Marine wood-boring Bivalvia are well known in European waters for a long time and according to the latest findings, nine different shipworm species (Teredinidae) are currently known in this region (Borges et al., 2012). They are typically associated with natural and man-made wood such as mangrove roots in tropical regions, coastal protection and harbor facilities as well as floating structures such as wooden ships and driftwood. Since these xylotrophic organisms decompose wood much faster than marine fungi or bacteria (Borges, 2007), they have caused considerable damage in Europe and worldwide over the last centuries (Distel et al., 2011). One of the most widespread and most destructive wood-boring Teredinidae in European waters is the common marine shipworm Teredo navalis. Since the origin and natural habitat of T. navalis is still unclear this species is currently considered cryptogenic and invasive in European waters (Borges et al., 2014).

A closer look into the literature shows that reports of shipworms from the South of Europe are considerably more abundant and earlier noted than for the North (Redeke, 1912; Moll, 1914). The first descriptions from the Mediterranean Sea date back to the authors of the classical antiquity like Aristoteles, Ovid, and Pliny (Redeke, 1912; Moll, 1914). They mentioned different kinds of shipworms, but the species involved were unknown. The next observations of Teredinidae in the Mediterranean Sea dated considerably later from the sixteenth to seventeenth century (Moll, 1914; Streftaris et al., 2005) and thus coincide with reports from Central Europe (North Sea).

The first credible proofs of T. navalis for Central European waters are reported from several authors for the North Sea around 1700–1730, the years of the first documented mass outbreak in the coastal waters of the Netherlands (e.g., Vrolik et al., 1860; Van Benthem Jutting, 1943). All authors assigned the attacks of wooden coastal protection structures unanimously to T. navalis. For many non-indigenous marine species mass occurrences typically have been described some years after their introduction (Gollasch et al., 2009), often resulting in their discovery. Also the sample material for the species description of T. navalis by Linnaeus in the year 1758, was collected in the Netherlands during this time by Sellius (Redeke, 1912).

For the Baltic Sea the first observations of T. navalis were documented from the Bay of Kiel in 1835 (Meyer and Möbius, 1865). The last and ever-present mass occurrence for the Baltic Sea was recorded in 1993 (Sordyl et al., 1998). Since then a stable and self-reproducing population has been established at the southern Baltic Sea coast, as Sordyl et al. (1998) detected free-swimming larvae of T. navalis in the water column. The current distribution of this species in the North Sea and Baltic Sea can be considered as stable as reflected by several recent studies (e.g., Paalvast and van der Velde, 2011; Appelqvist et al., 2015a; own unpublished data).

Although, much is still unknown about the ecology and especially the physiology of T. navalis, temperature, salinity and water currents determine to a great extent the distribution of this species. Regarding the required abiotic conditions for reproduction and growth, the optimum salinity ranges between 7 and 35 (Blum, 1922; Nair and Saraswathy, 1971) while the optimum water temperature is between 15 and 25°C (Roch, 1932; Kristensen, 1969). Teredo navalis is a protandric hermaphrodite species reaching sexual maturity seven to 8 weeks after metamorphosis (Grave, 1942). The larvae are released into the water column ~5 days after fertilization with a high release capacity of up to five million larvae per female and year (Kaestner, 1982). This short maturation period in combination with a high larvae release leads to a high reproductive potential. In combination with a short generation time this could result in about three to four generations per year. For the cold temperate waters in Central Europe, however, only one generation per year is expected for this species (see Section Genetic differentiation of populations).

Little is known so far whether T. navalis was absent between the mentioned mass occurrences or survived in reduced populations. This leads to the question whether the reported periodic mass occurrences of T. navalis in the North Sea and the Baltic Sea during the nineteenth and twentieth century, are reflected in their current distribution and population structure. Therefore, in the present study it was evaluated, whether and to what degree T. navalis is genetically structured. The conceivable existence of a self-containing Baltic Sea population separated from a North Sea/Atlantic Ocean population was a second question to be addressed. Furthermore, a third aim was to assess the possibility of a demographic expansion of T. navalis in the past together with the attempt to determine the point of origin of this expansion in time.

