The genus Grateloupia C. Agardh (Halymeniaceae, Rhodophyta) in the Thau Lagoon (France, Mediterranean): a case study of marine plurispecific introductions

M. Verlaque, P.M. Brannock, T. Komatsu, M. Villalard-Bohnsack and M. Marston. 2005. The genus Grateloupia C. Agardh (Halymeniaceae, Rhodophyta) in the Thau Lagoon (France, Mediterranean): a case study of marine plurispecific introductions. Phycologia 44: 477–496. Based on morphological data and molecular analyses [Nuclear ribosomal internal transcribed spacer (ITS), rbcL and mitochondrial cox2-cox3 spacer sequences] of Grateloupia spp. populations in the Thau Lagoon (France, Mediterranean) we demonstrated that at least five exotic species of Grateloupia were introduced. These include: (1) Grateloupia asiatica, a recently described species that was previously misidentified as G. filicina in Japan and Grateloupia sp. in the Thau Lagoon; (2) G. lanceolata from Japan; (3) G. luxurians, a Pacific species described as G. filicina var. luxurians; (4) G. patens from Japan; and (5) G. turuturu, a Japanese species previously misidentified as G. doryphora in the NE and NW Atlantic and Mediterranean Sea. These nonnative species probably were introduced in the Thau Lagoon in the 1970s along with the massive importations of Japanese oysters, Crassostrea gigas, into Europe for mariculture purposes. Since their introduction, they all have established large, reproductive populations with the exception of G. patens. The Mediterranean Grateloupia specimens are genetically and morphologically similar to Pacific specimens of the same species, although in the Thau Lagoon, G. asiatica specimens are morphologically more variable than those found in Japanese populations. This is the first report of G. asiatica in the Mediterranean Sea and Europe. Based on morphological data and molecular analyses (rbcL sequences) G. subpectinata is placed in synonymy with G. luxurians.


INTRODUCTION
The Thau Lagoon (France) has become one of the major hotspots of marine species introductions in the world (Verlaque 2001(Verlaque , 2002Verlaque et al. 2002). The likely vector of species introductions is the importation and transport of adult oysters and spat (Hamon & Pichot 1994). With a standing stock of 25,000 t of oysters [Crassostrea gigas (Thunberg 1793)] and an annual production reaching 12,000-13,000 t, the Thau Lagoon is by far the leading site for mollusc shellfish aquaculture in the Mediterranean Sea (Hamon & Tournier 1990;Trousselier et al. 1991).
was genetically similar to specimens from Brittany, France; Portsmouth, England; Tholen Island, The Netherlands; and some specimens from the Thau Lagoon (Marston & Villalard-Bohnsack 1999. In the past, foliose specimens in warm seas belonging to the genus Grateloupia have been regarded as conspecific with G. doryphora (Ardré & Gayral 1961;Dawson et al. 1964;Irvine & Farnham 1983). On the basis of a morphological study and biogeographical considerations, Verlaque (2001) suggested that the alga previously reported as G. doryphora could be a misidentification of the Japanese taxon G. turuturu Yamada. A subsequent rbcL sequence and morphological analyses (Gavio & Fredericq 2002) confirmed that the G. doryphora introduced in the NE and NW Atlantic corresponds with G. turuturu described from Japan. Thus it is likely that G. doryphora from the Mediterranean Sea is also G. turuturu. In addition to G. turuturu, Verlaque (2001) also reported four other introduced species: a foliose species, G. lanceolata (Okamura) Kawaguchi, and three profusely branched species, G. filicina (J.V. Lamouroux) C. Agardh var. luxurians A. Gepp & E.S. Gepp, G. patens (Okamura) Kawaguchi & Wang (as Prionitis patens Okamura) and an unidentified taxon, Grateloupia sp. Comparative gene sequence analysis confirmed the occurrence of two distinct foliose species of Grateloupia in the Thau Lagoon, which were referred to as G. doryphora and Grateloupia sp. (Marston & Villalard-Bohnsack 1999. In addition to the nonnative Grateloupia spp., several native taxa of Grateloupia have been reported in the Mediterranean Sea: G. coriacea Kützing, G. cos- Table 1. Collection sites, sources, and specimens used in the molecular analyses (NA: data not available). New accession numbers associated with this study are given in bold.  phora and other Grateloupia spp. from the Thau Lagoon and performed a comparative morphological and molecular genetic study of populations of Grateloupia spp. from the Thau Lagoon, the Atlantic and Indo-Pacific. The objectives of the present study were (1) to unravel the taxonomic status of exotic Grateloupia spp. in the Thau Lagoon; (2) to confirm the misidentification of G. turuturu as G. doryphora in this coastal lagoon; and (3) to examine the possibility that the introductions of Grateloupia spp. were from Japan or Korea.

