5. DISCUSSION
5.1. Genetic diversity in Hediste diversicolor
Hediste diversicolor samples analysed in this study consisted of two main genetically divergent lineages, one including populations from Atlantic Ocean, the other including populations from Western Mediterranean, with no shared haplotypes between the two regions. This pattern is consistent with the hypothesis proposed by Patarnello et al. (2007) for other marine invertebrates. These authors stated that the Atlantic-Mediterranean genetic divergence is a consequence of vicariant events favoured by restricted gene flow between the two basins which promotes allopatric divergence. Similar patterns are reported for the blue mussel Mytilus galloprovincialis (Ladoukakis et al. 2002) and for the chaetognath Sagitta setosa (Peijnenburg et al.
2004, 2006). Virgilio et al. (2009) suggested that the constriction of the Gibraltar Strait (or Almeria-Oran front) is not a major biogeographical barrier for H. diversicolor.
In their molecular study, these authors found the presence of phylogenetically related haplotypes in populations from the Atlantic Ocean (North European Coasts, Portugal, and Morocco) and from the Mediterranean coast of France (Marseille). This evidence was interpreted as a proof of the occurrence of gene flow between E-Atlantic and W- Mediterranean (Virgilio et al., 2009). Moreover, the presence of a major phylogeographic break located between the coast of Tuscany/Sardinia and the Mediterranean coasts of France was proposed (Virgilio et al., 2009). In the present study one population from Southern Corsica (Figari bay, France), and one from Eastern Corsica (Migliacciaru, France) were analysed. These populations showed clear affinity with populations from Sardinia and Tuscany; hence, the phylogeographic break proposed by Virgilio et al. (2009) should be put north-west of the two sites in Corsica sampled in this study. The study of populations from the Mediterranean coasts of Spain and France, the Balearic Islands (if any), and from North-Western Corsica (if any) could be crucial to clarify this fundamental issue and eventually to reject the hypothesis, discarded by Virgilio et al. (2009), of anthropogenic introduction for Marseille sample.
Different analyses revealed that the Atlantic and Mediterranean lineages are further structured in haplogroups. Some of the detected haplogroups clustered together populations from the same geographical regions. Two haplogroups were
detected for the Atlantic Ocean: France (Roscoff), and Portugal (Furadouro); four haplogroups were detected for Western Mediterranean: Western Sardinia (Cabras and Calich ponds, Italy), Bonifacio strait (Cala Petralana and Figari bay, respectively Italy and France), Eastern Corse (Migliacciaru, France), and Tuscany (Fiume Morto and Coltano, Italy). Marked population structuring is a common feature of brackish- water species, mainly due to their limited dispersal capabilities, partial enclosure of brackish-water environments, and the occurrence of local selection (Cognetti &
Maltagliati, 2000; Bilton, 2002). H. diversicolor lacks planktonic larval stage and larvae exhibit brooding behaviour that can account for the high level of genetic structuring at different spatial scales, from a few meters (Virgilio & Abbiati, 2006), to hundreds of kilometers (Abbiati & Maltagliati, 1996; Röhner et al., 1997; Breton et al., 2003; Audzijonyte et al., 2008; Virgilio et al., 2009). The presence of shared haplotypes among closely sited populations could imply that homogenisation through gene flow is occurring on a scale of a few tens of kilometres within the same geographical region. A more plausible explanation for the presence of shared haplotypes could be the retention of common ancestral haplotypes. Co-ancestry and recent differentiation was proposed also by Abbiati & Maltagliati (1996) to justify the similarity of two Tuscan populations that where too distant to invoke contemporary gene flow. If not common, yet migration and consequent gene flow cannot be ruled out. Given that adults of H. diversicolor can survive in sea water (Scaps, 2002), we cannot exclude that migration may occur at least over short distances during exceptional floods or rainfalls by drifting or rafting. A possible case of exceptional gene flow could possibly justify the presence in the population of Cabras (Western Sardinia, Italy) of a haplotype that is more closely related to the Bonifacio Strait haplogroup (haplogroup II, Figari Bay and Cala Petralana, France and Italy, respectively).
Selection is supposed to play an important role in shaping the distribution of gentic diversity (e.g. see Belfiore & Anderson, 2001) and some authors proposed that the effect of selection could justify patterns of allozyme polymorphism in H.
diversicolor (Fong & Garthwaite, 1994; Abbiati & Maltagliati, 1996; Röhner et al., 1997; Virgilio et al., 2003; Virgilio & Abbiati 2004). Studies based on mtDNA were aimed toward the disclosure of different topics (Breton et al., 2003; Virgilio et al., 2006) or found no effects of selection (Audzijonyte et al., 2008; Virgilio et al., 2009).
Consistently with the latter studies, the populations analysed in the present study
appeared in a condition of selective neutrality, and the only instance of statistical significance for Tajima’s (1989) neutrality test (Cabras pond, Western Sardinia, Italy) is supported by a weak probability value (P = 0.05).
