Taxonomic and morphometric analyses of the Haplosyllis spongicola complex ( Polychaeta : Syllidae : Syllinae ) from Spanish seas , with redescription of the type species and descriptions of two new species

This is the first contribution to a worldwide taxonomic revision of the closely related genera Haplosyllis (the main goal is to describe species within the Haplosyllis spongicola complex) and Geminosyllis. The type species, Haplosyllis spongicola, is re-described based on the syntypes and other material collected from Spanish seas. A combined taxonomic and morphometric (i.e. PCA and discriminant analysis) approach was used to study species of the Haplosyllis spongicola complex from Spanish seas. Two of them are described here as new: Haplosyllis carmenbritoae n. sp. from the Canary Islands, and Geminosyllis granulosa n. sp. from Iberian and western Mediterranean waters, and which is attributed to Geminosyllis due to the presence of a pharyngeal trepan. The third species, which also belongs to Geminosyllis, is morphologically very close to, and morphometrically indistinguishable from, H. spongicola, except for the presence of the trepan. This casts serious doubts on the relevance of this structure for separating the two genera and thus prevents us from describing this last species as new.


INTRODUCTION
Haplosyllis spongicola (Polychaeta: Syllidae: Syllinae) was described from specimens collected in the Adriatic Sea by Grube (1855).Since then, the species has been widely reported in temperate, tropical and subtropical seas (Martin et al., 2003).It is commonly known as a sponge symbiont (Martin and Britayev, 1998;López et al., 2001), although it has also been reported on many different hard substrates.Despite being a frequent component of faunistic lists and papers on descriptive benthic ecology, very little is known about its ecology and behaviour (but see Martin andBritayev, 1998 andLópez et al., 2001).Moreover, a recent study on inter-population variability (Martin et al., 2003) casts serious doubts on the accuracy of identifications of H. spongicola for populations associated with more than 40 currently known host sponges all around the word (López et al., 2001).
According to the literature, the specimens of Haplosyllis spongicola seem to have a high morphological variability, and several authors described varieties, subspecies, or even similar species, mainly based on differences in cirrus length and chaetal shape.For instance, H. spongicola var.tentaculata was characterized by the presence of long (20-30 articles) cirri (Marion, 1879), like the H. spongicola of Fauvel (1923), whose drawings (Fig. 95 a-d) corresponded to H. hamata (Claparede, 1868) from Naples, a species with short cirri which Fauvel (1923) sinonymized to H. spongicola.These kinds of mistakes have generated confusion in identifying the species, which have often been repeated throughout the literature.As a consequence, H. spongicola is commonly reported as a short-cirri species.In turn, Rioja (1941) described another species with short cirri from Acapulco, México, H. spongicola brevicirra, and Cognetti (1955Cognetti ( , 1957Cognetti ( , 1961) ) described two subspecies, Syllis (Haplosyllis) spongicola spongicola, with short cirri, and S. (H.) spongicola tentaculata, with long cirri, both from the Mediterranean.The same subspecies were later reported by Imajima (1966) from Japanese waters.
Nowadays, there is strong evidence that marine species with worldwide distributions represent species complexes (Hilbish, 1996), which are sometimes very difficult to distinguish on the basis of morphological characters.However, recent studies have demonstrated that, once distinguished, appropriate morphological characters allow some of these cosmopolitan species to be subdivided into many valid species.In turn, their geographical ranges of distribution usually became narrow and, in the case of symbionts, their associations tend to demonstrate a high degree of specificity.The Harmothoe lunulata complex, for instance, was separated into 21 species belonging to three genera in addition to Harmothoe (i.e., Malmgreniella, Paragattiana and Wilsoniella) by Pettibone (1993).
More recently, the Owenia fusiformis complex has generated many papers describing new species all around the world including California (Blake, 2000), the Yellow Sea (Koh and Bhaud, 2001), the North Atlantic (Koh and Bhaud, 2003;Koh et al., 2003), Australia (Ford and Hutchings, 2005), and the Persian Gulf (Martin et al., 2006).
