 1930 already pointed out the similarities between all of them. Later authors have struggled with these species, to the extent that two of them (S. aviculifera and S. pungens) have been uncertainly synonymized with S. imperati (Barroso 1935;Gautier 1962;d'Hondt 1978;Zabala 1986;Zabala & Maluquer 1988;d'Hondt & Ben Ismail 2008).
The goal of this work is thus threefold. Firstly to clarify the imperati-group problem in order to solve the present taxonomic puzzle; secondly to redescribe the resulting Atlanto-Mediterranean species, as well as those that have rarely been found and are poorly described; and thirdly to build a key for all the species of the genus supposedly inhabiting the Atlantic-Mediterranean region. To this purpose, we re-examine the original type specimens and the historical collections holding Atlantic and Mediterranean material, and perform a morphometric study of the colonies, based on photographs using scanning electron microscopy (SEM).

Museum collections
The studied material comprises specimens from the bryozoan collections in the Department of Life Sciences of the Natural History Museum, London (NHMUK), the Musée océanographique de Monaco (MOM) and the Departments of Zoology and Paleontology of the Musée national d'Histoire naturelle in Paris (MNHN), as well as additional own material from the western Mediterranean, which is housed in the Museum of Natural Sciences in Barcelona (MZB).
We followed the bryozoan systematic classification in WoRMS (World Register of Marine Species) compiled by Bock & Gordon (2017).

Metric analysis
All specimens were photographed, uncoated, with either a Quanta 200 FEI, XTE 325/D8395 SEM, a LEO VP-1455 SEM or a HITACHI SU3500 SEM, using the backscattered electron (BSD) mode, under variable pressure. Some of our own material was treated with diluted domestic bleach on selected specimens.
All measurements were made on SEM pictures using the image software ImageJ (Schneider et al. 2012) and are given in microns (µm). The orifice in certain species is obscured by the operculum; therefore, it is termed 'aperture' to distinguish it from the primary orifice in autozooids. Measured characters are identified by the following acronyms and abbreviations: Qualitative analysis Some of the most common discriminant characters for the determination of bryozoans are useless in the present analysis. Thus, all the studied species share the presence and number of spines, shape of the aperture (condyles, sinus, denticles), presence of giant frontal avicularia and ovicell shape. By contrast, some qualitative characters show enough intra-colonial stability and inter-colonial variability to discriminate between species. In order to make a clear splitting of these qualitative characters we used the following categorical attributes: -abfrontal oval adventitious avicularia: absent, rare, frequent -abfrontal surface: smooth, rugose, tuberculate -abfrontal triangular adventitious avicularia: absent, rare, frequent -abfrontal vicarious avicularia: absent, rare, frequent -adventitious avicularia: absent, rare, frequent -frequency of frontal pores: based on average number of pores -frontal shield surface: smooth, finely rugose, rugose -ovicell: longer than broad, broader than long -peristome shape: rounded, triangular We also recorded both the conservation status (alive/dead) and the developmental stage (young/old) of the fragment of colonies under study, in view of the suspicion that ontogenetic development and conservation have played a role in previous misidentifications.

Multivariate statistical analysis
Multivariate statistical analyses were performed on both qualitative and metric characters to test similarities between colonies. All analyses for metric and non-metric characters were carried out using PRIMER 6 statistical software (Clarke & Gorley 2006).
This study comprises all the museum specimens originally labelled as pertaining to one of the species grouped in the imperati-group: S. aviculifera, S. granulosa, S. imperati, S. longisetae and S. pungens (Fig. 1). We also include S. hassi in the morphometric analysis because of its unexpected similarity with S. aviculifera.

