Human-mediated global dispersion of Styela plicata ( Tunicata , Ascidiacea )

Styela plicata (Lesueur, 1823) is a solitary ascidian found in shallow, protected environments in tropical and warm-temperate oceans. Its origin is uncertain, given that it has already been identified in several oceans since it was first described, showing a very broad geographical distribution. Although S. plicata has been historically classified as a cosmopolitan species, in the past few decades it has been considered as an introduced or invasive species in some regions of the world. The present study investigated the genetic variation among populations of S. plicata. A total of 51 samples were obtained from locations on the Atlantic and Pacific coasts of the USA, Japan, and southern and southeastern Brazil. The amplification of a fragment of the cytochrome oxidase subunit I gene (COI) generated nine distinct haplotypes. There was considerable variation in the geographical distribution of haplotypes, yet the highest nucleotide and haplotypic diversities were clearly found in the Pacific samples. Interestingly, one of the haplotypes showed more than 3% divergence in relation to the remaining haplotypes, suggesting the possibility of a cryptic species. These results, together with historical records, indicate that commercial shipping could be the main cause for the global distribution of S. plicata. The northwestern Pacific region is hypothesized as the center of distribution of the species.


Introduction
Ascidians are considered as an excellent model system for biogeographical studies (Monniot 1983), given that the adult stage is sessile and natural dispersal occurs through gametes or larvae (Lambert 2005).Under natural conditions, colonial ascidian larvae usually do not disperse more than a few meters.Larvae of solitary species, on the other hand, can remain swimming freely for periods longer than 12 h, causing a consequent broader distribution (Ayre et al. 1997).
Styela plicata (Lesueur, 1823) is a solitary ascidian found in shallow, protected environments in tropical and warm-temperate oceans (Figure 1).Its geographical distribution is broad, yet its origin is unknown (Lambert 2001).The type-specimen was found attached to the hull of a ship in Philadelphia, although no other individual was detected in natural substrates in the region (Van Name 1945).Other records of the presence of S. plicata on ship hulls have been made in the Bay of Hann, Senegal, in 1950 (Pérès 1951) and on the USS Palos after a voyage through the Pacific, coming from either China or Japan (Tokioka 1967).
The species has been recorded in warm, temperate waters of the Atlantic Ocean and the Mediterranean (Annex 1, Harant 1927;Harant and Vernières 1933).Although it has been found throughout much of the eastern coast of North America since the beginning of the 20 th century (Van Name 1912;Huntsman 1912), it had been considered very rare on the west coast until the mid 1940's (Van Name 1945).It had not been recorded in Bermuda at the beginning of the 20 th century (Van Name 1902), until it was first reported as a single specimen in 1945 (Van Name 1945) and as a small population in 1972 (Monniot 1972).It was considered as an introduced species in the Gulf of Mexico (Lambert et al. 2005).The earliest record in the Pacific Ocean is in the Sydney harbor, Australia, in 1878 (Heller 1878), but it is also considered as an introduced species in that region (Kott 1985;  Berents and Hutchings 2002;Wyatt et al. 2005), whereas it was first reported in New Zealand in 1948 (Brewin 1948).There were no occurrence records on islands and reefs in tropical waters in either the east or west Pacific until 1980 (Kott and Goodbody 1980).It is invasive in southern Brazil (Rocha and Kremer 2005) and southern California (Lambert and Lambert 2003).
Styela plicata thrives on brackish and polluted waters, frequently being found in estuarine environments (Kott 1952(Kott , 1972a;;Kott and Goodbody 1980) or in shallow waters surrounding the mainland where large flows of freshwater reduce local salinity (Sims 1984).It is also found in disturbed areas such as in the proximity of refineries, power plants, and fishing harbors (Carballo and Naranjo 2002).It can be considered as an indicator species in areas that have experienced intense stress (substrate transformation, water stagnation, and excessive sedimentation) for extended periods of time (Naranjo et al. 1996).It can adhere to several types of substrate, particularly artificial substrates, and is also found in epibiosis.Styela plicata occurs on the shells of bivalves and competes with them for resources (Perera et al. 1990), and can also prey on their larvae (Bingham and Walters 1989).These features cause the presence of S. plicata to be highly undesirable in shellfish aquaculture.
The goal of the present study is to investigate the evolutionary relationships among seven populations of S. plicata, including samples from both the Atlantic and Pacific oceans, through the study of sequence variability in a fragment of the cytochrome oxidase subunit I gene.The comparison of samples of varying geographical distances is used to understand the dispersion patterns of the species, as well as to infer its geographical origin.