Haplosyllis spongicola was also regarded as a cosmopolitan species (San Martín, 2003).Nevertheless, the different populations reported around the world under the species name H. spongicola actually consist of a complex of cryptic species, which show enough taxonomically robust differences to be formally described as different species (Martin et al., 2003).Recently, some papers have already described new species of Haplosyllis, which are closer to the "spongicola" complex but different enough to prevent confusions caused by the imprecise description of the type species and the posterior intricate network of synonymies.For instance, Haplosyllis loboi Paola, San Martín and Martin, 2006 from Argentina (Paola et. al., 2006) and Haplosyllis crassicirrata Aguado, San Martín andNishi, 2006 from Japan (Aguado et. al., 2006).However, the present paper can be considered as the first contribution to a worldwide taxonomic revision of the genus Haplosyllis.The type species of H. spongicola is re-described here based on type material and newly collected Iberian specimens.This creates a solid formal basis that will allow further work on the species complex.As a consequence, two new species, one Haplosyllis and one Geminosyllis, are also described from Spanish seas (Atlantic, Cantabric Sea, western Mediterranean and Canary Islands).A second species of Geminosyllis is recognized but not described as new because its distinction from H. spongicola remains doubtful.Combined with classical, taxonomically robust morphological observations, a morphometric approach is used to identify the characters to distinguish species within the H. spongicola complex.

MATERIAL AND METHODS
The syntypes of Haplosyllis spongicola were examined from a loan of the Museum Przyrodnicze Universitetu Wroclawskiergo.The samples of the populations of the Spanish Atlantic and Cantabric Seas and from the Balearic Columbretes and Alborán Islands were collected during the "Fauna Ibérica" I, II, III and IV cruises, and examined as a loan from the Museo Nacional de Ciencias Naturales of Madrid (MNCN).Samples from Nerja were collected by the staff of the Zoology Department at Malaga University (San Martín and Aguirre, 1991).Samples from Almería and Cape of Palos (Murcia) belong to the personal collections of G. San Martin and were collected from rhizomes of Posidonia oceanica (San Martín and Viéitez, 1984;San Martín et al., 1990).Specimens from the Cape of Creus were collected by C. Alós from calcareous concretions (Alós, 1988;1989) and those from the Canary Islands were collected by J. Núñez from coralline substrates associated with Dendrophyllia ramea (Núñez, 1990).The specimens from Blanes and Cadaqués belong to the personal collection of D. Martin.
Scanning electron microscope images were taken at the "Servicio Interdepartmental de Investigación" (SIDI) of the "Universidad Autónoma de Madrid" (UAM) and at the "Servei de Microscopia Electrònica" of the "Institut de Ciències del Mar" of Barcelona (ICM, CSIC).
Morphometric analyses were based on 40 different measures of the body, appendages and proventricle, together with six chaetal measurements based on Martin et al., (2003) (Table 4).Measurements were standardized by dividing by the individual's body length.Measurements of 16, 34 and 13 specimens respectively of Haplosyllis spongicola, Geminosyllis granulosa sp.nov and Geminosyllis sp. 1 (depending on the available material) were included in the analyses.Haplosyllis carmenbritoae n. sp. was not considered because of the low number of specimens.
The morphometric analyses were carried out using STATISTICA 6.0 (Stat Soft, Inc) for Windows.The relationships between the three species under study were assessed based on 40 morphometric measurements using a Principal Component Analysis (PCA).The morphometric distinction between the three species was assessed by means of a discriminant analysis using a forward stepwise method; for the characters that most significantly contributed to this discrimination, a standard statistic Wilks' lambda (ranging from 1, no discriminatory power, to 0, perfect discriminatory power) was used.The partial Wilks' lambda index was used to assess the individual contribution to the betweengroup discrimination for each character respectively.The graphical representation for the distinction between the three species was performed by a canonical analysis.Description of syntypes.Three specimens all in bad condition.Cylindrical, broad body, thinner from mid-body to posterior end, measuring 2.4 cm long (77 segments, complete specimen), 1.7 cm long (52 segments) and 2.5 cm long (53 segments), and from 2 to 1 mm wide in anterior and posterior ends respectively.Yellow in colour.Prostomium and palps not seen.Dorsal cirri whip-shaped, long through all body.All observed chaetae bidentate (posterior-most).Proventricle long, rectangular, extending through 10 to 12 chaetigers.