Results
The imperati-group problem

Morphometric analysis
Metric attributes (Fig. 3) confirmed the segregation shown in the qualitative analysis. The first two PCA axes explain 73.7% of the variability. From the metrical point of view, the first species to separate is S. hassi, which is larger for all measurements. The second detached group corresponds to the smallest specimens that noticeably groups all the Mediterranean and Atlanto-Moroccan specimens previously labelled as S. imperati, S. longisetae and S. pungens. The third group of species is more disperse and encompasses S. imperati (Busk, 1884), and S. 'imperati' sensu O'Donoghue & de Watteville 1939. The fourth most scattered group encloses S. aviculifera and S. granulosa, the specimens showing greater interspecific variability. The most discriminating character is the zooidal size (i.e., AzL), which separates the western Mediterranean specimens of the imperati group (smaller species) from S. hassi, S. imperati Busk and S. imperati sensu O'Donoghue & de Watteville,1939 (larger species). The first principal component is strongly and positively correlated with the AzL, and secondly with avicularian size (i.e., AbAL and VAL). AbAL and VAW also grow with increasing values of the first principal component, suggesting that these two criteria vary together. The second component is related to the AzW and is the most important factor to discriminate between S. aviculifera and S. hassi.
In summary, the present analysis resolves that the imperati group must be partitioned into four different species. The true S. imperati of Busk (1884) is a different species from the Mediterranean material, and at present it is only known from the Cape Verde Islands in the Atlantic. All the specimens previously assigned to S. imperati from the Mediterranean belong to the same species, described by Canu & Bassler as two separate species: S. longisetae and S. pungens. The latter name has priority. On the other hand, S. aviculifera and S. granulosa from Canu & Bassler are the same species. There is a striking similarity of S. aviculifera with the recently created S. hassi ). Finally, the Egyptian specimen of S. 'imperati' sensu O'Donoghue & de Watteville 1939 seems to separate as a different species. The existence of a sole specimen lacking ovicells, however, suggests that a complete description must wait for more material to be collected.  (Canu & Bassler, 1930), S. granulosa (Canu & Bassler, 1930), S. hassi Harmelin, Bitar & Zibrowius, 2007, S. imperati (Busk, 1884), S. imperati sensu O'Donoghue & de Watteville 1939, S. imperati of other authors, S. longisetae (Canu & Bassler, 1928) and S. pungens (Canu & Bassler, 1925) .

Systematic account
In this section, we fully re-describe the species found in the northeast Atlantic and Mediterranean Sea, with the exception of S. dentata, the type specimen of which, housed at MOM, is badly damaged; and of S. hassi and S. serratimargo, which were recently described and redescribed respectively Reverter & Fernández-Pulperio 2007), as Schizotheca serratimargo and are not controversial.
Primary orifice obscured by a tubular peristome, only its proximal side being visible through the spiramen. Proximal border provided with sharp, flat condyles drawing a U-shaped shallow sinus; distal border not observed (Fig. 4B).
Peristome well-developed. In young zooids it grows as two lobes, slightly widening at its free edge, embodying 6-8 spiniform processes on each side and converging at the anterior end (Fig. 4B,E), but remaining cut by large circular spiramen, shaped like ace of spades ( Fig. 4A-B, E); in older zooids, distal end of lobes appear just denticulate and spiramen disappears (Fig. 4D).
One to three very large acute avicularia (usually paired), placed over large cystid in middle of frontal shield (when single) on both sides of aperture (when paired), mostly directed proximally, but sometimes distally (Fig. 4E). Rostrum acute and distally hooked, with stout bar without columella. One round pore on each side of cystid, 2 to 4 for entire frontal.
The ovicell in the type material (incompletely developed?) is hyperstomial, resting distally on adjacent zooid frontal shield, longer than wide; cut by large frontal fissure, broader at its distal end, smooth and imperforate, non-cleithral (Fig. 4A,E).
Abfrontal side consisting of sheets of irregular kenozooids separated by raised sutures; surface rippled, with numerous avicularia of two types: one similar to those on frontal, large and acute, 1-3 per kenozooid, becoming occluded by secondary calcification; and small elliptical avicularia rarely present (might be the anterior part of the large avicularia from which distal end has been occluded by calcification) (Fig. 4F).
Schizoretepora calveti is a deep-water species that has only been reported from the Azores archipelago, 599-712 m deep.

Remarks
This species was first figured by Calvet (1931) and considered as a variety of Schizellozoon tessellatum. D'Hondt (1975) noticed the differences in the material dredged by the Jean-Charcot during the Biaçores mission and that figured by Calvet (1931). After comparison with the type material sent by the British Museum he erected a new species: Schizoretepora calveti. D'Hondt (1975) did not designate a lectotype from his material; therefore, we formally designate here as type material the specimen MNHN-IB-2008-7565.
Looking at the abfrontal avicularia, this species resembles the Mediterranean species Reteporella pelecanus (López de la Cuadra & García Gómez, 2001). However, the latter species is a true Reteporella Busk, 1884, possessing frontal labial avicularia and ovicells with an elongate median fissure not observed in the genus Schizoretepora. Nonetheless, the aspect of the frontal side with the circular spiramen and long peristome makes S. calveti unmistakable. (Risso, 1826) Fig. 5, Table 2 Retepora  Autozooids hexagonal, longer than wide, convex, separated by distinct sutures. Frontal slightly rugose, with two areolae, placed on the proximal part of the zooid (Fig. 5C).

