To name but a few: descriptions of five new species of Terebellides (Annelida, Trichobranchidae) from the North East Atlantic

Abstract The number of described species of the genus Terebellides Sars, 1835 (Annelida, Trichobranchidae) has greatly increased in the last years, particularly in the North East Atlantic. In this context, this paper deals with several putative species recently delineated by molecular means within a well delimited clade of Terebellides. Species are characterised here by a combination of morphological characters, and a complementary nucleotide diagnostic approach. Three species were identified as the nominal species T. stroemii Sars, 1835, T. bigeniculatus Parapar, Moreira & Helgason, 2011 and T. europaeaLavesque et al., 2019. Five species are described as new: T. bakkenisp. nov., T. kongsrudisp. nov., T. norvegicasp. nov., T. ronningaesp. nov. and T. scoticasp. nov. The distinctive morphological characters refer to the branchial shape, absence or presence of papillae on lamellae of anterior margin of branchial dorsal lobes, absence or presence of ciliated papillae dorsal to thoracic notopodia, geniculate chaetae in one or two chaetigers, and the morphology of thoracic and abdominal uncini teeth. Furthermore, the description of T. bigeniculatus is revised and complemented after examination of type specimens. An updated identification key to all species of the genus in NE Atlantic and a proposal of a classification of different types of abdominal uncini to be used in taxonomy are also included.


Introduction
The species richness in the genus Terebellides Sars, 1835 (Annelida, Trichobranchidae) in the North East Atlantic (NEA hereafter) seemed to be well known after several taxonomic studies (Holthe 1986;Jirkov 1989Jirkov , 2001Gagaev 2009;Parapar et al. 2011Parapar et al. , 2016cJirkov and Leontovich 2013;Parapar and Hutchings 2014). Nevertheless, molecular taxonomy approaches performed recently in a comprehensive sample of NEA Terebellides have substantially changed the understanding of the species diversity hidden within members of this genus in European waters. Studies by Nygren et al. (2018) and Lavesque et al. (2019) showed a number of genetic lineages, compatible with the species concept -independently evolving entities that are genetically (and phenotypically) distinct (Barraclough 2010). As a result, the total number of species in the NEA has increased dramatically from seven to 32 (Nygren et al. 2018;Lavesque et al. 2019), but some of these still remain unnamed or not formally described.
Terebellides is the most species-rich genus of trichobranchids, with 82 nominal species (Parapar et al. 2020;Read and Fauchald 2020) but fairly homogeneous morphologically. It is distinguished from other members in the family by their characteristic branchiae with a single mid-dorsal stalk on segment 3. However, species identification presents some difficulties as there are no clear boundaries between the intraspecific and interspecific variability of some of the morphological attributes considered of high taxonomic relevance. Species diagnostic features mainly rely on details of the branchiae, shape and size of anterior thoracic lateral lobes, and uncinal morphology (Parapar and Hutchings 2014;Parapar et al. 2016aParapar et al. , 2016b. Surprisingly, analyses of DNA sequences showed a large genetic diversity within the group, especially in mitochondrial markers, and while the genetic intraspecific divergence in the universal barcoding marker cytochrome c oxidase subunit I (COI) ranged from 0 to 3.4%, the interspecific distance between species varied from 8.8 to 22.9% (Nygren et al. 2018).
Phylogenetic analyses consistently showed that the NEA Terebellides are divided into four major clades, named Groups A-D in Nygren et al. (2018). The aim of the present paper is the systematic revision of members of Group A (according to Nygren et al. 2018), and the morphological characterization of the species assessed after phylogenetic and species delimitation analyses of DNA sequence data (Nygren et al. 2018). Given that there are some species complexes, with scarce morphological differences between the species, if any, a list of apomorphic nucleotides (present in all sequences of a certain species and unique of that species) is also provided as a complementary diagnostic feature (Rach et al. 2008;Wong et al. 2009).