Specimen collection and herbariums
Observations and sampling in the Thau Lagoon (7000 ha, mean depth 3.8-4.5 m, maximum depth 10 m; Fig. 1) were performed from September 1994 to April 2003. Specimens were hand-collected in shallow water (0 to Ϫ 1 m mean low water) and preserved in buffered 4% formaldehyde-seawater. Previous collections from 1984, 1988, 1990, 1993, 1997 were also re-examined. Fixed material and dried specimens were studied under the light microscope. Materials were sectioned manually with a razor blade. Transverse sections were stained in 1% aqueous Aniline Blue, washed, and then acidified with a drop of 1 N HCl. Photomicrographs were made using a Nikon Optiphot-2 (Nikon, Tokyo, Japan).   1). Chloroplast-encoded rbcL sequences were amplified using the primer pair F8-R1150 (Wang et al. 2000). The internal transcribed spacer ITS1, 5.8S ribosomal DNA (rDNA), and ITS2 regions were amplified via polymerase chain reaction (PCR) using the primer pair TW81-AB28 (Goff et al. 1994). A noncoding region between the Cytochrome Oxidase subunit 2 and subunit 3 genes in the mitochondrial genome (cox2-cox3 spacer) was amplified using two degenerate primers: cox2-for and cox3-rev (Zuccarello et al. 1999). Amplifications were carried out in 25 or 50 l volumes as previously described (Marston & Villalard-Bohnsack 2002).
For each set of reactions, a control sample containing all reagents but lacking template DNA was included. Identical cycling parameters were used to amplify chloroplast rbcL, ITS and mitochondrial cox2-cox3 spacer sequences and included 4 min denaturation at 95ЊC, followed by 45 s at 95ЊC, 45 s at 55ЊC, and 1 min at 72ЊC for 32 cycles and a final extension of 3 min at 72ЊC. Products of all PCR reactions were visualized on a 1% agarose gel stained with ethidium bromide. The products from two or three different PCR reactions all containing the same primer and template combination were pooled prior to cloning. Products were cloned using a TOPO TA cloning kit (Invitrogen, Carlsbad, CA, USA) following the vendor's instructions. Plasmid DNA was isolated using a Qiagen plasmid mini kit (Qiagen Inc, Valencia, CA, USA). For each individual and region, two to five clones were sequenced using an ABI Prism 377 automated sequencer (Applied Biosystems, Foster City, CA, USA). M13 forward and reverse primers were used in the sequencing reactions. In a few cases, single nucleotide differences were observed between clones from the same individual. In these instances, consensus sequences were used. All sequences have been deposited in GenBank (Table 1).

Sequence alignment and phylogenetic analyses
The rbcL, ITS and cox2-cox3 spacer sequences were aligned for phylogenetic analyses using the Clustal X version 1.83 computer program (Thompson et al. 1994). Additional se-quences included in the alignments and phylogenetic analyses were obtained from GenBank (Table 1). The rbcL alignment was 1030 bp representing positions 120 to 1149 of the full rbcL gene and contained no gaps. The ITS alignment was 816 bp and contained many insertions and (or) deletions (indels) due to species-specific length variations of the ITS1 and ITS2 regions (i.e. the ITS1, 5.8S rDNA, and ITS2 regions in these species ranged from 645 bp in G. doryphora to 725 bp in G. asiatica.). The cox2-cox3 spacer sequence alignment of 349 bp included only one gap of 1 bp. Primer sequences were not included in the alignments or in any of the analyses. The final sequence alignment for each region was used to calculate uncorrected pairwise distances.