Key to the Spanish genera and species of
Description of Iberian individuals.Body cylindrical, broad (Fig. 2 A-B), abruptly thinner from mid-body to posterior end.Size variable, up to 6.5 cm long and 1.5 mm wide.Yellow (small specimens) to brownish (large specimens) in colour.Dark, thin, transverse bands sometimes present on anterior segments.Pair of lateral ciliated nuchal organs located between peristomium and prostomium, only visible under SEM (Fig. 2H).Prostomium pentagonal (Fig. 1A), with two pairs of small eyes arranged trapezoidally.Antennae long, progressively thinner from base to tip; median antenna inserted on middle of prostomium, longer than lateral ones, up to 0.8 mm for 40 articles.Lateral antennae inserted on anterior margin of prostomium, 0.26-0.80mm long for 13-23 articles, up to 0.80 mm.Palps long, broadly triangular, fused at their bases, but clearly divergent all along their length (Fig. 1A, 2G); sensory organs as two ciliate bands on ventral side of each palp; lateral sensory organs absent (Fig. 2 C-D).Tentacular segment well defined, shorter than subsequent ones.Tentacular dorsal cirri longer than ventral ones, but shorter than dorsal cirri of first chaetiger, up to 1.10 mm long for 16 to 30 articles.Dorsal cirri whipshaped, similar to antennae; first cirri longer than remaining ones, up to 1.40 mm for 24 to 40 articles.Second cirri smaller than subsequent ones, up to 0.52 mm long for 26 articles.Third, fourth and fifth cirri slightly longer (Table 1).Ventral cirri digitiform, shorter than parapodial lobes.Chaetae all bidentate (Fig. 1B, 1D, 1F, 2F, 2I), the anterior-most with a thin small main fang (Fig. 1B).Medium and posterior chaetae wider, with a longer, well-defined main fang (Fig. 1D, 1F, 2F, 2I).Small spines on upper side of main fang.Two chaetae per parapodia (occasionally 1 or 3), both with same morphology, but frequently of different sizes.Aciculae broad, either enlarged with a pointed tip (Fig. 1C), with a hooked tip directed upwards or with 90°bent, rounded tip (Fig. 1E, 1G), 1 to 6 per parapodia in anterior-most segments and only 1 in posterior-most segments.Pharynx orange, similar in length to proventricle, with large tooth (Fig. 2C) surrounded by crown of soft papillae and an inner ring of short, numerous cilia (Fig. 2E).Proventricle long, rectangular, extending through 9 to 10 chaetigers (occasionally only 6 or reaching up to 13), with about 46 to 56 muscular cell-rows in the inner region.Pygidium with two long anal cirri, with 20 articles each.Annus surrounded by a ciliated region.
Reproduction.Haplosyllis spongicola reproduces by means of acephalous stolons with one pair of blackish ocular spots per parapodia.The presence of ocular spots may be observed early during the formation of the stolon, with the reproductive segments still attached to the adult body.Stolons may have more than 30 segments, reaching up to 0.6 cm long (Martin et al. 2003).
Ecology.It is a common species from intertidal to circumlittoral and bathyal environments, and occurs both inside and outside all kinds of sponges and hard substrates.According to the present redescription and in comparison with the known data on other populations, mainly tropical, the species seems closer to a sponge specialized pred-SCI.MAR., 71(3), September 2007, 551-570 ator than to a strict sponge endo-or ectosymbiont (Martin and Britayev, 1998).
Distribution.Mediterranean and European Atlantic waters; other records in temperate and tropical seas must be reviewed.