Materials and methods
This paper is based on the study of 132 specimens identified as belonging to Group A as defined in Nygren et al. (2018) and corresponding to several putative species. This material is deposited in the Zoological Museum Bergen (ZMBN, Bergen, Norway), Göteborg Natural History Museum (GNM, Goteborg, Sweden), the Norwegian University of Science and Technology, University Museum (NTNU-VM, Trondheim, Norway; Bakken et al. 2020) and the Senckenberg Museum Frankfurt (SMF, Frankfurt, Germany).
The sampling area covered in this paper is mostly the Norwegian and Swedish continental shelf but also includes some samples from the Irish and Celtic seas, North Sea, Barents Sea, Greenland Sea, South Icelandic coast and the Arctic Ocean (Suppl. material 1: Table S1; Nygren et al. 2018).
Light microscope images were obtained by means of an Olympus SZX12 stereomicroscope equipped with an Olympus C-5050 digital camera. Line drawings were made with an Olympus BX40 stereomicroscope equipped with camera lucida. Specimens for Scanning Electron Microscopy (SEM) were prepared by critical point drying, covered with gold and examined and photographed under a JEOL JSM-6400 electron microscope at the Servizos de Apoio á Investigación (SAI, Universidade da Coruña, Spain).
Methyl green (MG) staining patterns and thoracic uncini morphology were characterised based on the classification proposed by Schüller and Hutchings (2010) and Parapar et al. (2020) respectively; specimens of similar/comparable size were used.
The species dealt within the present study are quite homogenous morphologically. Therefore, common traits shared by all members of Group A are described first in order to avoid repetition of the same characters in each species description.
For each species, the list of the museum registration numbers and collection details (geographic area, locality, coordinates, depth, collecting date and habitat) is provided in Suppl. material 1: Table S1. Unless specified, each registration number holds a single specimen; associated GenBank DNA sequence accession numbers are provided in Suppl. material 2: Table S2.
The present systematic account follows the phylogenetic hypothesis presented by Nygren et al. (2018), after phylogenetic analyses of mitochondrial COI (ca. 658bp) and 16S rDNA (ca. 440 bp), and the nuclear ITS2 (290-419 bp) and 28S rDNA (ca. 760 bp) sequences from 513 specimens of Terebellides species from the NEA. In their topology, four strongly supported major clades were recovered, and named Groups A-D. We are herein dealing only with members of Group A. Other subgroups (A1-A4) within Group A were established after analyses of combined datasets ( Fig. 1; Nygren et al. 2018). In the present study comparison of the morphological traits of species within these subgroups were performed in order to find potential characteristic diagnostic features.
The COI universal barcoding gene proved to be very informative for species delimitation purposes alone, but insufficient to resolve deeper relationships in the Terebellides radiation (Nygren et al. 2018). However, in the present study further analyses based on this mitochondrial marker alone have been performed in order to assess diagnostic nucleotides for each of the species and establish genetic distances between them. Phylogenetic analyses of COI Terebellides sequences in GenBank generated by Nygren et al. (2018) and Lavesque et al. (2019) were performed, using Trichobranchus roseus (Malm, 1874), Polycirrus sp., and Pista cristata (Müller, 1776) as outgroups (Nygren et al. 2018). Four hundred and seventy-one sequences were aligned with MAFFT version 7.017 (Katoh et al. 2002), and with default parameters, trimming some starting nucleotides of the sequence of Terebellides sp. (MN207188) to become 659 bp alignment. Best-fit model according to Bayesian information criterion -BIC (TVM+F+I+G4), was calculated with IQTREE version 1.6.11 (Nguyen et al. 2015). Maximum likelihood phylogenetic analyses were also run in IQTREE version 1.6.11 (Nguyen et al. 2015), with ultrafast bootstrap (Hoang et al. 2018). Tree topology and support values for the nodes are found in Fig. 2. Given the morphological homogeneity in the Terebellides Group A species, GenBank accession numbers (COI sequences) are provided for each species, indicating those belonging to type series. Moreover, unequivocal nucleotide diagnostic characters are provided as the positions in the alignment (nucleotide), with the alignment available in Suppl. material 2:
nov., T. ronningae sp. nov. and T. scotica sp. nov. The remaining five species will be dealt with in future studies.

Terebellides Group A (sensu Nygren et al. 2018)
Description. The morphological features shared by all studied species in Group A are itemized below. Some of these are also shared by Groups B, C and D as defined in Nygren et al. (2018) (see Remarks below).
Remarks. Among the aforementioned characters, branchial features might serve to distinguish most of Group A species (except for A3 species) from those in Groups B-D. Those include branchial size, lobes size (i.e., whether dorsal and ventral are of similar size or differ), presence of terminal papilla/filament on posterior lobes, and presence of ciliary structures (rows, tufts or buttons) on lamellae. Other taxa described or reported worldwide bear similar branchiae including T. stroemii sensu Parapar et al. (2011) from Iceland and sensu Parapar et al. (2013) from the Adriatic Sea, T. kerguelensis McIntosh, 1885 andT. longicaudatus Hessle, 1917 from Antarctic latitudes Moreira 2008a, 2008b), and T. kobei Hessle, 1917 from Japan (Imajima and Williams 1985).
The other species groups as found in Nygren et al. (2018) were not studied in depth here and will be the aim of a subsequent study. However, Group B seems to be characterised by having a shorter body and free branchial lobes; these features are shared with T. atlantis Williams, 1984 andT. irinae Gagaev, 2009 as already suggested by Nygren et al. (2018). Members of Group C are apparently not defined by any unique shared morphological character but show the same geographic distribution as T. irinae. Finally, the three putative species in Group D were related to T. gracilis Malm, 1874and T. williamsae Jirkov, 1989by Nygren et al. (2018 even though the latter was proposed to be synonymised with the former by Parapar et al. (2011). These species seem characterised by having ventral white colouration in a number of anterior chaetigers and similar-sized branchial lobes; these characters are not shared with Group A.
Regarding Group A, six morphological characters have been considered to delineate subgroups and species (Table 1). Two characters can be determined with the aid of the STM: 1) general branchial shape, 2) number of thoracic chaetigers with geniculate chaetae; four characters require SEM examination: 3) presence of papillae on lamellae of dorsal branchial lobes, 4) presence of ciliated papillae dorsal to thoracic notopodia, 5) features of thoracic and 6) abdominal uncini shape dentition. Branchial typology (1) is defined according to Parapar et al. (2016c) and thoracic uncini (5) follows Parapar et al. (2020). Typology of abdominal uncini (6) is described here (see Discussion).
Furthermore, species will be also characterised according to geographic and bathymetric distribution according to available data.

Subgroup A1
Analyses of molecular data found low or no support for monophyly of this clade (Figs 1, 2) and there is no apparent morphological synapomorphy supporting this clade either. Cohesion of members of this group needs to be studied further, but meanwhile, it is considered herein as a morphologically homogenous gathering of species 10-13 and 18-19 (Figs 1, 2). As it was indicated above, only species 10, 11, and 13 will be described herein, of which 10 and 13 are new to science and 11 corresponds to T. stroemii; some comments on species 12 (Terebellides sp. 1 hereafter) are also provided.