The PAUP software package (Version 4.0b10, Swofford 2002) was used to reconstruct phylogenetic gene trees from the aligned data sets of rbcL, ITS (including the 5.8S rDNA) and cox2-cox3 spacer sequences. Maximum parsimony (MP) and maximum likelihood (ML) methods were used to estimate phylogenetic relationships for each data set. Halymenia floresia (Clemente y Rubio) C. Agardh (Wang et al. 2000) was used as an outgroup for rooting the rbcL trees. Based on its basal position in the rbcL tree topology, G. filicina was used to root the ITS and cox2-cox3 trees. In the parsimony analysis, trees were constructed using the heuristic search option with 500 random sequence additions. Tree bisection-reconnection branch swapping was performed, and all characters and character state transformations were unweighted. To compare relative support of the branches, 1000 bootstrap replications of full heuristic searches (Felsenstein 1985) were performed. For the ML analyses, the computer program Modeltest 3.06 (Posada & Crandall 1998) was used to estimate parameters and find the model of sequence evolution that best fit each data set. This program uses a hierarchical hypothesis-testing framework. The optimal model selected for the rbcL dataset was a General Time Reversible model with a gamma distribution (GTR ϩ G). For the ITS and mitochondrial cox2-cox3 data sets the optimal model was Hasegawa-Kishino-Yano (HKY) (Hasegawa et al. 1985) with a gamma distribution (HKY85 ϩ G). The model of sequence evolution and estimated parameters were then imported into PAUP to estimate phylogenetic relationships using heuristic ML searches (10 random additions). For the ITS and cox2-cox3 spacer sequences, bootstrap resampling support for ML was based on 500 iterations. Due to computational limitations, a ML bootstrap analysis of the rbcL data set was limited to 250 iterations. Thalli exhibit substantial morphological plasticity in texture, number, and branching patterns of axes and lateral branchlets; upright axes, 12-15 cm long, are dark red in colour and gelatinous to cartilagineous in texture; percurrent axis are compressed to flattened, 1-5 mm wide, up to 620-630 m thick, tapering above; simple or dichotomously branched once to three times, set with numerous lateral branchlets with pinnate arrangement and occasionally on the surface; branchlets are simple or dichotomously branched with pinnate proliferations along the margins and occasionally on the surface (Figs 2-4); marginal proliferations, 0.2-13.0 cm long and 1-2 mm wide, are compressed to oval in section and usually with second-order proliferations (Fig. 5); thallus is multiaxial and consists of a compact cellular cortex and a loose filamentous medulla (Fig.  6); medullary filaments are 5-8 m in diameter; cortex is 7-9 cells thick (outer cortex 4-5 cells thick); outer cortical cells, 3-5 m in diameter by 3-9 m in length, are ovoid to slightly cylindrical (Figs 7,8); medulla is lax but not hollow; gametophytes produce reproductive structures over the entire thallus except for the basal portion; auxiliary cell ampullae are initiated in the inner cortex; they are small and composed of 3-4 simple ampullary filaments, 5-10 cells long; mature auxiliary cells are oval in shape, slightly larger than other ampullary cells (Fig. 9); carpogonial ampullae are with 2-3 simple secondary filaments and a two-celled carpogonial branch (Fig. 10); cystocarps are spherical, not protruding, 160-200 m broad including pericarp; enveloping filaments are largely derived from ampullary cells (Figs 11,12); tetrasporangia are ellipsoidal, 29-40 m long and 15-20 m wide, and scattered over the entire thallus except for the basal portion, arising laterally from cortical cells in the fourth to fifth cortical cell layers from surface; cortical cells tending to become elongated as paraphyse cells (Figs 13, 14); spermatangial thalli were not observed. Upright blades, 10-70 cm long, dull rose-reddish to brownish in colour, membranaceous, lubricous in texture, are attached to substratum by means of discoidal holdfast of 7-20 mm diameter; blades, lanceolate foliose, 500-700 m thick, 3-15 cm broad, are stipitate, simple, branched dichotomously to palmately and complanate; margin entire or sometimes finely serrate in old fronds ; thallus is multiaxial and consists of a compact cellular cortex and a filamentous medulla (Fig. 18); medullary filaments, 3-8 m in