Genealogical partitioning and phylogeography of Colpomenia peregrina (Scytosiphonaceae, Phaeophyceae), based on plastid rbcL and nuclear ribosomal DNA internal transcribed spacer sequences

G.Y. Cho, S.M. Boo, W. Nelson and M.N. Clayton. 2005. Genealogical partitioning and phylogeography of Colpomenia peregrina (Scytosiphonaceae, Phaeophyceae), based on plastid rbcL and nuclear ribosomal DNA internal transcribed spacer sequences. Phycologia 44: 103–111. Colpomenia peregrina shows a large morphological variation, and two morphotypes have been described. We used the protein-coding plastid rbcL and the nuclear ribosomal internal transcribed spacer (ITS) region to investigate whether these morphotypes constitute distinct species and to explain the current distribution of the species. Here, we sequenced the rbcL gene from 38 specimens (32 C. peregrina and six putative relatives) and the ITS region from 33 specimens of C. peregrina, including an outgroup taxon. The C. peregrina specimens were variable, having up to 1.17% intraspecific divergence and nine haplotypes in the rbcL gene, and up to 11.01% intraspecific divergence and 21 haplotypes in the ITS region. Independent analyses of the rbcL and ITS data sets produced highly congruent but not identical results. Colpomenia peregrina is monophyletic, but is partitioned into two deeply divergent clades (‘lineage I’ and ‘lineage II’) that we interpret as different species. Lineage I consists of 27 specimens, in both rbcL and ITS data sets, and lineage II contains six specimens. Both lineages occur together in Australia, Korea, New Zealand and USA. Lineages I and II correspond to the epiphytic and epilithic forms, respectively, recognized by Clayton. Our rbcL and ITS data sets corroborate the recent anthropogenic dispersal event between the northwest Pacific and northeast Atlantic Oceans, and also suggest some natural dispersal events during the Pleistocene between the North and South Pacific Ocean.


INTRODUCTION
Colpomenia peregrina (Sauvageau) Hamel is a scytosiphonacean brown alga that occurs in temperate waters of both the northern and southern Pacific Ocean, as well as in the North Atlantic Ocean. It is recognized by extensive irregular sori that lack cuticles, and a thin thallus of three to four layers of colourless medullary cells (Clayton 1975). It superficially resembles C. sinuosa (Mertens ex Roth) Derbès & Solier, but the latter has punctate sori with a cuticle and commonly four to six layers of medullary cells (Clayton 1975). The type locality of C. sinuosa is Cadiz in Spain, and the lectotype locality of C. peregrina is Morbihan in France (Yoshida 1998).
Colpomenia peregrina is an annual, and has a heteromorphic life history in which a parenchymatous gametophyte alternates with a pseudoparenchymatous sporophyte (Clayton 1979;Vandermeulen 1986;Kogame & Yamagishi 1997). The gametophyte is globular to irregular and 5-10 cm in diameter, whereas the sporophyte is a prostrate crust, 1-3 mm in diameter (Kogame & Yamagishi 1997). The saccate thalli of C. peregrina are hollow and filled with water and air, which provides buoyancy.
In the North Pacific, C. peregrina occurs from Korea (Cho et al. 2001) and Japan (Kogame & Yamagishi 1997), through Vladivostok in Far-East Russia (Adrianov & Kussakin 1998), to Alaska, British Columbia, Oregon and California in North America (Scagel et al. 1989), where it grows intertidally in spring (Vandermeulen 1986;Kogame & Yamagishi 1997). The species also occurs in the South Pacific in Australia and New Zealand (Blackler 1967;Clayton 1975;Parsons 1982). Although C. peregrina occurs from Norway to the Mediterranean in the North Atlantic, the species is considered to be introduced, possibly from the northwest Pacific (Farnham 1980;Blackler 1981;Fletcher 1987;Lüning 1990). Clayton (1975) described two morphotypes of C. peregrina from southern Australia: a relatively small globose form (with thallus diameter 1-5 cm), and a larger irregular form (thallus diameter 7-10 cm) with a deeply infolded surface. The globose form is usually epiphytic and occurs in tidepools, whereas the irregular-shaped form occurs mainly on open rock surfaces in the intertidal to upper subtidal zones. The globose form is the type most commonly found in Europe (Blackler 1967), whereas the irregular form predominates in winter in southern Australia (Clayton 1975). However, Clayton (1975 suggested that variability in the two forms did not justify separate species. The objectives of the present study were to test Clayton's (1975) hypothesis that two morphs can be distinguished in C. peregrina, using the large subunit of the Rubisco spacer (rbcL) and the nuclear ribosomal DNA (nrDNA) internal transcribed spacer (ITS) region, and to explain the current distribution of the species. We determined the rbcL sequences of biogeographically representative specimens of C. peregrina and its putative relatives. The rbcL gene is useful for identifying species because it is generally highly conserved at the species level (Draisma et al. 2001), although intraspecific variation has been  Table 1, and haplotypes are indicated (rbcL/ITS). Black arrowheads represent specimens of lineage I, and white arrowheads those of lineage II.