Characters present only in subgroup A1
None (Table 1).
Character/s shared with subgroup A2 • Branchiae of type 1 (stroemii-type, comma-shaped), all four lobes fused for approximately half of their length and ventral ones usually obscured by dorsal ones (Fig. 11A-C).
Character/s shared with subgroup A3 • Border of anterior region of dorsal branchial lamellae not provided with papillary projections.
Character/s variable within subgroup A1 • Abdominal uncini type 1 (Fig. 6G) and 2 ( Fig. 7G) (see Conclusions Section). Lavesque et al. (2019) describe several species from French waters similar to those of Group A in terms of body and branchial shape. Among them, Terebellides gralli Lavesque, Hutchings, Daffe, Nygren & Londoño-Mesa, 2019 is described as lacking papillary projections on branchial lamellae, but no mention is made to whether or not ciliated papillae are present dorsal to thoracic notopodia. The sequences of this species do not relate with those of any putative species as defined in Nygren et al. (2018). Moreover, T. gralli differs morphologically from other congeners in having longer branchiae that may reach TC4-6 (Lavesque et al. 2019: 169, fig. 12A) instead of only reaching TC3-4.
Distribution and bathymetry. Barents Sea, Greenland Sea, northern Norwegian coasts from the Lofoten Islands to Trondheim; at depths of102-378 m (Nygren et al.  Table S1). One specimen found in North Iceland at 1,250 m deep.
Etymology. This species is named after Dr. Torkild Bakken, from the NTNU-University Museum, Trondheim (Norway), housing institution of some of the specimens used in the present study, for his dedication to the study of Norwegian polychaetes and his friendship.
Remarks. Terebellides bakkeni sp. nov. is a small-sized species, maximum-sized specimens reaching 20.0 mm in length (n = 3). This species is characterised by the presence of ciliated papilla dorsal to thoracic notopodia, lack of papillae on the margins of branchial lamellae and presenting abdominal uncini of type 1. Most of these features are also shared by the closest relative, T. stroemii (species 11 herein), but they differ in the morphology of the abdominal uncini, being of type 2 in T. stroemii and type 1 in T. bakkeni sp. nov. (Table 1). One specimen studied with SEM showed ciliary tufts in the inner side of the branchial lamellae ( Fig. 5D). If this feature is not an artefact and is confirmed in all members of the species -so far only two specimens were examined under SEM -it would be an autapomorphy for the species. A similar feature was found in the non-closely related T. gracilis, that is also present in NEA. The ciliary tufts in T. bakkeni sp. nov. are, however, connected by rows of cilia ( Fig. 5D), while in T. gracilis they are confined to isolated tufts (Parapar et al. 2011: 12, fig. 9c). On the other hand, there are no clear morphological differences between T. bakkeni sp. nov. and T. kongsrudi sp. nov. (species 13). These sympatric species differ in the southern limit of their geographic distribution: T. bakkeni sp. nov., as T. kongsrudi sp. nov. are present above 65°N (Fig. 8A, C) while the latter and T. stroemii reach more southern latitudes, such as the Skagerrak and Bergen respectively ( Fig. 8B, C).
Of the 462 sequences, including all NEA species, and 659 positions in the COI alignment, the 12 sequences assigned to T. bakkeni sp. nov. hold two unique nucleotides positions, and an additional one only shared by a single specimen from another clade (see Suppl. material 2: Table S2). The species also showed 0-1.9% of intraspecific divergence in the COI marker, and a minimum of 11.5% uncorrected genetic distance with congeners (in this case T. stroemii) (Nygren et al. 2018).
Nucleotide diagnostic features. There are no unique apomorphic nucleotides in the fragments of COI analysed for T. stroemii, when considering all Terebellides species present in the NEA (Suppl. material 2: Table S2). However, when comparing homologous nucleotide positions with members of only Group A (183 sequences in the COI alignment), the following autapomorphies arise: 174 (C), 183 (C), 453 (A), 612 (C).
Distribution and bathymetry. Terebellides stroemii was traditionally considered as a cosmopolitan species, but its known distribution seems in fact restricted to the Norwegian coastline (Parapar et al. 2011;Parapar and Hutchings 2014;Lavesque et al. 2019). Specimens examined by Nygren et al. (2018) and in the present paper, obtained after comprehensive sampling in the NEA, were found only in W Norway, between 115 and 388 m deep (Figs 8B, 10; Suppl. material 1: Table S1).
Remarks. In the five sequences belonging to this species, there were four haplotypes showing 0-1.1% of intraspecific divergence, and a minimum of 11.5% uncorrected genetic distance with members of the closest relative, T. bakkeni sp. nov. (Nygren et al. 2018).
Terebellides stroemii is a large species, reaching up to 52 mm in length (Parapar and Hutchings 2014) and is characterised by the presence of ciliated papilla dorsal to thoracic notopodia, lack of papillae on margins of branchial lamellae, thoracic uncini of type 3 and abdominal uncini of type 2. All these features are shared with T. kongsrudi sp. nov.; T. bakkeni sp. nov. is also very close morphologically to T. stroemii but they differ in the morphology of the abdominal uncini as explained above. Nygren et al. (2018) misidentified species 6 as T. stroemii, but this was later corrected by Lavesque et al. (2019) who pointed out that the molecular sequences of these specimens fit with those of T. europaea.