Schizoretepora solanderia
Primary orifice always obscured by the peristome, longer than wide, distal rim with rounded denticles, proximal border with U-shaped sinus flanked by short, smooth and wide condyles (Fig. 5B). Peristome well-developed, forming tubular high collar; with drop-shaped long spiramen and upper slit cutting it in two plates (young zooids) (Fig. 5C), becoming shorter and smoother in older zooids (Fig. 5D).
Two types of frontal triangular avicularia. Adventitious large avicularia on raised cystid, laterally directed, irregularly present, placed on the median line of frontal surface (Fig. 5D). Giant vicariant avicularia hooked, regularly present on lateral sides of branches, facing upwards and distolaterally directed, sometimes with up to three piling cystids (Fig. 5F). Both types with robust crossbar and lacking columella.
Abfrontal layer with two rows of large, rugose, polygonal kenozooids, with one to three pores and raised sutures. Abfrontal avicularia large, acute, triangular, sporadically present, more frequent near base of colony (Fig. 5H). Remarks Until now, four oral spines have consistently been reported for S. solanderia (Calvet 1902;Gautier 1962). As shown in our specimens, however, the number of spines changes with ontogenetic development, with a higher number of spines on apical autozooids (up to six spines). Schizoretepora solanderia is easily distinguished from all other species of Schizoretepora by the well-developed peristome forming a high collar with a drop-like spiramen. Schizoretepora solanderia is rather common in the Mediterranean and north-eastern Atlantic on coralligenous and detritic bottoms from 30 to 300 m deep.

Remarks
This species has never been found again after its description. Calvet (1931) remarked the similarities of S. dentata with S. solanderia, pointing to the distinctive backed zooids and the presence of two avicularia on each autozooid (Fig. 6A). We examined the type specimen housed at the MOM that, unfortunately, is badly damaged and does not allow a detailed description or measurements (Fig. 6B). There is no further available material for this species. The lack of some features in the described type specimen, which are diagnostic of the genus (i.e., complete ovicell) makes the ascription of this species to Schizoretepora uncertain. Fig. 6. Available characteristics of Schizoretepora dentata (Calvet, 1931). A. Original drawing from Calvet (1931). B. Actual condition of the type housed at MOM. Scale bar: 1 mm. (Canu & Bassler, 1930) Fig. 7, Table 3 Schizellozoon aviculiferum Canu & Bassler, 1930: 60-62, pl. VIII, figs 1-13. Schizellozoon granulosum Canu & Bassler, 1930: 62-63, pl. VII, figs 6-8.

Material examined
Holotype (
Primary orifice semi-elliptical, as long as wide, distal rim with 16-18 blunt denticles, proximal border with U-shaped sinus (wider than long), framed by two acute denticles that form a gutter on the proximal peristomial rim, and flanked by smooth condyles, sloping towards the edges of the sinus (Fig. 7B).
Two types of frontal avicularia. Adventitious avicularia almond-like, with triangular hooked tip, typically proximo-lateral to orifice and directed disto-laterally (Fig. 7E), with slender crossbar and without columella. Giant avicularia on raised cystid (with 2-5 pores) (Fig. 7F); orientated perpendicular to frontal plane and directed proximo-laterally; similar in shape to almond-like adventitious avicularia, triangular but with straight borders and a strongly hooked tip; crossbar robust, without columella. Giant and almond-like adventitious avicularia rarely occur in same zooid, the former being more abundant on the central part of branches.
Ovicell typically globular, non-cleithral (Fig. 7F), slightly wider than long, hyperstomial in young zooids, subimmersed with secondary calcification; surface rugose imperforate, with large central fissure, widening at the base. Abfrontal convex and notably nodular (Fig. 7H), with large polygonal kenozooids separated by raised sutures, and numerous pores (2 to 8). Abfrontal avicularia always triangular, with straight margins, hooked, variable in number but occasionally very numerous (Fig. 7G), of two different sizes, larger on the basis of the fenestrae; with slender crossbar, wide triangular lacuna and without columella. Base of the colonies with numerous kenozooids bearing small oval avicularia.