Taxonomic remarks. The morphology of the Iberian specimens clearly agrees with that of the
Haplosyllis spongicola syntypes.Thus, H. spongicola is re-defined here as the large-sized morphotype having long cirri.The Mediterranean specimens with short cirri and trepan, up to now frequently confused with H. spongicola, clearly belong to Geminosyllis and will be discussed in the following section.We will also discuss the differences between H. spongicola and the other Haplosyllis species described here.Description.Body sub-cylindrical, ventrally flattened, small (1.25-1.65 mm palps length plus prostomium plus first 10 chaetigers), 0.29 to 0.55 mm width (excluding parapodia).Specimens without the most posterior segments, pale cream, with brown tonalities depending on the presence of dorsal granules.Two nuchal organs as pore regions at both sides of body, between prostomium and peristomium (Fig. 4E-F).Prostomium rounded, with two pairs of small eyes in trapezoidal arrangement (Fig. 3A), small specimens apparently without eyes.Antennae similarly wide from base to tip, with translucid edges and brown marks on each article.Median antenna inserted on middle of prostomium, slightly longer than lateral ones, up to 0.54 mm for 14 to 24 articles.Lateral antennae inserted on anterior margin of prostomium, 0.22 to 0.43 mm long for 11 to 17 articles.Palps long, broad, fused at their bases but clearly divergent all along their length (Fig. 3A), slightly longer than prostomium; with sensory organs as pore areas on ventral side (Fig. 4A-B) and pore holes on lateral sides (Fig. 4C-D).Tentacular segment well defined, shorter than subsequent ones.Tentacular and dorsal cirri similar to antennae.Dorsal tentacular cirri longer than ventral ones and shorter than first dorsal cirri, up to 0.46 mm long for 16 to 20 articles.Cirri from the first to fifth chaetigers with a strongly specific length-pattern: first cirri longer than remaining ones, up to 0.59 mm for 20 to 23 articles; second smaller than the others, up to 0.33 mm with 10 to 15 articles; third and fourth similar and slightly longer, up to 0.59 mm for 8 to 24 articles; fifth shorter, up to 0.36 mm for 6 to 15 articles (Table 1).Ventral cirri shorter than parapodial lobes.All chaetae bidentate (Fig. 3B, 3D, 3F, Fig. 4H-L), anteriormost thin, with short main fang (Fig. 3B) and midbody and posterior-most ones wider, with long main fang, longer when more posterior (Fig. 3D, 3F).Distal and proximal teeth clearly separated, long and well defined.Upper side of main fang with 4 to  6 long spines followed by several smaller spines near the tip, visible both under SEM (Fig. 4H) and light microscope (Fig. 3B, 3D, 3F).Two chaetae per parapodium (sometimes 1 or 3) similar in shape but different in size.Aciculae broad, some enlarged, with rounded tip (3C, 3E) or with a hooked, pointed tip, directed upwards (Fig. 3G). 1 to 6 anterior aciculae per parapodia, only 1 in posterior ones.Pharynx orange, similar or longer than proventricle, long and thin, with a small tooth, surrounded by a crown of soft papillae and an inner ciliary ring (Fig. 4G).Proventricle long and rectangular, 0.42 to 1.09 cm long and up to 0.35 mm wide, extending through 4 to 5 segments.Few last segments, pygidium and stolons not seen.
Etymology.The name refers to Mari Carmen Brito, polychaetologist from the University of La Laguna (Tenerife, Canary Islands).
Haplosyllis villogorgicola Martin et al. 2002 is geographically the nearest species, but it has 3-5 hooked chaetae per parapodium, with a very short main fang and the apical teeth are much more enlarged than in H. carmenbritoae n. sp.The mor-phology of the appendages (antennae and cirri) is also different, more enlarged and sharper.It has also been described as a specific symbiont of the gorgonian Villogorgia bebrycoides. of 9 to 11 teeth and a large mid-dorsal tooth.Simple chaetae, from 3 to 7 in each parapodium.Capillary chaetae absent.