Specimens examined here bear thoracic uncini that are most similar to other members of Group A; SEM examination showed, however, that some uncini have a rostrum distal tip that is distinctly bent downwards (deformity?) (Fig. 7E, arrow) as already described for the type specimens by Parapar and Hutchings (2014: 8, fig. 7F, G), and attributed to preservation for too long in EtOH. However, we have found similar bent rostrum among specimens of T. kongsrudi sp. nov. (Fig. 12D, arrow), T. ronningae sp. nov. (species 7) (Fig. 21C, arrows) and T. bigeniculatus (species 20 + 28) (Fig. 26E, frame) suggesting this may not be related to preservation. The abdominal uncini are quite similar to those described in Parapar and Hutchings (2014: 9, fig. 8C-E) also showing a small gap among the anteriormost teeth of rostrum (Parapar and Hutchings 2014: 8-9, fig. 8F; Fig. 7G); these features are not shared by other species of subgroup A1, i.e., T. bakkeni sp. nov. and T. kongsrudi sp. nov. In all, species 11 agrees well with the redescription of T. stroemii. Geographic and bathymetric distribution of our specimens also agree with that of T. stroemii (see Parapar and Hutchings 2014), with Manger (Norway) (i.e., type locality of T. stroemii; Fig. 10) being its southernmost distribution limit. The other three taxa, i.e., species 5, T. europaea and T. bigeniculatus, were also found near Manger, but all can be clearly distinguished morphologically from each other (see above and below for T. europaea  Sars, 1835 (species 11; non-type specimen, ZMBN 116399), SEM micrographs. A anterior end, right lateral view B TC6 to TC8, lateral view C geniculate chaetae D TC4 and TC5, nephridial papillae E, F thoracic uncini (arrow in E pointing to rostrum curved at distal end) G abdominal uncini. Abbreviations: bdl -branchial dorsal lobes; dpn -dorsal projection of notopodium; np -nephridial papilla; TC -thoracic chaetiger; tdp -thoracic dorsal papilla; tm -tentacular membrane. and T. bigeniculatus) and species 5 belongs to Group B and seems closer morphologically to T. atlantis. On the other hand, type specimens of T. stroemii come from depths of 55-110 m (Parapar and Hutchings 2014) as well as specimens belonging to T. europaea, T. ronningae sp. nov., T. scotica sp. nov. (species 9) and species 12 (<200 m), and therefore they seem to constitute a shallow-water assemblage of species from an ecological point of view.
Finally, the Icelandic specimens reported as T. stroemii by Parapar et al. (2011) might not correspond to this species. In fact, it is likely that they represent at least two different species, namely T. bakkeni sp. nov. and T. kongsrudi sp. nov., both reported here to the North and East of Iceland. Therefore, the aforementioned specimens deserve further revision.   (Fig. 11A). Geniculate chaetae in TC6, acutely bent, with low marked capitium (Fig. 12A, B). Two pairs of nephridial pores in TC4 and TC5 and ciliated papilla dorsal to thoracic notopodia (Fig. 11D, E). Thoracic uncini in one row with rostrum/capitium length ratio approximately 2 : 1 and capitium with a first row of 2-5 medium-sized teeth, followed by several smaller teeth (  Table S1).
Etymology. This species is named after Dr. Jon Anders Kongsrud, Department of Natural History, Zoological Museum Bergen-ZMB (Norway), housing institution of some of the specimens used in the present study, for his dedication to the study of Norwegian polychaetes and his friendship. Remarks. This is a large species reaching up to 50.0 mm long, and is characterised by the presence of ciliated papilla dorsal to thoracic notopodia, lack of papillae on the margins of branchial lamellae, thoracic uncini of type 3 and abdominal uncini of type 1. These features are also shared by species 12 (sensu Nygren et al. 2018), which will be described elsewhere (Gaeva and Jirkov, pers. comm.). Terebellides kongsrudi sp. nov. is also morphologically similar to T. bakkeni sp. nov. (see above) but T. kongsrudi sp. nov. and species 12 show a wider geographic distribution; on the contrary, species 12 is present at shallower depths (<200 m) while T. kongsrudi sp. nov. extends to deeper depths (>500 m).
Finally, in the 26 sequences belonging to this species (see Suppl. material 2: Table  S2), there were fourteen haplotypes showing 0-1.9% of intraspecific divergence, and a minimum of 8.2% uncorrected genetic distance with members of species 12 which is the closest relative (sensu Nygren et al. 2018  Remarks. This species will be described elsewhere by D. Gaeva and I. Jirkov (pers. comm.). In order to confirm characters here used to link species within each subgroup, two specimens were examined under the SEM that share with subgroup A1 the following features: branchiae type 1 sensu Parapar et al. (2016c) (Fig. 13A), lack of papillae on border of branchial lamellae (Fig. 13B), geniculate chaetae on TC6, ciliated papilla dorsal to thoracic notopodia (Fig. 13C, D), and thoracic uncini of type 3 (Fig. 13E). Nevertheless, abdominal uncini are of type 2 (Fig. 13F), as it occurs in T. stroemii and differently to T. bakkeni sp. nov. and T. kongsrudi sp. nov., that are the most similar species within subgroup A1 (Table 1).