Remarks
The descriptions by Canu & Bassler of S. aviculiferum and S. granulosum are unfortunate, because they are based on very old material, abraded and fragmented. In particular, they failed to observe the total number of spines and other details of young zooids. The specimens of S. aviculiferum collected by Gautier (MNHN-IB-2008-11126) are the best-preserved and show all the characters well. Gautier mistakenly ascribed the specimen MNHN-IB-2008-11127 to S. longisetae (see Fig. 7C).
To date, S. aviculiferum has been considered a synonym of the former S. imperati (now S. pungens), as suggested by Barrosso (1935), Zabala (1986), d'Hondt & Ben Ismail (2008 and Rosso & Di Martino (2016). Barroso (1935) noted that the differences between S. imperati and S. aviculiferum (and also S. pungens) are limited and of "secondary order", suggesting that they were probably the same species. Table 3. Measurements of Schizoretepora aviculifera (Canu & Bassler, 1930 Schizoretepora granulosa was never reported again after its description and it was recently considered an equivocal report by Rosso & Di Martino (2016). Canu & Bassler (1930) in their description of S. granulosum pointed out that this species is very similar to S. pungens, the only differences being the grainy frontal and the absence of abfrontal avicularia. Despite most fragments being very calcified, thus the abfrontal avicularia are probably embedded by secondary calcification, an accurate revision of Canu & Bassler's material (Fig. 7H) reveals the presence of the characteristic triangular avicularia.
Overall, the descriptions of S. aviculiferum and S. granulosum were thus based on different parts of the colony and different growth stages that rendered equivocal characters to identify the species. The detailed examination of museum specimens together with the morphometric analysis presented here indicate that both species are the same, characterized by having only one type of frontal adventitious avicularia (triangular) and numerous abfrontal triangular avicularia.
Schizoretepora hassi (Harmelin, Bitar & Zibrowius, 2007) shows great similarities with S. aviculifera, particularly with the presence of frequent abfrontal avicularia (more abundant at the base of the colony in S. hassi), the adventitious avicularia typically located proximo-laterally to the orifice and the nodular aspect of the frontal. Differences between S. hassi and S. aviculifera are mainly based on the morphology of the colony and also on measurements, generally larger in S. hassi.
Schizoretepora aviculifera is very similar to S. pungens and differences between both species are very subtle. These differences are explained below (see remarks for S. longisetae).

Remarks
Schizoretepora hassi occurs in the eastern Mediterranean (Lebanon coast) only and is found in shaded rocky habitats between 3 and 36 m deep. Its origin is unresolved, being either a steno-endemic in the Levantine Basin or an Erythraean immigrant ). This species shows a great variability in colony shape, from fenestrate to ramose and bilaminated as well as a large variability in the zooidal characters, frontal shield and number of avicularia depending on their position, age and locality ). Schizoretepora hassi is very similar to S. aviculifera (see remarks above). (
Primary orifice semi-elliptical, longer than wide, distal rim with 16-18 blunt, triangular denticles, proximal border with U-shaped sinus (wider than long) and flanked by smooth condyles, sloping towards the edges of the sinus (Fig. 8B).  (Busk, 1884 Peristome thin, slightly raised on both sides of orifice (Fig. 8C). Autozooids with two oral spines (only the bases observed), one with thicker diameter (Fig. 8B). Giant avicularia on a large cystid (with 2 pores), in the middle of frontal plane, rostrum triangular with slightly concave margins and hooked tip, laterally directed, crossbar robust and without columella (Fig. 8C).

Remarks
Spines were not reported by Busk (1884) in the original description, but the study of his own material reveals the presence of the remaining holes. The giant abfrontal avicularium is rare in the material provided (syntypes); however, Busk (1884) already reported their presence, as did Waters (1895). Since no young or apical parts remain in the type material, it cannot be ascertained whether young zooids present the two triangular flaps characteristic of other nearer species (i.e., S. pungens, see Figs 9E, 10C); thus, this difference could also be due to ontogenetic changes.