Genus
Remarks.The presence of trepan as a diagnostic character for Geminosyllis is doubtful, as some specimens with a trepan share more characters with Haplosyllis spongicola than with Geminosyllis species.The situation of Geminosyllis in Japanese waters has some parallels with that of the Spanish species of the H. spongicola complex, as the presence of H. spongicola spongicola and H. spongicola tentaculata has been reported in addition to G. ohma (Imajima and Hartman, 1964), which was described as a new species from Japanese waters).In light of the present data, the situation does not seem clear enough, and will require further studies to assess whether these two genera are present in Japanese waters or not, and whether the species are the same as, or different from those in the Mediterranean.Description.Body long, sub-cylindrical (Fig. 5A), small to medium size: 0.82 to 2.08 cm long (palps, plus prostomium plus 10 fist chaetigers), 0.17 to 0.50 mm width (excluding parapodia); pale yellow, without colour marks and with pores connected to internal structures giving granulose aspect to body surface (Fig. 5A).Pore holes also visible under SEM (Fig. 6A).Nuchal organs as a pair of lateral ciliated regions between peristomium and prostomium, only visible under SEM (Fig. 6A, C).Prostomium rounded to oval (Fig. 5A, 6A), with two pairs of small red eyes in trapezoidal arrangement, anterior pair larger than posterior pair.Antennae slightly longer than palps; median antennae inserted near the mid-prostomium, between posterior eyes, reaching up to 0.7 mm long for 9 to 30 articles; lateral antennae inserted near anterior margin of pros-tomium, reaching up to 0.43 mm for 6 to 17 articles.Palps long, broadly triangular, fused at their bases, but clearly divergent all along their length (Fig. 5A); sensory organs as two ciliated regions on ventral side of each palp; lateral sensory organs absent.Tentacular segment well defined, shorter than subsequent ones.Cirri and dorsal cirri similar to antennae.Dorsal tentacular cirri longer than ventral ones and shorter than first dorsal cirri, up to 0.60 mm long for 17 to 22 articles (only 7 in small specimens).Cirri from first to fifth chaetigers with strongly specific length pattern: first cirri longer than the remaining ones, up to 0.77 mm long for 18 to 30 articles (occasionally 11); second and third much shorter and similar in size, up to 0.46 mm long for 8 to 16 articles; fourth slightly longer, up to 0.57 mm long for 7 to 25 articles; fifth small, up to 0.38 mm long for 5 to 14 articles (Table 3).Mid-body long dorsal cirri with 11 to 17 articles (only 6 in some specimens).Mid-body short cirri with 4 to 13 articles.Ventral cirri shorter than parapodial lobes.Chaetae bidentate; anterior ones thin, with small main fang (Fig. 5C); mid-body and posterior ones wider, with longer and welldefined main fang (Fig. 5F; 6E-F  main fang with small spines apparently forming 2-3 rows (difficult to observe with light microscope).Some specimens with 2-3 chaetae per fascicule, others 3-7, all of same morphology but different in size (Fig. 6E-F).Aciculae broad, some with an oblique end and pointed tip clearly directed upwards, others with straight tip (Fig. 5D-E).Specimens from Cape of Palos have 90°bent end with rounded tip.Usually one acicula per parapodium.Pharynx long, similar in length to proventricle, broad and cylindrical.
Stolons not seen.
Etymology.The specific name refers to the granulose aspect of the dorsal surface of the specimens, which is due to the presence of internal glands with pore openings.

Ecology. Most specimens have been found on calcareous concretions of Mesophyllum lichenoides
and Lithophyllum expansum, between 3 and 25 m in depth and among rhizomes of Posidonia, between 2 and 20 m in depth.Also found at 80 to 85 m in depth, associated with detritic bottoms.
Taxonomic remarks.Geminosyllis granulosa n. sp. has been repeatedly confused with the Mediterranean specimens identified as H. spongicola (San Martín, 2003: Fig. 179B), but careful observations reveal that the former can be clearly distinguished by the presence of 3 to 7 chaetae per parapodia and, more specifically, by the presence of a trepan.Moreover, G. granulosa n. sp. has a slender, medium-sized body, with numerous pores on the dorsum, while H. spongicola has a more robust body reaching a larger size.The dorsal cirri of G. granulosa n. sp. are short, broad, with the same width from base to tip and show a specific length pattern from the first to fifth anterior cirri.Seen under light microscopy, the chaetae strongly differ: in H. spongicola the two distal teeth are short and wide, with a narrow angle and rounded tips, while in G. granulosa n. sp.they are enlarged and slender, with a wide angle and pointed tips.The aciculae seem to have differences at the population level, but not between species.The aciculae of both species from Nerja are very similar, but those of G. granulosa n. sp.from Cape of Palos are large, with a 90 degreebent end.At this stage, it is not possible to assess whether this single difference can be considered species specific, which would lead to the existence of an additional new species in the complex.