diameter, are compactly interlaced; cortex is 9-13 cells thick (outer cortex 5-6 cells thick); outer cortical cells, 2-3 m in diameter by 6-10 m in length, are cylindrical elongated, (Figs 19,20); gametophytes produce reproductive structures over the entire thallus except for the basal portion; auxiliary cell ampullae are initiated in the inner cortex; they are long, conical and composed of 3-5 ampullary filaments, 10-22 cells long; mature auxiliary cells are oval in shape, slightly larger than other ampullary cells (Fig.  21); carpogonial ampullae are long, conical and composed of 3-6 secondary ampullary filaments and a two-celled carpogonial branch (Fig. 22); cystocarps, scattered on both surfaces are grouped irregularly; they are spherical, not protruding, 205-350 m broad including the well-developed pericarp; ostiole is slightly protruding (Figs 23-25); tetrasporangia are oblong, divided cruciately, occasionally irregularly, 40-60 m long and 20-34 m wide, and scattered over the entire thallus except for the basal portion, arising laterally from cortical cells in the fourth to sixth cortical cell layers from surface (Figs 26, 27); spermatangial thalli were not observed. Upright axes, up to 20 cm long, are purplish red to dull brownish red in colour, and cartilaginous and becoming firmer in texture when dried; percurrent axis are compressed, up to 5 mm wide, up to 490-575 m thick, tapering above; di-or trichotomously branched, set with numerous lateral branchlets produced from the margin in a pinnate manner; branchlets, simple or dichotomous, linear-lanceolate or oblong-lanceolate, some of them continuing to grow and produce lateral proliferations, as the main branches (Figs 41, 42); thallus is multiaxial and consists of a cellular cortex and filamentous medulla (Fig. 43); medulla consists of densely intermeshed and mainly periclinal oriented filaments, 2-7 m in diameter; cortex is 8-12 cells thick (outer cortex 5-9 cells thick); outer cortical cells are small, 2-3 m in diameter by 2-5 m in length, and ovoid (Figs 44,45); reproductive structures are confined to sorus-like groups in lateral branchlets and ultimate segments of main branches; gametophytes are monoecious; spermatangia cutt off from outer cortical cells (Fig. 46); auxiliary cell ampullae are initiated in the inner cortex; they are conical and composed of 2-3 simple ampullary filaments, 5-13 cells long; mature auxiliary cells are oval in shape, significantly larger than other ampullary cells (Fig. 47); carpogonial ampullae are small with 2-3 simple filaments and a twocelled carpogonial branch (Fig. 48); cystocarps are spherical, not protruding, 200-250 m broad including pericarp; enveloping filaments are largely derived from ampullary cells; ostiole is not to slightly protruding (Figs 49, 50); tetrasporangia are ellipsoidal, 44-50 m long and 13-22 m wide, arising laterally from cortical cells in the fourth to fifth cortical cell layers from surface (Fig. 51).  (Figs 52, 53); thallus is multiaxial and consists of a compact cellular cortex and a loose filamentous medulla; medullary filaments, 3-6 m in diameter, are loosely interlaced, some of them tending to be periclinal in young thalli (Fig. 54), but becoming randomly arranged in older individuals; cortex is 5-6 cells thick (outer cortex 3-4 cells thick); outer cortical cells, 4-9 m in diameter by 6-14 m in length, are ovoid to cylindrical (Figs 55, 56); gametophytes are monoecious and produce reproductive structures over the entire thallus except for the basal portion; auxiliary cell ampullae are initiated in the inner cortex; they are small, conical and composed of 2-3 ampullary filaments, 10-11 cells long; mature auxiliary cells are oval in shape, significantly larger than other ampullary cells (Fig. 57); carpogonial ampullae are with 2-3 simple secondary filaments and a two-celled carpogonial branch; cystocarps are spherical, not protruding, 220-290 m broad including the pericarp; the ostiole is not protruding to slightly depressed (Figs 58-60); spermatangia are cut off from outer cortical cells, 4-5 m in diameter, are produced per 2-3 by the mother cells; tetrasporangia are oblong, divided cruciately, 29-40 m long and 19-29 m wide, and scattered over the entire thallus except for the basal portion, arising laterally from cortical cells in the third to fifth cortical cell layers from surface without any modification of the cortex (Fig. 61).