The COI (cytochrome c oxidase subunit I) locus has been proven to be a useful marker in population genetics and was employed in many studies for species analyses of the Bivalvia (e.g., Sirna Terranova et al., 2006; Sá-Pinto et al., 2012; Vergara-Chen et al., 2013). While in a previous study specific COI primers were provided for the first time and different regions in Europe and North America were investigated with respect to species identification, Weigelt et al. (2016) showed that no sibling species of T. navalis exist in the North Atlantic and the adjacent European seas. The present study investigated T. navalis in Central Europe marine waters at the population genetic level. Therefore, the genetic diversity of 352 T. navalis specimens collected along a salinity gradient was determined by comparing different sampling sites from Brittany in the west to the southern Baltic Sea in the east.

In a previous study specific COI primers for T. navalis were introduced and COI sequence data were obtained for the first time of molecular species identification on samples collected in different regions in Europe and North America (Weigelt et al., 2016). In the present study the main goal was to evaluate the population structure of T. navalis in Central European waters.

The two calculated median-joining networks showed a “star-like” phylogeny and the frequency of the central haplotype (Tn1) was dominating the complete network in both cases. The shared haplotypes in DS 1 (27.4%) are arbitrarily distributed over all locations and without clear geographic distribution pattern, mainly because all sampling sites were represented in the central haplotype (Figure 2). The proportion of three shared haplotypes in DS 2 (9.6% of all haplotypes) for a small scale sampling area between the two sampling sites Kühlungsborn and Zingst (~100 km apart) over the time scale of 16 years appeared to be low. This might be explained by the loss of haplotypes through genetic drift or possibly rather the gain of new haplotypes in this area by gene flow. The proportion of unique haplotypes in both networks are 72.6% (DS 1) and 80.7% (DS 2), respectively. These comparatively high proportions of unique haplotypes of both networks in relation to all haplotypes can be interpreted as one indicator for an expanding population (Slatkin and Hudson, 1991; Fu, 1997). It is most likely that all of these haplotypes originated from the predominant ancestral haplotype Tn1 after the expected sudden population expansion. Such a pattern and interpretation has been reported before for other marine bivalves, such as Mya arenaria (Lasota et al., 2016).

Some unique haplotypes from the initial study (Weigelt et al., 2016) are no longer unique since in the present study they could also been recorded at different sampling sites. This could be an indication for a high genetic diversity which might be reflected by high abundances. Up to 700 animals per dm² have been observed in pine wood panels during the investigation period (own unpublished data), which is quite high compared to other regions like the Kattegat (Appelqvist et al., 2015b). Moreover, the network analyses of both datasets revealed no differentiated substructures or demes. These results corresponded with findings from a previous study (Weigelt et al., 2016) in which the determined shape of the network was almost identical.

Since the network analyses displayed no signs of a population structure, the genetic diversity of the different sampling sites was investigated. The haplotype diversity ranged from 0.692 to 0.913 which are common values for Bivalvia (e.g., Mao et al., 2011; Li et al., 2015). Thus, no decrease of the haplotype diversity toward the eastern distribution border of T. navalis along the salinity gradient in the Baltic Sea could be demonstrated. Even the easternmost sampling site Dranske (Rügen) showed the highest haplotype diversity and the second highest nucleotide diversity of all sampling sites. At this point it can be summarized, that indications for a founder effect, a bottleneck effect or a generally reduced genetic diversity at the edge of the distribution of T. navalis in the Baltic Sea could not be confirmed by the present study.

Also the calculated pairwise F_ST values showed no differentiation between the 13 sampling sites since the values ranged only from −0.0150 to 0.0172 for DS 1. These are common F_ST values for Bivalvia, since Lasota et al. (2004) found for Mya arenaria average F_ST values of 0.01 to 0.03 between populations which were hundreds of kilometers apart. Following Wright (1978), F_ST values from −1 to 0.05 indicate only a very slight genetic differentiation between populations. Negative F_ST values have to be considered as zero values which means that there is no genetic subdivision between the sampling sites especially as none of the values of the present study were significant. Therefore, it is not possible to identify a population structure for T. navalis by the calculated F_ST values of pairwise differences. Likewise, no differentiation in the population structure can be detected by the AMOVA analysis in DS 1 since almost 100% of the variation are within the sampling sites and not among them (Table 4).

