Ancient globetrotters—connectivity and putative native ranges of two cosmopolitan biofouling amphipods

The geographic distributions of some coastal marine species have appeared as cosmopolitan ever since they were first scientifically documented. In particular, for many benthic species that are associated with anthropogenic substrata, there is much speculation as to whether or not their broad distributions can be explained by natural mechanisms of dispersal. Here, we focused on two congeneric coastal crustaceans with cosmopolitan distributions—the tube-dwelling amphipods Jassa marmorata and Jassa slatteryi. Both species are common elements of marine biofouling on nearly all kinds of artificial hard substrata in temperate to warm seas. We hypothesized that the two species’ modern occurrences across the oceans are the result of human shipping activities that started centuries ago. Mitochondrial DNA sequences of the CO1 fragment of specimens from distinct marine regions around the world were analysed, evaluating genetic structure and migration models and making inferences on putative native ranges of the two Jassa species. Populations of both species exhibited considerable genetic diversity with differing levels of geographic structure. For both species, at least two dominant haplotypes were shared among several geographic populations. Rapid demographic expansion and high migration rates between geographically distant regions support a scenario of ongoing dispersal all over the world. Our findings indicate that the likely former native range of J. marmorata is the Northwest Atlantic, whereas the likely former native range of J. slatteryi is the Northern Pacific region. As corroborated by the genetic connectivity between populations, shipping still appears to be the more successful vector of the two species’ dispersal when compared to natural mechanisms. Historical invasion events that likely started centuries ago, along with current ongoing dispersal, confirm these species’ identities as true “neocosmopolitans”.


INTRODUCTION
Human-mediated dispersal of organisms can exceed natural mechanisms, carrying species over large distances and often across natural barriers (Ashton et al., 2007;Cohen & Carlton, 1998;Galil, Froglia & Noël, 2002;Gollasch, 2002;Molnar et al., 2008). For example, for as long as vessels have navigated oceans across the globe, non-native marine species have been transported and introduced into new regions as part of the fouling communities on vessel hulls and, for the last 100± years, in ballast water (Carlton, 1985;Thiel et al., 2003).
Thanks to the rise of scientific documentation and description, human-mediated introductions of marine species across the globe within the last half-century have been increasingly documented. In clear contrast, the dispersal of marine species in earlier centuries is nearly undocumented and probably a great many introductions since the 1820's have yet to be recognized. Many coastal benthic species that occur in anthropogenic habitats, especially those associated with the hulls of wooden ships (i.e., fouling and boring), had probably been distributed across the oceans by human shipping activities centuries before their geographic distributions were scientifically documented for the first time. As a consequence, modern geographic distributions of some marine species show cosmopolitan occurrences (e.g., Conlan, 1990;Thiel & Gutow, 2005), classified as "neocosmopolitan" (sensu Darling & Carlton, 2018). There is much speculation as to the likely former native ranges of coastal benthic species, with many still not reliably assigned native or introduced status in parts of their global range (Geller, Darling & Carlton, 2010). In parallel, there is mounting evidence from modern taxonomy and molecular research studies which increasingly cast doubt on the existence of marine benthic "eucosmopolitan" species (sensu Darling & Carlton, 2018;Havermans, 2016).
Here, we focused on two congeneric coastal species that are widely distributed in the world's oceans: the tube-dwelling amphipod crustaceans Jassa marmorata Holmes, 1905 and Jassa slatteryi Conlan, 1990. Despite both species being common elements of marine fouling communities in harbors ranging from temperate to warm seas, their geographic distributions are not well characterized (Marchini & Cardeccia, 2017). Jassa species are effective colonizers of clean hard substrata through immigration of nearby adult animals, but primarily by recruitment of drifting juveniles through the water column (Beermann, 2014;Franz & Mohamed, 1989;Havermans et al., 2007). By these means, Jassa species can dominate fouling communities particularly on artificial hard substrates such as navigational buoys, harbour walls, offshore wind farms and ship wrecks, easily reaching densities of more than 1,000,000 individuals/m 2 and exceeding other taxa in terms of abundance and biomass (Coolen et al., 2018;Franz, 1989;Scinto et al., 2007;Zintzen et al., 2008).