Specimen collection
Seven populations of S. plicata were sampled across four geographically distinct regions.Two populations were in the Pacific Ocean and the remaining populations were in either North or South Atlantic Oceans (Table 1).Samples of two species were used as outgroups in the analyses: Polycarpa pomaria (Savigny, 1816) (obtained from GenBank under accession number AY600984.1)and Styela canopus (Savigny, 1816) (collected on August 11, 2006 in Bocas del Toro, Panama).
Specimens were dissected in the field or in the laboratory for the removal of tissue fragments (muscles and gonads), which were stored in 95% ethanol and kept at approximately -18°C (Hillis et al. 1996).Universal primers HCO1498 (5'-TAA ACT TCA GGG TGA CCA AAA ATC -A 3') and LCO2198 (5'-GGT CAA CAA ATC ATA AAG ATA TTG G -3') were used to amplify approximately 700 pb of the cytochrome oxidase subunit I gene, according to Folmer et al. (1994).PCR amplification was carried out in 25-µL solutions using the following final concentrations: 1 X reaction buffer, 3 mM of MgCl 2 , 0.4 mM of dNTP (Biotools BeM Labs), 50 mol of each primer, 1 U of Taq Platinum DNA polymerase (Invitrogen) and 10-50 ng of template DNA.Thermocycling conditions had the following specifications: 2 min at 95°C, followed by 35 cycles of 40 s at 92°C, 40 s at 57°C, and 40 s at 70°C, and a final extension for 5 min at 70°C.After purification with ammonium acetate (7.5 M), sequencing reactions were carried out in a 10-µl volume containing 0.5 µl of BigDye (Terminator v3.1 Cycle Sequencing Kit -Applied Biosystems), 1.0 µl of primer (1.6 mol), 0.5 µl of buffer solution, 0.5 µl to 6.0 µl of DNA and ultrapure water, with the following thermocycling conditions: 1 min at 96°C, followed by 35 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C.Sequencing was carried out on a 3130 Applied Biosystems automatic sequencer.

Data analysis
Electropherograms were reconciled using the software BioEdit v.5.0 (Hall 1999) and aligned unabiguously due to the absence of indels.
A phylogenetic tree of the studied haplotypes was inferred using the neighbor-joining (NJ) method (Saitou and Nei 1987), with topology and bootstrap support values (1000 replicates) being calculated using the software PAUP* 4.0 (Swofford 2000) based on a GTR+ model of sequence evolution.This model was selected by comparing the likelihood values of alternative models using the Akaike information criterion, as implemented in Modeltest (Posada and Crandall 1998).These results were complemented by an additional analysis using Bayesian Inference (BI) using to the GTR++I model (2.500.000generations, with a burn-in of 1000 generations), as implemented in the software MrBayes 3.0 (Ronquist and Huelsenbeck 2003).
Diversities at both nucleotide () and haplotype (Hd) levels (Nei 1987) were measured in each population using the software DnaSP v.4.1 (Rozas et al. 2003).The hypothesis of a nonrandom distribution of haplotypes (as opposed to panmixia) among populations of different locations was assessed using the exact test for population differentiation (Raymond and Rousset 1995), with its significance being measured from 10.000 random permutations in the software Arlequin 3.1 (Excoffier et al. 2005).