Geminosyllis granulosa sp.nov differs from G. ohma (Imajima and Hartman, 1964), the single previously known species of the genus, in being smaller, with short cirri.In addition, the prostomium is pentagonal in G. ohma and H. spongicola, while in G. granulosa it is round to oval.G. ohma has a smaller main fang and the aciculae have rounded tips.There is no mention of the presence of spines on the upper edge of the main fang in G. ohma, but it is not possible to dismiss their presence as the species has never been observed under SEM.G. ohma does not have pores on the dorsum, which is an important character for distinguishing the two species.Description.Body long, sub-cylindrical (Fig. 7A), medium-sized: 0.79 to 2.26 cm long (palps, plus prostomium plus 10 first chaetigers), 0.27 to 0.98 cm wide (excluding parapodia).White-cream, without colour marks.Nuchal organs as pair of lateral ciliated regions between peristomium and prostomium, only visible under SEM (Fig. 8D).Prostomium oval to sub-pentagonal (Fig. 7A), 1 to 2 times wider than long, with two pairs of small eyes in trapezoidal arrangement, anterior pair larger than posterior pair.Antennae longer than palps, slightly thinner at the tip.Median antennae inserted near mid-prostomium, up to 0.68 mm long for 16 to 26 articles; lateral antennae inserted near anterior mar-gin of prostomium, up to 0.60 mm for 14 to 20 articles.Palps long, broadly triangular, fused at their bases, but clearly divergent all along their length (Fig. 7A, 8A); ventral sensory organs not seen; lateral sensory organs absent.Tentacular segment well defined, shorter than subsequent ones.Dorsal tentacular cirri longer than ventral one and shorter than first dorsal cirri, up to 1 cm long for 20 to 35 articles (sometimes only 15).Ventral tentacular cirri shorter than dorsal ones.Dorsal cirri similar to antennae (Fig. 7A).Cirri from first to fifth chaetigers with strongly specific length pattern: first cirri longer than remaining ones, 0.62 mm long in average for 21 to 38 articles; second and third much smaller and similar in size, up to 0.92 mm long for 15 to 30 articles (sometimes up to 38); fourth slightly longer, reaching more than 1 cm with 13 to 34 articles; fifth small, up to 0.92 mm long for 11 to 32 articles (Table 3).Mid-body long dorsal cirri with 13 to 26 articles (reaching up to 37); mid-body short dorsal cirri with 7 to 20 articles.Ventral cirri shorter than parapodial lobes.Chaetae bidentate; anterior ones thin, with small main fang (Fig. 7B); mid-body and posterior ones wider, with long and well-defined main fang (Fig. 7C-D; 8E-F).Under light microscopy, distal and proximal teeth small, short spines present on upper side of main fang; under SEM, teeth seem longer, spines decrease gradually in size towards the apex (Fig. 8E).Two chaetae per parapodium (occasionally three), with same morphology, but different sizes.Aciculae broad, some with oblique end and tip clearly directed upwards, others with straight tip (Fig. 7 E-G); two aciculae per parapodium.Pharynx long, wide and cylindrical; anterior end surrounded by soft wide papillae and an inner ciliary ring.Trepan with 9 to 10 triangular teeth (very small under light microscopy) and sharp, sub-terminal mid-dorsal tooth (Fig. 8B).Proventricle long and rectangular, extending through 5 to 14 segments.Pygidium with two short anal cirri.Stolons not seen.Taxonomic remarks.It is very difficult to distinguish Geminosyllis sp. 1 from the above re-described Haplosyllis type species, H. spongicola.Their morphologies are almost identical and they also both have 1-3 chaetae per parapodium.The two species differ specifically in the presence/absence of the trepan (up to now considered as a diagnostic character for Geminosyllis) and also because Geminosyllis sp. 1 does not have a whip-shaped cirrus, the chaetal main fang is longer and the spines on its upper side are enlarged.As mentioned before, the existence of the specimens of Geminosyllis sp. 1 casts some doubts on the distinction between Geminosyllis and Haplosyllis, based on the presence of the pharyngeal trepan.Until this situation is resolved, we have decided not to give a formal name to the taxon.With the current data, it is not possible to discard the idea that specimens of Geminosyllis sp. 1 could simply be juvenile forms of H. spongicola with a trepan.