S. imperati (Busk) differs from S. pungens in:
-lacking oval adventitious avicularia both frontal and abfrontal -the shape of abfrontal adventitious avicularia, triangular in S. imperati (Busk) (Fig. 8F) but oval or almond-shaped in S. pungens (Fig. 10E-F) -in the shape of the peristome, thinner in S. imperati. Indeed, in the type material we did not observe the distinctively pointed lateral flaps characteristic of S. pungens (Figs 9C-D; 10C), though these were included in the original description (Busk 1884), and it could be a character associated with young zooids The abfrontal of S. imperati (Busk) is closer to that of S. aviculifera (as described here) as both have numerous triangular avicularia, but they differ on the rugosity of the surface (Figs 7H, 8F), being smoother in S. imperati (Busk). They differ on the frontal face because of the presence of almond-like avicularia in S. aviculifera (Fig. 7E), always missing in S. imperati (Busk) (Fig. 8C-D).
Primary orifice semi-elliptical, longer than wide, distal rim with 16-18 (18-24 in pungens type) blunt, triangular denticles, proximal border with U-shaped sinus (wider than long) framed by two acute denticles that form a gutter on proximal peristomial rim, and flanked by smooth condyles, sloping towards edges of sinus (Fig. 9B).
Two to four long oral spines, articulated and carinate, with basal parts thickly fused in distal young zooids, one of them particularly large and thick, club-like (Fig. 9C, E). Spines fragile and easily broken, only two of their bases remaining in older zooids (Fig. 9B, D).
Three types of frontal avicularia. Adventitious avicularia with two shapes, oval (round distal mandible) and almond-like (triangular distal mandible), both irregularly present, scattered proximally to the orifice and variably directed, with slender crossbar and without columella (Figs 9F, 10B-C). Giant vicarious avicularia on a large cystid (with 2-5 pores), in the middle of the frontal plane and at the basis of the fenestrae (Fig. 9G); rostrum triangular with hooked tip (Fig. 10A), variably (but never distally) directed, crossbar robust, without a columella (Fig. 9H).
Abfrontal layer slightly rugose, with large polygonal kenozooids separated by raised sutures and pores (Fig. 10D). Three types of abfrontal avicularia. Adventitious both oval and almond-like small avicularia (similar to those on the frontal), scattered around the kenozooids, variable in number (often more than 3) and orientation (Fig. 10E-F). Giant vicarious avicularia typically located close to fenestrae, triangular, with thin and large mandibles, and often distally hooked (Fig. 10F). Base of colonies with numerous kenozooids bearing small oval avicularia.

Remarks
All material coming from the Mediterranean Sea previously identified as S. imperati (Busk) (except S. "imperati" sensu O'Donoghue & de Watteville 1939, see description below) as well as all material from the eastern Atlantic examined for S. longisetae belongs to S. pungens.

Remarks
This specimen seems closely related to S. imperati, with a similar general shape, a relatively low peristome, an orifice, spines and frontal as well as abfrontal avicularia. Nevertheless, there are subtle differences in the appearance: the surface is more nodular and the giant avicularia are somehow different, with straight or slightly convex margins and a more acute profile (VAL/VAW > 2). The most remarkable differences, however, are in the biometric analysis (Fig. 3), with it being smaller than the rest of species studied. The rugose aspect of the frontal is reminiscent that of S. aviculifera, but it differs in the lack of frontal adventitious, almond-like avicularia.
O' Donoghue & de Watteville (1939) identified this species as Retepora imperati, though expressing doubts about the number of existing species of Retepora at that time. They already pointed out the difficulties in identifying specimens of Reteporidae, taking into account the ontogenetic differences, age of colony and differential criteria used by each author.
Overall, the little existing material, the lack of an ovicell in the examined specimen, and the sole location (Alexandria; south-eastern corner of the Mediterranean) renders the status of this possible new species dubious.

Discussion
The imperati-group problem The present study shows that all the species grouped as the imperati group (S. aviculifera, S. granulosa, S. imperati, S. longisetae and S. pungens) are in fact very similar, both in size and morphological features. Thus, the past puzzle and the existent taxonomical confusion are not only comprehensible but also justified.

Ontogenetic changes responsible
Probably most of the muddle is due to the enormous ontogenetic variability of all members of the family Philoporidae. In any complete colony of lace coral, one can observe a striking transformation of the zooidal morphology, from young zooids located in apical margins to older ones placed in the basal parts. Young, just budded, zooids usually have a smooth frontal shield, exhibit numerous oral spines, present well-developed or large peristomes, and frequently lack other structures such as giant avicularia and ovicells (Figs 5C, 7C, E, 9A, C, E, 10A, G). When moving down to the basis of the colony, zooids begin to lose spines and peristomes, increasing calcification erodes edges and other structures (such as giant avicularia) and ovicells become frequent (Figs 5D, 7D, F, 9B, D, F, 10C, H).
In consequence, it becomes difficult to decide on the diagnostic characters to distinguish the different species. Descriptions coming from the observation of just a part of these colonies, as often occurred in the original descriptions of species of the imperati group, can lead to disparate results and explain misidentifications. Another consequence is that their complete descriptions become exceptionally complex, requiring a comprehensive depiction of the different stages.