Results
A total of 51 sequences (562 bp) of the studied COI fragment were obtained for S. plicata and were classified into nine distinct haplotypes (Table 2, GenBank accession numbers EU708881-EU708889).Twenty-three polymorphic sites were identified (4.1%), four of which represented nonsynonymous substitutions (Figure 2).Transitions accounted for approximately 77% of the substitutions, of which 77.3% were located on the third codon position and 22.7% on the second position.The mean AT content of the sequences was 63.2%.Nucleotide diversity () and haplotype diversity for the entire dataset were 0.0095±0.002and 0.784±0.05,respectively (mean ± SD).
The phylogenetic relationships among the haplotypes of S. plicata are shown in Figure 3.The results from BI and NJ analyses were highly concordant, except for the expected instability in the very short nodes (indicated in Figure 3 as the nodes without support values).There was considerable geographical variation in the distribution of haplotypes, with the same haplotype being found in very distant locations, as one would expect in a highly invasive species.Interestingly, haplotype 9 is highly divergent in relation to all other haplotypes and might in our opinion represent a distinct species.Therefore, henceforth the text will focus only on the remaining 8 haplotypes.
The studied haplotypes could be divided into two groups.The first group includes haplotypes 1-4 and is most commonly found in the West Atlantic, except for two occurrences of haplotype 2 in Japan.The second group includes haplotypes 5-8 and is monophyletic using the NJ method (but with only 33% bootstrap support) but not under the BI method (Figure 3), possibly due to the fairly high divergence between haplotype 9 and the remaining sequences.The haplotypes in the second group tend to be more geographically widespread, such as haplotypes 7 and 8 (Table 2).Such differences are also reflected in a statistically significant heterogeneity in the distribution of haplotypes (P<0.05) based on an exact test for population differentiation (Raymond and Rousset 1995).Interestingly, although the dataset included more samples from the Atlantic, the highest haplotype frequencies were observed in the Pacific, particularly in California (Hd=0.800)and Japan (Hd=0.806).The same pattern was observed for nucleotide diversities, which were nearly twice that in the Atlantic.These results are consistent with the Pacific being the source for the dispersal of S. plicata.Moreover, the occurrence of South American samples in both groups of haplotypes (Figure 3) might indicate two separate invasions of S. plicata into the Atlantic coast of South America.