SubGroup A2
Molecular analyses of mitochondrial and nuclear markers recovered a strongly supported subgroup A2 (Fig. 1). This subgroup is composed by species 6, 7, 8, and 9 (sensu Nygren et al. 2018). Analyses of the COI dataset alone also find support for this clade, and incorporate the recently described T. lilasae Lavesque, Hutchings, Daffe, Nygren & Londoño-Mesa, 2019 (Fig. 2). There are several morphological features that are shared, and exclusive to, all members of subgroup A2, and includes other NEA species (see below). Three (7,8,9) of these four species are described herein as new to science and the fourth species (6) corresponds to T. europaea. Character/s present only in Group A2 • Border of anterior region of dorsal branchial lamellae provided with papillary projections (Figs 15C, 20C, 22C).

Terebellides europaea
Nucleotide diagnostic features. All sequences belonging to T. europaea share the unique apomorphic nucleotide in position 240 (C) of the alignement.  Table S1). Lavesque et al. (2019) included the Ría de Ferrol (Galicia, NW Spain) as part of the Bay of Biscay, but this locality belongs to the northern Galician Rias that are out of the western limit of this bay.
Remarks. This species is characterised by the combination of the following features: presence of papillary projections over the edge of the anterior border of dorsal branchial lamellae, lack of ciliated papilla dorsal to thoracic notopodia, thoracic uncini of type 3 and abdominal uncini of type 2. The original description states that body length is less than 17 mm, but maximal length of specimens examined here was up to 46.0 mm. Examination of live and preserved specimens has revealed that the size ratio between the ventral and dorsal branchial lobes is similar in all specimens; however, their arrangement differs among specimens, i.e., the ventral lobes are visible in some while in others are hidden behind the dorsal lobes.
Terebellides europaea was misidentified as T. stroemii by Nygren et al. (2018; species 6) due to their morphological similarities and coexistence near the type locality of the latter (Fig. 9). Nevertheless, Lavesque et al. (2019) found that members of species 6 have papillae on the edge of the dorsal branchial lobes, unlike the neotypes of T. stroemii described by Parapar and Hutchings (2014). Molecular analyses show that the sequences of specimens found in the Bay of Biscay belong to species 6 (Lavesque et al. 2019); examination of all specimens also confirmed the presence of the aforementioned papillae. Moreover, T. europaea is generally found in bottoms above 100 m deep while T. stroemii is present in deeper environments (>100 m) (Fig. 9). ; non-type specimens, GNM15116 and GNM15118), SEM micrographs. A anterior end, right lateral view B buccal tentacles and branchiae, left lateral view C branchial lamellae, detail. Abbreviations: bdl -branchial dorsal lobe; bdltp -branchial dorsal lobe terminal papilla; blp -branchial lamellae papillae; bst -branchial stem; bt -buccal tentacles; bvltp -branchial terminal lobe terminal papilla; cr -ciliary row; dpn -dorsal projection of notopodium; gc -geniculate chaetae; gr -glandular region; loli -lower lip; SG -segment; TC -thoracic chaetiger; tll -thoracic lateral lobes.

Terebellides ronningae
Etymology. This species is named after Dr. Ann-Helén Rønning, Head Engineer of the Department of Technical and Scientific Conservation, Natural History Museum-NHMO (Oslo), for her help and friendship.
Remarks. Terebellides ronningae sp. nov. is characterised by the lack of ciliated papilla dorsal to thoracic notopodia and the presence of papillary projections pointing over the edge of the dorsal anterior border of branchial lamellae, thoracic uncini of type 1 and abdominal of type 2 (Table 1). It is distinguished from the closest relatives of subgroup A2 by the presence of thoracic uncini type 1 instead of type 3 (Table 1).
Nucleotide diagnostic features. All sequences of T. norvegica sp. nov. share the unique apomorphic nucleotides in positions 48 (C) and 285 (G) of the alignement.
Etymology. The name of the new species refers to the country where members of this lineage were found, along the Norwegian coast from the Barents Sea to the Skagerrak Strait.
Remarks. Terebellides norvegica sp. nov. is characterised by the presence of marginal papillae in the anterior region of branchial dorsal lamellae, thoracic uncini of type 3 and abdominal uncini of type 2, and by lacking ciliated papilla dorsal to thoracic notopodia (Table 1). These features are shared with species of subgroup A2: T. europaea, T. ronningae sp. nov. and T. scotica sp. nov. (Table 1), apart from the thoracic uncini type that is different in T. ronningae sp. nov. Furthermore, T. norvegica sp. nov., T. europaea and T. scotica sp. nov. also show the same variability in whether ventral branchial lobes are hidden or not by dorsal lobes. Therefore, it seems that members of these three species can only be distinguished according to the DNA sequences. However, they show little overlapping in their geographic distribution and bathymetric ranges (Figs 9, 18A, C, D). Terebellides norvegica sp. nov. inhabits deep-water habitats (mostly below 200 m) along the Norwegian coast; its distribution only overlaps with that of T. europaea in southern waters (Skagerrak). As stated before, T. europaea has a broader distribution reaching to the South NW Iberian Peninsula and is generally found in shallower habitats (<100 m) similarly to T. scotica sp. nov. Ciliate epibionts attached over dorsal body surface were also observed (Fig. 23F).
On the other hand, the internal anatomy of T. norvegica sp. nov. has been examined by transparency in one alive specimen (Fig. 14D). The digestive tract is divided in an oesophagus clearly distinguishable between TC1 and TC3, that is followed by the stomach and the associated digestive gland (TC4-TC7) and then by the intestine (from TC11). Regarding the circulatory system, a double dorsal blood vessel is present in anterior body end from which arise four afferent vessels at the level of branchial stem and into the branchiae; the coelomic cavity bears oocytes from TC11. All these internal features agree with those described by Jouin-Toulmond and Hourdez (2006) and Parapar and Hutchings (2014) for other species of the genus.