Phylogenetic analyses
In this study, the rbcL gene from 16 specimens, the ITS region from 10 specimens, and the mitochondrial cox2-cox3 spacer region of 14 specimens were sequenced to use in phylogenetic analyses along with sequences obtained from GenBank (Table  1). Grateloupia patens was not included in the phylogenetic analyses because thalli have not been found since September 1995 and no material was available for DNA extraction. For each of the three regions (rbcL, ITS, and cox2-cox3 spacer), the overall topologies of the phylogenetic trees obtained from the MP and ML analyses were congruent. Phylogenetic analyses of the chloroplast rbcL, nuclear ITS sequences, and mitochondrial cox2-cox3 spacer sequences confirmed the identity of four nonnative Grateloupia spp. in the Thau Lagoon: G. asiatica, G. lanceolata, G. luxurians and G. turuturu. The Japanese and Thau lagoon G. asiatica specimens form a well-supported clade in the rbcL tree (Fig. 62). The rbcL sequence divergence between the G. asiatica specimen from Japan and specimens from the Thau Lagoon was 0.2%. The three G. asiatica specimens from the Thau Lagoon had identical rbcL and identical mitochondrial cox2-cox3 spacer sequences, whereas the sequence divergence of the ITS region ranged from 0.0% to 0.1% (Figs 62-64).
In the rbcL and cox2-cox3 gene trees, the G. lanceolata specimens from Japan and the Thau Lagoon fell into a wellsupported clade (Figs 62, 64). The two G. lanceolata specimens from the Thau Lagoon had rcbL sequences that were identical to one another and differed from the two Japanese G. lanceolata specimens by 0.0%-0.1%. The cox2-cox3 spacer sequences of Thau Lagoon G. lanceolata and Japanese G. lanceolata were identical. The ITS sequences of two Thau Lagoon specimens were identical (Fig. 63).
The Thau Lagoon, Atlantic France, and Australian specimens of G. luxurians and the Japanese specimens of G. subpectinata grouped together in the rbcL gene tree (Fig. 62). The Thau Lagoon, Atlantic France, and Australian specimens had identical rbcL sequences that differed from the Japanese specimens by 0.7%. The G. luxurians specimens from Thau Lagoon had identical mitochondrial cox2-cox3 spacer sequences and ITS sequences that differed by 0.4% (Figs 63,  64).
Specimens of G. turuturu from Japan, Thau, and the USA formed a well-supported clade in all three gene trees . The rbcL and ITS sequences of Thau Lagoon G. turuturu specimens were identical to those of the Japanese G. turuturu specimen. Intraspecific variation in mitochondrial cox2-cox3 spacer sequences, ranging up to 0.6%, was observed among the G. turuturu specimens.
The native G. filicina specimens form a robust clade in the rbcL and cox2-cox3 gene trees that is clearly distinct from the other clades of Grateloupia spp. (Figs 62, 64). The G. filicina rbcL sequences differ from all other rbcL sequences used in the analyses by 4.5-10.0%. The ITS and cox2-cox3 sequences of G. filicina differ from the sequences of all other species by 11.4-21.8% and 9.7-13.5%, respectively. The other Mediterranean species, G. dichotoma, also forms a distinct clade in rbcL and cox2-cox3 trees (Figs 62, 64). The rbcL sequence divergence between specimens in the G. dichotoma clade and all other species in the analysis is 7.1-11.0%. Interestingly, in all three trees, there is an undetermined isolate (Grateloupia sp.) from the Thau Lagoon that falls within the G. dichotoma clade (Figs 62-64) but that differs from the open sea G. dichotoma specimens by rbcL, ITS   and cox2-cox3 spacer sequence divergences of 1.3%, 1.6% and 3.4%, respectively.

Distribution and seasonality
In the Thau Lagoon, G. asiatica, G. lanceolata, G. luxurians and G. turuturu are common, but not invasive. They are found along the north coast of the Thau Lagoon, from Marseillan to Sète and in the aquaculture facilities (Parcs A-C). Grateloupia turuturu, however, was not observed in Parcs B and C (Fig.  1). Grateloupia patens, which has not been found since September 1995, was observed only along the north coast of the Thau Lagoon, between Bouzigues and Mèze, and in Parcs B. Thalli were observed from February to December (G. turuturu), March to December (G. asiatica and G. lanceolata), May to October (G. luxurians) and from September, October and April (G. patens). Field observations were not conducted in January. Thalli grow from 0 to Ϫ 1 m Mean Low Water, attached to hard substrata: bedrock outcrops, loose stones, man-made rocky structures, aquaculture facilities and shellfishes (mussels and oysters).