The long-standing difficulties in the taxonomy of the genus Jassa led to misidentifications and thus, unreliable and questionable species records (Conlan, 1989(Conlan, , 1990. Furthermore, with no historical documentation of the species' native ranges prior to scientific documentation, it was next to impossible to retrace the origin and native ranges of these species. Although the historical invasion pathways of J. marmorata and J. slatteryi are difficult to reconstruct, given their ecology, we hypothesized that the current "cosmopolitan" distribution of these species is probably the result of "ancient" invasive events (i.e., prior to scientific documentation and description), mediated by transoceanic shipping activities that started centuries ago.
Molecular approaches have often been used to retrace species' native ranges, particularly in the case of ecologically harmful invasive invertebrate species such as woodwasps (e.g., Boissin et al., 2012) and ladybird beetles (e.g., Lombaert et al., 2011). By using mitochondrial sequence data or shorter fragments such as microsatellites or SNPs, the origins and pathways of invasion or colonization can be retrieved, even in cases of complex histories, independent introductions and subsequent admixtures (e.g., Estoup & Guillemaud, 2010). To this purpose, we analysed mitochondrial DNA sequences of the CO1 fragment of the two globally distributed Jassa species in order to evaluate their genetic structure and infer their putative former native ranges.

Specimen collection and taxonomic identification
Samples were collected opportunistically from artificial substrates (e.g., harbor walls, pontoons, offshore constructions) within distinct marine regions via different sampling methods. A total of 419 specimens were collected, 291 of which were identified morphologically as J. marmorata and 128 of which were identified as J. slatteryi, following the original descriptions provided by Conlan (1990) (Table 1).

DNA isolation, amplification and sequencing
If available, at least 10 specimens (including males, females and juveniles) from each sampling location were used for molecular analysis. Two to three pereopods (pereopods 5-7, depending on the animal's size) of each specimen were used for DNA isolation. DNA extractions were performed using the QIAmp DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocols. Polymerase chain reaction (PCR) amplifications at the cytochrome c oxidase 1 fragment were carried out using the universal primers LCO1490 (5′ GGTCAACAAATCATAAAGATATTGG 3′) and HCO2198 (5′ TAAACTTCAGGGTGACCAAAAAATCA 3′) (Folmer et al., 1994). The 25 µl reaction mix consisted of 0.2 mM dNTPs, 0.5 µM forward and reverse primers, 10 µM PCR buffer, 0.02U/µl Hotmaster Taq (5 Prime GmbH, Hamburg, Germany), 3 µl template DNA, and mol grade water to bring the mix to the final volume of 25 µl. PCR thermal cycling conditions for amplifying CO1 sequences consisted of an initial denaturation at 94 C for 2 min, followed by 36 cycles of 94 C for 20 s, primer annealing at 42 C for 20 s, extension at 65 C for 1 min, and a final extension at 65 C for 15 mins. The PCR product quality was assessed on a 2% agarose gel, sequencing was performed at a contracting sequencing facility (EUROFINS, Germany). Material contributed by J. Coolen was processed using the same protocol as in Luttikhuizen et al. (2019).

Sequence editing and alignments
The software CodonCode Aligner v.8.0 (CodonCode Corporation, Deham, MA, USA) was used to check electropherograms for ambiguous bases, remove primers, and to check for stop codons to prevent inclusion of pseudogenes in the analyses. Clustal W (Thompson, Higgins & Gibson, 1994) implemented in MEGA 7.0 (Kumar, Stecher & Tamura, 2016) was used to align the resulting sequences and the basic local alignment search tool (BLAST) provided by the National Center for biotechnology information website (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used as a sequence control. DNA barcode data from all genetically analyzed specimens were made publicly available in the project "JASSA" in the Barcode of Life Data System (BOLD; Ratnasingham & Hebert, 2007) (dx.doi.org/10.5883/DS-JASSA). In addition to the resulting 330 newly edited CO1 sequences (199+131) for the two species obtained for this study, 128 previously published sequences of J. marmorata and 16 sequences of J. slatteryi were retrieved from GenBank and BOLD. Thus, the complete dataset consisted of 458 DNA barcodes. Additionally, 44 haplotype sequences of J. herdmani (Walker, 1893) from a previous work (Luttikhuizen et al., 2019) were added as outgroup (GenBank accession numbers: MH052599-MH052642). Detailed information on the specimens and sequences are provided in Table S1. The full alignment of 657 base pairs (bp) was collapsed to unique haplotypes, subsequently treating gaps as missing data using ALTER (Glez-Peña et al., 2010). A shorter alignment of 589 bp which excluded sequences shorter than 500 bp was chosen for the population analyses that tend to be sensitive to missing data. Finally, an alignment for each species was produced manually, preserving rare haplotypes and deleting shorter sequences of common haplotypes.
Intra-and interspecific species distances were calculated in MEGA 7.0 based on the Kimura-two-parameter (K2P) model with pairwise deletion using the full length of 657 bp sequences; these were then used to estimate genetic divergence between taxa. The K2P model was preferred over more complex and fitting models in order to ensure comparability of the results with those of published literature and other species delimitation analyses (Del Pasqua et al., 2018).
In the program DNASP 5.10 (Librado & Rozas, 2009), metrics of genetic diversity within populations of at least three sequences per region and species were estimated by computing indices such as number of haplotypes (H), haplotype diversity (Hd), and nucleotide diversity (π).
To test for departure from neutrality, Fu (1997), Tajima (1989) and mismatch distribution were computed in ARLEQUIN 3.5 (Excoffier & Lischer, 2010) for three different spatial scales (population, region and species). The mismatch distribution was tested against the predicted outcome of the models under demographic and spatial expansion in order to give further support to the neutrality tests. Departure from neutrality as well as demographic and spatial expansion would support the hypothesis of neocosmopolitanism whereas stability would support an eucosmopolitan scenario. Fixation indices were used to analyze genetic differentiation between populations (Fst), among regions of populations (Fct), and among populations within regions (Fsc; Wright, 1949). The Fst values were visualized in non-metric multidimensional scaling (nMDS) plots for each species in PRIMER 6 (v. 1.0.3). The significance of uncorrected pairwise Fst values were tested by performing 10,100 permutations in ARLEQUIN with the null hypothesis of no differentiation.
In order to test for a possible differentiation among regions or populations, one-way and two-way analyses of molecular variance (AMOVA; Excoffier, Smouse & Quattro, 1992) were performed in ARLEQUIN with 10,100 permutations for each species. A high degree of differentiation would suggest low connectivity and low genetic flow between populations, disproving the hypothesis of a neocosmopolitan distribution.
As possible differences between populations can be related to the remoteness of some locations, the relationship between geographic and genetic distances was tested with a Mantel test implemented in Alleles in Space (AIS; Miller, 2005) with 10,000 permutations. Geographic distances were calculated between coordinates regardless of geographic barriers (such as continents, currents). This was set to provide good approximation and strength to the other tests and to exclude simple spatial correlation.
In addition to a putative poor differentiation, a high degree of migration between populations can provide further information on the present connectivity between populations, supporting a neocosmopolitan scenario. Migration rate among regions was assessed using MIGRATE-N 3.6.11 (Beerli, 2006) on the CIPRES Science Gateway (Miller, Pfeiffer & Schwartz, 2010). By using a coalescent approach and Bayesian Markov chain Monte Carlo (MCMC) method, MIGRATE-N quantified both population size (θ = 2N e m for haploid mtDNA) and migration rate (M = m/m). The product is the number of effective migrants per generation (2N e m). The probabilities were estimated using 10 replicates, wherein each replicate was run for 50,000,000 generations and sampled every 1,000 steps, thereby recording a total of 50,000 steps. Each replicate included four MCMC chains with relative temperatures of 1.0, 1.5, 3.0 and 100,000. Convergence was assessed to ensure that the effective sample size (ESS) was higher than 200. Migration rates were only considered when 95% of the confidence intervals (CIs) were different from zero.
A haplotype network was computed for each species using an alignment of 372 sequences and 636 bp for J. marmorata, and 85 sequences and 595 bp for J. slatteryi with the median-joining method (Bandelt, Forster & Röhl, 1999) in the software Popart (v. 1.7).

Phylogenetic analyses and ancestral range estimations
Phylogenetic relationships among the Jassa species were reconstructed using only unique haplotypes with maximum likelihood (ML) and Bayesian inference (BI) of the CO1 gene fragment. The ML was inferred with PhyML (Guindon et al., 2010) using 1,000 bootstraps for the branch support. The best substitution model (GTR + I + G) was tested with the SMS routine in PhyML using both AIC and BIC as optimality criteria (Lefort, Longueville & Gascuel, 2017).
A time-calibrated BI phylogeny was reconstructed in BEAST 2.5.2 (Bouckaert et al., 2014) on XSEDE (Towns et al., 2014). The GTR model of evolution, with proportion of invariant (I) and gamma shape parameters (G), determined with bModelTest (Bouckaert & Drummond, 2017), and the Yule speciation model were set for priors. An uncorrelated relaxed clock with log-normal distribution was applied following the specifications proposed by Copilaş-Ciocianu, Sidorov & Gontcharov (2019); that is, rates ranged from 0.7% to 1.7% Ma −1 with a starting value of 1.2%. Three runs each of 20,000,000 iterations of MCMC sampled each 1,000 iterations were performed. All runs were examined using Tracer v1.7.1 and all sampled parameters achieved sufficient sample sizes (ESS > 200). Tree files were combined using LogCombiner on XSEDE (1.8.4) with 15% of burn-in. The maximum clade credibility tree was generated using TreeAnnotator on XSEDE. All the XSEDE analyses were performed on the CIPRES Science Gateway (Miller, Pfeiffer & Schwartz, 2010).
The ancestral ranges of the haplotypes were estimated with the R package BioGeoBears (Matzke, 2013;R Development Core Team, 2018). This package is used to perform biogeographic inferences such as the estimation of ancestral ranges of species or Operational Taxonomic Units (OTUs) under different hypotheses (e.g., dispersion, founder-event speciation). Furthermore, BioGeoBears allows for the implementation of a third parameter (J) which permits a "jump speciation" in the daughter lineage, resulting in a possible different area from the direct ancestor. This feature accounts for the biology of Jassa species in this study by virtually including anthropogenic dispersal. The analyses were run using the DEC and DEC+J models (Matzke, 2014) and a maximum of five (J. marmorata) and four (J. slatteryi) areas of possible occurence as this was the maximum number of regions where a single haplotype was found for each species respectively. In order to avoid overinterpretation, only the nodes with at least 70% of bootstrap support and 90% of posterior probabilities were considered. The best-fitting model was selected based on Akaike's information criterion (AIC).

RESULTS
The distance between species was 15.2% ± 1.7 and the average distance within species accounted 0.17% ± 0.07 for J. marmorata and 0.57% ± 0.17 for J. slatteryi.
Within J. marmorata, the haplotype diversity was variable throughout marine regions, ranging from 0.884 in the NW Atlantic region, to 0.305 in the NE Pacific region. Nucleotide diversity was low among the regions, ranging from π = 0.001 in the NE Atlantic to π = 0.006 in the NW Atlantic region.
The haplotype diversity of J. slatteryi also exhibited a wide range. The highest regional value of Hd = 0.731 occurred in the NE Pacific (California) while the lowest was detected in the SE Pacific (Hd = 0.433). Nucleotide diversity ranged from π = 0.022 in the NW Pacific to π = 0.002 in the SE Pacific (Table 2).

Demographic analysis
Tajima's D values were negative and significant for each respective species in total; this indicates an excess of rare nucleotide site variants which suggests a departure from the Wright-Fisher neutral model (Wright, 1931; Table 2). The Fu's F S test, which is based on the distribution of haplotypes, showed a significant negative value for J. marmorata in total, indicating an excess of rare haplotypes. The mismatch distribution analysis was in concordance with D and Fs results, for the two species, which supports the spatial and demographic expansion scenarios for both species (Table S2; SSD and Raggedness index).
Jassa marmorata did not display departure from the null hypothesis of demographic expansion for any other region except for Peru and the Northern European Seas where no significant spatial expansion was detected (Table S2; SSD and Raggedness index). In contrast, J. slatteryi displayed a significant deviation from the null model of demographic expansion for the Mediterranean Sea as well as the North and South East Pacific, whereas there was no significance for a spatial expansion (Table S2; SSD and Raggedness index).

Geographical structure
The one-way AMOVA revealed significant differences among regions for both species. However, when exploring the variation among populations in the two-way test, the within-region variation was not significant (Table 3). While the main contribution to molecular variance was found within populations, the variation among populations within the same regions was also significant in both species.
In J. marmorata, specimens from the SW Atlantic showed the highest level of differentiation within NE Pacific and Northern European seas (Fst = 0.71 and 0.81 respectively). In the comparison of populations, two main groups were distinguished within J. marmorata: a general group with the majority of the populations, and another with Argentina and Chile. The Virginia population showed the highest average differentiation and appeared to be an outgroup (average Fst = 0.776; Fig. 1; Tables S3 and S4). Jassa slatteryi showed an overall low level of differentiation on the regional scale which was only significant for the comparison between Lusitania (Atlantic Spain) and the NE Pacific (California) (Tables S3 and S4). Moreover, there was no obvious pattern with Japan which displayed the highest average Fst (Fst = 0.576; Fig. 1; Table S4).
For both species, only few migration rates met the assumptions (95% CI > 0; Table S5) and were considered in the migration model among regions. The migration between populations displayed contrasting patterns. In J. marmorata, no emigration was observed from Chile, but from almost every other population. In contrast, for Connecticut and Virginia there was only one population donor although multiple receivers were detected ( Fig. 2; Tables S5 and S6).
For J. slatteryi, neither immigration nor emigration for Mediterranean samples (Spain) and for the NW Pacific populations met the assumptions of the migration model; conversely, more emigrants were found from South Korea to Japan (Fig. 2; Table S5).
The haplotype network of J. marmorata was star-like with two main haplotypes, one of which occurred in almost every region sampled except the NW Pacific and SW Atlantic (Fig. 3). The second abundant haplotype occurred mostly in the NE Atlantic with additional records from the NW and SW Atlantic. Moreover, only three haplotypes were separated by more than two mutations from either of the two central haplotypes (i.e., SE Pacific: 10, NW Atlantic: 8, NW Pacific: 4), while the highest distance was observed between SE Pacific and NW Atlantic haplotypes (19 steps). On the other hand, the network of J. slatteryi was much more dispersed in comparison and did not show any clear pattern of (geographic) segregation but included multiple missing haplotypes and three haplotypes which were present in more than two regions (Fig. 3). There was no significant correlation between genetic and geographic distances in either J. marmorata or J. slatteryi (p > 0.1).
The phylogeny was poorly resolved, showing high support (posterior probabilities = 1; bootstrap support > 95) for the cluster grouping J. mamrorata and J. slatteryi as well as each species (Fig. S1).
The  (Table S7). In J. marmorata, while the reconstruction of the most recent common ancestor's (MRCA of all the haplotypes) geographic range was inconclusive, it seemed to exclude the Mediterranean Sea and SW Atlantic but an ancestral divergence of a clade in the NW Atlantic was evident (Fig. S2).
Similar to J. marmorata, it was not possible to determine the ancestral geographic range of the MRCA of J. slatteryi even though in the Pacific it appeared to be comparatively more diverse. One clade was restricted to the Mediterranean Sea (Fig. S3).

DISCUSSION
Using the molecular mitochondrial marker CO1, we were able to reconstruct putative native regions (i.e., prior to their "ancient invasion") of two cosmopolitan amphipods that are common elements of marine biofouling. The examination of the genetic variability between these two species in our analyses indicated high levels of intraspecific divergence within the J. slatteryi populations, as well as comparable variation within populations and among populations within regions for J. marmorata. A high haplotype number was detected within J. slatteryi when compared to J. marmorata, and at least two main haplotypes within each species were shared among several geographic populations.

Genetic population structure and connectivity of J. marmorata
The results indicated that the overall percentage of variation present in J. marmorata populations was derived almost equally from both variation within populations and variation among populations within marine regions. The highest genetic diversity (accounting for both haplotype and nucleotide diversity values) was found in the NW Atlantic region, and the lowest occurred in the NE Pacific region.
Previous research involving J. marmorata samples taken from assumed potential source populations along the Atlantic coast of the United Stated showed varying results: one population from coastal Connecticut revealed no genetic diversity at all in the population, whereas the samples analyzed from coastal South Carolina were shown to have been derived from a separate Jassa species-level lineage (Pilgrim & Darling, 2010;J. Beermann, A. Desiderato, 2019. Pacific North American coast populations of J. marmorata are classified as invasive and are assumed to be transfers from its native range of the Atlantic coast of North America (Chapman, 2007;Conlan, 1990).
In contrast to Pilgrim & Darling (2010), the results of the current study showed high genetic diversity in the presumed native range of J. marmorata as well as in populations from the SW Atlantic region. Nevertheless, the presence of a shared haplotype between the SW Atlantic and the SE Pacific region could signal a connection between the two. This is in accordance with the recent first record of J. marmorata in Argentina (De Pina, 2005).
Given that introduction events transfer only a subset of native genetic diversity, invasive populations are frequently seen to exhibit lower diversity than the native source populations (Dlugosch & Parker, 2008). The gene pool of introduced populations is expected to be limited due to the stochastic process of introduction mechanisms (Holland, 2000). J. marmorata's low genetic diversity in the NE Pacific is thus consistent with this expected founder effect (Pilgrim & Darling, 2010). Overall haplotype and genetic diversity were lower in J. marmorata populations when compared to J. slatteryi populations despite a larger sample size. The overall genetic diversity of J. marmorata populations was characterized by high haplotype diversity and low nucleotide diversity, indicating only small differences between haplotypes. This is evident from the minimum spanning haplotype network which showed mostly single nucleotide differences between haplotypes in J. marmorata.
The obtained data for J. marmorata indicated differing levels of geographic structure within the sampled populations; there appeared to be substantial geographic structure within some populations (e.g., Virginia, Chile, Argentina, New York, the North Sea and Oregon) whereas other populations were less structured (Spanish Mediterranean coast and Connecticut). These high levels of structure suggest a more limited gene flow or different selective pressures. Some J. marmorata populations were not panmictic throughout a marine region, but displayed high Fst values, indicating an absence of genetic connectivity between them. Populations from Chile-Peru as well as Connecticut-Virginia, on the other hand, could each be assigned to the same marine regions, and were characterized by high heterozygosity values. J. marmorata populations from California and Oregon exhibited genetic connectivity with populations in Norway and Iceland, which is unlikely to result from natural dispersal. Furthermore, negative Tajima's D, Fu's Fs statistics, as well as the results of the mismatch distributions for J. marmorata indicated a population size expansion, perhaps after a bottleneck. It can be assumed that the amount of samples was rather limited considering that the genetic diversity of these populations has probably been shaped largely by stochastic events. A lack of samples from entire marine regions remains an open issue that impedes clear inferences on the connectivity of globally distributed populations.
Genetic population structure and connectivity of J. slatteryi The magnitude of genetic variation within J. slatteryi was highest within populations and lowest among marine regions. In J. slatteryi, there was discordance between haplotype and nucleotide diversity. Haplotype diversity was higher in the Mediterranean Sea and the NE Pacific whereas nucleotide diversity was higher in the NW Pacific. The lowest genetic diversity was found in the SE Pacific region.
Jassa slatteryi had been previously considered to be either cryptogenic or introduced to the NE Pacific (Boyd, Mulligan & Shaughnessy, 2002). Populations of J. slatteryi from the Pacific North American coast had been found to show similar genetic diversity indices, haplotype diversity measures, and intraspecific genetic distances as the presumed native species from that region, J. staudei (Chapman, 2007;Pilgrim & Darling, 2010). Both species exhibited diversity measures much higher than J. marmorata in the same area. The authors suggested that perhaps the Pacific coast of North America could represent the native range of J. slatteryi even though their study was geographically restricted to the continental U.S. coasts.
Our results corroborate the previous assumption that J. slatteryi originated from Northern Pacific regions (NE and NW). Surprisingly, there was also high genetic diversity in the populations from the Mediterranean Sea region. The highly divergent haplotypes found in the Mediterranean Sea and the private haplotypes from the SW Atlantic might be artifacts due to small sample sizes and a resulting undersampling in the putative native regions due to the nature of our opportunistic sampling campaign. The overall genetic diversity of J. slatteryi populations was different from that of the J. marmorata populations as the overall nucleotide diversity was higher, indicating larger differences between haplotypes. This was displayed by the haplotype network where the respective haplotypes were separated by more mutations than in J. marmorata.
Populations of J. slatteryi also varied in their level of geographic structure, ranging from highly structured populations (Japan) to lower levels of structure (California, Spanish Atlantic coast, Chile and South Korea) and higher gene flow between populations. In contrast to J. marmorata populations which showed only limited or no gene flow between populations of the same marine region, J. slatteryi populations did not reflect this structure between populations within the same regions. Furthermore, our findings indicate gene flow between geographically remote populations over large distances which is highly unlikely via natural dispersal (e.g., Mediterranean and Spanish Atlantic coasts with South Korea and/or Chile). The mismatch distribution for J. slatteryi suggested a geographic expansion but not a demographic expansion.
We identified potential connectivity and dispersal differences between J. marmorata and J. slatteryi populations in regions where both species co-occurred. J. slatteryi populations in Chile, California, South Korea, the Spanish Atlantic and the Mediterranean coasts were particularly closely related to each other (evident by very low Fst values), whereas the J. marmorata population from Chile showed more structure with very little gene flow between this and other populations. The Fst values also indicated that, for example, while there is measurable gene flow between J. marmorata populations from the neighboring regions of Chile and Argentina, populations of J. slatteryi from the same regions are not panmictic. However, both Jassa species showed evidence of gene flow between the Mediterranean coast of Spain and California. Both J. marmorata and J. slatteryi exhibited high haplotype diversity and low nucleotide diversity, indicating rapid demographic expansion from a small effective population size (i.e., contrasting the mismatch results in J. slatteryi; Avise, 2000).
The scenario of ongoing dispersal and continuous input of new propagules in different marine regions all over the world is also supported by the inconclusive results of the ancestral range reconstruction of both species, confirming these species' true identities as "neocosmopolitans".

Origin and possible pathways of introduction
Maritime shipping over the past centuries has facilitated dispersal of attached fouling species on the hulls of vessels, and for the last 100± years, transported larvae and other propagules in ballast water around the world. Consequently, this facilitates gene flow between populations, particularly those with non-planktonic larvae (Carlton & Hodder, 1995;Carlton, 1985Carlton, , 1987Williams et al., 1988). These unintentional introductions have increased rapidly in recent decades as transit times have diminished, ship-carried trade has become globalized, and ballast-water volumes have risen (Carlton & Geller, 1993;Ruiz et al., 1997;Williams & Grosholz, 2008). Despite the increase in introduction of non-native species by means of maritime traffic, shipping pathways for species dispersal remains poorly understood on a global scale (Mack et al., 2000). Seebens et al. (2013) identified major shipping routes that are high-risk pathways for bioinvasions and also determined that the highest invasion risks were concentrated in a small number of ports. These invasion pathways help explain the geographic distribution of the Jassa populations in our study.
The ability of Jassa to build tubes on anthropogenic substrata and the absence of planktonic larval stages are excellent prerequisites for human-mediated transport by ships. Accordingly, specimens of both J. marmorata and J. slatteryi have been reported from hull fouling of ships and boats in different parts of the world (Castro et al., 2020;Martínez-Laiz et al., 2019;Peters, Sink & Robinson, 2019). J. marmorata has long been recognized in the literature from worldwide locations (Marchini & Cardeccia, 2017). Although the native origin of J. marmorata was uncertain, the worldwide distribution of currently known Jassa species indicated that it could be native to the NW Atlantic region where J. marmorata occurs as the only species of the genus (Conlan, 1990;Marchini & Cardeccia, 2017). This hypothesis is supported by our findings. Although already Conlan (1990) noted a wide global distribution of J. slatteryi, records of this species after Conlan's original description have increasingly been reported in the past recent years-for example, from Europe (Bonifazi, Mancini & Ventura, 2018;Gouillieux, 2017) or South America (Rumbold et al., 2015). This is probably due to an increasing awareness of this group's taxonomy, which is corroborated by the standardized utilization of DNA barcoding for species identification. Hence, J. slatteryi had been classified as cryptogenic species with uncertain native origin (Marchini & Cardeccia, 2017) but our findings indicate that the species' likely former native range is the Northern Pacific.
Jassa species were most probably transported by ship traffic between NE Atlantic ports and ports in South America (e.g., Argentina, Chile, Peru). The results of this study indicate that these shipping routes played a major role for the establishment of both J. marmorata and J. slatteryi populations in the SE Pacific and SW Atlantic regions. The models of Seebens et al. (2013) also suggest that although the northern European Seas (NE Atlantic in our study) are recently most strongly connected to tropical and subtropical ecosystems in terms of marine traffic, with the adjacent NW Atlantic waters providing similar enough climatic conditions. Therefore, the North American Atlantic coast still dominates as the major source region of invasions into the NE Atlantic for coastal benthic species. The northern Pacific waters are also source regions for invasions to the NE Atlantic. This could explain the presence of both J. marmorata and J. slatteryi in the NE Atlantic region. Our data indicate both Jassa species occur along the West and East coast of the United States, which is likely due to the connection of both coasts via shipping routes through the Panama Canal. This high invasion risk canal is also passage for shipping routes to and from the NE Atlantic regions (Seebens et al., 2013). The Jassa populations from Argentina could have been transported there by way of the Spanish Atlantic and Mediterranean coast, as well as North Africa, and there is a high invasion probability between the NE Pacific region and the SE Pacific region (e.g., Peru and Chile).

CONCLUSIONS
As a result of speculative knowledge of "ancient" invasion events and historically unsettled taxonomy, we cannot determine with certainty through which series of events these Jassa species achieved their neocosmopolitan distribution. It is most likely that the species' distributions are a result of human-mediated introductions which started centuries ago.
Our findings indicate that the likely former native range of J. marmorata is the NW Atlantic region, and the likely former native range of J. slatteryi is the Northern Pacific (NW and NE) region. Due to the relatively fast travel speeds, extensive surface area, and independence of stochastic ocean currents (Haydar, 2012), shipping still appears to be the more successful vector of Jassa species dispersal on a global scale as compared to natural dispersal on a local level (e.g., rafting or drifting; Thiel & Haye, 2006). As with these two successful biofoulers, the occurrence of many other coastal benthic species with wide modern geographic distributions is probably the result of human shipping activities over the last centuries.