All data presented in this study clearly indicate a lack of genetically differentiated populations in T. navalis between Brittany and the Baltic Sea, and hence might be interpreted as a regional European panmictic population, which, by definition, assumes that there exists no mating restrictions upon this population. The theoretical framework for a panmictic population of T. navalis is, however, not highly probable, because an unrestricted mating between the sessile adults from Brittany to the Baltic Sea along a coastline of ca. 3,900 km seems not to be possible. In addition, water exchange between the North Sea and Baltic Sea is restricted, which also hampers unrestricted mating. Otherwise no biogeographical break could be estimated between the highly saline North Sea and the brackish Baltic Sea and a panmictic population structure of T. navalis can be observed. Nonetheless, the determined panmixia for T. navalis stands in contrast to recent studies on other marine Bivalvia in the North Sea and Baltic Sea which always displayed a sharp differentiation in the genetic population structure between inside and outside the Baltic Sea (Johannesson and André, 2006; Johannesson et al., 2011). In comparison, the two median values of the haplotype as well as the nucleotide diversity of the Baltic Sea are slightly larger than those from the North Sea (Table 2). In this context, the results of the present study of T. navalis cannot confirm the findings of a species-by-species meta-analysis of Johannesson and André (2006). These authors showed that a majority of Baltic Sea populations lost genetic diversity compared to populations of the same species in the Atlantic Ocean and adjacent waters. The only exception found in their meta-analysis was Macoma balthica which was 30% more diverse inside the Baltic Sea than outside because of two genetically distinct lineages (Johannesson and André, 2006), an Atlantic lineage in the western and a Pacific lineage in the eastern part of the Baltic Sea (Nikula et al., 2008). However, the occurrence of different genetic lineages can be excluded for T. navalis (Weigelt et al., 2016).

Since all sampling sites were not differentiated from each other, the sites seem to be connected by a high gene flow. As restricted gene flow has been reported to cause genetic population structuring (Sá-Pinto et al., 2012), in the present study unrestricted gene flow seems to be still ongoing as no genetic population structure has been determined in T. navalis. Also for other marine invertebrates with a high dispersal potential and high abundances such as Holothuria polii, unrestricted gene flow is followed by a low population genetic structure over small spatial scales (Vergara-Chen et al., 2010). High dispersal potential and high gene flow of marine invertebrates favors genetic homogenization of populations over broad geographical distances (Maltagliati et al., 2010). Genetic homogeneity and a limited genetic structure has also been observed in other bivalve species, for instance in the invasive soft-shell clam Mya arenaria (Lasota et al., 2004; Strasser and Barber, 2009).

The observation that all sampling sites hardly differ genetically is a strong indication that all animals originate from only one source population, which is still unknown. Since the first reports on the occurrence of T. navalis in Central Europe date back to the years 1700–1730, the first settlement of this species has presumably be taken place at the coast of the Netherlands. This nation was one of the world's leading global trading nations at that time. Since all other reports on T. navalis from this area were mentioned later, a successive range expansion from this starting point in the North Sea to the adjacent waters is the most likely explanation. At this point, the life history strategy of T. navalis is helpful as it is very effective during the successful colonization of new patchy habitats (Borges et al., 2014) and the associated further spread into new areas. Due to the pelagic larval stadium a range expansion into the Baltic Sea via the Skagerrak, the Kattegat and through the Danish Straits seems very likely. Natural range expansions from the North Sea into the Baltic Sea are well known from other marine organisms, e.g. for the dinoflagellate Prorocentrum minimum (Hajdu et al., 2000) or the polychaete Marenzelleria viridis (Essink and Kleef, 1993; Blank et al., 2008).