The specimen collected in 1984 and attributed to G. filicina in Ben Maiz et al. (1986) belongs to G. asiatica. This was the first collection of this species in the Mediterranean Sea. Grateloupia asiatica specimens from the Thau Lagoon are morphologically more diverse than Asiatic specimens. All the transitional forms were observed; some specimens possess a morphology similar to Japanese specimens of G. asiatica (gelatinous in texture; percurrent axis flattened, simple or dichotomously branched once or twice; numerous flattened lateral branchlets with pinnate arrangement, see , whereas other specimens possess a cartilaginous thallus, percurrent axis almost cylindrical and dichotomously branched similar to other Japanese Grateloupia species (e.g. G. divaricata Okamura) (Fig. 3); the most extreme forms have a percurrent axis dichotomously branched up to seven times, entirely covered with short dichotomous branchlets similar to the branchlets of the Prionitis spp. (now included in the genus Grateloupia, Wang et al. 2001) (Fig. 4). Despite this morphological variation, no differences in the molecular markers considered here were observed between the different forms of G. asiatica in the Thau Lagoon. The G. asiatica specimens from the Thau Lagoon all had identical rbcL sequences that were only 0.2% divergent from the rbcL sequences of the Japanese G. asiatica specimen. This level of rbcL divergence is within the intraspecific rbcL variation observed for other Grateloupia species (Wang et al. 2000Gavio & Fredericq 2002;Faye et al. 2004;De Clerck et al. 2005;Wilkes et al. 2005), thus confirming the identity of these specimens as G. asiatica. Although the initial introduction may have been restricted to a few individuals (small inoculate), new phenotypic variations could be the result of inbreeding, interspecific hybridization or phenotypic plasticity associated with different environmental conditions (Ellstrand & Schierenbeck 2000;Holland 2000;Lee 2002).
Recently G. subpectinata Holmes, a taxon described from Japan [Holmes 1912; type locality: not specified in protologue; Enoshima, Kanagawa Prefecture according to Okamura (1936)] and placed in synonymy with G. filicina, was reinstated (Faye et al. 2004). This species is morphologically similar to G. luxurians (Table 2). Furthermore, the rbcL sequence divergence between G. luxurians and G. subpectinata is only 0.7% (present study), a value far lower than the interspecific divergence values reported for the genus Grateloupia (i.e. reported values range from 1.5% to 10.0%, but are usually between 5.0% and 8.8%; Wang et al. 2000Wang et al. , 2001Gavio & Fredericq 2002;Faye et al. 2004;De Clerck et al. 2005;Wilkes et al. 2005). The morphological comparison between the Australian, Mediterranean and Atlantic specimens of G. luxurians and some Japanese specimens of Grateloupia previously identified as G. subpectinata (H7060-7061; Figs 65, 66) did not reveal any differences. As a result, G. subpectinata and G. luxurians appear to be conspecific and G. subpectinata Holmes (1912) is considered as a later taxonomic synonym of G. filicina var. luxurians A. Gepp & E.S. Gepp (1906) (now G. luxurians). Consequently, the geographical distribution of G. luxurians is extended to Japan.
The re-examination of the N. Ben Maiz collections confirmed the hypothesis proposed by Verlaque (2001) and Gavio & Fredericq (2002) that specimens attributed to G. doryphora in Ben Maiz et al. (1986) belong to G. turuturu. Moreover, according to the description and the illustrations given in Riouall et al. (1985), it is probable that the first observation of G. doryphora in the Thau Lagoon in 1982 was a misidentification of either G. turuturu (thallus linear foliose, undulate with margin entire or provided with proliferations, see figs 2, 3, in Riouall et al. 1985) or G. lanceolata (thallus foliose, palmately branched in one plane with margin entire, cortex   2002). The identity of these populations awaits molecular determination.