Nucleotide diagnostic features.
There are no unique apomorphic nucleotides in the fragments of COI analysed for T. scotica sp. nov., when considering all Terebellides species present in the NEA (Suppl. material 2: Table S2). However, when comparing homologous nucleotide positions with members of only Group A (192 sequences in the COI alignment), the following autapomorphies arise: 279 (G), 444 (C), 517 (A), 630 (C).
Remarks. Among A2 species, T. scotica sp. nov., T. europaea and T. norvegica sp. nov. have thoracic uncini of type 3 and show ventral branchial lobes that may be hidden in between dorsal lobes in some specimens. As stated previously, these species can only be distinguished according to DNA sequences.
The specimen studied under SEM shows a small knob near the notopodial lobe of TC1 (nop, Fig. 24C); its biological role is unknown and it may correspond to an artefact. Two different sequences (see Suppl. material 2: Table S2; 0.2% distance) have been attributed to this species (Nygren et al. 2018). As stated above, the closest NEA congener is T. norvegica sp. nov., at 10.5% genetic distance.
Character/s present only in subgroup A3 • Branchiae stroemii-type but irregular in many specimens, with all four lobes slightly fused; ventral lobes shorter and slimmer than dorsal ones and not hidden in between.
Character/s shared with subgroup A1 • Border of anterior region of dorsal branchial lamellae not provided with papillary projections.
Material examined herein corresponds to a few small and incomplete specimens. Therefore, the list of diagnostic characters given was developed with the aid of the type specimens re-examined and the original description.
Nucleotide diagnostic features. All sequences of T. bigeniculatus share the unique apomorphic nucleotides in positions 67 (G) and 138 (G) of the alignement.
Distribution and bathymetry. Around Iceland at both sides of the GIF Ridge; 179-968 m deep (Parapar et al. 2011). Material examined here also confirms its presence in shallow and deep bottoms of Norway and Barents Sea (Fig. 8D).
Remarks. In some of the species delimitation analyses performed, Nygren et al. (2018) were able to distinguish between two closely related lineages, clades 20 and 28, but some analyses of nuclear and mitochondrial datasets lump them together in a single entity. Given that all specimens examined share characteristic features that are distinct from other Terebellides species studied herein, clades 20 and 28 have been considered in the present study as a single species and identified as T. bigeniculatus.
As stated above, the sequenced specimens are small and not well preserved, hindering the examination of relevant morphological features with taxonomic value (i.e., branchial type). However, this species is characterised by having geniculate chaetae on TC5 and TC6 instead of only on one chaetiger (Parapar et al. 2011: 7) as in congeners listed in the Key of the present study. Furthermore, T. bigeniculatus is characterised by the low fusion of the usually irregularly-shaped branchial lobes (Parapar et al. 2011: 7-8, figs 4, 5a, b), ventral lobes are not obscured by dorsal ones, the lack of marginal papillae in the anterior region of the branchial dorsal lamellae, the presence of ciliated papilla dorsal to thoracic notopodia, and by having thoracic uncini of type 3 and abdominal uncini of type 1. However, it is likely that the irregular shape of the branchiae may correspond to an artefact related to fixation/preservation; other specimens show instead well-defined branchiae that agree with those of A1 and A2 species but less developed (Fig. 26A, B; Parapar et al. 2011: 8, fig. 5a). Regarding the four branchial types as defined by Parapar et al. (2016c), branchiae of T. bigeniculatus might correspond therefore to type 3 but with lobes showing a more variable shape.
The original description states that nephridial papillae are located on TC3-TC4 or TC4-TC5 (Suppl. material 1: Table S1; Parapar et al. 2011: 7-9, figs 5c, 6d). Examination of the holotype and several paratypes confirmed that pores are on TC4 and TC5, as in other Group A species. Nephridial pores, as found in most Terebellides species, are usually flat and can be easily overlooked when examined with STM and even SEM; those of T. bigeniculatus are larger and easier to distinguish comparatively with STM (Parapar et al. 2011: 9, fig. 6d).
Members of species 21 (see below, as Terebellides sp. 2) also bear geniculate chaetae in two chaetigers; this feature had been considered as unique to T. bigeniculatus regarding other NEA species. However, species 21 is present in Arctic waters (cf. Nygren et al. 2018: fig. 6) while the distribution of members of species 20 + 28 and identified here as T. bigeniculatus agrees with that of the type specimens (see Fig. 8D).
Remarks. As explained for Terebellides sp. 1, two specimens were examined under SEM; these share with T. bigeniculatus the irregular shape of branchial lobes (Fig. 27A), the presence of geniculate chaetae on TC5 and TC6 (Fig. 27C-E) and abdominal uncini of type 1B (Fig. 27G). They share with subgroup A1 the presence of one ciliated papilla dorsal to thoracic notopodium (Fig. 27B) and thoracic uncini of type 3 (Fig. 27F).
On the other hand, species 18 and 19 of A1 (not described here because of the few specimens being available) and 23 (A4) have a geographic distribution similar to that of T. bigeniculatus but their position in the cladogram by Nygren et al. (2018: fig. 5) suggests that they may not bear geniculate chaetae in two chaetigers.
There are no unique diagnostic nucleotide positions that are shared by the two haplotypes (in 18 sequences) in COI. Eighteen sequences, in one single haplotype, have been attributed to this species (Nygren et al. 2018). Members of this species show a minimum of 3.0% uncorrected genetic distance, with its closest relative being T. bigeniculatus (Fig. 1).

Key to European species of Terebellides
The following key of European Terebellides species is based on Lavesque et al. (2019) and updated by including all species of Group A (in bold) apart from those that will be described elsewhere. The known geographic or bathymetric distribution has been used when there is a lack of discriminatory morphological characters between some species (e.g., subgroup A2).