observed . We also analysed the ITS region from 33 specimens including C. sinuosa as outgroup. The ITS region is useful for comparing local populations within scytosiphonacean taxa (Cho et al. 2002) and congeneric species in brown algae (e.g. van Oppen et al. 1993;Peters et al. 1997;Stache-Crain et al. 1997;Sasaki et al. 2001;Kim & Kawai 2002). However, we did not include other Colpomenia (Endlicher) Derbès & Solier species because ITS sequences were too divergent to be aligned in our study.

Taxon sampling
Thirty-two individuals of C. peregrina from 29 localities in the Pacific and Atlantic Oceans were sampled ( Fig. 1; Table  1). For the rbcL analysis, six putative relatives (one C. bullosa, one C. phaeodactyla and four C. sinuosa) were also included. Additional sequences of Colpomenia spp. were downloaded from GenBank. Scytosiphon lomentaria was used as outgroup for the rbcL data set (Kogame et al. 1999;Cho et al. 2001Cho et al. , 2003. We examined the ITS region from C. sinuosa as the outgroup taxon as well as 32 C. peregrina specimens. Collection sites and GenBank accession numbers are listed in Table 1. All samples were air-dried and preserved with silica gel prior to extraction of genomic DNA. Voucher specimens were deposited at the Research Center of Chungnam National University, Daejon, Korea.

Deoxyribonucleic acid extraction, amplification and sequencing
Dried samples were ground to fine powder in liquid nitrogen. Approximately 0.01 g of the powder was used for genomic DNA extraction using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Extracted crude DNA was stored at Ϫ20ЊC and used for polymerase chain reaction (PCR) amplification of rbcL and ITS.
The rbcL region was amplified and sequenced using the primers PRB-F0, F2, F3, R1A, R2, R3A (Kogame et al. 1999), and RS1 and RS2 (Yoon & Boo 1999). The ITS 1 and 2, parts of the 3Ј-terminus of the 18S and the 5Ј-terminus of the 26S ribosomal RNA genes, and the complete 5.8S gene were amplified using the LB1 and LB2 primers of Yoon et al. (2001). For the ITS region, the primer pairs LB1/BC2 (Saunders & Druehl 1993) and YB1 )/LB2 were used in the sequencing reactions. PCR and sequencing reaction concentrations followed Kogame et al. (1999). The PCR products for both regions were purified using the High Pure PCR Product Purification Kit (Roche, Indianapolis, IN, USA), according to the manufacturer's protocol. The sequences of the forward and reverse strands were determined for all taxa, using an ABI PRISM 377 DNA Sequencer (Applied Biosystems, Foster City, CA, USA). The electropherogram outputs for each sample were checked using Sequence Navigator v. 1.0.1 software (Applied Biosystems). The rbcL sequences were collated using the multisequence editing program SeqPup (Gilbert 1995), and aligned visually with those published previously (Kogame et al. 1999). A total of 43 rbcL sequences, which included five previously published sequences, were used for the phylogenetic analyses. All 33 new ITS sequences, including that of C. sinuosa, were aligned by eye.