Mismatch distributions suggested that for all the analysed populations effective population sizes may have been reduced and that recent population expansions have accelerated the coalescence of lineages and increased the degree of phylogeographical subdivision. A similar mechanism was proposed for many organisms with Atlanto-Mediterranean distribution (Patarnello et al., 2007) and for H.
diversicolor by Virgilio & Abbiati (2006) and Virgilio et al. (2009). Such results could be related to recurrent mortality events followed by recruitment peaks (Gillet &
Torresani, 2003). Nonetheless, some authors stressed out that mismatch distribution can be biased by the presence of genetic substructuring (Marjoram & Donnelly, 1994; Ptak & Przeworski 2002), or by mutation rate heterogeneity (Aris-Brosou &
Excoffier, 1996). Furthermore, mismatch distribution can be outperformed by other tests, R2 statistics in particular, especially when dealing with relatively small sample sizes (Ramos-Onsins & Rozas, 2002). Both Fu’s (1997) Fs and Ramos-Onsins &
Rozas’ (2002) R2 produced non-significant and sometimes inconsistent outcomes across the populations, with the exception of Cala Petralana (North-Eastern Sardinia, Italy), which resulted statistically significant for all tests. These results suggested that the hypothesis of demographic expansion and its relative effect on genetic structure and divergence among H. diversicolor populations should be taken with great care.
The phylogenetic reconstruction obtained from the alignment of a 270 bp fragment of the COI sequence from our data-set with data from Audzijonyte et al.
(2008), and Virgilio et al. (2009) (Appendix III) revealed a clear monophily for the Mediterranean Sea. Within this clade, haplotypes from the Adriatic Sea and from Western Mediterranean form two clearly divergent haplogroups, and in the latter haplogroup, several monophyletic subgroups relative to Tuscany, Bonifacio Strait and Eastern Corsica were recognisable. For the Atlantic Ocean samples, haplotypes from Furadouro (Portugal) formed a monophyletic group, with other haplotypes from Portugal and Morocco, while haplotypes from Roscoff (France) clustered sparsely with haplotypes from northern Europe. Jolly et al. (2005) proposed that the English Channel could represent a phylogeographic break for the polychaete Pectinaria koreni. In this species, populations living at the two sides of the English Channel, preset different haplotype composition, suggesting the presence of both historical
and contemporary obstacles to connectivity due to hydrological featurers. In this perspective, the divergence among haplotypes from the Morocco/Portugal group and the rest of Northern Europe could have arisen in allopatry during the Pleistocene, when the Channel was partially closed (Smith, 1989), and population from Roscoff (France, this study) and Severn estuary (UK, Virgilio et al., 2009) could be the result of re-colonization from the North Sea after the catastrophic opening of the Dover strait in the late Pleistocene (Smith, 1989). The analysis of samples from the area between Portugal and the English Channel would help to clarify this intriguing issue.
5.2. Genetic diversity in Mytilaster minimus
Populations of Mytilaster minimus living in brackish-water environments seemed to have suffered dramatic demographic decrease during the past 10 years.
During the sampling campaign of this study it was found that most of the populations analysed by Camilli (2000) and Camilli et al. (2001) were virtually extinct. Moreover, serious technical difficulties were encountered during the optimization of DNA extraction procedure, and PCR reaction were often difficult, yet the results of this study did present interesting findings.
MDS of the Tamura and Nei (1993) genetic distances, Median-joining network and Neighbour-joining tree, evidenced the presence of three distinct haplogroups, separated by 12 and 13 fixed mutations. Given the adopted sampling design, no clear pattern of geographical structure could be outlined, and two of the detected lineages were found at more than one sampling localities. Haplotype H_1 was present both at S’Ena Arrubia and Bocca d’Arno (Western Sardinia and Tuscany respectively, Italy). The haplogroup containing the major part of the haplotypes (H_2, H_3, H_4, H_5, H_6, H_7, H_8) is present at S’Ena Arrubia, Calich, Lesina and Venice (Western Sardinia, Apulia, Veneto, respectively, Italy). The presence of divergent lineages has been previously reported for other mytilid bivalves such as Brachidontes exustus (Lee & Ó Foighil, 2004) and Brachidontes variabilis (Sirna- Terranova et al., 2007). Both in Lee & Ó Foighil (2004) and Sirna-Terranova et al.
(2007) the different lineages were found in allopatry and the presence of a species complex was invoked to justify the observed divergence between haplogroups.
Whenever divergent haplogroups are found in sympatry, like in the present study, the interpretation is that of vicariance and secondary contact (Avise, 2000). Another
possible explanation could be the retention of ancestral polymorphism, as proposed by Wilding et al. (2000) for rough periwinkle (Littorina spp.) and Schneider-Broussard et al. (1998) for stone crabs (Menippe spp.). Alternative hypotheses may be provided by the presence of different biological phenomena such as gene introgression or doubly uniparental inheritance (DUI) (Zouros, 2000), which are common phenomena in bivalves (Passamonti & Ghiselli, 2009). In presence of DUI, sequence divergence between female and male mtDNA can reach levels as high as 30% (Passamonti &
Ghiselli, 2009). Given it was impossible to determine sex of sampled individuals, DNA was extracted from the foot tissue to avoid contamination from male mtDNA, yet the chance that male mtDNA is present in somatic tissues is possible (Zouros, 2000).