Group A species: taxonomy and distribution
The comprehensive study by Nygren et al. (2018) revealed that the genus Terebellides holds a large species diversity in NEA waters regardless its morphological homogeneity. Over 25 molecular entities that meet the requirements to be recognized as species were recovered forming four main and robust clades (A-D); Group A is composed, in turn, by thirteen species. Among the latter, members of only three species were identified herein as current nominal species: T. stroemii, T. bigeniculatus, and T. europaea; the remaining ten represent undescribed taxa. Within Group A, three subgroups (A1-A3) can be defined based on molecular data, being only A2 and A3 well supported and congruent among all molecular analyses and datasets (Figs 1, 2;Nygren et al. 2018) but also by morphological features. A1 and A2 gather species morphologically similar to T. stroemii, while species included in subgroup A3 share morphological features with T. bigeniculatus. The original description of T. stroemii by Sars (1835) lacks detailed specific diagnostic features as are recognised nowadays in many closely related species, most of them described in the last years. On the contrary, T. bigeniculatus belongs to a small group of species bearing geniculate chaetae in two thoracic chaetigers (TC5 and TC6) instead of one (TC6), a distinct morphological trait for the group; T. bigeniculatus was described from deep Icelandic waters by Parapar et al. (2011), and only later reported NEA by Nygren et al. (2018). Terebellides europaea was recently described after molecular analyses by Lavesque et al. (2019) and fits within species of A1+A2. Other species from NEA, namely T. gracilis, T. atlantis, T. williamsae, T. irinae and T. shetlandica Parapar, Moreira & O'Reilly, 2016, differ from members of Group A in shape and body length, ventral colouration in a number of thoracic chaetigers, branchiae shape and degree of fusion and relative size of dorsal/ventral lobes (see Holthe 1986;Jirkov 2001;Parapar et al. 2011Parapar et al. , 2016c. The aforementioned species fit either within groups B, C, or D sensu Nygren et al. (2018) and will be dealt with in a forthcoming paper.
The characters considered to delineate morphologically the aforementioned subgroups (A1-A3) should be taken with care because there are limitations due to number of specimens available to be studied and their condition of preservation. However, considering the variety and origin of the material examined we were able to elucidate some general patterns on taxonomy and distribution of the studied species. Thus, all studied species seem quite homogeneous in terms of general body features and share many characters; however, presence/absence of some macroscopic/microscopic characters has allowed their organization in the subgroups proposed above. Nevertheless, some species could not be differentiated according to morphological characters but genetic data. On the other hand, geographic distributions of species do not show apparent gaps; some species have a wider distribution and were more frequent in samples such as T. norvegica sp. nov. and T. kongsrudi sp. nov.; this suggests that many previous reports of T. stroemii in NEA might correspond to the aforementioned species. Other species apparently show a more restricted distribution, i.e., T. bakkeni sp. nov. in northern Norway or have their limit of distribution in southern Norway, as T. europaea. Similarly, there are no gaps in the bathymetric distribution of species, but some seem to appear typically at shallow depths, reaching the continental shelf (0-200 m) such as T. europaea, T. ronningae sp. nov. and T. scotica sp. nov. On the contrary, T. bigeniculatus and T. norvegica sp. nov. are found at depths of below 200 m while T. stroemii, T. bakkeni sp. nov. and T. kongsrudi sp. nov. show a wider bathymetric distribution.
Given the morphological homogeneity, DNA sequences have been shown to provide advantageous data and support when it comes to species delineation in Terebellides. The most informative markers in previous studies are COI and ITS (Nygren et al. 2018;Lavesque et al. 2019). In the present study, analyses have been mainly based on mitochondrial COI, the universal barcoding gene, because it offers no ambiguities in the alignment process, and is the most commonly used in molecular taxonomy in annelids (e.g., Borda et al. 2013;Tomioka et al. 2016;Álvarez-Campos et al. 2017;Aguado et al. 2019;Grosse et al. 2020) and other taxa (e.g., Kekkonen and Hebert 2014). After species delimitation, identification to the correct nominal species level is ideal, as species names allow the communication, study, quantification, classification, use and management of life on the planet. This has been the motivation of recognising unequivocal diagnostic nucleotides in specific positions for the species described in the present study. As with morphological traits, molecular diagnostic characters are tested continuously when additional intraspecific and interspecific variation within the groups has been found. Nevertheless, and as pointed out by previous studies, diagnostic nucleotides may be an effective and relatively simple way for species identification (Rach et al. 2008;Wong et al. 2009).

Comparisons with other NE Atlantic Terebellides
Lavesque et al. (2019) described eight new species of Terebellides from continental France considering an integrative taxonomy approach. Those species could be informally grouped in two assemblages: 1. Species similar to Group A sensu Nygren et al. (2018) regarding body colour and shape, and branchiae features: T. bonifi, T. europaea, T. gentili, T. gralli and T. lilasae.
The first five species were already discussed above. Regarding the remaining three species, only T. ceneresi was sequenced by Lavesque et al. (2019) and according to their phylogenetic analyses, it is not related to any species of Group A; in fact, it differs from Group A species: a) in having a very distinct MG staining pattern corresponding to a solid stain manifested in the first ten thoracic chaetigers, being lighter in TC4; b) the anterior branchial lobe (5 th lobe) is not present; c) the outer edge of branchial lamellae bears tufts of cilia. These characters would relate T. ceneresi to Group D sensu Nygren et al. (2018). This species was described with 'eagle head'-shaped thoracic uncini, which are similar to those of T. stroemii, T. ronningae sp. nov. and T. kongsrudi sp. nov. as described here and T. stroemii sensu Parapar and Hutchings (2014). However, as explained above (see Remarks for T. stroemii), the taxonomic value of this character should be viewed cautiously and its consistent presence across the three aforementioned species needs to be assessed.
Terebellides parapari differs from Group A species in the shape and arrangement of branchial lobes that are free from each other, and by the presence of terminal filament in ventral lobes. These features and its short body length relate T. parapari to T. shetlandica and Group B sensu Nygren et al. (2018). Finally, T. resomari is unique among NEA Terebellides because of having "not well packed (separated) disposition of the branchial lamellae" (Lavesque et al. 2019: 177, fig. 18B) and therefore branchiae seem lacking a defined shape. In addition, this species also shows the "upper lip very elongated with convoluted margins" (Lavesque et al. 2019: 177, fig. 18C), that was also reported by Parapar et al. (2020) for Terebellides sp. from the Atlantic African coast. Therefore, these unusual features do not allow for the allocation of T. resomari to any group as defined by Nygren et al. (2018).