Vector, date and origin of the introductions
From 1971 to 1976, there were massive importations of Japanese oysters, C. gigas (Thunberg), from Japan (Sendai, Miyagi Prefecture, NE Honshu) to the Thau Lagoon (Grizel & Héral 1991;Grizel 1994). The presence of the introduced species of the Thau Lagoon in Japan and the dates they were first observed in the lagoon (i.e. G. lanceolata and G. turuturu: possibly 1982; G. asiatica: 1984; and G. luxurians: 1990) (Verlaque 2001(Verlaque , 2002Verlaque et al. 2002;Kim et al. 2003). In the case of G. patens, its discovery in 1994 and its apparent fast disappearance could be due to an introduction in the early 1990s by illicit oyster importations from Japan, as was probably the case with Chondrus giganteus Yendo f. flabellatus Mikami Verlaque & Latala 1996). A bibliographical analysis of the introduced flora of the Mediterranean lagoons (M. Verlaque, unpublished observations) shows that (1) most of the introductions of marine macroalgae occurred in the lagoons harbouring shellfish aquaculture facilities (oysters and (or) mussels) and (2) more than 80% of the introduced taxa were native to Japan and (or) Korea, the world's two major oyster exporters. However, in the case of G. luxurians and G. turuturu, it is possible that the introduction occurred via oyster-spat importations from populations of Japanese oysters already established in the Atlantic because both species seem to have been introduced in the NE Atlantic before the 1970s (see Farnham &Irvine 1968, 1973, as G. doryphora andG. filicina var. luxurians). On the basis of this study, it is not possible to determine the exact source population(s) of the introduced species. It is particularly difficult considering that there were probably multiple introductions of each species into Europe and that these species have a very wide distribution in the warm and temperate seas of Japan and Korea (Okamura 1935;Kawabata 1962;Lee & Kang 1986;Noda 1987;Chihara 1990;H.B. Lee & I.K. Lee 1993;Tokuda et al. 1994;Yoshida 1998;Faye et al. 2004).

Probability of spread and risks of invasion
The five introduced Grateloupia species in the Thau lagoon constitute a case study of multiple introductions. Numerous studies of plants have documented the occurrence of intraand interspecific hybridizations after species introductions (Ellstrand & Schierenbeck 2000;Lee 2002). The risk of hybridization increases as the number of introduced allopatric populations of the same taxon or different taxa increases. Although in this study none of the specimens we examined appears to be a hybrid (i.e. gene trees based on nuclear, chloroplast, and mitochondrial sequences do not exhibit well-sup-ported incongruities in the relative positions of taxa), the presence of multiple, closely related, introduced species makes the Thau Lagoon a 'melting pot' from which new genetic combinations could arise. Although hybridization in Rhodophyta is extremely rare (Rueness 1978;Brodie et al. 1993Brodie et al. , 1997, this should be of concern because there are many examples of plants in which hybridization preceded the emergence of successful invasive populations (Ellstrand & Schierenbeck 2000).
In the NW and NE Atlantic, G. turuturu (as G. doryphora) is an invasive species (Harlin & Villalard-Bohnsack 2001;Simon et al. 2001;Villalard-Bohnsack & Harlin 2001;Bárbara & Cremades 2004). This species withstands a large range of salinities, temperatures, and other environmental conditions (Simon et al. 1999(Simon et al. , 2001Harlin & Villalard-Bohnsack 2001). In the Thau Lagoon, four Asiatic Grateloupia species (e.g. G. asiatica, G. lanceolata, G. luxurians and G. turuturu) have been successfully established for the past 20 to 25 yr and have developed reproductive populations without becoming invasive. Nevertheless because the Thau Lagoon is an important site for the exportation of living bivalve molluscs (C. gigas, Ostrea edulis, Mytilus galloprovincialis, Tapes spp.), these species are likely to spread, if they have not already, to other regions and countries in the Mediterranean and throughout Europe.

NOTE ADDED IN PROOF
According to Art. 11.2 of the ICBN (Greuter et al. 2000), priority only applies in the same rank. G. luxurians was described as a variety of G. filicina and hence cannot take priority over G. subpectinata. Consequently, the correct name for this species is G. subpectinata Holmes.