Geminosyllis
Geminosyllis sp. 1 differs from other members of the genus as follows: it does not have the typical pores of G. granulosa n. sp., its body is much stronger, and the cirrus longer.Geminosyllis sp. 1 differs from G. ohma because this species has 3 to 7 chaetae per parapodia, a smaller main fang and aciculae with rounded tip.
Multivariate analysis.Nearly 56% of the interspecific variability was explained by the selected morphological parameters used in the PCA (Fig. 9A).All mid-body chaetal parameters except the mid-joining point between teeth, the length of the palps and prostomium length and width, accounted for the highest weights in factor 1 (Table 4), and are more correlated to Geminosyllis sp. 1 (Fig. 9A).
The number of articles of first to fifth cirri and long and short cirri, showed the highest weights in fac-tor 2 (Table 4), and are more correlated to H. spongicola (Fig. 9A).The length of the fourth dorsal cirri and mid-body long and short cirri are negatively correlated to Geminosyllis granulosa n. sp.The PCA clearly differentiates the specimens of Geminosyllis granulosa n. sp.from those of the other two species, which always remain mixed in the plot.(Fig. 9B).The 18 parameters selected by the discriminant analysis using the forward stepwise method, presented significant discriminatory power (Wilk's Lambda < 0.05, F value >1) (Table 5).These parameters contributed greatly to discriminating between the three species.Figure 10 shows the clearly distinguishable clusters, each one corresponding to the specimens of a single species.

Taxonomic study
Based on the syntypes and additional material from the Spanish seas, the type species of Haplosyllis, H. spongicola, is re-described and two new species are distinguished within the complex.The first one, Haplosyllis carmenbritoae n. sp., is characterized by its small size, short cirri, chaetal shape and type of sensory and nuchal organs.The second species was assigned to genus Geminosyllis due to the presence of pharyngeal trepan.For the same reason, a third group of specimens was also assigned to Geminosyllis.However, they closely resemble H. spongicola in all remaining characters, casting some doubts on the taxonomic relevance of the presence of trepan at the genus level.This paper also contributes to clarifying the doubtful taxonomic position of some H. spongicola subspecies.San Martín ( 2003) described Mediterranean H. spongicola including two different forms, with long cirri and short cirri.The former is assigned here to the type species, H. spongicola, while later it is considered as Geminosyllis granulosa n. sp.Besides the fact that the type material of two Mediterranean subspecies, Syllis (Haplosyllis) spongicola spongicola and Syllis (Haplosyllis) spongicola tentaculata, described by Cognetti (1955Cognetti ( , 1957Cognetti ( , 1961)), were lost during a fire, it is also very probable that they could be synonyms of Geminosyllis granulosa n. sp. and H. spongicola respectively.
The size and shape of first to fifth dorsal cirri and tentacular dorsal cirri was revealed to be very important to effectively differentiate Haplosyllis carmenbritoae sp.nov and Geminosyllis granulosa n. sp.from H. spongicola.Conversely, these structures do not allow us to discriminate between Geminosyllis sp. 1 and H. spongicola.
The species specific chaetal differences existing between species of the Haplosyllis spongicola complex (Martin et al., 2002) are also useful for distinguishing Haplosyllis from Geminosyllis.The distal and proximal teeth are short with a small angle in H. spongicola, slightly longer in Geminosyllis sp. 1, and much longer in Geminosyllis granulosa n. sp. and Haplosyllis carmenbritoae.The size of the main fang is clearly longer and wider in Geminosyllis sp. 1.Finally, the spines of the main fang revealed to be the most species specific chaetal feature.
The differences in chaetal morphology are frequently used to identify the species of both genera   (Table 2).Like in the case of the Haplosyllis spongicola complex, however, recent studies have revealed species specific variability that was overlooked in the original descriptions (e.g.Martin et al., 2002 for Haplosyllis anthogorgicola), which shows that careful, classical studies often provide strong clues for clarifying the taxonomic status of the different species.