Data analysis
We used only haplotypes of both rbcL and ITS sequences for phylogenetic reconstructions. Analysis of phylogenetic relationships between haplotypes of each data set was conducted using PAUP* 4.0b10 (Swofford 2002). For the rbcL data, a maximum likelihood (ML) analysis was conducted using the TrN ϩ G model, which was selected by ModelTest (v. 3.06;Posada & Crandall 1998). The parameters of the model were as follows: estimated substitution rates: T ϭ 1, and shape parameter: 0.014413. Tree likelihoods were estimated using a heuristic search with 100 random sequenceaddition replicates, and tree bisection-reconnection (TBR) branch swapping. Maximum parsimony (MP) analysis was done using a heuristic search algorithm with the following settings: 100 random sequence-addition replicates, TBR branch swapping, MulTrees, all characters unordered and unweighted, and branches with a maximum length of zero collapsed to yield polytomies. For the ITS data, the optimal model was the TrN ϩ I ϩ G model. The parameters used were as follows: estimated substitution rates: invariable sites: 0.518586, and shape parameter: 0.607407. The MP analysis followed the same method as was used for the rbcL data set. For both rbcL and ITS trees, nonparametric bootstrap values for nodes in ML and MP phylograms were calculated based on 100 and 1000 resamplings, respectively. The neighbour-joining analyses for both rbcL and ITS sequence data were done with the same parameters used for the ML analyses but, because results were congruent with the ML and MP analyses, they are not shown.
In C. peregrina specimens, all substitutions except one were at the third codon position and silent. However, in the Gellibrand specimen (AU1), at position 751, which is a first codon position, adenine was changed to guanine, and thus valine was replaced by methionine.
We found nine haplotypes among the 33 rbcL sequences of C. peregrina, including one previously published sequence (Figs 1, 2; Table 1). Haplotype 1 was found in 19 specimens from Korea, Japan, Russia, Australia, New Zealand, France and the UK. Haplotypes 2 and 7 were observed in both Korea and USA. Haplotype 8 occurred only in Australia, and hap-lotype 9 only in New Zealand. Haplotypes 1 and 6 occurred together at Anin, Korea, and haplotypes 1 and 9 were found at Weller's Rock, New Zealand.
Colpomenia bullosa specimens from Japan and New Zealand had identical rbcL sequences, as did those of C. phaeodactyla from Korea and Japan. Among five specimens of C. sinuosa, two haplotypes, differing by a single base pair, were found: one from Korea, Japan, and Australia and the other from Japan.
The tree produced by the ML analysis based on the rbcL haplotypes is illustrated in Fig. 3, and is identical with the single most parsimonious tree [Tree length ϭ 139, Consistency index (CI) ϭ 0.855, and Retention index (RI) ϭ 0.916]. The trees consistently showed a monophyletic clade of all 33 C. peregrina specimens, which were separated into two major assemblages designated 'lineage I' and 'lineage II'. Each of the two lineages was strongly supported by bootstrap values. Lineage I contained specimens from Korea, Australia, France, Japan, New Zealand, Russia, the UK and the United States. Within lineage I, haplotypes 1, 3 and 5 formed a weakly supported clade [bootstrap support (BS) 61% for both ML and MP]. Lineage II contained the remaining specimens from Korea, Australia, New Zealand and USA. However, lineage II consisted of two subclades; one subclade contained haplotypes 6 (KE2) and 8 (AU1) (BS 86% for ML and 93% for MP), and the other consisted of haplotypes 7 (KE4 and US1) and 9 (NZ2 and 3) (BS 95% for ML and 98% for MP).
Colpomenia sinuosa was consistently the sister clade to C. peregrina with strong support. Colpomenia bullosa and C. phaeodactyla formed a single clade with maximum support.
Twenty-one ITS haplotypes were found within C. peregrina: 13 (haplotypes 1-11, 17 and 18) from Korea, two (haplotypes 12 and 19) from Australia, four (haplotypes 13-15 and 20) from New Zealand, and two (haplotypes 16 and 21) from USA ( Fig. 1; Table 1). Haplotype 1 also occurred in Russia, France and the UK. Two different ITS haplotypes were found at one location: 1 and 17 at Anin, Korea, and 13 and 20 at Weller's Rock, New Zealand.
The pairwise divergences of the ITS in C. peregrina haplotypes ranged from 0.11% (e.g. between haplotypes 1 and 7) to 11.01% between haplotypes 16 and 18. The pairwise divergences averaged 4.7% among all 21 haplotypes.