Nonetheless, high levels of divergence between different haplogroups were detected in the mytilid Brachidontes variabilis in absence of DUI (Shefer et al., 2004, Sirna- Terranova et al., 2007).
Interestingly, significant levels of genetic divergence was detected among population sampled in marine and brackish-water sites with different statistical approaches. This result confirmed the outcome of previous studies by Camilli (2000) and Camilli et al. (2001), where high levels of divergence between marine and brackish-water populations of M. minimus were revealed by mean of allozyme electrophoresis. However, the origin of genetic divergence among different habitats remained undisclosed (Camilli, 2000; Camilli et al., 2001; this study). Possible explanations included larval retention (Bilton et al., 2002) and differential selective pressures (Cognetti, 1994; Cognetti & Maltagliati, 2000). Moreover, a bias induced by sampling error cannot be excluded a priori.
5.3. Genetic diversity in Xenostrobus securis
Samples of X. securis analysed in this study revealed high levels of within- population genetic diversity. Such levels of polymorphism are unexpected for an invasive species (Nei et al., 1975), which is supposed to experience bottlenecks and founder effects, and are comparable with those of natural population of other mytilids like Brachidontes exustus (Lee & Ó Foighil, 2004) and Mytilus californianus (Ort &
Pogson, 2007). Moreover, as previously reported for M. minimus, the presence of three largely divergent haplogroups was revealed by mean of different analysis. High level of polymorphism detected for this species and levels of divergence between
haplogroups led to exclude the hypothesis of a single colonization event followed by a demographic expansion. It appears more probable that several colonization events took place at different times, or that colonization involved numerous genetically divergent individuals, at least for the populations of Venice and Scolmatore canal (Venice Lagoon, Veneto and Scolmatore canal, Tuscany respectively, Italy) where haplotypes from each of the three haplogroups were found. The virtual absence of haplogroup III in the Arno river (Tuscany, Italy) suggested that this biotope has been colonized by individuals bearing just type-I or II haplotypes, yet this remains an hypothesis at this stage and will be confirmed sequencing a higher number of individuals from this location. Moreover, the scarce number of sequenced individuals from the Morto river (Tuscany, Italy) and Olbia (Eastern Sardinia, Italy) did not permit any phylogeographical consideration on these localities. As previously proposed for M. minimus, levels of divergence detected among haplogroups could be the result of allopatric divergence and secondary contact (Avise, 2000), retention of ancestral polymorphism (Schneider-Broussard et al.,1998; Wilding et al., 2000), introgression or doubly uniparental inheritance (DUI) (Zouros, 2000; Passamonti & Ghiselli, 2009).
As expected, no genetic structure at all was found among the analysed populations. The absence of genetic divergence reflected the fact that the analysed populations had not enough time to diverge, since their first establishing. In fact, first records in the Mediterranean dated no more than 20 years ago (Lazzari & Rinaldi, 1994; Sabelli & Speranza, 1994; Russo, 2001). Moreover, the populations of Venice and Scolmatore could be connected by ship-mediated migrations. At this regard, an intense naval traffic is expected between the north Adriatic ports and the port of Leghorn (Tuscany). Most marine invertebrates have long-lived planctonic larvae that guarantee high dispersal capabilities. Larvae of X. securis are planctonic but they cannot survive in marine water (Wilson 1968), excluding the possibility of larval dispersal for the colonization of new environments or high levels of gene flow between different brackish-water basins through the sea. Colonization of different areas can involve adults rafting attached to vegetation fragments (Wilson 1969). This hypothesis can explain the spreading of this species between spatially proximate sites (e.g. Scolmatore canal – Arno river – Morto river), but this mechanism is not realistic for long range migrations. The presence of X. securis in the Adriatic Sea was commonly associated with the intense aquaculture activity present in this area, in the sense that there is the possibility that adults individuals have been accidentally
introduced as secondary transport with commercial species (Occhipinti-Ambrogi, 2000; Zenetos et al., 2004). From the Adriatic Sea X. securis could have undergone spreading to the coasts of northern Tuscany, mediated by maritime traffic to the port of Leghorn. As reported by Giusti et al. (2008), from here the species had the possibility to spread into the Scolmatore canal, which is directly connected to the port. In recent years, other alien bivalves as Theora (Endopleura) lubrica Gould, 1861 and Musculista senhousia Benson in Cantor, 1842 were reported in the port of Leghorn, which can be considered a 'hotspot' of alien species (Giusti et al., 2008).
The Scolmatore canal is part of a large brackish-water system connecting the sea with the Arno river. From there, X. securis had the possibility to reach the Morto river which is currently the species’ northernmost limit along the Pisan coast.