Morphometric analysis
The multivariate morphometric approach complements the morphological observations, and both contribute to a more accurate taxonomic assessment.Both the PCA and discriminant analysis segregate Haplosyllis spongicola, Geminosyllis granulosa and Geminosyllis sp. 1.This shows their usefulness in distinguishing cryptic species and defining the discriminant characters.
The discriminant analysis and the PCA showed the same patterns for the selected parameters.The chaetal characters have a high weight in the first factor of the PCA, and were also selected in the discriminant analysis, with the measurements related to the main fang length being particularly determinant for identifying the species.This supports the relevancy of chaetal features in traditional studies of Haplosyllis.However, they must be observed in detail (using SEM if possible), taking into account their morphometric measurements and their variability along the body.The length of the proventricle has also been selected as a discriminant character, in agreement with the high variability previously pointed out (Martin et al., 2003).The number of articles of both cirri and antennae highly contributed to the second factor, while their length had a lower weight.However, it is recommended to use the former instead of the latter because the appendages can bend easily, and are strongly affected by different contraction levels depending on the fixation process.Moreover, their length-associated patterns of variability seem to be species specific in many cases.In fact, the comparison between H. chamaeleon and H. villogorgicola, used to assess the description of the latter as a new species (Martin et al., 2002), already demonstrated the potentially high value of the appendix-length pattern in Haplosyllis as a discriminant character.
Morphometric approaches are very rare in studies dealing with polychaetes.A recent attempt made on worldwide populations of the H. spongicola complex revealed some inter-population segregation by means of a PCA based on a few morphometric measures, which also evidenced some allometric intra-population relationships (Martin et al., 2003).The biogeographical range covered by that study, however, was more extensive than that in the present paper, so that a higher number of species could be potentially involved.Lately, three new species of Owenia from Australian waters were defined based on classical taxonomical descriptions, clearly supported by robust morphometric analyses (Ford and Hutchings, 2005).The morphometry of uncini in the genus Polycirrus was intensively explored in order to characterize the different types, which probably has interesting phylogenetic implications (Glasby and Glasby, 2006).
In other invertebrate groups, PCA and discriminant analysis based on morphological/morphometrical characters are habitually performed.For instance, clear results have been obtained using genitalia morphology of butterflies (Wakeham-Dawson et al., 2004).Velásquez de los Ríos and Colmenares (1999) found the PCA to be very useful for segregating hymenopterans species from Venezuela.Therefore, multivariate statistical analysis has proved to be a useful tool for helping to distinguish cryptic species, as well as selecting the morphological characters to be used in identifying them, not only for insect taxonomists, but for polychaetes too.
FIG. 2. -Haplosyllis spongicola.A, anterior region, lateral view; B, mid-body, lateral view; C, anterior end, ventral view with the arrow indicating the position of the ciliated sensory organs from ventral side of palps; D, detail of the ciliated sensory organs; E, ciliary band on pharyngeal papillae; F, Mid-body chaetae; G. anterior end, dorsal view; H, ciliated nuchal organs; I, detail of mid-body chaetae.

FIG. 6
FIG. 6. -Geminosyllis granulosa sp.nov.A, anterior end, dorsal view, with the white arrow pointing to the nuchal organ and the black arrow to a pore hole; B, anterior end, ventral view; C, nuchal organs; D, pharynx and trepan, the arrows pointing to teeth; E-F simple bidentate mid-body chaetae.
FIG. 9. -PCA diagrams for the first two factors, F1 and F2.A, Biplot of the 39 morphological characters and relative position of Haplosyllis spongicola (S), Geminosyllis granulosa sp.nov.(G) and Geminosyllis sp 1 (A).B, Distribution of all specimens of the three species.

TABLE 4 .
-Parameters and abbreviations used in the morphometric analysis.Factor 1 and 2 indicate the respective weights of each parameter in the PCA.The highest weights are indicated in bold.The parameters are represented by the corresponding numbers on Figure 9A.