The tree constructed by the ML analysis based on the ITS haplotypes is illustrated in Fig. 5, and is identical with one of 158 equally most parsimonious trees (Tree length ϭ 252, CI ϭ 0.849 and RI ϭ 0.908) in producing two lineages. The composition of both lineages was the same as that of the rbcL analyses. Lineage I was supported by 73% BS for ML and 100% BS for MP, and lineage II was supported by 64% BS for ML and 100% BS for MP. In lineage I, most of the branches between haplotypes were not well resolved. Lineage II consisted of two subclades; the first included haplotypes 19 (AU1), 20 (NZ2 and 3), and 21 (US1), and the second contained haplotypes 17 (KE2) and 18 (KE4). One subclade was supported by 62% BS for ML and 93% BS for MP, and the other by 59% BS for ML and 97% for MP.

DISCUSSION
The rbcL tree highlights a single evolutionary origin for C. peregrina. Both the rbcL and ITS trees reveal partitioning of C. peregrina into two groups (lineages) with the same constituent specimens. Each lineage is supported by high bootstrap values for both markers. In the rbcL data, the two groups of C. peregrina have greater divergence (0.89-1.17%, 13-17 bp) than that between North and South Pacific populations of C. sinuosa (0.07%, 1 bp), and between C. bullosa and C. phaeodactyla (0.34%, 5 bp). These levels of divergence within C. peregrina are equal to or surpass those between other species of scytosiphonacean algae (0.5-5.9%; Kogame et al. 1999).
Similar high divergence values (9.20-11.01%) are found between the two ITS lineages of C. peregrina. This genetic divergence within samples of C. peregrina is in marked contrast to the lower levels of variability recorded between specimens of Petalonia binghamiae (J. Agardh) Vinogradova from Korea and USA (0.33%; Cho et al. 2002) and between arcticantarctic populations of Desmarestia viridis (O.F. Müller) Lamouroux -D. willii Reinsch (0.09%; van Oppen et al. 1993). The divergence value within C. peregrina is higher than those (0.5-6.4%) among five species of Fucus Linnaeus (Leclerc et al. 1998), and higher than or equal to those (3.8-17.4%) among species of Desmarestia Lamouroux (Peters et al. 2000).
It is therefore clear that C. peregrina consists of two deeply divergent and evolutionarily distinct lineages, which we interpret as two different species. A full taxonomic and nomenclatural revision of the species will appear elsewhere. Here, we will only briefly discuss some morphological and ecological characteristics. In our samples from Korea and elsewhere, we observed features discriminating the two lineages of C. peregrina, as described by Clayton (1975). Lineage I corresponds to her epiphytic group and lineage II to her epilithic group (see Table 1). Lineage II specimens were irregular in shape, usually found subtidally, and always epilithic, whereas the lineage I samples were often but not always globose, usually intertidal, and always epiphytic. Other morphological characters such as thallus diameter need further study before diagnosing the lineages.
Colpomenia sinuosa was consistently the sister taxon to C. peregrina in our rbcL tree. These results are congruent with similarities of both species in morphology and life history: both have pluri-and unilocular zoidangia on prostrate thalli (Kogame et al. 1999) and form plurilocular zoidangia in longday culture conditions, but mostly unilocular zoidangia under short-day conditions (Kogame 1997). Biogeographically, C. peregrina and C. sinuosa overlap in the North and South Pa-  Table 1). Numbers above the sequences indicate sites, and dots denote the same nucleotide as haplotype 1; dashes represent alignment gaps. cific (e.g. Wynne & Norris 1976;Parsons 1982;Kogame & Yamagishi 1997); however, the former species occurs commonly in cold-temperate waters (Clayton 1975), whereas the latter inhabits warm-temperate waters (Kogame 1997). These observations lead us to suggest that C. peregrina probably originated at the northern margin of the taxon that also gave rise to C. sinuosa. The genetic variability within C. peregrina, mentioned above, suggests that the evolutionary divergence of the two lineages in C. peregrina may have a long history.
Colpomenia peregrina is common in the North Pacific and in Australasia (Clayton 1975;Parsons 1982;Kogame & Yamagishi 1997), but is notably absent in South America and South Africa (Ramírez & Rojas 1991;Stegenga et al. 1997). This distributional pattern suggests that the northwest Pacific may be the distribution centre of the species. In our rbcL and ITS data sets, the two lineages of C. peregrina overlap in Korea, Australia, New Zealand and USA, and both occur at Anin, Korea and at Weller's Rock, New Zealand. This result indicates that the populations from the North Pacific are not separated evolutionarily from those of Australasia, and the cooccurrence of the two lineages in these regions may be due to congruent dispersal events from north to south or vice versa. In both rbcL and ITS data sets, haplotypes from Korea appear more diverse than those from Australasia; however, further systematic and unbiased sampling will decide the genetic diversity centre.