The importance of oceanographic connectivity for the population structure of marine invertebrates with pelagic larval stages was documented by several authors (e.g., Seebens et al., 2013; Wrange et al., 2016). Since populations of marine bivalves are often characterized by high gene flow among different sampling sites, the absence of a population structure over hundreds of kilometers is not uncommon (Vierna et al., 2012, and references therein). It seems that oceanographic connectivity explains regional genetic structure, rather than geographic distance (Wrange et al., 2016), especially in the context with the dispersal potential of bivalve larvae. As previously mentioned, the short generation time in combination with the high output of larvae and a pelagic lifestyle of up to 34 days in the water column (Nair and Saraswathy, 1971) could lead to a wide spreading via water currents. However, this would not explain a continuous exchange of larvae between the most distant regions investigated in this study (Brittany vs. Baltic Sea). Appelqvist et al. (2015a) simulated in a biophysical model a maximum spreading of T. navalis larvae by about 400 km for four different locations in the southern Baltic Sea. In addition, there are also temporal warm water inflows of sea surface water documented for the Baltic Sea, particularly in summer such as in July 2002 and 2003 (Feistel et al., 2004). Such irregular water masses could act as vector for an increased input of T. navalis larvae from the North Sea to the Baltic Sea.

Theoretically, also anthropogenic activities via ballast water transport might contribute to a steady distribution of T. navalis larvae between different biogeographic regions. Previous studies on marine invertebrates have proven that long distance dispersal can be enhanced by human-mediated vectors (Wrange et al., 2016) and hence have a great influence on the biogeographic distribution and finally on the population genetic structure of a species. Additionally, the location of all sampling sites in an area with a high transport ship traffic (e.g., the area of the Kadetrinne in the Baltic Sea is on average passed by 50,000 ships every year; WSA, 2014) and the previously mentioned presence of T. navalis larvae in ballast water (Gollasch, 2002) may play a very important role in the distribution of teredinid species (Borges et al., 2014). A spread of larvae via ballast water also against the main prevailing currents (Seebens et al., 2013) might be responsible to transport larvae across biogeographical boundaries (Canales-Aguirre et al., 2015), which finally favors an increased gene flow. In the light of our data, however, a range expansion from the North Sea into the Baltic Sea by the prevailing water currents is the most likely explanation for the observed population genetic structure of T. navalis in both the North Sea and the Baltic Sea.

In the subsequent analyses of the demographic history, the entire dataset (DS 1) was treated as one panmictic population. Fragmentation of habitats in the past may cause population structuring e.g., in form of a genetic bottleneck followed by a sudden demographic expansion (Strasser and Barber, 2009; Lasota et al., 2016). As stated before all analysis of the historical demographic situation of T. navalis came to the same result indicating a sudden population expansion after a historical decrease event. The significant negative values of Tajima's D and Fu's F test as well as the significant R₂ statistics (Table 5) support this assumption (Maltagliati et al., 2010). Also the data of the mismatch distribution analysis support these result. With the determined τ-value (1.5) in combination with the different substitution rates of 1.0 and 2.0% per million years for the COI gene and the assumed generation time of one per year, the calculated starting point for the sudden expansion was ~55,556–111,111 years ago. The geographical starting point of the expansion of T. navalis cannot be determined with the data of this study, but is unlikely to be located in the Central Europe marine waters due to the last glaciation of Europe. The last great glaciation in Europe ended ~10,000 years before present. At that time parts of the North Sea and the whole Baltic Sea were covered by glaciers (e.g., Grunewald and Scheithauer, 2010). Thus, they could not act as a refugium for a surviving population of T. navalis since all marine life was extinguished at that time. As some authors even assume a glacial ice cover south to the Iberian Peninsula (Grunewald and Scheithauer, 2010), the hypothetical territorial starting point of this sudden expansion, if it has been in Europe, could only be in the Mediterranean Sea or south-west of the Iberian Peninsula. However, the place of origin cannot be verified with the present data set. Therefore, the status of T. navalis should still remain cryptogenic as defined by Carlton (1996).

In conclusion, this study shows a high level of genetic diversity and gene flow in the investigated area for the examined mtDNA COI gene. This is reflected by the presence of a regional panmictic population of T. navalis in the waters of Central Europe. The lack of any genetic population structure indicates for the expansion history only one source population. Nevertheless, in case of T. navalis it seems that the use of this marker is not suitable to answer all questions of the population structure and the expansion history. For a deeper understanding of the colonization history of European marine waters by T. navalis, the use of highly polymorphic markers such as e.g. microsatellites would be recommended for future investigation.

RW, HL, and RB collected the samples. RW and RB acquired the sequence data and all authors analyzed and interpreted the data. RW drafted the work and all authors critically revised the work and approved the final manuscript.