Discriminant vs. non-discriminant body characters in species delineation
This study has revealed that some of the traditionally morphological-based taxonomic characters are not appropriate for Terebellides species identification. The number of species in the genus is now large and their morphological homogeneity high. Regarding Group A, two macroscopic characters have, however, been useful: 1) presence of geniculate chaetae in one or two chaetigers (A1+A2 vs A3), 2) presence of papillary projections in the border of branchial lamellae (A2 vs A1+A3). On the contrary, we found that the development of lateral lappets and the presence of a dorsal projection on the anterior thoracic notopodia seem dependent on size/age and preservation, and therefore these characters should be taken with care for species identification. Similarly, the species in Group A seem quite homogeneous when considering branchial morphology, particularly within A1 and A2. Some of the morphological differences observed between Terebellides species rely in the exposure of the ventral lobes (hidden or not behind the dorsal lobes). However, we have also observed some degree of variability between specimens belonging to the same species and could be due to size or the contraction of specimens after fixation.
Morphology of thoracic and abdominal uncini seems useful for species identification; such features need to be examined under SEM and are being considered in descriptions of Terebellides in the last years. Recently, Parapar et al. (2020) describe tentatively several types of thoracic uncini. The uncini of the NEA species treated here are quite similar because of their phylogenetic proximity, being T. ronningae sp. nov. the only species that differ in uncini type from other congeners of subgroup A2. There were, however, differences in abdominal uncini that correspond to two morphologies that agree well, in turn, with groups of species as defined by molecular-based phylogenetic analyses. Following Parapar et al. (2020), we propose here the use of similar criteria for the characterization of abdominal uncini, that are based on the rostrum vs. capitium length ratio (RvC), and the number of the capitium teeth and their relative size. Therefore, considering our results after SEM examination and other previous work, two main types of abdominal uncini can be defined:
On the other hand, we observed differences in whether the capitium is defined or not in geniculate chaetae of TC5/TC6, as previously highlighted by Parapar et al. (2011Parapar et al. ( , 2013Parapar et al. ( , 2016aParapar et al. ( , 2016bParapar et al. ( , 2016c. For instance, T. ginkgo Schüller & Hutchings, 2012 shows a well-defined capitium conformed by many large-sized teeth whereas other species bear an almost inconspicuous capitium (e.g., T. bakkeni sp. nov., T. kongsrudi sp. nov.) (Schüller and Hutchings 2012: 10, fig. 5a-c; Figs 6G, 12G); Parapar et al. (2011) also reported from Iceland several species with conspicuous capitium, i.e., T. atlantis, T. gracilis and T. stroemii. In this sense, the specimens of T. stroemii examined here bear a low capitium in comparison to those aforementioned from Iceland (Parapar et al. 2011); this suggests that the latter might not correspond to T. stroemii but to other taxa as explained above. Again, the taxonomic value of this character should be tested in other species considering potential intraspecific variation.

Methyl Green staining pattern
The MG staining pattern was mostly similar across the studied species and according to type 1 sensu Schüller and Hutchings (2010), being solid in three to five anterior chaetigers, TC1-TC3(5), striped in subsequent seven or eight chaetigers, i.e., TC4(6)-TC10(11), and fading towards the end of the thorax at TC18; minor observed differences can be attributed to body size, degree of contraction and preservation of specimens. Parapar et al. (2011) reported a similar pattern for specimens identified as T. stroemii from Iceland: solid in the first six chaetigers after turning into a striped pattern and fading in the posterior thoracic segments, while for T. bigeniculatus staining is solid from TC1 to TC11, striped between TC12 and TC14, and then fading in the following segments. The first pattern only partially agrees with that of T. stroemii (species 11) and the second one would match better with that of T. bigeniculatus (species 20 + 28) as examined here. Parapar and Hutchings (2014) reported a MG staining pattern for neotypes of T. stroemii being solid from TC1 to TC3, striped from TC4 to TC12 and fading in the last thoracic segments; this is exactly the same pattern as observed in T. stroemii from Norway (Suppl. material 1: Table S1).

Nephridial papillae
Schüller and Hutchings (2010) and Parapar et al. (2011), among others, suggest that position of thoracic papillae (nephridial/genital) should be considered as of taxonomic value. We agree with this and have found that papillae are present always in TC4 and TC5 in the species/clades studied here. This position has also been reported in T. gracilis sensu Parapar et al. (2011Parapar et al. ( , 2013, T. mediterranea, T. kerguelensis, and T. hutchingsae Parapar, Moreira & Martin, 2016. On the contrary, other species reported elsewhere have such papillae in TC1 instead, including T. persiae Parapar, Moreira, Gil & Martin, 2016, T. mediterranea, and T. hutchingsae.

Conclusions
To sum up all results and according to the discussion of the aforementioned characters, the general characteristics for each subgroup of Group A sensu Nygren et al. (2018) are listed below. A1 and A2 are particularly close to each other and were informally designed by Nygren et al. (2018) as "stroemii-group"; subgroup A3 is the most dissimilar, with T. bigeniculatus as the typical species.

Subgroup A1
Species are similar morphologically and differ from A2 in lacking papillae on branchial lamellae and in having ciliated papillae on thoracic notopodia. Regarding morphology and distribution, T. bakkeni sp. nov. and T. kongsrudi sp. nov. are closest to each other than to T. stroemii. Terebellides stroemii (as species 11 here) shows also a similar geographic and bathymetric distribution (Table 1), but seems less frequent across Norway and differs in abdominal uncini type (cf. Fig. 7G vs. Figs 6G, 12G).

Subgroup A2
The subgroup is morphologically homogeneous. It differs from A1 in having lamellae papillae and by the lack of thoracic ciliated papillae (at least not observed with SEM). The most recognisable species is T. ronningae sp. nov. because of having thoracic uncini of type 1, a long rostrum and a capitium provided with long first row teeth; the other three species bear thoracic uncini of type 3 and differ of each other in the geographic (T. europaea, T. scotica sp. nov.) and bathymetric distribution (T. norvegica sp. nov.).

Subgroup A3
This subgroup is composed by T. bigeniculatus (species 20 + 28) and species 21 (not formally described here). Branchial shape is irregular and geniculate chaetae are present in two thoracic chaetigers (TC5 and TC6). Other features are shared with A1 such as lack of lamellae papillae; thoracic uncini type 3 or presence of thoracic ciliated papillae. The bathymetric distribution of species is similar to A1.