Our rbcL and ITS data are considered to reflect dispersal events of the C. peregrina taxa. In lineage I, the occurrence of identical rbcL haplotypes but different ITS types between Korea and USA and between Korea and Australia-New Zealand leads us to infer natural dispersal events in the Pacific Ocean. The continuous distribution of C. peregrina from Korea, Far-East Russia and northern Japan, through Alaska and British Columbia to Oregon and California, USA, indicates that this species might have spread naturally. The species is speculated to float on the warm Kuroshio Current from the west to the east side of the Pacific Ocean or on the North Equatorial Current from the east side to the west side. Between the North and South Pacific Oceans, the species is inferred to have crossed a land bridge between Australia and Indonesia during the Pleistocene, when seawater temperatures in this area were 6-8ЊC lower than they are today (Clayton 1984;Raven et al. 2002). This is consistent with the views of van den Hoek (1982), who considered that brown algal species, such as S. lomentaria and Petalonia fascia (O.F. Müller) Kuntze, may have crossed the tropics during cool periods of the Pleistocene. In contrast, there is no evidence that C. peregrina migrated along the east coast of the Pacific Ocean at this time because its absence in Chile (Ramírez & Rojas 1991).
Although it was identified from diverse localities (Australia, Korea, New Zealand and USA), lineage II is represented by only six out of 32 samples. The subtidal habitat occupied by most populations of lineage II means that its haplotypes were under-represented in the present study. Lineage II consists of two subclades, supported by high bootstrap values. However, KE4 and US1 occur in the same subclade in the rbcL tree, whereas the former is linked with KE2 in the ITS tree. This incongruence is difficult to explain in our data, although the phenomenon may be due to a higher substitution rate in the noncoding ITS region compared to the coding rbcL region. The close relationship between NZ2/3 and US1 suggests the possibility of human-mediated dispersal between New Zealand and USA.
We interpret, as evidence of recent introduction, the occurrence of identical haplotypes of both rbcL and ITS sequences between two areas. Despite the biogeographical barrier between the Pacific and Atlantic Oceans, specimens with rbcL haplotype 1 and ITS haplotype 1, belonging to lineage I, occur in France and UK and Korea and Far-East Russia. This is consistent with previous reports that the species was introduced anthropogenically from the northwest Pacific to Europe in the early 1900s (Farnham 1980;Blackler 1981;Fletcher 1987). According to Lüning (1990), C. peregrina was probably introduced from Japan through oyster cultures.
All specimens of C. sinuosa from Korea, Japan and Australia that we analysed have the same rbcL haplotypes, except a specimen from Hyogo Prefecture, Japan (Kogame et al. 1999), which has a single base difference. Investigations of C. sinuosa, which has a world-wide distribution (Silva et al. 1996), using the ITS region, will provide an interesting comparison with our results for C. peregrina.
According to Parsons (1982), C. bullosa was recently introduced into New Zealand on ships from either Japan or North America. In the present study, C. bullosa from Japan and New Zealand had identical rbcL haplotypes. This result leads us to recognize C. bullosa in New Zealand. Although Ramírez & Rojas (1991) and Adams (1994) considered C. bullosa (type locality: Pacific Grove, California) as a later synonym of C. durvillei (Bory) M.E. Ramírez, a molecular study of material from Concepción, Chile, the type locality of C. durvillei, is required to confirm synonymy of the two species.
In conclusion, this study has contributed important details to our understanding of the phylogeny of a brown alga, C. peregrina, which can be seasonally conspicuous in the intertidal zone at some localities. Within the species, we have identified two distinct monophyletic groups, which have additional distinctive characteristics, and which we interpret as two different species. Failure to recognize two evolutionarily distinct groups of C. peregrina could lead to the underestimation of brown algal diversity. The rbcL and ITS data corroborate a recent anthropogenic dispersal event between the northwest Pacific and the North Atlantic Oceans, and also suggest some natural dispersal events during the Pleistocene between the North and South Pacific Ocean. Finally, a full taxonomic and nomenclatural revision of this species is clearly needed, but is beyond the scope of this paper and will appear elsewhere.