Morphometric analysis
The most commonly used characters for the determination of cheilostomatous bryozoans have shown no value in the present work due to the large similarity in shape and size of all the species from the imperati group studied. That is the case with the aperture (shape, condyles, sinus, denticles), spines (presence and number), giant avicularia and ovicell (shape). Finally, the most revealing character is the presence, shape and position of adventitious avicularia. Other more qualitative characters (e.g., rugosity of the frontal) that have shown to be helpful are difficult to classify and are also subject to ontogenetic changes.
In summary, few characters showed intra-colonial stability and enough inter-colonial variability to discriminate between species.
Furthermore, the species making up the imperati group are very similar in size. Paired analysis of measurements did not allow clear metric limits to be set between species. Nevertheless, when considered in group, the multivariate analysis (Fig. 3) shows interesting results. On the one hand, it confirms the identity among all the Atlanto-Mediterranean records of S. imperati, S. longisetae and S. pungens, which appear close to each other, but far from the true S. imperati from the Cape Verde Islands (Eastern Atlantic). On the other hand, it also discriminates the group S. aviculifera-granulosa that is located between the former species and S. hassi, which is the largest one. Finally, it also supports the idea that S. imperati sensu O'Donoghue & de Watteville 1939 from Alexandria in Egypt might be a different and undescribed species.
The large similarities observed in this complex of species suggest a recent speciation in the South Mediterranean-Atlantic area (Fig. 12).

Historical revision
The core of the problem of the imperati group begins with the original description of S. imperati itself. Busk (1884) gave this bizarre name to the material collected by the Challenger expedition at the Cape Verde Islands (Port Praia) by assuming that it belonged to Retepora cellulosa Linnaeus, 1758, a species abundant in the Mediterranean, but poorly described (lack of details in the drawings and misperception of characters) by previous authors such as Johnston; Ellis & Solander and Imperato, to whom he dedicated the species.
Later, Waters (1925) provided more details on the description of R. imperati, based on Mediterranean material (from Oran). Waters (1925) pointed out that the peristome always hid the sinus observed by Busk (in fact Waters was 'redescribing' a different species, later described as S. longisetae). This first misidentification of S. imperati (Busk, 1884) led to subsequent confusion.
The problem was further exacerbated by the contributions of Canu & Bassler (1925, 1928, 1930 that described four new species, three of them ascribed to the genus Schizellozoon (S. aviculiferum, S. granulosum and S. pungens) and Schizoretepora longisetae from the eastern Atlantic-south Mediterranean area. Apparently, when Canu & Bassler (1925) described S. pungens they did not have their own material of S. imperati, and they based the differences of the new species on a comparison with the original description of Busk (1884). When in 1928 Canu & Bassler again compared these two species, they had their own material of S. imperati from the Vanneau expedition (in the Morrocan Atlantic) and the material sent to them by Waters from Oran (Mediterranean). In that work, they insisted on the differences between S. imperati from Busk (1884) and the Mediterranean material as already highlighted by Waters (1925). Moreover, they stressed the similarity between their S. imperati and S. pungens. These precisions where adopted by more recent authors who synonymized S. pungens with S. imperati (Barroso 1935;Gautier 1962;d'Hondt 1978;Zabala 1986;Zabala & Maluquer 1988;d'Hondt & Ben Ismail 2008). However, Canu & Bassler (1928) also described S. longisetae in the same work. The material used for the description of this species was sparse. When looking for differences compared to S. imperati and S. pungens, they pointed to the number of spines (4 instead of 2), longer autozooids and the presence of a small sinus on the proximal border of the peristome. As noted above in our description, the differences are simply ontogenetic, as they are characteristic of younger colonies of S. pungens, where all these traits are more conspicuous.
Schizoretepora aviculifera and S. granulosa, the other two species described by Canu & Bassler (1930) from the Tunisian coast (Mediterranean), have generated less historical controversy because they were rarely cited, and most authors considered (equivocally) that they must be synonymized with S. imperati.
In summary, after this revision of the so-called imperati group (S. aviculifera, S. granulosa, S. imperati, S. longisetae and S. pungens) four different species remain: