Canals and invasions: a review of the distribution of Marenzelleria (Annelida: Spionidae) in Eurasia, with a key to Marenzelleria species and insights on their relationships

Recent invasions of the North and Baltic Seas by three Marenzelleria species have extensively altered benthic communities in the region. Despite several studies on the morphology and biology of the worms, their morphological identifications are often challenging. Here we summarize and map the available records of Marenzelleria from Eurasia, distinguishing those based on morphology versus molecular data. Based upon the genetic similarity ( p = 0.08% for COI ) between individuals from the Baltic Sea and individuals from the Barents and White Seas we propose, for the first time, a possible route for the invasion of the Baltic Sea by M. arctia from the White Sea through the White Sea–Baltic Sea Canal. At the same time, our analysis of the sequences of COI fragments showed a significant genetic distance ( p = 4.28– 4.29%) between individuals identified as M. arctia from the Baltic, Barents and White Seas and those from the Kara Sea. This genetic distance, as well as the isolated estuarine habitats of these Arctic worms, and the large geographic distance between the type locality of M. arctia in the Beaufort Sea (Alaska) and northern Europe, raise doubts about the conspecificity of North American, North European and Northwest Pacific populations. We report M. neglecta for the first time for the British Isles (River Thames). We also review the evidence for the role of the Baltic Sea−Volga Canal and the Volga−Don Canal in facilitating the dispersal of M. neglecta to the Caspian Sea and the Sea of Azov, respectively. We further provide new insight on the phylogeny of Marenzelleria , an updated diagnosis of the genus and a key for morphological identification of Marenzelleria adults greater than 1.2 mm wide.


Introduction
1896 is a small group of spionid polychaetes that likely evolved in estuaries bordering the Arctic Ocean (Sikorski and Bick 2004;Bick 2005;Blank and Bastrop 2009). Subsequent dispersal, population isolation, and regional adaptation in isolated habitats on the Atlantic coast of North America apparently led to additional speciation in the Northwest Atlantic Ocean. Two basal species, M. wireni Augener, 1913 and M. arctia (Chamberlin, 1920), have been considered indigenous in the Arctic region, and three derived species are regarded indigenous in the Northwest Atlantic: M. viridis (Verrill, 1873), M. bastropi Bick, 2005 and M. neglecta Bick, 2004 (Blank andBastrop 2009;Radashevsky et al. 2021). The age of Marenzelleria has not yet been determined, but the high morphological similarity of larvae and adults of different species indicates a relatively recent divergence in this group.
In North European waters (North and Baltic Seas), alien Marenzelleria first appeared in the late 1970s-early 1980s (Atkins et al. 1987;Elliott and Kingston 1987;Kleef 1988, 1993;Bick and Burckhardt 1989). The rapid growth of the populations of these worms in the 1990s-2000s and their impacts on local benthic communities stimulated abundant studies on their distribution, morphology, reproductive biology, physiology, ecology, bioturbation and genetics (reviews by Bastrop et al. 1997;Blank et al. 2008;Blank and Bastrop 2009). Genetic analyses using either allozyme electrophoresis, PCR/sequencing (fragments of the mitochondrial 16S, COI and Cytb genes) or combined PCR/RFLP analysis distinguished three species: M. viridis, M. neglecta and M. arctia (Bastrop et al. 1995(Bastrop et al. , 1997Röhner et al. 1996a, b;Bastrop and Blank 2006;Blank et al. 2008;Blank and Bastrop 2009). Marenzelleria viridis and M. neglecta are considered to have been introduced into European waters from the Atlantic coast of the United States. Marenzelleria arctia is only known in Northern Europe from the Baltic Sea, where its precise origin remains unclear.
The phylogenetic relationships of Marenzelleria species have been explored by the analyses of mitochondrial DNA by Blank and Bastrop (2009), Syomin et al. (2017), and Radashevsky et al. (2021). Worldwide records of M. viridis and records of Marenzelleria spp. from North America were recently summarized and mapped by Radashevsky et al. (2021). The purpose of the present study was to further explore the phylogeny of Marenzelleria based on an analysis of a larger set of data, including nuclear genes. We also review earlier reports, provide new records of Marenzelleria spp. (except M. viridis) from Eurasia, and, using molecular data, verify conspecificity of disjunct populations of M. arctia. In so doing, we update and refine our knowledge of the distribution of Marenzelleria species. Our additional purpose was also to hypothesize a possible route along which M. arctia could have been transported from the Arctic into the Baltic Sea, and to review the role of canals in facilitating the dispersal of Marenzelleria worms in Eurasia.

Material
Collections were made in the intertidal in the Kandalaksha Gulf (White Sea, Russia). Sediments collected for this study were washed in the field on a 500-µm mesh sieve, and Marenzelleria worms retained in the residue were removed and examined alive under light microscopes in the laboratory. For molecular analysis, worm fragments were preserved in 95% ethanol. After morphological examination, worms were fixed in 10% formalin solution, rinsed in fresh water, transferred to 70% ethanol, and then deposited in the polychaete collections of the Museum of the A.V. Zhirmunsky National Scientific Center of Marine Biology (MIMB), Vladivostok, Russia, and the White Sea Branch of the Zoological Museum of the Lomonosov Moscow State University (ZMMU_WS), the White Sea Biological Station, Poyakonda, Russia.
We also examined museum samples of Marenzelleria spp. collected by recent expeditions of the Russian Academy of Sciences to the Arctic and Northwest Pacific. Ethanol-fixed specimens of M. arctia from the Kara Sea (Russia) were provided by Alexandra N. Stupnikova, and M. neglecta from Taganrog Bay (Sea of Azov, Russia) were provided by Vitaly Syomin. To map the distribution of Marenzelleria species, we considered reliable records made by earlier authors based on morphological characters, and records by Bastrop and Blank (2006), Blank et al. (2008), Bastrop (2009), andSyomin et al. (2017) based on genetic data. Complete information on newly collected material, museum samples examined during this study and by other authors, and records by other authors for which no museum deposits were noted, is provided in Supplementary material Tables S1-S3. Records by other authors are annotated in Tables in the Comments. A list of the museums and other collections (and their acronyms) holding the examined or reported specimens of Marenzelleria spp. is in Table S4.
When no coordinates were provided for sampling sites from other studies, they were estimated using Google Earth Pro according to the original descriptions of the locations. Sampling locations of Marenzelleria spp. are plotted on maps using QGIS 3.20.0 software and the geodata provided by the OpenStreetMap Project (https://osmdata.openstreetmap.de). Final maps and the plates were prepared using CorelDRAW®2019 software.

DNA extraction, amplification and sequencing
We used the ReliaPrep gDNA Tissue Miniprep System (Promega Corporation, Madison, WI, USA) for DNA extraction and purification with standard protocol for animal tissue. Polymerase chain reaction (PCR) amplification of nuclear 18S rDNA, D1 region of 28S rDNA and Histone 3, mitochondrial 16S rDNA and cytochrome C oxidase subunit 1 (COI) gene fragments was accomplished with the primers and conditions described by Radashevsky et al. (2014Radashevsky et al. ( , 2016Radashevsky et al. ( , 2020. Purified PCR products were sequenced in both directions on an ABI Prism 3500 Genetic Analyzer (Applied Biosystems) using the BrilliantDye Terminator v3.1 Cycle Sequencing Kit (NimaGen) and the same primers as for PCR. Sequence editing and contig assembly were performed using SeqScape 2.5 (Applied Biosystems). GenBank accession numbers and brief information about sequences used in the present analysis are shown in Table S5. To link sequences with complete corresponding data, unique numbers from the first author's database (VIR) are given to samples in Tables S1-S3 and S5.

Data analysis
We aligned DNA sequences using the ClustalW method implemented in the MEGA 5.1 software (Tamura et al. 2011). Ambiguous positions and gaps were excluded from subsequent analysis using GBlocks (Castresana 2000) with settings for a less stringent selection. Pairwise distances (p, see Nei and Kumar 2000) both within and between groups were calculated in MEGA 5.1 (Tamura et al. 2011). We concatenated DNA data partitions using SequenceMatrix (Vaidya et al. 2011) and specified substitution models for each partition individually. The best-fitting nucleotide substitution models for Bayesian analysis (TVM+G for COI, GTR+G for 16S, SYM+I for 18S, TVM for 28S, and HKY+I+G for Histone 3) were selected in MrModeltest version 3.7 (Posada and Crandall 1998) using Akaike Information Criterion (AIC).
We used MrBayes 3.2.7 (Huelsenbeck and Ronquist 2001;Ronquist and Huelsenbeck 2003) via the CIPRES web portal (Miller et al. 2010) for the Bayesian analyses of 10,000,000 generations, four parallel chains and sample frequencies set to 500, in two separate runs. Based on the convergence of likelihood scores, 25% of sampled trees were discarded as burn-in.
We performed two Bayesian analyses of sequences of M. arctia from the Kara and White Seas, including those obtained by Radashevsky et al. (2014Radashevsky et al. ( , 2021, and sequences of M. bastropi, M. neglecta, M. viridis and M. wireni provided by Bastrop et al. (1998), Bastrop and Blank (2006), Blank et al. (2008), Bastrop (2009), andSyomin et al. (2017). A general analysis included available sequences of five genes: COI, 16S, 18S, 28S and Histone 3. This analysis was done to formulate hypotheses about the phylogenetic relationships of Marenzelleria species for the first time using concatenated set of both mitochondrial and nuclear genes. The resulting tree was rooted using the sequences of Spiophanes cf. bombyx (Claparède, 1870) (provided by Radashevsky et al. 2020) according to a preliminary phylogenetic analysis of molecular data for spionid polychaetes, where Spiophanes Grube, 1860 appeared basal to Marenzelleria clade (Radashevsky et al. unpubl. data). We also performed an analysis of mitochondrial COI and 16S genes of Marenzelleria species only. This analysis allowed to obtain more accurate p-distance estimates between Marenzelleria samples by reducing loss of data during exclusion of ambiguous positions and gaps after aligning Marenzelleria sequences with outgroup. It was rooted using sequences of M. arctia that in the general analysis appeared most basal among Marenzelleria.  Bastrop and Blank (2006), Blank et al. (2008), and Blank and Bastrop (2009). WS followed by numbers refer to the voucher specimens deposited at the ZMMU_WS collection. AZOV followed by numbers refer to the haplotypes of M. neglecta from the Azov Sea reported by Syomin et al. (2017). The numbers without letters preceding collecting locations are unique numbers from the VIR database linking the individuals on the tree with the sampling data in Tables S1-3, S5; numbers of individuals are separated from sample numbers by dots. Marenzelleria species of Arctic origin are shown in blue; species of Northwest Atlantic origin are shown in purple. Bastrop et al. (1998), Bastrop and Blank (2006), Blank et al. (2008), Bastrop (2009), andSyomin et al. (2017) registered unique haplotypes in GenBank, but not all of the sequences obtained in their studies. Therefore, some of the sequences from different geographic locations shown in Table S5 are numbered the same as each other.

Molecular analyses
Five-genes analysis ( Figure 1) The aligned sequences of Marenzelleria spp., with gaps excluded, comprised in total 2482 bp, including 612 bp (100% of original aligned sequences) for COI, 302 bp (93.2%) for 16S rDNA, 980 bp (96.3%) for 18S rDNA (5'-and 3'-ends of the fragments; middle parts were excluded), 297 bp (96.4%) for 28S rDNA, and 291 bp (100%) for Histone 3. The Bayesian analysis of the combined dataset resulted in a fully resolved consensus tree ( Figure 1). The average p-distances for the individual gene fragments between groups of specimens are given in Table S6.
All ten 18S sequences of M. arctia (five from the Kara Sea, three from the White Sea, one from the Baltic Sea, and one from the Barents Sea) were identical. Eight 28S sequences (five from the Kara Sea and three from the White Sea) also were identical. Histone 3 sequences of M. arctia from the Kara and White Seas differed by three substitutions (average p = 1.03%). No variability was found among the Histone 3 sequences of conspecific individuals from the same location (Table S6).
The Bayesian analysis of the combined dataset of two mitochondrial markers resulted in a fully resolved consensus tree ( Figure S1). It revealed two groups among specimens identified by morphology as M. arctia. One group included specimens from the Baltic, Barents and White Seas; the other included specimens from the Kara Sea. Three COI haplotypes and three 16S haplotypes were identified in each group, but none of them was common to specimens from both groups. In the first group, specimens from all locations had one common COI haplotype and one 16S haplotype. The maximum ingroup p-distances in each group were 0.33% for COI and 0.62% for 16S. The p-distances between the two groups ranged from 4.09% to 4.58% (28 variable sites) for COI and 0.62% to 1.24% (6 variable sites) for 16S (Table S7).

Marenzelleria arctia (Chamberlin, 1920) (Figures 2, 3)
Scolecolepides arctius Chamberlin, 1920: 17-18, pl Remarks on the identity of Marenzelleria arctia. Scolecolepides arctius was first described from a lagoon at Collinson Point (Camden Bay, Beaufort Sea, Alaska, USA) by Chamberlin (1920). The species was largely forgotten until Sikorski and Buzhinskaya (1998) redescribed it based on the type material (paratypes MCZ ANNb-2194, 2195) and transferred it to the genus Marenzelleria. Earlier reports of M. arctia from the Kandalaksha Gulf by Stolyarov (1994), , and  were based on the identifications by Andrey V. Sikorski. It is noteworthy that various aspects of the biology of M. arctia have been studied in North European populations, whereas the American population remains unexplored. In the present study, we compared the sequences of gene fragments of individuals from the Baltic, Barents and White Seas and individuals from the Kara Sea, which were all morphologically identified as M. arctia. Genetic distances between European and the Kara Sea groups of specimens were significant (p = 4.28-4.29%) for COI fragments (in contrast to the distances between individuals from the same group p = 0.05-0.13% for COI), while they were 0.84-0.88% for 16S, 0.93% for Histone 3, and 0.0% for 18S and 28S (see Table S6). The high genetic distances between the COI fragments (which evolve faster), the low distances between the 16S and Histone 3 fragments, and the sequence identity of 18S and 28S can be interpreted as a result of the isolation of the Kara Sea population and ongoing speciation. At the same time, the high genetic distance between COI fragments, isolated estuarine habitats of these Arctic worms, and the large geographic distance between the type locality of M. arctia in the Beaufort Sea (Alaska, USA) and Northern Europe raise doubts about the conspecificity of the North American and North European populations. The phylogenetic relationships and the systematic position of these populations require careful further study.
Remarks on Laonice annenkovae Zachs, 1925. Zachs (1925 described Laonice annenkovae Zachs, 1925 from the Tuloma River estuary (Kola Bay, Barents Sea, Russia), and Uschakov (1939Uschakov ( , 1950Uschakov ( , 1953Uschakov ( , 1955 reported this species from the White Sea and the Amur Liman (on the border between the Okhotsk and the Sea of Japan). Sikorski and Buzhinskaya (1998) placed L. annenkovae into synonymy with M. arctia and for the first time reported this species from the Bering Sea in the Chukchi Peninsula and Kamchatka. Although Sikorski and Buzhinskaya (1998) noted that Uschakov's material was lost, they expanded the distribution of before-only-Arctic M. arctia along the Asian Pacific coast southward to the Amur Liman. New specimens of Marenzelleria from the Amur Liman and Sakhalinsky Gulf were collected by an Expedition of the Institute of Marine Biology, FEB RAS, in 2005 (MIMB 17905, 36659-36665). The worms from the Amur Liman are similar to those from Baffin Bay, Canada, but differ somewhat in the later start of sabre chaetae in neuropodia from chaetigers 10−12 instead of chaetiger 4. We also identified as M. arctia specimens from the Sea of Okhotsk collected in 1955 and 1997 (ZISP 13723 and 10/49457, respectively) that have not been previously reported in the literature. They are similar to Marenzelleria from the Amur Liman and Sakhalinsky Gulf. Note worthily, Sikorski and Bick (2004, p. 273) assumed that "In the Far East [Northwest Pacific], M. arctia and possibly M. neglecta occur." In the Amur Liman, Marenzelleria is an important part of the prey of the Amur sturgeon Acipenser schrenckii and the Kaluga sturgeon Huso dauricus (Kolobov et al. 2013). The systematic position of Marenzelleria from the Northwest Pacific requires further study. Sikorski and Buzhinskaya (1998) reported that designated a lectotype (ZISP 01/2210) and 23 paralectotypes (ZISP 02/2211 and 03/13765) from a type series of Laonice annenkovae. In reality, on the examination by one of us (VIR) on 12 Dec 2019, ZISP 01/2210 contained anterior fragments of 15 large worms. Therefore, one of these specimens, ca. 85-chaetiger anterior fragment, was designated by VIR as the lectotype (ZISP 01/2210) whereas the other 14 specimens were designated as paralectotypes and catalogued as ZISP 02a/50775 (12 specimens) and MIMB 42145 (2 specimens). Sample ZISP 03/13765 was not mentioned by Zachs (1925) in the original description of L. annenkovae and therefore cannot be considered as a part of the type series, as erroneously Sikorski and Buzhinskaya (1998) Invasion of the Baltic Sea by M. arctia. In the Baltic Sea, M. arctia was first identified by genetic analysis of specimens collected in 2005 in two Swedish locations in the western part of the sea in Söderhamn and the Isle of Askö (Bastrop and Blank 2006: fig. 1). Soon after that, Blank et al. (2008: fig. 1) reported M. arctia from nine sites in the northern Baltic Sea (Sweden and Finland) and obtained new sequences of specimens from six sites in the Gulf of Bothnia, Åland Sea, and westernmost part of the Gulf of Finland (Figure 3). In 2009, many mature individuals of M. arctia were first found in the eastern part of the Gulf of Finland, after a series of hypoxic-anoxic events that led to the decline of native benthic communities (Maximov 2010(Maximov , 2011(Maximov , 2015(Maximov , 2018. Currently, M. arctia dominates in the eastern deepwater (down to the depths of 70-80 m) part of the Baltic Sea (Maximov et al. 2014Golubkov et al. 2021;Kocheshkova and Ezhova 2018). Remarkably, the species has not been reported from the North Sea.
Two Marenzelleria specimens were collected from Trosa Archipelago (Baltic Sea, Sweden) and photographed by Fredrick Pleijel in June 2008 ( Figure 2D, E). One of these specimens was preserved (SIO BIC A5893), and, although not examined in this study, according to the features shown on the picture (i.e., nuchal organ length, arrangement of branchiae), we refer it to M. arctia. Blank et al. (2008) showed that Baltic M. arctia shared haplotypes with specimens from the Tuloma River (Kola Bay, Barents Sea, Russia) and suggested an introduction by ship ballast water from the European Arctic to harbours in the central or northern parts of the Baltic Sea. The exact route of that introduction remained unknown, however, because of most likely delayed record of the first appearance of the species in the Baltic Sea due to difficulties in identification of Marenzelleria specimens based on morphological characters, and possible inadequate genetic characteristics of populations within the range of distribution of M. arctia.
Here, for the first time, we propose a possible route for the invasion of the Baltic Sea by M. arctia from the White Sea though the White Sea-Baltic Sea Canal. The Canal was opened in 1933 and connects the White Sea with Lake Onega, and then with Lake Ladoga, which further connects with the Gulf of Finland (Figure 3). The most likely vector of this invasion is the transportation of larvae with ship ballast. The proposed route seems to contradict the first reports of this species from the western part of the Baltic Sea, but not from the Gulf of Finland. However, this may be due to the delayed record of the first appearance of the species in the area, as well as due to the locations of ballast water discharge by ships from the White Sea and further movement of larvae by local currents and subsequent successful settling of larvae, phenomena about which we have no data at this time. Remarks. Bastrop et al. (1995) and Röhner et al. (1996a) examined Marenzelleria populations from northern Europe using allozyme electrophoresis and suggested the presence of two different species in the region: one in the North Sea and the other in the Baltic Sea. In an attempt to determine the origin of these worms, Röhner et al. (1996b) compared their allozymes to Marenzelleria from the Atlantic coast of North America. Three species were distinguished and referred to as Marenzelleria Types I, II and III. The North Sea population was found similar to Marenzelleria Type I from the US coastal waters between Barnstable Harbour (Massachusetts) and Cape Henlopen (Delaware). The Baltic Sea population was found similar to Marenzelleria Type II from the US coastal waters between Chesapeake Bay (Trippe Bay) and Ogeechee River (Georgia). Therefore, it was suggested that the North and Baltic Seas were colonized by two Marenzelleria species from the North American Atlantic coast. Bastrop et al. (1997Bastrop et al. ( , 1998 confirmed this assumption when analyzing 16S rDNA fragments from the same populations. Sikorski and Bick (2004) revised the genus Marenzelleria and described Marenzelleria Type II as a new species M. neglecta. Darss-Zingst-Boddenchain (Germany) was chosen at the type locality of the species (Figure 4). Based on new material from Currituck Sound (North Carolina, USA), Bick (2005) described Marenzelleria Type III as the new species M. bastropi. The distribution of M. neglecta and M. bastropi in North America was reviewed and mapped by Radashevsky et al. (2021).

Marenzelleria neglecta
In the first half of the 1990s, M. neglecta began to spread in the Gulf of Finland in Estonia and Finland (Norkko et al. 1993;Stigzelius et al. 1997;Kotta and Kotta 1998). In 1996, rare Marenzelleria (initially identified as M. viridis) were first discovered in the eastern part of the Gulf in Russia (Lyakhin et al. 1997), but the next year the worms spread over large areas and soon became a common component of both shallow and deepwater benthic communities (Maximov and Panov 2003;Maximov 2011Maximov , 2015. Dramatic changes in the zoobenthos of the eastern part of the Gulf of Finland took place in 2009, shortly after the devastating hypoxic-anoxic events in 2003 and 2006. Alien M. arctia quickly occupied most of the deepwater zone and became the dominant species in the communities (Maximov 2015). A similar distribution of the two Marenzelleria species was found in the southeastern part of the Baltic Sea in 2001-2014: M. arctia was mainly found in relatively deep (down to the 70-80 m depth), mesotrophic areas with salinity above 5‰, whereas M. neglecta inhabited shallow, eutrophic and hypertrophic, brackish waters of the Vistula and Curonian lagoons (Kocheshkova and Ezhova 2018).
In February-March 2014, the adults and larvae of Marenzelleria sp. were first found in the Don River and the Taganrog Bay of the Sea of Azov (Syomin et al. 2016). Their morphological characters varied greatly and corresponded to both M. neglecta and M. arctia. However, genetic analysis (mainly COI sequences, but also 16S, 28S, cytb, and Histone 3) showed that only M. neglecta was present in the region (Syomin et al. 2017). Soon after the first find in Taganrog Bay, M. neglecta became dominant in the region and also appeared in the centre of the Sea of Azov, in the Strait of Kerch, and on the Caucasian coast of the Black Sea (Taman Peninsula; Syomin et al. 2017). Syomin et al. (2016) suggested that, most likely, the worms could have entered the Don River and Taganrog Bay with the ballast waters of ships coming from the Baltic Sea through the Baltic Sea−Volga Canal, the Volga River, the Volga−Don Canal, and then the Don River (Figure 4). Syomin et al. (2017) warned about the further spread of M. neglecta in the Black Sea, as well as about the possible invasion of the Caspian Sea by this species. Indeed, in April 2016, M. neglecta was first collected off the western coast of Crimea (Boltachova and Lisitskaya 2019;Boltachova et al. 2021), and in October 2018, Marenzelleria was first found in the northern part of the Caspian Sea (Mikhailova et al. 2021). Mikhailova et al. (2021) noted that Caspian worms were similar to M. arctia, but since M. neglecta was already identified by molecular methods in the neighbouring Sea of Azov, they referred them to as Marenzelleria sp., pending reliable identification by molecular analysis. Because Syomin et al. (2016) also identified by morphology some worms from the Don River and Taganrog Bay as M. arctia, but later, based on molecular data, re-identified them as M. neglecta, here we tentatively refer the Caspian worms to M. neglecta. The worms could have entered the Caspian Sea with ballast waters of ships sailing from the Baltic Sea through the Baltic Sea−Volga Canal, and then through the Volga River (Figure 4).
Here we report for the first time M. neglecta in the British Isles (River Thames; MIMB 36644), recording the further spread of this species in the North Sea. Complete information on M. neglecta records is given in Table S2 (mapped in Figure 4). However, due to the lack of good material for a complete description, he reported that it was Scolecolepis sp. In revising the Spionidae family, Mesnil (1896) used Wirén's (1883) and Marenzeller's (1892) descriptions to create a new genus, Marenzelleria, although he did not give a species name for the corresponding material. Following the descriptions, Mesnil (1896: p. 117) mistakenly diagnosed the new genus as having branchiae beginning from chaetiger 2. Augener (1913) received three specimens collected by the Scottish Jackson-Harmsworth Expedition of Professor W.S. Bruce near the Cape Flora (Franz Joseph Land, northeastern Barents Sea) in 1896. Augener (1913: p. 265) re-examined Wirén's (1883) and Marenzeller's (1892) materials, found them similar to his material from the Franz Joseph Land, and named it after its first discoverer as Marenzelleria wireni.
Marenzelleria wireni/M. cf. wireni/Microspio wireni were reported from the North Sea (Wohlenberg 1937;Elliott and Kingston 1987;Schiedek 1999;Essink and Dekker 2002;Wolff 2005), but these reports were misidentifications. Until now, there is no genetically confirmed record of M. wireni from the North and Baltic Seas, as well as from North America. Sikorski and Buzhinskaya (1998) and Sikorski and Bick (2004) provided new records of M. wireni for the Barents, White, Pechora, Kara, Laptev, East Siberian and Chukchi Seas (all in the Arctic Russia). Complete information on M. wireni records is given in Table S3 (mapped in Figure 3).

Phylogeny of Marenzelleria
Earlier analyses of phylogenetic relationships among Marenzelleria species were based on sequences of the mitochondrial DNA only. Bastrop (2009) andSyomin et al. (2017) used 16S, COI and cytochrome b sequence data, whereas Radashevsky et al. (2021) used only 16S and COI data. All three analyses suggested a basal position for two Arctic species, M. arctia and M. wireni, and, thus, an Arctic origin of Marenzelleria. However, they resulted in slightly different hypotheses about relationships among boreal Northwest Atlantic species. Blank and Bastrop (2009: fig. 1) proposed sister relationships of M. bastropi with M. viridis: neglecta (bastropi+viridis), while Syomin et al. (2017: fig. 6) proposed sister relationships of M. bastropi with M. neglecta: viridis (bastropi+neglecta). The analysis by Radashevsky et al. (2021: fig. 1) suggested sister relationships of M. neglecta with M. viridis: bastropi (neglecta+viridis). The latter hypothesis was supported by the present analysis of five genes and seems more plausible given that the nodes in the phylogenetic tree received higher support.
Tree topology (Figure 1) and the distribution of Marenzelleria along the Atlantic coast of North America (see Radashevsky et al. 2021: fig. 2) allow us to assume that the boreal Northwest Atlantic species had evolved from a common ancestor that was originally widely distributed along the Atlantic coast of North America. Population isolation and adaptation in isolated habitats apparently led to the divergence of the ancestral population and the origin of three species: M. bastropi, M. neglecta and M. viridis.

Identification of Marenzelleria species
Although Marenzelleria includes only five described species, identifying worms by morphological characteristics is often challenging. Diagnostic characters such as the length of the nuchal organs, the distribution of branchiae, hooded hooks and sabre chaetae are age-dependent and individually variable (Sikorski and Bick 2004: table 1, fig. 6;Syomin et al. 2016: table), resulting in complex correlations between body size (width, length) and morphological features (see Radashevsky et al. 2021: figs. 5, 6). The usual presence of only anterior fragments in samples further complicates their identification. Hence, molecular data is an important complementary diagnostic tool (Röhner et al. 1996a;Bastrop and Blank 2006;Blank et al. 2008). Blank et al. (2008) developed a PCR/RFLP protocol and provided a molecular identification key for three Marenzelleria species from the Baltic Sea. The protocol used the restriction fragment length polymorphism (RFLP) method, when the polymerase chain reaction (PCR) products of two mitochondrial DNA gene segments (16S, COI) were cut using restriction enzymes. Sikorski and Bick (2004) provided the first key for the morphological identification of five Marenzelleria species. The key was mainly based on species specific arithmetic differences of three numerical characters: the first chaetiger with neuropodial (ventral) hooded hooks (VHH), the first chaetiger with notopodial (dorsal) hooded hooks (DHH), and the last chaetiger with branchiae (Br). Because all these characters, as well as the nuchal organs length (the fourth important diagnostic character), were found to be size/age-dependent, Sikorski and Bick (2004) suggested that the key be used only for specimens more than 1.0 mm wide. Bick (2005) modified and updated the first morphological key, but noted that the revised version was best for specimens larger than 1.2 mm wide. Below is a further update to the key, which includes morphological data recently published by various authors.

Conclusions
Polychaetous annelids, and especially spionids, are among the lists of alien species in the various regions discussed in the present work. Ballast water and hull fouling have been major vectors for the introduction of polychaetes, including spionids, worldwide (Çinar 2013). Because polychaetes are mainly marine or estuarine, transoceanic or long-distance coastal movements have been considered as main transportation routes. Ships transporting ballast via canals have rarely been reported and may be underestimated. The invasion of the Sea of Azov by M. neglecta thus likely occurred via transport of worms (larvae or adults or both) from the Baltic Sea via the Baltic Sea−Volga−Don Canal (Syomin et al. 2016(Syomin et al. , 2017. Similarly, M. arctia may have invaded the Baltic Sea by transport from the White Sea via the White Sea−Baltic Sea Canal. The Arctic M. arctia has been extensively examined based on the North European populations located far from the type locality of the species in the Beaufort Sea (Alaska). However, molecular analysis of specimens, initially identified morphologically as M. arctia, found significant genetic distances between the Baltic, Barents and White Seas and those worms from the Kara Sea. This genetic distance, as well as the isolated estuarine habitats of these Arctic worms, and the large geographic distance between the type locality of M. arctia in the Beaufort Sea (Alaska) and northern Europe, raises doubts about the conspecificity of North American, North European and Northwest Pacific populations. Sequence data from the North American population are urgently needed to characterize the molecular identity of the species and to verify the conspecificity of the disjunct populations. A similar situation may exist for M. wireni, which has been reported from isolated estuaries along much of the Arctic coast. Molecular data are only available for the Spitsbergen population of M. wireni from the Greenland Sea, which is approximately 800 kilometers from the proposed type locality of the species in Franz Josef Land (Kara Sea) and thousands of kilometers from the Eurasian mainland. Further molecular studies will be required to ensure a systematic revision of Marenzelleria and further development of the hypothesis of the origin and evolution of these spionid polychaetes.
The following supplementary material is available for this article: Table S1. Sampling location data and museum registration numbers of Marenzelleria arctia. Table S2. Sampling location data and museum registration numbers of Marenzelleria neglecta. Table S3. Sampling location data and museum registration numbers of Marenzelleria wireni. Table S4. List of the museums and collections (and their acronyms) holding the examined or reported specimens of Marenzelleria spp. Table S5. Taxa, sampling location data, references, and GenBank accession numbers of sequences used in the present study. Table S6. Uncorrected pairwise average genetic distances (p, in %) between Marenzelleria spp. calculated in the analysis of five genes. Table S7. Uncorrected pairwise average genetic distances (p, in %) between Marenzelleria spp. calculated in the analysis of two mitochondrial genes. Figure S1. Majority rule consensus tree of the Bayesian inference analysis of the combined COI (612 bp) and 16S (322 bp) sequences (934 bp in total) of Marenzelleria spp. rooted with sequences of Marenzelleria arctia.