Discussion
The results clearly indicate considerable geographical variation in the distribution of haplotypes of Styela plicata.Two scenarios could be invoked to explain this pattern: either the species is very ancient, with a widespread distribution followed by several local extinction events, or the species has experienced significant transport through human vessels, such that the observed pattern would be the result from multiple local introductions.Two lines of evidence support the second scenario.The species was recorded for the first time in the Pacific, Atlantic and Mediterranean only in the 19 th century.Moreover, an approximate date of introduction can be inferred in several locations, such as California (Ritter and Forsyth 1917), Bermuda (Van Name 1945), regions to the west of the Sydney Harbor (Kott 1972a), and New Zealand (Brewin 1948), based on the fact that previous surveys had failed to indicate the presence of the species.Styela plicata is a large species with straightforward identification characters, suggesting that its presence would likely not remain unnoticed.In addition, most records have been made in regions near harbors, particularly on artificial substrates (Annex 1), which also supports its status as an invasive species.
Styela plicata has several biological characteristics that might have contributed to its dispersion capacity.Under experimental conditions (26ºC and salinity 34‰), the larva takes on average 4.5 h to attach to a substrate after the hatching from its egg.However, they can continue to swim for up to two days without compromising their metamorphosis or post-larval development (Thiyagarajan and Qian 2003).
Even damaged embryos can still develop into larvae and potentially generate functional adults (Nakauchi and Takeshita 1983).Within only four days after settlement, the functional ascidian is completely formed in temperatures around 24-25 o C. The animals are sexually mature at 40 mm, and live for five to nine months (Yamaguchi 1975).The species does not seem to be sexually active during the colder months, but can still attain two to three reproductive generations per year in Japan (Yamaguchi 1975).The rapid growth and the high reproductive rates allow the species to rapidly and intensively colonize new substrates, leading to large populations (Morris et al. 1980).The species can be highly resistant to anti-fouling paints (Raftos and Hutchinson 1997).Finally, the presence of secondary metabolites on the body wall of S. plicata causes it to be unpalatable to predators, particularly fish (Pisut and Pawlik 2002).
Although the dispersion of many ascidian species is a natural phenomenon on smaller scales, particularly through water currents or attached to floating objects (Carlton 1987), anthropogenic transportation has been the main cause of successful invasions of exotic ascidian species (Lambert 2005).The presence of ascidians in harbors and the movement of vessels contribute considerably to their transport between continents (Monniot et al. 1985), with adults attached to vessel hulls being the most likely vectors (Carlton and Geller 1993;Lambert 2001).More recently, the transport inside seachests has become increasingly important, given that the water friction during travel is minimal (Coutts and Dodgshun 2007).The historical records demonstrating the occurrence of S. plicata in several harbor regions and the presence of individuals of the species traveling attached to ship hulls strongly support a link between its dispersal and global navigation.International maritime transport is associated with nearly 80% of the world's commerce (Carlton 2001).In southern Brazil, more than 22,000 ships went through the Paranaguá and Antonina harbors between 1995 and 2006, with a nearly 70% increase in activity in this period (http://www.portosdoparana.com.br).The traffic in the Itajaí harbor more than doubled between 1999 and 2006, with nearly 6,400 moored ships.This harbor has weekly maritime connections with northern Europe, the Mediterranean, the Gulf of Mexico (USA and Mexico), the western Atlantic (USA and Canada), the eastern Pacific (EUA), the Caribbean (Jamaica, Puerto Rico, Bahamas, Dominican Republic), and Japan (http://www.portoitajai.com.br).These data reflect the growth tendency for maritime commerce and the increased risk of new introductions.
The genetic structure of an invasive species depends on several factors, including the effective size of the introduced population and the genetic diversity of the source population.In general, invasive populations have low genetic variability, particularly if the introduction takes place as a single event and involves few founders (Holland 2000).This seems to be the case in the invasive ascidian Botryllus schlosseri Pallas, 1766, which was introduced to the harbors of southern Europe (López-Legentil et al. 2006).The fairly high haplotypic diversity observed in the present study (average Hd=0.737±0.03,without haplotype 9) seems therefore to contradict that prediction.However, recent studies have demonstrated that it is possible to find high genetic variability in invasive species, both sessile (Turon et al. 2003) and vagile (Barbaresi et al. 2003).For instance, the crab Carcinus maenas Linnaeus, 1758 showed similar levels of haplotypic diversity in its native (Roman and Palumbi 2004) and introduced regions (Roman 2006).Therefore, the expected effects of genetic drift following the introduction can be mitigated if the founder individuals are sufficiently diverse genetically, or if the process involves several independent rounds of invasion.In fact, the historical record of the colonization of S. plicata in different regions (1823 and 1843, in the north Atlantic; 1883 in the south Atlantic and Mediterranean; 1894 in the North Pacific, and 1877 in the South Pacific) and the distribution of haplotypes observed in this study suggest the possibility of complex rounds of invasion through different routes.
The exceptionally high levels of nucleotide and haplotypic diversity in the Pacific samples (Table 2) suggests that this region represents the center of dispersion of S. plicata.In particular, the sample from Japan showed not only the highest genetic variability levels, but also includes the most ancient haplotype (haplotype 8, Figure 3).Indeed, it is a common pattern inferred from the neutral theory that the most common haplotype is probably the most ancient (Watterson and Guess 1977).A connection between Asia and California had been suggested before (Ruiz et al. 1997), and the very high genetic variability in California might be due to repeated invasion from the Northeastern Pacific.
A surprising result from the phylogenetic analysis of S. plicata haplotypes is the possibility that haplotype 9 might in fact represent a distinct species.Several lines of evidence corroborate this claim.First, it has consistently been identified as the most divergent haplotype (nearly 3% divergent from the remaining haplotypes).Second, it has been consistently placed outside the clade including all the remaining haplotypes with very strong support from all analyses.Finally, it has been found sympatrically with other haplotypes in several locations, indicating that such high level of divergence is not due to geographical differentiation.Further studies, particularly including nuclear markers, will be instrumental to corroborate this hypothesis and to elucidate the relationship of haplotype 9 to S. plicata.If the existence of this new cryptic species is indeed confirmed, all the previous records of S. plicata should be reviewed, but we suspect that most of them are correctly assigned to S. plicata, given that most of the haplotypes are evidently from this species.
Several studies have demonstrated the effectiveness of mtDNA in bioinvasion studies to elucidate dispersion routes and centers of origin (Geller et al. 1997;Turon et al. 2003;Mackie et al. 2006).However, analyses based on a single molecular marker should be regarded with caution.The present study indicates a complex pattern of dispersal of haplotypes of S. plicata.Molecular and historical evidence suggest that the broad geographical distribution of S. plicata has an anthropogenic origin, with northwestern Pacific being the possible center of origin of the species.Future studies, including more extensive sampling of the northeastern Pacific (Japan, Korea, China, and Russia) and additional loci will allow for a better description of the origin and dispersion of S. plicata from that region.

Table 2 .
Styela plicata.Haplotype frequencies, nucleotide diversity () and haplotype diversity (Hd) within populations, and number of polymorphic sites in the studied populations The haplotype 9 was not considered in obtaining to , Hd and polymorphic sites.
Figure 2. Sequence variation among the studied haplotypes of Styela plicata.Above numbers represent the position of the variable sequences, and asterisks indicate nonsynonymous substitutions.The number of individuals of each haplotype are shown in parenthesis.See text for details.